Co-evolutionary Landscape Ecology Framework for Analyzing Human Effects on

Co-evolutionary Landscape Ecology Framework for Analyzing Human Effects on
UNIVERSITY OF PRETORIA
Pretoria, South Africa
A Co-evolutionary Landscape Ecology
Framework for Analyzing Human Effects on
KwaZulu-Natal Province Landscapes and its
Relevance to Sustainable Biodiversity
Conservation
by
Dean Howard Kenneth Fairbanks
Submitted in partial fulfillment of the requirements for
the degree
in the Faculty of Natural & Agricultural Sciences
University of Pretoria
Pretoria
Copyright by
Dean Howard Kenneth Fairbanks
November 2000
I gratefully aclrnowledge the assistance of the many following individuals in providing me
with guidance for this study. In particular, the support provided by Dr. Albert van Jaarsveld in
both the conceptual development and his ongoing interest was critical to the accomplishment of
this work. I also give special thanks to Dr. Richard Norgaard (UC Berkeley), Dr. Kurt Riitters
(North Carolina State Univ.), Dr. John Estes (DC Santa Barbara), Dr. David Everard (CSIR
Environmentek), Dr. Les Underhill (Avian Demography Unit, Univ. of Cape Town), Dr. Keith
Bevon (University of Pretoria), Dr. Bob Pressey (New South Wales National Parks, Australia)
and Dr. Raymond O'Connor (Univ. of Maine, Orono) for their early support and criticisms of my
understanding of the work at hand. Special thanks as well for the exceptional contributions of Dr.
Mrigesh Kshatriya (Univ. of Pretoria), Dr. Andries Engelbrecht (Univ. of Pretoria), Grant Benn
(KwaZulu-Natal Nature Conservation Services) and Belinda Reyers (Univ. of Pretoria) in
assisting with some of the analyses presented in this document. A timely completion would not
have happened without full time financial support granted to me by the South African
Biodiversity Monitoring and Assessment Program for the last two years of the study.
I also wish to thank my two independent external reviewers for their thorough examination and
constructive criticisms of this thesis: Dr. Peter August (Univ. of Rhode Island) and Dr. Amanda
Lombard (Univ. of Cape Town).
Increasingly, I have to aclrnowledge, I learned from this thesis that a project proposal and a
final scientific document are two different things. The way I work is that I piece together the
analyses like a jigsaw puzzle, look at it, and figure out what's missing. At that point, it's more
about how the science output logically flows together rather than how the output must fit with the
original proposed science plan. Evolution within bounds is good and hindsight is, as always,
20/20.
The duration of these studies and my working period for the CSIR Division of Water,
Environment and Forestry Technology in South Africa was marked by growth on a personal level
as well as in the academic domain. I aclrnowledge my international friends in South Africa and
especially friends and family in the USA, for their support during this long period and in
broadening my perspectives of life and what sustainability really means to an individual and in
the global culture. To the South African gang who kept me in good humor: Mike Adam, Mark
Thompson, Jane Thompson, Phil Plarre, Kevin Higgins, Ennio Macagnano, Stuart Martin, Patrick
McKivergan, Mike Musgrave, Brett Harrison, Rose Smith, Thorsten Rosener, Barend Erasmus,
Stephanie Koch, Marinda Dobson, Ian Meikeljohn, Berndt van Rensburg, and the many South
Africans I have met along the way. Thanks to my parents Ron and Bonnie, and my sister Devin
for all their support, teasing, and reality checks that were needed often. Yes, yes, I am finally
done! Hurray for E-mail and the support of friends back in the States. Thanks Sy Henderson,
Ken McGwire, David Elliott, Kevin Elliott, Jason Rogers, Susan Sullivan, Mich Taniguchi, and
Eric Payne for all your support.
Most special of all I aclrnowledge the love and support of Portia Odessyl Ceruti. To have met
her and fallen in love has been one of the most fulfilling and defining moments in my life. Itsjust
amazing to me that we met so far from home, yet we grew up practically in each others
backyards. Life has been good to me and will only get better. I am done P and now its time for
me to help and support you.
1991
1993
2001
Bachelor of Arts in Geography, University of California, Santa Barbara, USA
Master of Arts in Geography, University of California, Santa Barbara, USA
Ph.D. in Sustainable Ecological Management, University of Pretoria, Pretoria, South Africa
Map and Imagery Laboratory, Davidson Library, University of California, Santa Barbara, USA
Student Research Assistant, 1989 to 1991
Remote Sensing Research Unit, Dept. of Geography, University of California, Santa Barbara, USA
Student Research Associate, 1990 to 1991
Graduate Student Researcher, 1991 to 1993
Council for Scientific and Industrial Research, Division of Water, Environment
Research Scientist, 1994 to 1999
SA Biodiversity Monitoring and Assessment
University of Pretoria, South Africa
Senior Research Officer, 1999 to 200 I
Program,
C/o Conservation
and Forestry, Pretoria, South Africa
Planning
Unit, Dept. of Zoology and Entomology,
Fairbanks, D.H.K. and Benn, G.A., 2000. Deriving the landscape structure of a region for biodiversity
case study from KwaZulu-Natal, South Africa. Landscape and Urban Planning, 50 (4):237-257.
Fairbanks, D.H.K. and McGwire, K.c., 2000. Coarse-scale gradient analysis of environmental
diversity for vegetation communities of California. Geographic Information Science, 6 (I): 1-13.
Fairbanks, D.H.K., McGwire, K.c., and Estes, J.E., 2000. Multi-temporal NDVI Relationship
Vegetation Communities of California. Global Ecology and Biogeography, (in preparation).
conservation
planning: a
factors in relation to plant species
to Patterns of Floristic Diversity in
Fairbanks, D.H.K., McGwire, K.C., Cayocca, K.D., and Estes, J.E., 1996. Sensitivity of floristic gradients in vegetation
communities to climate change. In M.F. Goodchild, L.T. Steyaert, B.O. Parks, C. Johnston, D. Maidment, M. Crane, and S.
Glendinning (Eds.), GIS and Environmental Modelling: Progress and Research Issues, pp. 135-140. GIS World Books, Fort
Collins, CO.
Fairbanks, D.H.K., Reyers, B., and van Jaarsveld, A.S., 2000. Species and environment representation:
retention of avian diversity in KwaZulu-Natal, South Africa. Biological Conservation, (in press).
Fairbanks,
D.H.K. and Thompson,
selecting reserves for the
M.W., 1996. Assessing land-cover map accuracy for the South African land-cover database.
South African Journal of Science, 92:465-470.
Fairbanks, D.H.K., Thompson, M.W., Vink, D.E., Newby, T.S., van der Berg, H.M., and Everard,
characteristics of South Africa. South African Journal of Science, 96: 69-85.
D.A., 2000.
Land-cover
Fairbanks, D.H.K. and van der Zel, D.W., 1996. Afforestation potential in South Africa. In D.W. van der Zel (Ed.), The South
African National Forestry Development Plan, pp. 109-125. Department of Water Affairs and Forestry, Government Printers,
Pretoria.
Hassan, R.M., Fairbanks, D.H.K., Magagula, G., and Faki, H., 1998. Analysing Comparative Economic Advantage of
Agricultural Production and Trade Options in Southern Africa: Guidelines for a Unified Approach, Technical paper No. 104,
Sustainable
Development
Publication
Series, Office of Sustainable Development,
Bureau for Africa, USAID, Washington
Scott, D.F., Le Maitre, D.C., and Fairbanks, D.H.K., 1998. Forestry and streamflow
system for assessing extent and distribution. Water SA, 24: 187-199.
reductions
in South Africa: a reference
Reyers, B., Fairbanks, D.H.K., and van Jaarsveld, A.S., 2000. South African vegetation priority conservation
approach. Diversity and Distributions, (in revision).
Reyers, B., Fairbanks, D.H.K., Wessels, K., and van Jaarsveld, A.S., 2000. A multicriteria
addressing long-term biodiversity maintenance. Biodiversity and Conservation, (in revision).
D.C.
approach
areas: a coarse filter
to reserve selection:
Thompson, M.W., Vink, D.R., Fairbanks, D.H.K., Balance, A., and Shackelton, C., 2000. Comparison of extent
tflm~formation of South Africa'~ woodland biome from two national databases. South African Journal of Science (in press).
and
A Co-evolutionary Landscape Ecology Framework for Analyzing Human Effects on
KwaZulu-Natal Province Landscapes and its Relevance to Sustainable Biodiversity
Conservation
Student: Dean H.K. Fairbanks
Supervisor: Prof. Albert S. van Jaarsveld
Department: Zoology and Entomology, University of Pretoria, South Africa
Degree: Doctor of Philosophy (Sustainable Ecological Management)
Abstract
The conservation
of biotic diversity is a growing challenge within southern Africa at the
beginning of the 21 st century.
Growing populations and trends toward a questionable Western
development model place demands on the use of land for food, fiber, and fuel production.
traditional establishment
The
and use of formal conservation areas is being challenged against the
needs of humans and the past unbalances created by colonial rule. Conservation areas, as isolated
islands in a sea of change driven by interconnected economic and social systems, may not be a
basis for sustainable biodiversity
species
diversity
Environmental
response
conservation.
to abiotic
This thesis examines characteristics
environmental
variables
and
land
of avian
transformation.
and land-use correlates of species gradients, species diversity patterns, and the
spatial patterning of bird assemblages varied with location.
The findings supported a conceptual
model of multi-scaled controls on bird distribution, and the related notion that local community
structure is the result of both regional environmental and local-scale landscape pattern that must
be taken in to account in regional conservation planning assessments.
including an landscape model, use of complementary-based
An analytical framework
reserve selection procedures, gradient
analysis, and inclusion of the total spatial economy and development needs of the KwaZulu-Natal
Province proved to be important for developing an integrated conservation plan for sustainable
avian conservation.
Pattern recognition results of the spatial economy and landscape pattern
revealed the strong dichotomy in Western economic versus rural African landscapes, which have
lead to strong differences in avian assemblage patterns.
The research described in this thesis
targets specific objectives of the Sustainable Biosphere Initiative by addressing requirements for
landscape level analysis of humans and ecosystems in an integrated analytical framework.
The
development of a co-evolutionary landscape ecology framework for examining human-ecosystem
interaction provides a strong basis for supporting targeted conservation planning within regions
rather than supporting a generic conservation planning framework.
Keywords: biodiversity, birds, conservation, co-evolution, landscape ecology, gradients, spatial
statistics, sustainability, KwaZulu-Natal Province, South Africa.
This thesis consists of a series of chapters and appendices that have been prepared for submission
to, or publications in, a range of scientific journals.
As a result, styles may vary between chapters
and appendices in the thesis and overlap may occur to secure publishable entities.
1.
Acknowledgements
.ii
Vita
iii
Abstract
iv
Disclaimer
v
List of Figures
x
List of Tables
xiv
Preface
xvi
Introduction
1
1.1. Current Biodiversity Conservation Strategies
2
1.2. Biodiversity Conservation Strategies in the Modem African Context
1.3. Methodological
2.
.4
Background
7
1.3.1. Study Site
7
1.3.2. Data Sets Used in Study
9
1.3.2.1.
Potential Vegetation
9
1.3.2.2.
Topography
12
1.3.2.3.
Climate
12
1.3.2.4. Avian Distribution and Diversity
14
1.3.2.5.
Land-cover/Land-use
15
1.3.2.6.
Road Effects Database
23
1.3.2.7.
Socio-economic Indicators
24
1.3.2.8.
Provincial Protected Areas Database
26
Database
1.4. Differing Aspects of this Study
26
1.5. Format
28
Developing of a Co-evolutionary Landscape Ecology Framework to Address Sustainable Biodiversity
Conservation
30
2.1. Sustainability and Resilience
"
31
2.2. Biodiversity Protection Strategies
32
2.3. Critique and Reconstruction of Problems
32
2.4. Evolutionary Pathways
36
2.4.1.
Co-evolutionary Framework
36
2.4.2.
Non-linear Dynamics
37
2.4.3.
Landscape Socio-ecodynamics
39
2.5. Landscape Ecology Principles to Ensure Sustainability
2.5.1.
Hierarchy, Scale, and Landscape Metrics
2.5 .2. Measuring the Ecological Effects of Landscape Pattern
2.6. Socio-Ecosystem
Interaction
2.7. Co-evolutionary Implications for Sustainable Biodiversity Conservation
.41
42
.43
.45
,48
3.
Identifying Regional Landscapes for Conservation Planning. '"
3.1. Methods
51
3.1.1.
Explanatory variables
51
3.1.2.
Approach
51
3.1.2.1.
Landscape Conservation Analysis
3.2. Results
3.2.1. Landscape Classification
55
3.2.2.
Validation
57
3.2.3.
Landscape Conservation Analysis
58
3.3. Discussion
64
3.3.1.
Landscape Scale and Structure
66
3.3.2.
Landscapes as an Element of Biodiversity for use in Prioritisation Procedures
66
68
Species and Environment Representation: Selecting Reserves for the Retention of Avian Diversity
.....................................................................................................................
4.1. Methods
70
71
4.1.1.
Ordination
'"
71
4.1.2.
Spatial Autocorrelation Analysis: Local Indicators of Spatial Association
72
4.1.3.
Conservation Area Selection
73
4.2. Results
74
4.2.1.
Ordination Analysis
74
4.2.2.
Spatial Autocorrelation Analysis
78
4.2.3.
Priority Conservation Areas
82
4.3. Discussion
84
4.3 .1.
Evaluation of the Techniques
84
4.3.2.
Practical Area Selection for Improved Conservation
88
4.4. Summary
5.
.53
55
3.4. Summary
4.
50
Human-Ecosystem
89
Co-evolution: Analysis of Bird Diversity and Structure with Human Land
Transformation
90
5.1. Factors Associated with Regional Variation in Species Composition
91
5.2. Influence of Geographical Extent and Location
92
5.3. Biological Indicators and Monitoring
92
5.4. Methods
93
5.4.1.
Diversity-evenness
and Human Transformation Patterns
93
5.4.2.
Geostatisical Analysis of Spatial Variation and Extent in Ecological Pattern
95
5.4.3.
Pattern and Process Measurement from Ordination Analysis
100
5.4.4.
Assessing Multi-species Temporal Distributional Changes
102
5.5. Results
5.5.1.
l 03
South African level
5.5 .1.1. Correlation Results
103
103
5.5.1.2.
5.5.2.
Semi-variogram Results
.106
KwaZulu-Natal Level
5.5.2.1.
108
Ordination Results
5.5.2.1.1.
108
Associations of Local and Regional Factors with Species Gradients
....................................................................................
6.
5.5.2.2.
Correlation Results
134
5.5.2.3.
Semi-variogram Results
136
5.5.204.
Association Analysis Results
.136
5.6. Discussion
139
5.6.1.
.151
Scope and Limitations
5.7. Summary
152
Analyzing Human Factors that Affect Biodiversity Conservation: The Co-evolutionary Model
155
6.1. Geographic Development Models, Cultural Landscapes, and Co-evolution
156
6.2. Methods
158
6.2.1. Data
158
6.2.2. Relationships Among Variables and Geographic Regions
159
6.2.3. Pattern Recognition
161
6.204.
Conservation Assessment
.161
6.3. Results
162
6.3.1. The Co-evolved Regions of Productivity in KwaZulu-Natal
163
6.3.2. The Socio-economic-environmental
163
Organization of Space in K waZulu- Natal
6.3.3. The Landscape Pattern Organization in KwaZulu-Natal
171
6.3 A. Regional Geographic Clusters
171
6.3.5. Pattern Recognition Results
.178
6.3.6. Implications for Planning Avian Conservation Persistence
7.
125
180
604. Discussion
187
6.5
191
Summary
Conclusions
.193
References
.196
Appendix A
217
Appendix B
220
Appendix C
245
Appendix D
270
Figure 1.1: (a) The multiple interplay between the three broad themes in sustainable development
analysis; and (b) a trade-off or regional optimization space curve. A given allocation of area to
conservation or development will result in a total cost and total forgone biodiversity, so that the
allocation can be plotted as a point in this space (Faith, 1995)
7
Figure 1.2: (a) Location of the KwaZulu-Natal Province study region within South Africa; and (b) major
place names and their economic hierarchy within KwaZulu-Natal Province
8
Figure 1.3: Functional vegetation types found within KwaZulu-Natal
types described by Low and Rebelo (1996)
Figure 1.4: Examples of environmental
landform index; and (c) growth days
Province based on vegetation
10
data layers used in thesis: (a) elevation;
Figure 1.5: The 1:50 000 map sheet system of grid cells for KwaZulu-Natal
distribution data during both survey periods
Figure 1.6: Spatial distributions of returned fieldcards and histograms
Robson (1980); and (b) Harrison et aI. (1997)
(b) topographic
13
used to record bird
16
distributions:
(a) Cyrus and
16
Figure 1.7: (a) Simplified map ofland-cover/land-use
distribution across KwaZulu-Natal province; and
(b) three level transformation map derived from Table 1.4. ..
.19
Figure 1.8: Conceptual model of the impacts of increasing levels of human developed
biodiversity and natural processes (modified from Reid et aI., 1994). .
land-use on
20
Figure 1.9: The 1:500000 KwaZulu-Natal road distribution network.
24
Figure 1.10: (a) Magisterial districts used for the 1996 Census; and (b) magisterial districts in relation to
the former KwaZulu and Transkei homeland boundaries (pre 1994, shaded gray)
25
Figure 1.11: Protected
Conservation Services
areas
of KwaZulu-Natal
Province
managed
by KwaZulu-Natal
Nature
28
Figure 2.1: Key resources appropriated through human action and the biota that are affected through a
hierarchical cascade
33
Figure 2.2: A shared learning framework for assessing human-ecosystem
sustainability dynamics
35
Figure 2.3: A co-evolutionary view of development (Norgaard, 1994)
37
Figure 2.4: Bifurcation diagram of the probable development of landscape pattern.
38
Figure 2.5: Conceptual diagram describing the process of co-evolution within landscapes
.40
Figure 2.6: The interaction among the major sectors affecting sustainable development.
action oversees and drives decisions and actions taken in the other sectors. .
Policy-political
.40
Figure 2.7: Landscape functionality as: (a) a continuum from functional to dysfunctional; and in relation
to
(b)
resistance
and
resilience
to
disturbance
(modified
from
Ludwig,
1999) .
..........................................................................................................................
.44
Figure 2.8: An overview of the multiple hierarchical indicator reading for analysing co-evolutionary
dynamics
.46
Figure 2.9: Linear regression relationship of male/female population ratio to percentage degraded land
per magisterial district in KwaZulu-Natal (N=52). Human population data from 1996 census and land
degradation assessment from the South African National Land-cover Database (Fairbanks et aI., 2000) .
. . . . . .. . .. . . . . .. . .. . . . . .. . .. . . .. .. . . . . .. . .. . . . .. . . .. . . . . .. . .. . . . . .. . . . . .. . . .. .. . . . . .. . . . . . . . . . . .. . .. . . . . . . . .. . . . . .. . .. . ....46
Figure 2.10: Economic core and poor rural periphery systems model of landscape development within
rural African communities in South Africa
.47
Figure 3.1: Analysis framework used to classify and identify the landscapes
51
Figure 3.2: Landscape classification (Level II; 24 classes) of KwaZulu-Natal Province, South Africa .. 57
Figure 3.3: Preliminary assessment of the level of transformation within the second level landscapes
relative to their areal coverage (See Figure 3.2 for number code descriptions.) .. ,
59
Figure 3.4: Scatter plot of current protection status vs. vulnerability for each landscape type (See Figure
3.2 for number code descriptions.)
59
Figure 3.5: Preliminary scores for irreplaceability (conservation value) and vulnerability to threatening
processes for the landscapes. Landscape types in the upper right-hand corner are conservation priorities
(See Figure 3.2 for number code descriptions.)
60
Figure 3.6: Landscape types classified by a 50% vulnerability status boundary and using the proposed
mCN 10% target for minimum protection of habitats
61
Figure 3.7: Priority ranks for landscapes and vegetation types as inclusion to rarity and richness-based
reserve selection algorithms
63
Figure 3.8: Selection order results for potential reserve networks based on either rarity or richness
procedures for birds; (a) and (b) results are based on a hierarchical mask of ranked landscape values
based on four quadrants derived from 50% cutoff points in Figure 3.8; and (c) and (d) results are based
on a hierarchical mask of ranked priority vegetation types based on current versus potential
transformation (see Reyers et al., in review)
64
Figure 4.1: Identified avian diversity communities derived from hierarchical classification of first two
axes of the detrended correspondence analysis results
74
Figure 4.2: Species-environment gradients identified from stepwise canonical correspondence analysis
with convex hulls of avian community biogeographic zones. (GTMEAN - annual mean of the monthly
mean temperature (0e) weighted by the monthly growth days; PSEAS _MN - precipitation seasonality
from the difference between the January and July means; DEMSTD - elevation heterogeneity;
GDMEAN - number of days per annum on which sufficient water is available for plant growth; and
EVANNMN - total annual pan evapotranspiration (mm»
77
Figure 4.3: Moran's I spatial autocorrelation results: (a) CCA axis 1; (b) CCA axis 2; and (c) combined
Moran's I axes 1 and 2
78
Figure 4.4: Graph of canonical correspondence analysis axis 1 Moran's I values relationship to the
variety of landscape-vegetation functional types found within each grid cell
80
Figure 4.5: Comparison of algorithm results: (a) species rarity-based algorithm; (b) species rarity and
beta diversity algorithm; (c) species richness-based algorithm; and (d) species richness and beta
diversity algorithm
82
Figure 4.6: Graph of algorithm efficiencies detailing species representation
required. (BD = beta diversity)
versus percent land area
83
Figure 4.7: Comparison of algorithm results based on an ideal network, i.e., not taking into account
current protected areas: (a) species rarity-based algorithm; (b) species rarity and beta diversity
algorithm; (c) species richness-based algorithm; and (d) species richness and beta diversity algorithm .
.......................................................................................................................
.... 84
Figure 5.1: (a) Bird species richness across South Africa; (b) evenness structure of birds across South
Africa; (c) Bird species richness using smoothed data; and (d) evenness structure using smoothed data .
.......................................................................................................................
.... 95
Figure 5.2: The separation between the transformation categories illustrates the spatial heterogeneity
found within South Africa, particularly highlighting the development differences between: (a) low
intensity transformation representing African ex-homeland areas; and (b) high intensity transformation
representing "White" developed South Africa
96
Figure 5.3: The vegetation biomes of South Africa based on the map by Low and Rebelo (1996) and
from the original classification work by Rutherford and Westfall (1986)
97
Figure 5.4: Some examples of forms of variograrns: (a) and (b) bounded variograrns; (c) unbounded
variogram; and (d) pure nugget variogram.
98
Figure 5.5: Model semi-variograrns of transformation level, bird richness, and community evenness in
South African biomes: (a) savanna woodland; (b) grassland; (c) shrub steppe; (d) succulent desert; and
(e) fynbos. (MPER - Low intensity transformation; TPER - high intensity transformation;
total transformation)
Figure 5.6: Assemblage
datasets
and TTOT 106
classifications
derived from ordination analysis of the CR life history bird
110
Figure 5.7: Assemblage classifications
datasets
derived from ordination analysis of the ADD life history bird
111
Figure 5.8: Patterns of variation in the first two axes of variation derived from detrended
correspondence analysis for each CR life history bird group, KwaZulu-Natal: (a) all birds; (b) summer;
(c) winter; (d) passerine; (e) non-passerine; (f) breeding; (g) non-breeding; (h) human; and (i) nonhuman. (Figure continued on next page.)
114
Figure 5.9: Biplots from canonical correspondence analysis of life history bird assemblages. All axes
have been rescaled to range from -1.0 to 1.0. Axes for explanatory environmental variables that were
not significant or that had very low correlations with the canonical axes are not shown
116
Figure 5.10: Patterns of variation in the first two axes of variation
correspondence analysis for each CR life history bird group
derived
from detrended
118
Figure 5.11: Biplots from canonical correspondence analysis of life history bird assemblages. All axes
have been rescaled to range from -1.0 to 1.0. Axes for explanatory environmental variables that were
not significant or that had very low correlations with the canonical axes are not shown
119
Figure 5.12: Biplots from canonical correspondence analysis of life history bird assemblages. All axes
have been rescaled to range from -1.0 to 1.0. Axes for explanatory landscape variables that were not
significant or that had very low correlations with the canonical axes are not shown
126
Figure 5.13: Patterns of variation in the first two axes of variation derived from detrended
correspondence analysis for each ADD ecological habitat bird group. Areas with no coverage of the
respective vegetation class are depicted in white
128
Figure 5.14: Biplots from canonical correspondence analysis of ecological habitat bird assemblages. All
axes have been rescaled to range from -1.0 to 1.0. Axes for explanatory environmental variables that
were not significant or that had very low correlations with the canonical axes are not shown
130
Figure 5.15: Biplots from canonical correspondence analysis of ecological habitat bird assemblages. All
axes have been rescaled to range from -1.0 to 1.0. Axes for explanatory land-cover class type variables
that were not significant or that had very low correlations with the canonical axes are not shown ..... 132
Figure 5.16: Kappa coefficient maps of each comparison between CR and ADD surveys and life history
bird assemblages
137
Figure 6.1: Detreneded correspondence analysis biplots: (a) two axes of magisterial district data space
(numbers match Figure 6.4 (a)); and (b) two axes offeature variable data space
.163
Figure 6.2: Factor patterns of variation derived from principal component analysis of the socioeconomic-environmental indicator data set, where shading indicates factor scores
167
Figure 6.3: Factor patterns of variation derived from principal component analysis of the landscape
mosaic pattern indicators data set.
172
Figure 6.4: Mapping of the clusters produced from hierarchical
procedures on the dimensions derived for each data set.
and k-means cluster classification
173
Figure 6.5: The landscape types identified in Chapter 3 are used to identify the dominant class for each
magisterial district based on a simple majority
175
Figure 6.6: Priority avian conservation areas from the "ideal" model developed in Chapter 4, associated
magisterial districts, and a general regionalization of the bird conservation areas by physiographic
boundaries
181
Figure 6.7: The following maps present a rating of the vegetation habitats: (a) to (d) based on patch size
and fragmentation, and (e) is the habitat connectivity rating considering all available vegetation types
residing in a magisterial district. The districts in a poor to moderate state (e) largely reside along the
coast and in the Midlands region. These areas were shown in the analyses of Chapter 5 to be
undergoing significant changes in bird assemblage structure because of high intensity transformation
.........................................................................................................................
184
Table 1.1: Codes and defmitions of explanatory variables, by variable subset, used in Chapters 3, 4, and 5.
. . .. . . . . .. .. . . . . . . . . .. . . . . . . . .. . . . . .. . . . . .. . . . . .. . .. . . . . .. . .. . . . . .. . . .. . . . .. . . . . .. . . . . . . . .. . . . . . . . .. . .. . . . . . . . .. . . . . .. . . . .. . . .. 11
Table 1.2: Functional vegetation classification of the 1:500 000 National Botanical Institute Vegetation of
South Africa, Lesotho and Swaziland (Low and Rebelo, 1996)
13
Table 1.3: Bird datasets and descriptions used in this thesis
17
Table 1.4: Land-cover/land-use classes used in the South African National Land-Cover (NLC) database and
the re-coded transformation classes used for this study
20
Table 1.5: Codes and defmitions of explanatory landscape mosaic indices used in Chapters 5 and 6, by
variable subset.
21
Table 1.6: Codes and defmitions of explanatory class level pattern indices used in Chapter 5, by variable
subset.
22
Table 1.7: Buffer widths assigned to road classes for calculating road effect zone (after Stoms, 2000)
Table 1.8: Codes and names of magisterial districts in Kwazulu-Natal
Province
24
25
Table 1.9: Names and descriptions of the protected areas managed by KwaZulu-Natal Nature Conservation
Services
28
Table 3.1: Landscape rarity, transformation, and protection
classification with accompanying importance ratings
classification
rules based on frequency
.54
Table 3.2: Pearson correlation matrix for environmental variables used in landscape classification (N =
4675). Correlations highlighted in bold violate the r> 0.50 multicolinearity limit defmed for this study ... 55
Table 3.3: Factor weights, eigenvalues, and total variance explained derived by the PCA analysis on the
chosen topographic and climatic variables. Values in bold denote the significant variable identified for each
axis
55
Table 3.4: Elevation, topographic landform index, and growth days index classification hierarchies
56
Table 3.5: Calculations of percent rarity, current transformation percentage, and percent protected in
managed nature reserves. The legend for the landscape numbers is given in Figure 3.2
58
Table 3.6: The values represent the percentage of each level II landscape type that is comprised of each
functional vegetation type. Values in bold represent vegetation types with >10% affiliated areas with level
II landscape types
60
Table 4.1: Avian bioindicators in order of importance based on Dufrene and Legendre (1997) indicator
species value measure for each identified avian community assemblage
75
Table 4.2: Eigenvalues and gradient lengths (1 standard deviation) for the first two axes from DCA and
DCCA of all bird species for Kwazulu- N ataI.
75
Table 4.3: Spearman's rank correlation of explanatory factors with axis scores from DCA and intraset
correlation coefficients from CCA that included all explanatory variables
76
Table 4.4: Summary of results from stepwise CCA.
Table 4.5: Percentage
community assemblage
of functional
vegetation
76
and land-cover/land-use
types per identified
avian
79
Table 4.6: Spearman's rank correlation coefficients of the Moran's I analysis and the diversity of landscape
definition types (see Table 1.1)
80
Table 4.7: Species conservation status and representation selection order based on algorithm type
85
Table 5.1: Setup of a 2 x 2 contingency table used to compare species sampling surveys per sampling unit
..............................................................................................................................
102
Table 5.2: Spatially corrected Pearson correlation coefficients (rs) for comparisons of species richness and
evenness against transformation classes among South African grid cells (only cells with records for all data
sets are included). Richness and human disturbance data were square root and log-transformed
analysis to improve normality
.,
., . .,
.,
.,
.,
before
.,103
Table 5.3: Spatially corrected Pearson correlation coefficients (rs) for comparisons of smoothed species
richness and evenness against transformation classes among South African grid cells (only cells with
records for all data sets are included). Richness and human disturbance data were square root and logtransformed before analysis to improve normality
.,
.,
.,
.,
., . .,
.,
., . .,104
Table 5.4: Spherical model estimates of range (lan) when a stable variance is reached for South African
biomes: species richness (8), Shannon diversity (H'), evenness (E), low intensity transformation (LI), high
intensity transformation (HI), and total transformation (IT) . .,., .. ., . .,., .. .,., . ., . .,.,.,.,., .. ., .,.,.,
107
Table 5.5: Eigenvalues and gradient lengths (1 standard deviation) for the first two axes from DCA and
CCA of all bird species groups in KwaZulu-Natal from the Cyrus and Robson (1970-1979) survey . ., .... 109
Table 5.6: Eigenvalues and gradient lengths (1 standard deviation) for the first two axes from DCA and
CCA of all bird species groups in KwaZulu-Natal from the ADU Bird Atlas (1987-1992) survey .
..... .,
.,
.,
.,
.,
.,
109
Table 5.7: Increases in total variation explained (TVE) by explanatory variables in stepwise canonical
correspondence analysis of CR bird species, by group type; the three greatest contributors to TVE in each
group type are show in boldface
.,
.,
.,
.,
113
Table 5.8: Increases in total variation explained (TVE) by explanatory variables in stepwise canonical
correspondence analysis of ADU bird species, by functional type; the three greatest contributors to TVE in
each group type are show in boldface
.,.,.,
.,., . .,., .. .,
., .. .,.,
.,.,.,.,.,
.,.,
., 113
Table 5.9: Proportion of total variation explained (TVE) by landscape variables while constrained by the
topography and climate variables chosen for each group type in partial canonical correspondence analyses
(CCAs) of ADU bird species; the three greatest landscape contributors to remaining TVE after constraining
by the topography and climate variables in each group type are show in boldface
125
Table 5.10: Increases in total variation explained (TVE) by explanatory variables in stepwise canonical
correspondence analysis of ADU bird species, by ecological type; the three greatest contributors to TVE in
each group type are show in boldface
129
Table 5.11: Proportion of total variation explained (TVE) by landscape variables while constrained by the
topography and climate variables chosen for each group type in partial canonical correspondence analyses
(CCAs) of ADU bird species; the three greatest landscape contributors to remaining TVE after constraining
by the topography and climate variables in each group type are show in boldface
131
Table 5.12: Pearson correlation coefficients for comparison of species richness (SR), Shannon diversity
(H'), and evenness (E) against transformation and disturbance variables among classified groupings of
ADU birds derived from ordination (DCA) and hierarchical classification. Human induced transformation
data were square root-transformed before analysis to improve normality . .,., .... .,.,., . .,., .. ., . .,.,., .... ., .. 134
Table 5.13: Semi-variogram derived distances (kilometers) of spatial dependence for species richness (SR),
Shannon diversity (H'), evenness (E), low intensity transformation (LI), high intensity transformation (HI),
total transformation (TT), road disturbance index (RI), and 1996 population density (PD96) among
classified groupings of ADU birds
.,., . ., . .,., .. ., . ., . ., . ., . ., . ., . ., . .,
.,., . ., . .,
.,.,
136
Table 5.14: Spearman's rank correlations of landscape variables with the five important climate and
topography variables from the ADU CCAs, by bird group. Values> 0.5 are in boldface
148
Table 6.1: Eigenvalues and cumulative proportion of variance explained by principal component analysis
for socio-economic-environmental
indicators and landscape pattern indicators, and eigenvalues and gradient
length for detrended correspondence analysis of LCLU
162
Table 6.2: Factor loadings from principal component analysis with varimax rotation for the socioeconomic-environmental
indicators based on the 1996 magisterial districts. (Table continued on next
page.)
.,
.,
.,
.,
.,
., . ., .163
Table 6.3: Factor loadings from principal component analysis with a varimax rotation for the landscape
pattern indicators derived from the 1996 magisterial districts
171
Table 6.4: If-then rules of landscape pattern indicators describing clusters developed by PCA classification
of the socio-economic-environmental
indicators
178
Table 6.5: If-then rules of socio-economic-environment
classification of the landscape pattern indicators
indicators describing clusters developed by PCA
178
Table 6.6: If-then rules of socio-economic-environment
classification of the LCLU abundance data
indicators describing clusters developed by DCA
179
Table 6.7: If-then rules of landscape pattern indicators describing clusters developed by DCA classification
of the LCLU abundance data
179
Table 6.8: Magisterial districts requiring landscape conservation plans for avian conservation, with
associated socio-economic factors that will need to be addressed for sustainable conservation (also see
Table 6.5 and 6.6)
182
Table 6.9: Magisterial districts requiring landscape conservation plans and the associated vegetation habitat
ratings derived from pattern indicators. Habitat connectivity rating is provided using all habitat types to
derive measure
183
The following is from a dialog between the late American journalist Bill Moyers and the late
Joseph Campbell, which seems to me to nicely tie together one of the great issues of society
and sustainable ecological management:
Moyers: Zorba says, "Trouble? Life is Trouble."
Campbell: Only death is not trouble. People ask me, "Do you have optimism
about the world?" And I say, "Yes, it's great just the way it is. And
you are not going to fix it up. Nobody has ever made it any better; it
is never going to be any better. This is it, so take it or leave it. You
are not going to correct or improve it."
Moyers: Doesn't that lead to a rather passive attitude in the face of evil?
Campbell:
You yourself are participating in the evil, or you are not alive.
Whatever you do is evil for somebody (or something). This is one
of the ironies of the whole of creation (and the paradox of
management).
Moyers: What about this idea of good and evil in mythology, of life as a conflict
between the forces of darkness and the forces of light?
Campbell: .. .In other traditions, good and evil are relative to the position in
which you are standing. What is good for one is evil for the other.
And you play your part, not withdrawing from the world when you
realize how horrible it is, but seeing that this horror is simply the
foreground of a wonder.
Therefore,
for conservationists
and others engaged in issues of sustainability
situation in the world may look sorrowful, it is necessary to participate
wouldn't
wonderful
be life if there were not temporality
opera set on a diverse geographic
though the
in the game.
It
involved, which is sorrow-loss.
It is a
backdrop--except
Within
that it hurts.
conservation and sustainability circles we must affirm that this is the way it is, the challenges
with re-integrating societies goals with the requirements of ecosystems will not be won or lost,
but will evolve through knowledge to something that is better than it was before but never to
the level that we want it to be.' Affirmation is difficult, and as a discipline, we are always
trying to affirm with conditions (i.e., I will affirm the world on condition that it gets to be the
way Aldo Leopold said it ought to be). By accepting the evolution of societies and ecosystems
and our role as conservationists,
landscape ecologists, and geographers as adding components
to its guidance, we will be able to make a difference in creating future landscapes with a level
of ecological integrity acceptable for that time. This may be all we can accomplish, however
this is a tremendous amount to accomplish, and therefore should not be seen as a loss. This
thesis work provides empirical evidence of how the human socio-economic-political
ecosystem response game has been played so far in the KwaZulu-Natal
Africa.
and
Province, South
This thesis examines the co-evolutionary nature of human development on landscapes and
the consequent shaping of species assemblages, which affect biodiversity conservation strategies
in Southern Africa.
environment.
A model is proposed to address the development nature of humans on the
Where this model may fit into current conservation biology principles and within
the field of landscape ecology is discussed.
This study then moves into a series of examinations
of the landscapes of KwaZulu-Natal Province, South Africa, focussing on an assessment of avian
diversity
conservation,
human development
patterns,
and human
action
in shaping
avian
communities, and an application of a co-evolutionary development model.
Techniques used include complementary-based
gradient
analysis,
geostatistics,
pattern
recognition
and mathematical
programs,
transformations
reserve selection algorithms, ecological
multivariate
of species
statistics,
assemblage
spatial
data.
statistics,
Timely data
products, such as the South African National Land-cover database (Fairbanks et al., 2000), the
1996 South African
Sustainability
Population
Census (Stats
SA, 1998), and the 1997 KwaZulu-Natal
Indicators Project (Kok et al., 1997), which records the regions socio-economic
and development status, were used to develop causal relationships.
African
Birds (Harrison
et al., 1997), representing
the results
The 1997 Atlas of Southern
from the largest biological
inventorying project conducted in Africa, and its predecessor, the 1980 Bird Atlas of Natal (Cyrus
and Robson,
1980) covering Natal and Zululand, are used as the biological
relation to the
biophysical and human development patterns.
A number of analyses are performed to describe attributes of the biodiversity hierarchy
(Noss, 1990) that are affected by evolved human development patterns, including impact on avian
distributions,
landscapes,
avian diversity variation, and spatial autocorrelation.
communities,
and species were studied in order to describe and attribute their
dynamics to human disturbance gradients.
assessing
regions
The organization scales of
for biodiversity
Typically, effort is given to studying structure when
conservation,
but this falls short of the main issue of
functioning, which is a dynamic product of changes in structure.
This research targets specific
objectives of the Ecological Society of America's Committee on the Application of Ecological
Theory to Environmental Problems by addressing requirements for understanding and monitoring
changes in biodiversity
associated with land-uses that are specifically associated with human
dimensions of global change (National Research Council, 1994; Lubchencho et al., 1991). These
include three ecological problems at two different levels of organization:
•
Community structure:
What do the collective properties of communities,
including various community indices, tell us about their
functioning?
•
Biotic diversity:
What are the patterns, causes, and consequences
spatial and temporal variation in species diversity?
of
How do land-use patterns influence the ecology of
component systems, including all levels of ecological
organization up to the scale of the landscape itself?
The biodiversity databases used in this study are of a coarser resolution than could be
implemented for local conservation assessment.
The described research, however, may provide a
valuable foothold for identifying commonalities
in the biodiversity
pattern-abundance,
spatial
expression of land-use/l and-cover classes, and relevant information, for the multistage effort that
would be required by such a local conservation assessment and planning effort. The application
of a number of different analysis strategies on the same data sets provides greater opportunity for
comparison and understanding than is available from the numerous unrelated case studies, which
have been performed to date. Typically, these case studies employ a presence/absence
species
database with a standard land-cover map, are limited by geographical variability in biological,
environmental and human response, treat human impacts in a limited fashion, and reflect on only
a local subset of the possible universe of human-ecosystem responses.
limited by many of the same considerations,
Though this study will be
the results provide a starting point from which to
assess the validity of applying general systematic reserve selection schemes to the developing
areas of Southern Africa.
Conservation
planning strategies rely on several contested methods (e.g., Mace et al.,
2000) to provide the best case for conservation
reserve
selection
algorithms,
gap analysis,
action.
These include complementary-based
species richness
"hot spots", keystone
species
surrogates, and environmental surrogates.
In the last decade, the conservation
community has made significant contributions
to
developing systematic reserve selection procedures (Bedward et al., 1992; Church et al., 1996;
Csuti et aI., 1997; Freitag and van Jaarsveld, 1995; Kirkpatrick, 1983; Lombard, 1995; Margules
et aI., 1988; Nicholls and Margules, 1993; Pressey et aI., 1996; Rebelo and Siegfried, 1992).
Conceptually, the need for systematic approaches to represent the protection of as many natural
features (i.e., species, communities, or environments) as possible is well acknowledged. The use
of the principles of complementarity,
flexibility, and irreplaceability
(see Pressey et al., 1993) for
selecting priority regions and regional reserves makes for computationally
elegant solutions.
These protocols for priority conservation area selection, however, have several weak points: use
of poorly surveyed taxa or habitat databases (Maddock and du Plessis, 1999); use of dangerously
simple surrogate information (Faith and Walker, 1996a; Reyers et al., 2000); and more to the
point, the efforts to date have generally not taken into account human influences, landscape
pattern and processes.
The systematic conservation techniques could also ignore interrelated
attributes and feedback's that a more thoughtful and comprehensive approach might illustrate.
In
some cases, the spatial pattern of development in an area might be biodiversity "friendly" (e.g.,
Gadgil et al., 1993; Norgaard, 1994; Dahlberg, 1996; Fairhead and Leach, 1996; Zimmerer and
Young, 1998; Shackelton, 2000) and have evolved with the resident human culture, but would not
be acknowledged in formal protection based approaches.
systematic
Increasingly, the shortcomings of the
reserve selection concepts to take into account the current or future biological
sustainability of the areas selected, or to have the ability to spread the risk of species extinctions
through proper spatial planning, is becoming evident.
Biological
conservation
strategies have traditionally
centered on biological
reserves,
where a reserve is 'an area with an active management plan in operation that is maintained in its
natural state and within which natural disturbance events are either allowed to proceed without
interference or are mimicked through management'
(Scott et al., 1993). The gap analysis school
of biodiversity protection planning attempts to identify the gaps in representation
of biological
diversity in areas managed exclusively or primarily for the long-term maintenance of populations
of native species and natural ecosystems.
It is proposed that once identified, gaps be filled
through new reserve acquisitions or designations, or through changes in management practices.
The goal is to ensure that all ecosystems and areas rich in species are represented adequately in
protected areas. Whereas the complementary reserve selection concept is an elegant and logical
solution, though unrealistic, the gap analysis procedure is simple, scale dependent, and assumes
that large tracts ofland are still available for conservation.
Large reserves (e.g., > 10000 ha) are
the most common strategy to maintain biotic communities over long periods in areas undergoing
large-scale conversion from natural vegetation to agricultural and urban systems (Shafer, 1990;
Noss et al., 1997). The gap analysis procedure can make only a partial contribution within South
Africa since the vast majority of land is under communal or private tenure (see Christopher,
1982), highly fragmented in the ecologically important biomes (see Fairbanks et al., 2000), and
the methodology does not provide for a representative (e.g., species, habitat) system. In areas of
extensive habitat conversion, as found in parts of South Africa, the design of reserve systems is
typically based on a model of reserves as isolated islands of habitat for native species (e.g.,
Rebelo and Siegfried, 1992; Lombard et al., 1997). The ultimate viability of a reserve system,
however, is based on the size, shape, and connectedness of these remnant habitat areas (Forman,
1995; Fahrig, 1997), which should be designed within associated environmental processes (e.g.,
Cowling et al., 1999).
To be sure, the most important consideration, which is typically ignored, in any of these
systematic methodologies is the role human societies, values, and economics playas threats and
protectors to biodiversity.
A logical framework for understanding
human threats has not been
considered in species or broad model approaches, but are root causes of the loss of biodiversity
(Ehrlich
and Wilson,
1991).
Conservation
planning
needs to incorporate
socio-economic
variables, as well as the landscapes, ecosystems, and species of an area, to be relevant within
developing countries. The case for integrated conservation planning in developing countries must
take into account all factors inherent in and relevant to the landscape environment, which includes
human needs.
The importance of flexibility in conservation
planning becomes important in
discussing issues of persistence, since there are typically many different complementary networks,
these can be exploited to reveal those networks that are currently sustainable based on their socioeconomic, cultural, and landscape ecological situation.
This thesis is concerned with the issue of sustainable biological
southern Africa.
conservation
within
In 1995, South Africa signed and ratified the United Nations Convention on
Biological Diversity (UNCED, 1992), its objectives are: the conservation
of biodiversity,
the
sustainable use of biological resources, and the fair and equitable sharing of benefits arising from
the use of genetic resources.
In 1997, the published response to this signing became the policy
and strategy document White Paper on the Conservation and Sustainable Use of South Africa's
Biological Diversity from the national government (South Africa, 1997).
Internationally
and
nationally, it is acknowledged that if there is to be global cooperation to conserve biodiversity,
recognition needs to be given to its uneven distribution around the world:
'Two-thirds of the world's biodiversity is located in developing countries,
collectively termed "The South", and provides an important resource for the
economic development of such countries. Biodiversity conservation thus
carries a heavier burden for developing countries than for the biologically
poorer "North", comprising the industrialized countries. Furthermore, it has
largely been private companies in industrialized countries, which have
benefited from the South's biological riches.
Thus, it is argued by
developing countries that issues such as the equitable sharing of benefits
from the conservation and use of biodiversity, must be included in any global
agreements concerning biodiversity.'
- White Paper on the Conservation and Sustainable Use of South
Africa's Biological Diversity, 1997
Under
this policy
climate,
rather
than promoting
the typical
objectives
for the
preservation of biological diversity (i.e., ecological health or biological integrity), which cannot
be (without much research and difficulty) formulated into scientifically
defensible biological
indicators, the principle goal of ecological management should be social, to maximize human
capacity to adapt to changing ecological conditions (Reid, 1994; Goodland, 1995). Reid (1994)
explains that in order to adapt to change, humanity needs both the diversity
from which
innovations can be created, and the productive ecological systems that provide biological and
economic capital to invest in those innovations.
for maximizing
humanities
Thus, maintaining biodiversity is a prerequisite
ability to respond to changing conditions,
as is maintaining
the
productivity of agricultural systems, the yield from forests and fisheries, clean water, and clean air
(e.g., Daily, 1997). Conserving biodiversity should not only be seen as a luxury or competing
land-use, but rather it is the embodiment of sustainable development within developing regions of
the world.
The socio:economic
situation in southern Africa can no longer promote large reserves
without social concessions (i.e., Peace Parks). A large portion of the common, public, and private
land is managed for renewable natural resources (i.e., livestock range, fuel wood, water, etc.), as
well as, dryland and irrigated agriculture, exotic tree plantations, and urbanization.
activities lead to some level of landscape fragmentation.
All of these
The use of a Biodiversity Management
Area (BMA) system in these various landscapes could serve as a model to provide quality core
habitat for many species (sensu Davis et aI., 1996). These ecologically managed areas would be
for those biological
components
that are negatively
impacted by human activities.
Their
arrangement on the landscape would be based on ideals of persistence as well as representation,
while acknowledging that human impacts and influences will be happening around them. In this
respect, BMAs are extended to communal, private, and public lands and across multiple habitats.
Human "quality of life" development should be allowed to go forward to the extent that they are
compatible with the goal of maintaining native species and ecosystem diversity. The concept of a
BMA is to monitor and manage in a hierarchical fashion from local ecosystem, to landscape, to
regional levels in order to reduce risk.
The goal of conserving biological diversity is to ensure population viability or persistence
over time within the required habitats.
integrated
fashion,
acknowledging
Sustainable conservation management must be seen in an
components
of population
biology,
landscape
ecology,
economics, and social needs. A major constraint to future biodiversity protection in South Africa
is that state land will not go towards conservation, but will be provided for retribution to those
landless individuals created by past British colonial and South African apartheid policies (South
Africa, 1997). In any case, the amount of available state land is low, as the almost total transfer of
land in the formerly White areas of South Africa from government to private ownership had
occurred by the mid 1930's (Christopher,
1982).
This is unique in colonialism,
as it did not
happen in other former British colonial areas outside of South Africa (i.e., Kenya, Australia,
Canada, USA, etc.). The current land ownership and land development patterns strongly reflect
the political and economic conditions of the apartheid era (Fairbanks et al., 2000).
In many regions, South Africa's biological conservation must also be viewed as managing
natural remnants. Fragments of natural landscape that are available for conservation have two
important
considerations:
Isolation primarily
configuration
isolation and human influence
affects the interior species.
are critical,
as are, corridor
from within the landscape
Therefore,
width,
matrix.
patch size; shape, number, and
and connectivity.
Patches
must have
characteristics adequate to support the interior species, and both corridors and patches must have a
configuration that permits rapid recolonization when an interior species becomes locally extinct.
Management of the flow of objects from the matrix to fragments or formal reserves is the
other focus. Human influence proceeds to minimize or eliminate these flows: there needs to be a
balance of structure.
Maintaining and creating large patches, and then surrounding these with a
high density of corridors and small patches containing edges may be a possible solution (Forman
and Collinge, 1995; Yu, 1996; Forman and Collinge, 1997). Recently, landscape ecologists and
conservation biologists have distilled their experiences into a number of conservation principles
that can be used as a basis for planning (Noss et al., 1997). These include: (1) species that are
well distributed across their historical range are less prone to extinction; (2) large patches that
support large populations support them for longer periods of time; (3) habitat patches that are
continuous (less-fragmented)
support long-term viability; (4) patches that are sufficiently close
together allow dispersal and thus support long-term viability; (5) patches that are connected by
corridors provide better dispersal; (6) patches of habitat that have minimal or no human influence
are better; and (7) populations that naturally fluctuate widely are more vulnerable than stable
populations.
Inevitably, this discussion leads to the appropriate integration of biodiversity protection
with competing economic pressure and social value. These three broad themes generally playoff
each other in an area, which then evolves landscapes into complex mosaics that natural resource
management institutions are faced with managing.
Faith (1995) and Faith and Walker (1996b;
1996c) present a multi-criteria trade-off analysis as one type of an analytical framework to assess
a region's
sustainability
(Figure 1.1). Regional sustainability
as defined by Faith (1995), will
reflect the region's success (or potential for success) in achieving effective trade-offs between
conservation and development (or other criteria).
-
l!!
::s
III
~
~
Co
.iij
•..c
·eo
.2: :;::
U
I
CII 0
i
'tlCll
I
I
.2 ~
c
I
COCII
CII III
C C
00
ClU
]1
1
...
_
Conservation (Diversity)
o
u.
!
I
I
I
I
I
!--------"
-""-~
!
Forgone Development
Opportunities
Figure 1.1: (a) The multiple interplay between the three broad themes in sustainable development
analysis; and (b) a trade-off or regional optimization space curve. A given allocation of area to
conservation or development will result in a total cost and total forgone biodiversity, so that the
allocation can be plotted as a point in this space (Faith, 1995).
Faith and Walker's
implement,
(1996b; 1996c) approach is simple, and probably
but it does clearly present the competing
goals commonly
encountered.
dissertation takes these ideas a step further by setting them in a co-evolutionary
which to view conservation planning.
model
is to establish
framework.
construct within
One of the principle aims of a co-evolutionary
human-ecosystem
interaction
within
This
dynamics
an interpretative/interrogative
This analytical framework allows for the integration of complex environmental and
socio-economic
indicators
sustainable biodiversity
allocations
dangerous to
(spatially)
to provide information
conservation.
for sometimes
needed to answer questions pertinent
for
However, persistence will depend not only on land-use
competing
land-uses,
but also the degree to which
appropriate implementation criteria and management for an area satisfies multiple goals.
This section outlines the study site, data acquisition, and initial data processing used in
this study.
The study site used in this thesis corresponds to the KwaZulu-Natal Province within the
Republic of South Africa (Figure 1.2). The province was chosen for its range of land-use/landcover, contrasting development patterns representing third world southern African and first world
Western influenced landscapes, availability of environmental and socio-economic data sets, and
access to both detailed historical and contemporary bird distribution databases.
fuwidt
•
Pietennartizburg
Stanga•
• Toogaat
~tllWlU1t
lOCrnrchy
•
I'<:<:rooriccoro
•
Mij<xm:.tropolitanccrtcr
•
l£vdqm::rt nab
•
Regicnal
rrarktt
cenlCrs
Pcriplayto\\TlS
Figure 1.2: (a) Location of the KwaZulu-Natal Province study region within South Africa; and (b)
major place names and their economic hierarchy within KwaZulu-Nata1 Province.
KwaZulu-Natal
Province is located on the east coast of South Africa and borders the
countries of Lesotho, Swaziland, and Mozambique.
KwaZulu-Natal covers just 7.6% of the land
area of South Africa, but contains the largest population base (20.7%) of any province (Stats SA,
1998). The province is an important sub-tropical agricultural and tree plantation region, and over
the last 20 years has seen increased development
pressure in direct conflict with its active
expansion of conservation based tourism (Thorrington-Smith
et aI., 1978; Armstrong et aI., 2000).
Climatically, the province is characterized by the influence of the Indian Ocean's warm
Agulhas current.
This creates a wide coastal region of sub-tropical climate, with high humidity,
high temperatures, and high summer rainfall. In southeast Africa the relief, which is in the form
of a number of ascending steps, is such that, in general, the inland isotherms tend to run in a
north-south direction, parallel to the coast.
Drakensberg
Escarpment,
KwaZulu-Natal's
western border is defined by the
which forms a marked climatic gradient.
There is a pronounced
difference in temperature between the hot eastern coastlands and the cooler interior highlands, and
at the same time, temperatures
along the coast increase gradually northwards.
The climatic
transition from the coast to the westerly plateau is, however, gradual.
Rainfall at the coast ranges from about 760 to 1400 mm per annum, and is heaviest at the
northern and southern districts of the area considered. Inland, on the seaward-facing escarpments,
rainfall is about 1750mm per annum, but on the intervening surfaces, it is considerably less. Most
of the rainfall is received during summer (September - March), but this characteristic is far more
pronounced inland than at the coast. Consequently, the region has warm, wet summers and cool,
dry winters.
The vegetation ranges from complex in the north-east, being made up of a number of
different ecological associations, which include mangrove forest, swamp forest, dune forest, sand
forest, coast forest, riverine woodland, and savanna woodland (Figure 1.3). To the south of this
area and towards the Drakensberg Escarpment, there is a marked thinning out of this complexity.
Bush clump grasslands and moist woodland dominant along the coast (south of St. Lucia),
grasslands interspersed with afromontane forests occur in the southern-central
the escarpment, dry thornwoodlands
interior and along
cover the western region of Zululand and a valley thicket
complex dominants the incised river valleys (e.g., Tugela River).
The multi-disciplinary
nature of this study required several strategic databases and used
many of the commonly available biophysical data layers.
Among the processes that have been
hypothesized to account for spatial patterns of species diversity are climatic extremes, climatic
stability, productivity, and habitat heterogeneity (Brown, 1995; Wickham et aI., 1997). Data were
compiled from existing sources to represent these processes (Table 1.1); they included climate
surfaces (Schulze,
1998) and a digital elevation model (Surveyor General 1993), as well as
potential vegetation (Low and Rebelo, 1996), and land-use/land-cover
2000) mapped in a raster-based
types (Fairbanks et aI.,
geographic information system (GIS; ESRI, 1998).
The GIS
database has a raster cell resolution of 1 km by 1 km. Both geographic and projected Albers equal
area cartographic systems were used.
Vegetation type is a primary determinant of ecosystem type (Peters, 1992), playing a
major role in determining
the associated fauna.
Two potential vegetation map products are
available for South Africa: Acock's (1953) vegetation types, which is largely based on the
agricultural potential of the vegetation, and Low and Rebelo's (1996) vegetation types, which is
based on both structure and floristics, but is essentially
a re-assessment
of Acocks.
vegetation potential map of Low and Rebelo (1996) was mapped at a scale of 1:500000.
vegetation types that occur in KwaZulu-Natal
The
The 26
were classified into eight functional community
groupings (Table 1.2; Low and Rebelo, 1996; Cowling et aI., 1997) for analysis (Figure 1.3).
Vegetation Functional Types
Settlement hierarchy
o
Economic core
o
Mljor metropolitan center
•
Mmtane furest
•
CoastallOrest
Arid woodland
Mlist \\OOdland (coastal bush)
Development nodes
Mxed \\Oodland (thom\\OodIand)
Regimal market centers
Thicket
Periphery towns
Highland grassland
Uplandllowland grassland
Figure 1.3: Functional vegetation types found within KwaZulu-Natal
vegetation types described by Low and Rebelo (1996).
Province
based on
Table 1.1: Codes and defInitions of explanatory variables, by variable subset, used in Chapters 3, 4, and 5.
Code
Defmition
Topography
DEMMEAN
DEMSTD
Climate
GDMEAN
MAP
GTMEAN
NGTMEAN
MAT
HOTMNTHMN
MINMNTHMN
EVANNMN
PSEAS MN
TSEAS MN
MXSEAS MN
Land Types
LANDt
LANDVEG
LANDVEGF
VEG
VEGF
LCLUTYFES
LCLULAND
Elevation (m)
Elevation heterogeneity (std. deviation)
Number of days per annum on which suffIcient water is available for plant growth
Mean annual precipitation (mm)
Annual mean of the monthly mean temperature (0e) weighted by monthly grow days
Mean temperature (0e) during negative water balance
Mean annual temperature (0e)
Mean temperature of the hottest month, usually January COe)
Mean temperature of the coldest month, usually July (0e)
Total annual pan evapotranspiration (mm)
Precipitation seasonality from the difference between the January and July means (mm)
Temperature seasonality from the difference between the January and July means (0e)
Maximum temperature seasonality from the difference between January and July (0e)
Variety
Variety
Variety
Variety
Variety
Variety
Variety
of
of
of
of
of
of
of
defmed landscapes from a maximum of 24
combined landscape and vegetation types from a maximum of 217
combined landscape and functional vegetation types from a maximum of 126
defmed vegetation types from a maximum of 26
defmed functional vegetation types from a maximum of 8
defmed land-cover/land-use types from a maximum of 29
combined landscape and land-cover/land-use types from a maximum of 334
t Landscapes derived from analysis presented in Chapter 3.
Table 1.2: Functional vegetation classifIcation of the 1:500000
South Africa, Lesotho and Swaziland (Low and Rebelo, 1996).
National Botanical Institute Vegetation of
Original potential vegetation types
Functional classifIcation
Afromontane forest
Coastal forest
Sand forestt
Montane forest
Coastal forest
Coastal forest
Arid woodland
Arid woodland
Arid woodland
Arid woodland
Arid woodland
Arid woodland
Arid woodland
Arid woodland
Moist woodland
Mixed woodland
Mixed woodland
Thicket
Upland/lowland grassland
Upland/lowland grassland
Upland/lowland grassland
Highland grassland
Highland grassland
Highland grassland
Highland grassland
Highland grassland
Highland grassland
Highland grassland
Highland grassland
Eastern thorn bushveld
Lebombo arid mountain bushveldt
Mixed lowveld bushveld
Natallowveld bushveldt
Sour lowveld bushveld
Subarid thorn bushveld
Subhurnid lowveld bushveldt
Sweet lowveld bushveld
Coastal bushveld grasslandt
Coastal hinterland bushveldt
Natal central bushveldt
Valley thicket
Coastal grassland
Moist upland grassland
Short mistbelt grasslandt
Afro-mountain grassland
Alti-mountain grassland
Moist clay highveld grassland
Moist cold highveld grassland
Moist cool highveld grassland
Moist sandy highveld grassland
North-eastern mountain grassland
Wet cold highveld grassland
tEndemic vegetation types to KwaZulu-Natal
Topographic position has been found in other studies to significantly influence ecosystem
variability patterns, especially the control of water movement (Kratz et aI., 1991; Forman, 1995).
A digital elevation model (DEM) of South Africa was available from the South African Surveyor
General (1993) with a horizontal resolution of 400 m by 400 m and a vertical resolution of 20 m
(Figure 1Aa). This was used to derive elevation information and a topographic landform index
(ridge, valley, slope) using standard GIS routines (Figure lAb; Fairbanks, 2000). The percent
slope surface was transformed to a surface representing flat-undulating «
4%) and ridge
landscapes (> 35%) and then a linear function scaled the slope data between the two extremes.
The principal controlling factor in southern African ecosystems is the soil water balance
(Cowling et aI., 1997; Scholes and Walker, 1993). The mean number of days per annum on
which sufficient water is available to permit plant growth was considered a biologically
meaningful index of water availability. Ellery et aI. (1992) developed such a water balance index,
which calculates the water budget from available climatology data. The index, called 'growth
days' (GD) is defined as the sum of the monthly ratios of precipitation to potential evaporation,
where the ratio is not permitted to exceed 1 in any given month (i.e., if rainfall is larger than
evaporation, it is not carried over into subsequent months, but is assumed to have been lost as
runoff). This is achieved by multiplying the monthly ratios by the number of days in the month
and summing over the year. Intuitively it can be thought of as the number of days per year when
soil moisture does not limit plant growth. The GD index was calculated on the 1 km by 1 km grid
covering the entire country (Figure lAc), from monthly mean rainfall (1960-1990) and the
monthly means of maximum and minimum daily temperatures (Dent et aI., 1989). The annual
mean of the monthly mean temperature weighted by the monthly growth days was recorded as
growth temperature (GT), giving an indication of energy supply during the growing season
(Ellery et aI., 1992), while no growth temperature (NGT) is derived from the months weighted by
no available growth days. The GT and NGT were calculated from available mean monthly
temperature surfaces (Schulze, 1998). Other climatic variables considered for use included
median annual precipitation, summed mean minimum and maximum rainfall for the driest and
wettest quarters, mean annual temperature, and mean minimum and maximum temperatures for
the coldest and hottest months. The seasonal variability with precipitation, temperature, and
evapotranspiration were calculated from these raw datasets (Table 1.1).
Elevation Classification (m)
D
D
m
m
~
Coastal plain
Coastal hinterland
Lowlands
Mid-lowlands
Upper lowlands
III
m
III
••
Low highlands
Mid-bighlands
Upper higblaods
Low Afro Alpine
Upper Afro Alpine
•
Topographic Classification
D
Ridge
Valley
Level (flat)
II
11
11
II
Footsl"!'e
Growth Days Classification
Midslope
Upperslope
SC3Ip
Dry
II
II
ModenItely dry
Moderately moiSl
II
••
Moist
Wet
Vet)'wet
Biological atlases had their precedent made when Perring and Walters (1962) published
the Atlas of the British Flora. Using a 10 kIn by 10 kIn gridded map, plant distributions were
plotted on a presence/absence
basis. This pointed the way for similar comprehensive and equally
objective mapping of the breeding birds of Britain (Sharrock, 1976).
This British tradition in
naturalist field biology was adopted during the 1970s in South Africa by the Natal Bird Club.
They developed a project whose aim was to map the distributions,
by presence/absence
per
month, of all bird species occurring in KwaZulu-Natal during the decade 1970-79 (with emphasis
on 1975-79), using the national quarter-degree grid (15 min x 15 min; -24 kIn x 28 kIn, hereafter
referred to as a grid cell).
topocadastral
Each of these grid cells represents one of the maps in the 1:50 000
map series produced by the South African Surveyor General (Figure 1.5). The
objectives were to present occurrences of birds in KwaZulu-Natal,
in the avifauna could be measured.
against which future changes
Data collection was conducted by means of fieldcards
submitted by club members, Natal Parks Board, and the authors of the atlas. In 1980, Cyrus and
Robson published the Bird Atlas of Natal, which represented a thorough account of the birds
found in the province during the 1970s.
Starting in 1987, the Southern African Bird Atlas project (Harrison, 1992) was initiated
by the Avian Demography Unit (ADU), University of Cape Town. The aims of their project were
the same as for the Cyrus and Robson (1980) survey, but designed to cover the entire Southern
African sub-region (South Africa, Lesotho, Swaziland, Namibia, Botswana, and Zimbabwe).
The
same procedures as used by Cyrus and Robson were adhered to (Nigel Robson was appointed as a
science steering committee
presence/absence
member), along with the continued
use of the grid cell.
The
of species was recorded during 1987-1992 (see Underhill et aI., 1991; Harrison,
1992; Harrison et aI., 1997 for details).
In the original forward to Cyrus and Robson (1980), Gordon Maclean (author of Robert's
Book of South African Birds, 1984) explained that the greatest apparent shortcoming of any
biological atlas is that it is out of date even as it comes off the press.
because
it illustrates
anthropogenic
impact.
the dynamic nature of biological
This is as it should be,
systems, especially
in the face of
Therefore, an atlas becomes increasingly valuable as it highlights the
changes that are constantly occurring.
Baselines for future comparisons become more necessary
every day, so an atlas of distribution in time and space becomes an invaluable tool in the hands of
planners, geographers, and conservation biologists.
KwaZulu-Natal
forms less than one percent
of the Afrotropical Region (Africa south of the Sahara), yet its economy in the late 1970s may
have been the largest per unit area, and its rate of progress close to the highest on the whole
continent.
Maclean made note, at that time, that a measure of the natural resources of KwaZulu-
Natal had become more critical than ever.
The Cyrus and Robson (CR) dataset comprises 33689 unique distribution records of 633
species covering 165 grid cells. The ADU dataset, clipped to cover the same number of grid cells,
includes 40036 unique distribution records of 604 species of resident and visiting birds, which
comprise 65% of the bird diversity recorded for the Southern African sub-region.
The reporting
rates for both datasets show observer bias in and around the Durban and Pietermaritzburg
areas,
and the Drakensberg and the Zululand game reserves (Figure 1.6). Nevertheless, for each survey
period>
90% of the grid cells had at least one fieldcard returned for recording for each month of
the year. In the case of the ADU survey the intensity of the recording during the 5-year survey
(1987-92) allowed for an average of 105 fieldcards returned per grid cell. This level of reporting
allowed the transformation of the number of times a species was recorded into relative abundance
values, which were used to analyze avian assemblage structure in Chapter 5. Unfortunately, this
type of data was not recorded within the CR atlas.
Investigations
of the patterns in these bird atlases have been conducted using several
biological and practical classifications.
For each atlas, the birds were first grouped by life history
class and then, for only the ADU atlas birds, grouped by primary ecological habitat requirement.
Waterbirds
were not analyzed separately as Guillet and Crowe (1985; 1986) had previously
examined them. Wetland and waterbody sites are also already protected under the South African
signing of the RAMSAR convention for wetland conservation (Cowan and Marneweck,
1996).
Table 1.3 describes each of these datasets and provides the dataset name, as it will be referred to
throughout
the thesis.
The conservation
reasoning, but instead on the requirements
dataset is the only dataset not based on biological
of the local conservation
authorities
for planning
purposes conducted in Chapter 4.
The South African National Land-Cover database (NLC; Fairbanks and Thompson, 1996;
Fairbanks
percentages
et aI., 2000) was used to derive land-cover/land-use
for each grid cell.
(LCLU) and transformation
This national database was derived using photo-interpretation
techniques from a series of 1:250,000 geo-rectified hardcopy satellite imagery maps, based on
seasonally
standardized,
single date Landsat
Thematic
Mapper
principally during the period 1994-95 (Fairbanks and Thompson,
satellite
1996).
imagery,
captured
It provides the first
single standardized database of current LCLU information for the whole of South Africa, Lesotho,
and Swaziland (see Fairbanks et al., 2000). For the purpose of this thesis, the 31 LCLU classes
were reclassified into three categories: un-transformed, low intensity transformation, and high
m>ID
17ZlDA
"DAD
mICA
"))(lJ
"DOC
Z731AD
"DID
mlOA
17)100
""OC
2l<31BA
2831fB
2932AD
2832(3'\
28)JOC
28:lJlD
2llZlDA
''''00
29DCA
lllD03
28:lJDA
29Doo
11129<D
28ZlOC
2929DO
28)J(C
20J(l)
20JOC
28)}OO
2llJICC
21131(l)
2929AD
2')Zl(3'\
2929ID
Zl:lJA!3
2')):)(3'\
29:1Jffi
Zl3lAA
2931A!3
292')0\
2')2')03
2')2'JDA
292900
29DW
29:1)00
2'))10\
2929<:£
"290)
292'JOC
Z:l29DO
29DW
2JJICC
D29AD
D290C
D291n
'12900
D29CA
:112103
:Il21DA
D2900
2831AC
21131AD
2l<31lD
2932AC
283100
2l<32CA
283100
1S32CC
2932AD
ZlJlBA
DDID
3UIAA
D29AC
"32CA
283IAO
lllDAD
Xl28lD
""OC
173203
1732(1)
28)}A(:
:Il2JA13
17l2AC
mlw
2929ID
2')):IOC
""IE
17310C
lllDB<
2'HKD
""B<
1732.'\D
1731CC
2l<2JOC
2J:mx
"32AD
mloo
28:lJA·\.
2829CC
"32M
1731DA
2929IE
2929CA
>;3200
173103
"<VB<
"DW
milD
>;32OC
1731(l)
172900
"Dm
Z7310C
>;32<D
Z73ICA
Z7ZlOC
2929AD
11129DO
m>oo
IDJAC
l>l2CC
DDOC
)))ICA
D:OW
J3))DA
DDO)
D29DO
31))AA
40
'--
0
40
80
160 Kilorreters
120
!JIII"""""'I!
Figure 1.5: The 1:50000 map sheet system of grid cells for KwaZulu-Natal used to record bird
distribution data during both survey periods.
I
Reporting rate
D
<20
20· 100
• •
100· 200
200>
Figure 1.6: Spatial distributions ofretumed fieldcards and histograms distributions: (a) Cyrus and
Robson (1980); and (b) Harrison et al. (1997).
Life History
All birds
All birds found in each dataset. 633 in the Cyrus and Robson (1980)
survey and 604 in the ADD (Harrison et aI., 1997) survey.
Summer
Birds recorded during the months September-March.
Winter
Birds recorded during the months April-August.
Birds classified as passerine in the descriptions provided by Harrison et al.
(1997). Chiefly altricial songbirds of perching habits.
Birds classified as non-passerine in the descriptions provided by Harrison
et al. (1997). Relating to an order of arboreal birds including the rollers,
kingfishers, hombills, cranes, raptors, etc.
Birds classified as breeding in South Africa and in particular to KwaZuluNatal as provided by Harrison et al. (1997).
Birds classified as not breeding in South Africa or KwaZulu-Natal as
provided by Harrison et al. (1997).
Birds classified as being positively influenced, usually by habitat and
therefore distribution, by human activity and/or land-use as described by
Harrison et al. (1997).
Non-human
influenced
Birds classified as either neutral to or negatively influenced by human
activity and/or land-use as described by Harrison et al. (1997).
Ecological habitat
Woodlandt
Birds primarily associated with savanna woodland habitat.
Forestt
Birds primarily associated with indigenous evergreen forest (afromontane,
coastal, and sand forest).
Thickett
Birds primarily associated with thickets, bushland, and bush clumps.
Grasslandt
Birds primarily associated with perennial grasslands.
Planning
Conservationt
Birds considered for representation in conservation efforts within
KwaZulu-Natal (derived from personal analysis; Important Bird Areas of
South Africa (1999); KwaZulu-Natal Nature Conservation Services).
f Relative abundances
derived from the reporting rate in the ADU dataset were used instead of presence/absence
; Only created from the more current ADU bird database.
measure.
intensity transformation land (Table 1.4; Figure 1.7a,b). Un-transformed class included all natural
vegetation, e.g., forest, woodland, thicket, and grassland.
Degradation, erosion, and subsistence
agriculture dominated the low intensity category. These areas have a very low vegetation cover in
comparison with the surrounding natural vegetation cover and were typically associated with rural
population
centers and subsistence level farming, where fuelwood removal, over-grazing, and
subsequent soil erosion were noticeable within the satellite imagery (Thompson, 1996; Fairbanks
et aI., 2000). The high intensity transformed category consisted of areas where the structure and
species composition were completely or almost completely altered, which includes all areas under
crop cultivation, forestry plantations, urbanized areas, and mines/quarries.
The LCLU classes are
essentially a measure of transformation status in the context of threats to biodiversity (Figure 1.8).
The developing
theoretical
field of landscape
basis for understanding
ecology has provided
landscape
structure,
function,
a strong conceptual
and
and change (Forman and
Godron, 1986; Urban et aI., 1987; Turner, 1989). Landscape ecology is largely founded on the
notion
that the patterning
characteristics,
of landscape
elements
including vertebrate populations.
(patches)
strongly
influences
ecological
Therefore, the ability to quantify landscape
structure is a prerequisite to the study of landscape function and change.
For this reason, much
emphasis has been placed on developing methods to measure landscape structure (e.g., O'Neill et
aI., 1988; Turner,
1990; Turner and Gardner,
1991; Li et aI., 1993).
While a number of
investigators have quantified landscape structure in a variety of ecosystems (e.g., Krummel et aI.,
1987; Turner
and Ruscher,
1988; Gustafson
and Parker,
1992), few have examined
the
relationship between landscape structure and landscape function (e.g., Romme, 1982; Franklin
and Forman, 1987; Baker, 1992; Baker, 1993).
The growing concern over the loss of biodiversity
has challenged
traditional
local
conservation strategy into developing better ways to examine and manage landscapes at a variety
of spatial and temporal scales.
Remote sensing developments have made it possible to analyze
and manage entire landscapes to meet multi-resource objectives. As part of this study, in addition
to LCLU proportions calculated per grid cell, a number of common landscape mosaic and class
type pattern metrics were calculated (Table 1.5 and 1.6) for use in Chapters 5 and 6. The program
FRAGSTATS (McGariga1 and Marks, 1995) was used to calculate the spatial configuration of the
LCLU within each grid cell and magisterial district. Landscape mosaic and class indices were
calculated using the raster grid option. The LCLU data was converted to a grid cell resolution of
100 m, which is considered appropriate for the NLC database (Fairbanks and Thompson, 1996),
development of pattern metrics (O'Neill et aI., 1996), and the coarse-scale of this study.
Twenty-eight
landscape mosaic indices of LCLU configuration
were used that were
considered appropriate for the land area of KwaZu1u-Nata1 (Table 1.5) and 28 class level indices
were calculated for each of the general vegetation types mapped (Table 1.6; woodland, forest,
thicket, and grassland). These pattern indices quantify different aspects of configuration, although
many are redundant
and simply represent
alternative
formulations
of the same formulation
(McGariga1 and Marks, 1995). The landscape boundary was considered the edge of the grid cell
or magisterial district for the purpose of calculating all the metrics.
procedure
means
that the true
sizes of patches
will decrease
The implications of this
because
of the closing
Land-cover/Land-use
II
•
••
••
Forest & Woodland
Forest
Thicket & Bushland
Shrubland
Grassland
Pature
Exotic plantations
Water
Wetlands
Bare rock
Bare soil (erosion)
Degraded
Degraded
Degraded
••
•
Degraded
Cultivated, permanent Irrigated
CUltivated, permanent. dryland
Sugarcane
Cultivated, temporary, Irrigated
Cultivated, temporary. dryland
Cultivated, subsistence. dryland
Urban/built-up
Urban/built-up
Urban/built-up
Urban/built-up
Urban/built-up
Urban/built-up
•
~.,,;
i..'
Urban/built-up
Mines & Quarries
No Data
Transformation rating
Un-transformed
Low intensity transformation
•
High Intensity transformation
No Data
Figure 1.7: (a) Simplified map of land-cover/ land-use distribution across KwaZulu-Natal
province; and (b) three level transfonnation map derived from Table 1.4.
Table 1.4: Land-cover/land-use classes used in the South African National Land-Cover (NLC)
database and the re-coded transformation classes used for this study.
Transformation Classes
Original NLC Classes
1
2
3
4
5
6
7
8
9
10
11
12
13-17
18-22
23
24
25-28
Un-transformed
Un-transformed
Un-transformed
Un-transformed
Un-transformed
Un-transformed
High intensity
High intensity
Un-transformed
Un-transformed
Un-transformed
Low intensity
Low intensity
High intensity
Forest and Woodland (savanna)
Indigenous Forest
Thicket, Bushland, or Bush Clumps
Low Shrubland and/or Fynbos
Herbland
Grassland
Improved Grassland (pasture, recreational fields)
Forest Plantations (exotic tree spp.)
Waterbodies
Wetlands
Bare Rock & Soil (natural)
Bare Rock & Soil (erosion surfaces)
Degraded Vegetation (NLC codes 1,3,4,5,6)
Cultivated lands (variations of commercial
permanent/temporary crops, irrigated/dryland, and sugarcane)
Cultivated lands (dryland subsistence)
Urban/bui1t-up land (residential)
Urban/built-up land (residential small holdings by subdivided
vegetation; NLC codes 1,3,4,5,6)
Urban/Built-up land (commercial)
Urban/Built-up land (industrial/transport)
Mines and Quarries
Low intensity
High intensity
Low intensity
High intensity
High intensity
High intensity
Biological Diversity
"!lIII---------------Evergreen Forests
Forest & Woodland
Grassland
Shrub lands & Low Fynbos
Thickets, Bushlands, Bush clumps
Commercial Agriculture
Plantation Forestry (exotic species)
....•----
Natural
-----------~
SYSTEMS
Artificial
Figure 1.8: Conceptual model of the impacts of increasing levels of human developed land-use on
biodiversity and natural processes (modified from Reid et aI., 1993).
Table 1.5: Codes and definitions of explanatory landscape mosaic indices used in Chapters 5 and
6, by variable subset.
Acronym
Landcover
POPTOT96
POPDEN96
FOR]ER
GRS]ER
WET]ER
LOWI]ER
PLNT]ER
DRY]ER
IRR]ER
URB]ER
M PER
T]ER
T_TOTAL
ROAD INDEX
Total population from 1996 census
Population density from 1996 census
Percent woody cover
Percent grass cover
Percent wetland and waterbody cover
Percent subsistence agriculture cover
Percent exotic plantation and woodlot cover
Percent commercial dryland agriculture cover
Percent commercial irrigated agriculture cover
Percent urbanization cover
Percent low intensity transformation
Percent high intensity transformation
Percent total transformation cover (i.e. combined low and high intensity transformation)
Percent road density cover
Patchiness
LPI
NP
PD
MPS
PSSD
CI
Largest patch index (%) - percent of landscape composed of the largest patch
Number of patches
Patch density (no.!1 00 ha)
Mean patch size (ha) - average size of patches in landscape
Patch size standard deviation (ha) - absolute measure of patch size variability
Contagion index - measure of dumpiness of patches within the landscape (continguity
across landscape).
Shape
MSI
AWMSI
FD
MPFD
Mean shape index - mean patch shape complexity; equals I when all patches are circular and increases as
patches become non-circular
Area-weighted mean shape index - similar to MSI, but patch shape index weighted by patch area
Fractal dimension - measure of shape complexity as a departure from simple Euclidean geometry
Mean patch fractal dimension - mean patch shape complexity; approaches I for simple geometric shapes
(e.g., circle, square) and 2 for complex shape; adjusted to correct for bias in perimeter
Area-weighted mean patch fractal dimension
Interior
MCAPP
PCASD
DCASD
DCACY
Mean core area per patch (ha) - sum of core areas divided by the number of patches
Patch core area standard deviation (ha) - square root of the sum of the squared deviations of each patch
core area from the mean core aras per patch, divided by the number of patches of the same type
Mean area per disjunct core (ha) - sum of the disjunct core areas of each patch, divided by the number of
disj unct core areas
Disjunct core area standard deviation (ha)
Disjunct core area coefficient of variation (ha)
Isolation
MNND
NNSD
MPI
Richness
CR
CRD
Mean nearest-neighbor distance (ha) - sum of distance to nearest patch divided by number of patches
Nearest-neighbor standard deviation
Mean proximity index - sum of patch area divided by nearest edge-to-edge distance squared between the
patch and the focal patch of all patches of the corresponding patchy type whose edges are within 500 m
Interspersion index (%) - measure of patch type adjacency against all other patch types (i.e., maximally
interspersed and juxtaposed to other patch types)
Class richness
Class richness density
Heterogeneity
SHDI
SOl
MSDI
Shannon diversity index
Simpson diversity index
Modified Simpson diversity index
Evenness
SHEI
Shannon evenness index
SEI
Simpson evenness index
MSEI
Modified Simpson evenness index
tSee McGarigal and Marks (1995) for a complete description and definition of each index.
I
115'7~LJ.~c;7
ble; 'A
)bblo"J
Table 1.6: Codes and definitions of explanatory class level pattern indices used in Chapter 5, by
variable subset.
Patchiness
LAND%
LPI
NP
PD
MPS
PSSD
pscv
Percentage of the landscape composed of the corresponding patch type
Largest patch index (%) - percent oflandscape composed of the largest patch
Number of patches
Patch density (no.llOO ha)
Mean patch size (ha) - average size of patches in landscape
Patch size standard deviation (ha) - absolute measure of patch size variability
Patch size coefficient of variation (%) - relative measure of patch size variability
Shape
MSI
AWMSI
MPFD
AWMPFD
Mean shape index - mean patch shape complexity; equals I when all patches are circular and increases as
patches become non-circular
Area-weighted mean shape index - similar to MSI, but patch shape index weighted by patch area
Mean patch fractal dimension - mean patch shape complexity; approaches I for simple geometric shapes
(e.g., circle, square) and 2 for complex shape; adjusted to correct for bias in perimeter
Area-weighted mean patch fractal dimension
Interior
CADI
NCA
CAD
MCAPP
PCASD
PCACV
MAPDC
DCASD
DCACY
TCA%
MCA%
Core area density index (%) - percentage of the landscape composed of core areas of the corresponding
patch type
Total core area (ha) - total amount of core area of the corresponding patch type; core areas were defined
by eliminating a 100 m wide buffer along the perimeter of each patch
Number of core areas - number of core areas, as defined above
Core area density (no.lIOO ha) - density of core areas, as defined above
Mean core areas per patch (ha)
Patch core area standard deviation (ha) - square root of the sum of the squared deviations of each patch
core area from the mean core aras per patch, divided by the number of patches of the same type
Patch core area coefficient of variation (ha)
Mean area per disjunct core (ha) - sum of the disjunct core areas of each patch, divided by the number of
disj unct core areas
Disjunct core area standard deviation (ha)
Disjunct core area CY (ha)
Total core area index (%) - total percentage of the class type that is core area
Mean core area index (%) - average percentage of a patch that is core area
Isolation
MNND
NNSD
NNCV
MPI
Mean nearest-neighbor distance (ha) - sum of distance to nearest patch divided by number of patches
Nearest-neighbor standard deviation
Nearest-neighbor coefficient of variation
Mean proximity index - sum of patch area divided by nearest edge-to-edge distance squared between the
patch and the focal patch of all patches of the corresponding patchy type whose edges are within 500 m
Interspersion index (%) - measure of patch type adjacency against all other patch types (i.e., maximally
interspersed and juxtaposed to other patch types)
tSee McGarigal and Marks (1995) for a complete description and definition of each index.
of the patches by an artificial study area boundary.
Since there is nothing simple that can be done
about this, conclusions drawn from the analysed data are appropriately tempered.
Several core
area indices were calculated based on a specified edge width, which, for the purpose of this study,
was defined as 100 m wide buffer along the perimeter of each patch.
This width represents a
somewhat arbitrary decision based, in part, on avian studies by Temple (1986), McGarigal and
McComb (1995), and studies by Laurance and Yensen (1991) and Laurance (1994). Edge related
metrics were not calculated for this study because of confounding using the arbitrary grid cell and
geopolitical magisterial district as sampling units. The use of the equal area grid cell, however,
did reduce the effects of area in the metric calculations for the analysis, eliminating the need for
regression area correction suggested in other landscape pattern metric studies (e.g., McGarigal
and McComb, 1995). However, this technique was used in Chapter 5 to remove the area effects
confounding the magisterial district metrics.
In addition to LCLU threats, one of the most widespread forms of alteration of habitats
and landscapes
(Trombulak
over the last century has been the construction
and Frissell, 2000).
and maintenance
of roads
Road networks affect landscapes and biodiversity
in seven
general ways: (1) increased mortality from road construction; (2) increased mortality from vehicle
collisions;
(3) animal behavior modification;
(4) alteration of the physical environment;
(6)
alteration of the chemical environment; and (7) increased alteration and use of habitats by humans
(from Trombulak and Frissell, 2000).
USA (Forman
and Alexander,
These networks cover 0.9% of Britain and 1.0% of the
1998), however the road-effect
zone, the area over which
significant ecological effects extend outward from the road, is usually much wider than the road
and roadside.
Thus, while the LCLU database provides a reasonable estimate of areas with high
current vulnerability to biodiversity loss due to existing anthropogenic land transformation; roadeffect zones can be used to provide another estimate of the threat to avian biodiversity.
Some evidence on the size of the road-effect zone is available from studies in Europe and
North America.
Reijnen et al. (1995) estimated that road-effect zones cover between 12-20% of
The Netherlands, while Forman (2000) illustrated that 19% of the USA is affected ecologically by
roads and associated traffic.
The road-effect zone for KwaZulu-Natal
was determined using a
similar method to that used by Stoms (2000) in which the spatial extent of road effects can be
used as an ecological indicator that directly represents impacts on biodiversity.
effect zone was used as a measure of the area potentially
For this, the road-
affected by roads.
The affected
distances were estimated from the reviews mentioned above, as well as from local published
studies (Milton and MacDonald, 1988), and unpublished data, which demonstrated that more than
80% of the transformed
approximately
area of KwaZulu-Natal
61 % of the untransformed
Province occurs within 2 km of a road, with
areas occurring within the same distance (Pers. Com.
Grant Benn, 1999). Therefore, national routes and freeways were assumed to affect biodiversity
for a greater distance from the roadway (1 km on each side) than dirt roads (50 m; Table 1.7).
Road segments from the South African Surveyor General (1993) 1:500 000 map series
files (Figure 1.9) were buffered to the distance related to its class. The roads in protected areas
were excluded from this analysis as the road-effect in nature reserves is of little concern in this
study. A road disturbance index was calculated within each grid cell by summing the total area of
the buffered roads and converting to a percentage of that grid cell.
Table 1.7: Buffer widths assigned to road classes for calculating road effect zone (after Stoms,
2000).
Buffer width (m)
National route
Freeway
Arterial
Main
Secondary (connecting and magisterial district roads)
Other (rural road)
Vehicular trail (4 wheel drive route)
1000
1000
500
250
100
50
25
Road network
N
N
N
National route
Freeway
Arterial
N
Main
j\/
Secondary
Other
,
"
",
"
'
"
Three databases of available social and economic indicators were examined for variables
that would cover the entire province using the latest magisterial district definition (Table 1.8) and
distributions from the 1996 Census (Figure 1.10). By limiting the data to the 1996 boundaries
used in the 1996 South African Census (Stats SA, 1998) a whole host of historical census and
economic data was made unacceptable for this study.
This is rather unfortunate, however, the
radical changes in districting that have accompanied the disbanding of the apartheid state have
seen the magisterial districts and boundaries change five times since the 1991 census. The 1996
census ushered in the first reliable geographic results of the countries demography.
Boundary
Figure 1.10: (a) Magisterial districts used for the 1996 Census; and (b) magisterial districts in
relation to the former KwaZulu and Transkei homeland boundaries (pre 1994; shaded gray).
Table 1.8: Codes and names of magisterial districts in KwaZulu-Natal Province.
HSRC
code
CD 200
CD 201
CD 202
CD 203
CD 204
CD 205
CD 206
CD_207
CD 208
CD 209
CD_210
CD 211
CD 212
CD 213
CD 214
CD_215
CD 216
CD 217
CD 218
CD_219
CD 220
CD 221
CD 222
CD 223
CD 224
CD 225
Magisterial district
Mount Currie
Alfred
Port Shepstone
Urnzinto
Ixopo
Polela
Underberg
Impendle
Richmond
Camperdown
New Hanover
Lions River
Pieterrnaritzburg
Mooi River
Estcourt
Ween en
Bergville
Umvoti
Kranskop
Durban
Inanda
Pinetown
Chatsworth
Kliprivier
Glencoe
Dundee
HSRC
code
CD 226
CD 227
CD_ 228
CD 229
CD 230
CD 231
CD_232
CD 233
CD 234
CD 235
CD 236
CD 237
CD 238
CD_503
CD 504
CD 506
CD 510
CD 513
CD 514
CD_515
CD 519
CD 520
CD 521
CD_522
CD 523
CD -552
Magisterial district
Dannhauser
Newcastle
Utrecht
Paulpietersburg
Vryheid
Ngotshe
Lower Tugela
Mtunzini
Eshowe
Mthonjaneni
Babanango
Lower Umfolozi
Hlabisa
Umbumbulu
Umlazi
Ndwendwe
Mapumulo
Nkandla
Nqutu
Msinga
Mahlabathini
Nongoma
Ubombo
Ingwavuma
Simdlangentsha
Urnzimkulu t
This district is managed by the Eastern Cape Province but has been included as part of KwaZulu-Natal
for this study.
t
problems with the ex-homelands,
especially in KwaZulu-Natal,
were finally removed, yet the
spatial landscape characteristics of their former presence was not.
The socio-economic
data was drawn from the 1996 census (Stats SA, 1998), 1996
KwaZulu-Natal Service Needs and Provision (Human Sciences Research
Council,
HSRC;
Schwabe et aI., 1996), and the 1997 KwaZulu-Natal Development Indicators (Human Sciences
Research Council, HSRC; Kok et aI., 1997) databases. The last two databases are unique in South
Africa, as KwaZulu-Natal
province is the only region to have rather recent social surveys
conducted for each magisterial district based on development
indicators (i.e., need for water,
sewer, etc.) that provide information on basic needs and tensions.
descriptive breakdown of the eighty-four socio-economic
Appendix A provides the
and environmental
indicators used in
Chapters 5 and 6.
KwaZulu-Natal
Nature
Conservation
Service provided
a spatial
database
of their
provincial protected areas (Figure 1.11). The protected areas database describes the boundaries of
provincial reserves, digitized from 1:50 000 maps.
descriptions of the protected areas.
Table 1.9 provides the names and basic
The spatial distributions of private conservation areas and
game farms were not available for the analyses.
The compilation
of a series of studies described in this thesis is unique from most
traditional landscape ecological and conservation biology studies in at least three major respects.
The first is the coarse-size of the geographical sampling unit from which the species distribution
information is derived; the second is the quantification of coarse-scale avian turnover related to
environmental
and landscape pattern gradients; and the third is the pattern analysis of socio-
cultural and economic data in relation to evolved landscape pattern.
Typically, most quantitative bird analyses have used plots or transect as sampling units.
The aim of such studies is to characterize local avian-vegetation relationships (e.g., Wilson, 1974,
Forman et aI., 1976; Cody, 1985; Opdam et aI., 1985; Opdam et aI., 1984).
samples have been used in coarse-scale
Small plot based
avian analysis for many years (e.g., Wiens, 1973;
Rottenberry and Wiens, 1981; Wiens, 1989a; McGarigal and McComb,
schemes rely on subjective choices to find representative
"homogeneous"
1995).
The sampling
vegetation plots in a
much larger vegetation community type or landscape within which a birds presences and relative
abundances are recorded.
Vryheid
Newcastle
'50
~35
Ladysmith
-~
&.
41
42
2~
,.26
27--
••
23
21'
",22
&.
Pietennartizburg
13~
10
11
tf:'
12
}.9
18. ~
17 '16
Figure 1.11: Protected areas of KwaZulu-Natal
Conservation Services.
When approached
Durban
Province managed by KwaZulu-Natal
Nature
from the regional scale, the plot sampling strategy leads to scale
problems (Wiens, 1981) and to substantial under representation ofless common species due to the
shorter survey periods (Preston, 1948). The plots can usually provide ecologists with an idea of
how species grade with the environment
on a fine scale, but comprehensive
bird species
information for a vegetation type or landscape is always limited by time and sampling effort. The
effort described
in this thesis is a trade-off
of spatial precision
for more comprehensive
community inventory in coarse mapsheet units. This study also has the added advantage of not
having to worry about high frequency spatial and temporal effects (Preston, 1960).
Table 1.9: Names and descriptions of the protected areas managed by KwaZulu-Natal
Conservation Services.
Map
Code
Name
Descri tion
1
2
3
4
Umtamvuna Nature Reserve
Mpenjati Nature Reserve
Skyline Nature Reserve
Oribi Gorge Nature Reserve
The Valleys Widlife Sanctuary
Mount Currie Nature Reserve
Bruce's Valley Natural Heritage Site
Vernon Crookes Nature Reserve
Soada Forest Nature Reserve
Greater Ingwangwana River
Greater Ingwangwana River
Greater Ingwangwana River
Coleford Nature Reserve
The Swamp Nature Reserve
Himeville Nature Reserve
Bluff Nature Reserve
Stainbank Nature Reserve
Paradiase Valley Nature Reserve
KrantzkloofNature
Reserve
Hazelmere Public Resort Nature Reserve
Doreen Clark Nature Reserve
Albert Falls Nature Reserve (dam)
Midrnar Dam Nature Reserve
Umgeni Vlei Nature Reserve
Fort Nottingham Heritage Site
Umvoti Vlei Nature Reserve
KarkloofNature Reserve
Blinkwater Nature Reserve
Weenen Nature Reserve (dam)
Wa endrift Nature Reserve (dam)
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Map
Code
Name
Descri tion
31
Natal Drakensberg Park
Tugela Drift Nature Reserve
Spoienkop Dam Nature Reserve
Royal Natal National Park
Chelmsford Dam Nature Reserve
Harold Johnson Nature Reserve
Amatikulu Nature Reserve
Dlinza Forest Nature Reserve
Entumeni Nature Reserve
Nkandla Nature Reserve
Qudeni Forest reserve
Tugela Gorge
Ngoye Forest Reserve
Richards Bay Game Reserve
Enseleni Nature Reserve
Lake Eteza Nature Reserve
Opathe Nature Reserve
Umfolozi-Hluhluwe Game Reserve
Ngomi Forest Reserve
Vryheid Mountain Nature Reserve
Pongola Bush Nature Reserve
ltala Game Reserve
Greater St. Lucia Wetland ParklMarine
Mkuzi-Pumulanga Game Reserve
Pongolwane Biosphere Reserve
Sileza Forest Reserve
Tembe Elephant Park
Ndumo Game Reserve
Umlalazi Nature Reserve
Ma utaland Bios here Reserve
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Nature
Reserve
This thesis is presented as chapters that document a set of studies that are stand-alone
papers. Several of the chapters have been published or are in press (Chapters 3, 4, and 5) and the
remaining Chapters (2, 5, and 6) are prepared for submittal. The chapters, however, are designed
around the central theme of conservation within human dominated systems, and thus the whole is
much greater than the parts. Chapter 2 provides the reader with a background and justification to
the theory of co-evolution and its integration within landscape ecology. The argument sets up the
analytical framework to be used in the subsequent chapters to show the need for this type of
approach for biodiversity conservation in developing nations.
Chapter 3 details the creation and
critiques the use of a landscape model for conservation area identification.
methodology
and procedures
complimentary-based
areas.
to use environmental
structure.
analysis
in conjunction
with
reserve selection algorithms to analyze and prioritize avian conservation
The effort in Chapter 5 looks at multi-scaled
community
gradient
Chapter 4 defines a
The study determines
spatial effects on avian diversity and
how changes
in landscape
structure
(both
composition and configuration)
affect bird populations in the spatially and temporally dynamic
landscapes at the extents of South Africa and of KwaZulu-Natal
Province.
pattern analysis of human dominated landscapes evolved in association
variables.
The emphasis is on the co-evolutionary
Chapter 6 provides
with socio-economic
model outlined in Chapter 2 and aimed at
assessing the ideal reserve system for birds developed in Chapter 4.
Thus, Chapter 6 brings
together lessons and results developed in all the previous chapters to lend support to a revision of
biodiversity threat models and analysis.
The thesis work is then rounded off with conclusions
providing a broader message on the impact of the results and delivering final thoughts on the
stated effort.
2.
Developing a Co-evolutionary Landscape Ecology Framework to
Address Sustainable Biodiversity Conservation
To understand the crisis with respect to the destruction of biodiversity we urgently require
an analytical framework, which takes into account socio-cultural values, economic systems, and
the biophysical theater in which this tragedy takes place.
process further urge the development
responses
of ecosystems
The growing rates of this destructive
of a conceptual framework aimed at understanding
to habitat destruction,
with associated
landscape
change.
the
Such
information will be available by the integration of both field and theoretical studies. What needs
to be articulated
for defined regions of the world are the principles
biodiversity
threats have evolved.
By developing
an appropriate
evolutionary
thought and landscape ecology principles, the likelihood
upon which the actual
framework
based on co-
of potential landscape
changes across a variety of systems may be assessed to guide conservation planning efforts.
Understanding the form, behavior, and historical context of landscape dynamics is crucial
to understanding
ecosystems and subsequent biological diversity at several temporal and spatial
scales (O'Neill et aI., 1986; Noss, 1990; Forman, 1995). This understanding and analysis should
not be limited to the physical or natural history of landscapes, but must include landscapes within
an anthropogenic
context first noted by Sauer (1925).
respect to biodiversity
conservation
In essence, sustainability research with
could be better addressed by way of a co-evolutionary
landscape ecology framework.
There have been numerous calls for the study of landscape or ecosystem diversity and
function for conservation purposes (e.g., Noss, 1983; Forman, 1989; Franklin, 1993; Forman,
1995; Walker, 1995; Risser, 1995; Folke et aI., 1996). Because conservation of species diversity
depends on conservation of the habitats and landscape ecosystems in which species live (Noss
1990; Franklin 1993), a greater attention should be given to understanding
economic,
social,
Fundamentally,
and cultural
diversity
of human
and examining the
groups in landscapes
within regions.
landscapes can be viewed as the critical spatial scale at which biodiversity is
minimized, as it is the scale where macro and microeconomic policies converge.
This chapter argues that problems related to biodiversity loss, landscape resilience and
ecosystem integrity have at their root a co-evolutionary response.
A conceptual development and
proposed research agenda to enhance the theoretical and application framework for biodiversity
conservation planning within developing country landscapes is expressed.
As the scale of the world's socio-economic situation continues to grow there is increasing
demand for land and its resources.
A firmer knowledge of changes in the diverse landscapes of
developing countries must aid the urgent need to join environmental management that is sound
with economic development that is viable in the long-term.
with-development
has been labeled
"sustainable
The imperative for conservation-
development"
(Goodland,
1995).
Much
publicized mandates for sustainable development echo forth as a sine qua non of conservation in
the developing countries of the world, a seeming panacea for the world's environmental problems.
Nonetheless,
sustainable development has remained a general concept and one that is subject to
unending debate (Redclift, 1987; Dovers and Handmer, 1993; Meffe and Carroll, 1997). Indeed
the more exact meanings of sustainability are typically lacking. The most widely used definition
of sustainability states: 'A sustainable condition is one in which there is resilience for both social
and physical systems, achieved through meeting the needs of the present without compromising
the ability of future generations to meet their own needs' (WCED, 1987).
Stability is substituted in the original statement for the operative term, resilience (Holling,
1973; 1986), which is required of a system to return to a "stable" state. For sustainability, the
concept of ecosystem
resilience
becomes crucial for biodiversity
conservation.
Resilience
represents the ability of ecosystems to recover from or adjust easily to disturbance, and the speed
with which they return to an attractor state (Pimm, 1984), which can be deemed "stability."
Following the work of Holling (1973; 1986), resilience can be used to identify the existence of
functions within systems that, at any given moment, are offset from anyone
of a number of
locally stable attractor states. Resilience in this sense is a measure of the perturbation that can be
absorbed before an ecosystem in the domain of one attactor state is dislodged into that of another
attactor state (Folke et al., 1996). It is essentially the capacity of the system to buffer disturbance.
The essential condition for the resilience of a system in order to persist is determined by spatial
heterogeneity and the associated biotic diversity, based on Elton's (1958) original hypothesis that
ecological stability should depend on biological diversity. There have been many conceptual and
empirical advances, and debates (e.g., Woodwell and Smith, 1969; Pimm, 1984; Holling 1986;
Tilman, 1996; Tilman et al., 1996) on the importance of diversity within systems.
At the landscape
structure and variability.
scale, biotic processes,
interacting
with abiotic ones, can control
This is also the scale range where human land-use transformations
occur, so that the area where plant and animal controlling interactions unfold is the same area
where human activities and population interact with the landscape.
The landscape concept is
appropriate for sustainable planning because it is sufficiently large to contain a heterogeneous
matrix of LCLU elements that provide a context for mosaic stability (Forman 1990; 1995).
2.2
BiodiversityProtectionStrategies
Efforts to conserve biodiversity remain largely rooted in the concept of species, a most
ephemeral part of an ecosystem.
Species-based approaches address only a small part of biological
diversity because they ignore different levels of organization and the functional linkages among
these levels (Noss, 1983; Pimm, 1991; Maddock and du Plessis, 1999). Broadening our view of
biodiversity into one of ecosystem hierarchy and diversity highlights that the species diversity of
an ecological system is a systems-related attribute (Noss, 1990; Jizhong et al., 1991). A focus on
ecosystem diversity underscores the inherent value of the systems, apart from which the myriad of
species cannot survive.
To be sure, the most important considerations,
which are typically directly ignored, for
any of the conservation methodologies outlined in Chapter 1 are the role human societies, values
and economies playas threats and protectors of biodiversity.
Conservation based public agencies
and academic conservation biology tends to disassociate themselves from the human-side of the
analysis and only focus on their biological domain science. Humans' are a part of natural systems
and
by their
fragmentation)
evolutionary
nature
disturb
"natural"
habitat
(e.g.,
through
habitat
loss,
and altercate key resources (e.g., water, soil, climate), which in turn affects the
species, community
assemblages
and food webs in the hierarchy
(Figure 2.1).
A logical
framework for understanding the interactions of human threats has not been considered in species
or broad model approaches of conservation planning, although they are the dominant causes of
biodiversity loss (Ehrlich and Wilson, 1991).
The multi-dimensional
and multi-disciplinary
field of sustainability
encompasses
traditional academic disciplines of ecology, economics, sociology, developmental
the
studies, and
philosophy (Norgaard, 1988; van Jaarsveld, 1996). It strives to integrate social, economic and
environmental goals into a single manageable framework capable of directing regional and global
development towards a more just and equitable future (Munasinghe, 1993).
Several problems
sustainability
continue to hamper the scientific
and the integration of the environment,
communities
ability to address
society and economics.
The following
problems currently challenge sustainability and biodiversity conservation:
•
The fallacy of "natural" nature. There is little point in regretting the history that
has made humans or exotic species part of the ecosystem they now inhabit (e.g.,
Cronon,2000).
Anthropogenic
altercation
Affect these in
part and whole
by a cascade
effect through
the hierarchy
Figure 2.1: Key resources appropriated through human action and the biota that are affected
through a hierarchical cascade.
•
The role of isolation,
development
i.e., one discipline
and conservation.
Problems
or segmented
disciplines
tend to be isolated,
driving
rather than
acknowledging their true connective nature.
•
The drive toward one correct analytical framework, when there are more than one
way of looking at solutions.
A continuation
of previous ideas and opening new ways of thinking and viewing our
current crisis with biodiversity
loss needs to be explored.
The following proposed profile of
reconstruction should continue the debate:
•
A co-evolutionary
understanding
of development
and the biodiversity
crisis in
especially developing countries.
•
The continued acknowledgement
and support of the role of resilience with the
adoption of adaptive environmental
management
follows on the realization that most ecosystems
(e.g., Holling,
1996).
This
are in various levels of dis-
equilibirium and that policy must remain flexible and evolutionary.
•
The need to develop models of macro/micro scale interaction in order to build
realistic conservation impact scenarios for planning and policy assessment (e.g.,
Dale et a1., 1994).
•
The continued
acknowledgement
of spatial variation
and scale as important
factors in understanding environmental systems (e.g., Wiens, 1989b)
One of the principle aims of a co-evolutionary
ecosystem interaction
critical, evaluative
dynamics model is to establish human-
within an interpretative/interrogative
methodology,
framework.
which stresses a multiple interpretative
consistent with the need for a multiple modelling strategy.
impossibility
We must develop a
of any single model
adequately
framework
and is
This is an acknowledgement
of the
encompassing
the diversity
of social
and
environmental phenomena, which comprise co-evolutionary systems.
Essentially all model characterizations
of human-ecosystem
processes are of necessity
both incomplete and proximate; thus we need a variety of model scenarios not only at different
temporal and spatial scales, but also at different levels of social and natural aggregation.
a research
observational
framework,
not
only
sets, but moreover,
capable
of encompassing
one in which empirical
qualitative
and
We need
quantitative
data can be situated within an
interpretive frame of reference. By proposing a co-evolutionary landscape ecology framework we
are looking for systematic ways of linking disparate bodies of knowledge, currently resident in
discrete
academic
boxes.
The conceptual
structure must be able to facilitate
and allow
interrogative dialogue between qualitative and quantitative data sets. To truly understand what is
needed for sustainable biodiversity conservation, a scheme should encompass three distinct areas
of knowledge acquisition interlinked and focussed on supplying information and knowledge for
shared learning (Figure 2.2).
This framework avoids reductionist
methods, which stress the
importance of arriving at a single unambiguous model as the basis of prediction. In contrast the
goal is directed at the representation of human-ecological
understanding
systems with a view to a more complete
of biodiversity threat and uniquely tailored regional action.
Therefore, we need
more, rather than fewer representations so as to create a more effective dialog.
Monitoring indicators should be able to provide information on long-term LCLU patterns
of the locale, along with attendant
management
social and political
strategies were implemented.
constraints
within which resource
The relationship between power structures and the
land provides new information on the rates of resource exploitation, human demography, and the
differential resilience of specific landscape units to support biodiversity.
The nature of this
framework should be able to generate a series of scenarios arriving, not at any single predictive
0'=00
Description
\~
~
?
\
J
Soils
!
Shared Learning
Pedology
Demography
~
Mutual Analytical
Discussion
\
Economics
History
Interrogative
Tools
model of unsustainability,
but at a series of potential
evolutionary
pathways
to which the
landscape or region is prone. In a sense, it is a mapping of the 'possibility spaces' within which
human settlement and ecosystem functioning can persist, and within which is nested a probability
space of human action.
Limited resources
between
land-use
for the protection
developments
of the environment
in developing
countries
call
and the rising competition
for appropriate
conceptual
frameworks and relevant methods for facilitating and analysing trade-offs and compromises.
The
methodology
in a
must encompass
economics,
socio-cultural
and environmental
attributes
disaggregated fashion to understand a regions overall vulnerability with respect to sustainability.
Van Jaarsveld (1996) offered one way of looking at the highly disaggregated
data that are
typically acquired from each of these sectors. It was suggested that the development of highly
aggregated indexes may have significant political advantages in communicating
with the public
and policy-makers, but they do not provide an ideal or adequate framework within which political
action should be prioritized. Instead a framework should be developed that would require an
evaluation of disaggregated data, and leaving the researcher with the problem of dealing with, and
interpreting,
complex environmental,
simplistic 'cause-effect'
understanding
social and economic data matrixes in the absence of a
of interactions between these variables or their social
values. In biodiversity conservation planning in developing nations, there is a need for analyses
that are able to answer questions of viability or security of conservation practices in the face of
anthropogenic
land-use changes driven by global economic policy.
environmental
and socio-economic
An integration of complex
indicators should provide the information needed to answer
questions pertinent for sustainable biodiversity conservation.
There need be little doubt that the landscapes
heterogeneous,
'wild' and humanized, fine-grained and coarse-grained),
which cultures
instruments.
we have today (homogeneous
interact
with nature, have been strongly influenced
and
and hence the ways in
by historic
economic
Increasingly aspects of social organization as well as the paths of knowledge and
technology advance affect the pathway landscapes assume (i.e., pattern and process).
Norgaard
(1988; 1994) presents this aspect of viewing these interactions between economics and other
factors by borrowing from evolutionary, and in particular from co-evolutionary,
change, portraying
organizational,
development
technological,
as a process
and environmental
of co-evolution
between
systems (Figure 2.3).
explanations of
knowledge,
values,
In Norgaard's
(1994a;
1994b) portrayal, each of these systems is related to each of the others, yet each is also changing
and affecting change in the others. Deliberate innovations, chance discoveries, random changes,
and chance introductions from other societies occur in each system which affect the fitness and
hence the distribution and qualities of components in each of the other systems. With each system
putting selective pressure on each of the others, they co-evolve in a manner whereby each reflects
the other. This type of thinking is consistent with landscape ecological theories which incorporate
the interaction
of humans and species immigrations,
emigrations,
pattern and process leading to state changes (Forman,
1995).
and populations
Co-evolution
everything is tightly locked together, yet everything is also changing.
effecting
explains how
This approach could be
used as a conceptual underpinning to understand current and future biodiversity threats and to
assess the sustainability of protected areas.
/
Values
~
Knowledge
\
Environment
Organization
..-.
Technology
/
Discoveries in the natural and physical sciences have demonstrated the evolutionary
pathways traced by non-linear systems, and their convergence towards a variety of stable, quasistable and unstable states. These trajectories are reached through a sequence of bifurcations,
during which the system undergoes qualitative change (Laszlo, 1987). A fundamental aspect of
human-environment relationships is the opposition or tension between temporal rhythms, which
are embedded in natural processes and those resident in societal structures; the asymmetries
between them provide the context for abrupt discontinuous transition through bifurcation (Figure
2.4). In this presented case the bifurcation is symmetric and represents the pathways of spatial
landscape change. Since humans are an intimate part of landscapes the process of landscape
pattern evolution begins with habitat perforation or dissection leading to fragmentation, shrinkage
of fragments and finally a lengthy process of attrition of the remaining fragments (Forman, 1995).
The land-use types that replace the natural habitat add to the diversity of the landscape till at some
point the attrition of the natural remnants is so great that the homogenizing forces of human
development at some defined analytical scale renders the landscapes simple again.
Human-environment systems are a prime example of the operation of non-linear
dynamical processes. They are governed by interlinked sets of non-linear processes, which resist
obvious disaggregation into systemic subsets- something which conventional reductionist
methodologies force upon them. An important property of such complex systems is the role
played by feedback mechanisms, which amplify or reinforce human physical and social processes.
For example, the development of economic core areas and a poor periphery appears to be the
process of cumulative and circular causation.
~r
~i
,,---(.:":,'.','
"0
"II
.•..~
.--.'::_.-
'"
'"
~
:---- ---(.~::.'.-.-,-.',-.'--.--'.'-'.',---.':::~--", "-
:e
'"
'"oc,;
••••
-"'_._(:----_._
1
..-.=--,._.---- .._.:~------_:
'-,
:1
... '
=1
~
u_ns_la_bl_e_Sla_l_eS
~!
.-.
Time
Simpler,closer
to equilibrium ~t --~l
More complex,
furtherfrom equilibrium~t
---~l
Simpler,further
from equilibrium
The existence of external economIes, economIcs of scale, and agglomeration
in core areas,
compounded by the provision of transportation networks, serve to enhance and capitalise upon
existing advantages of relative locations.
In the language of dynamical systems, two elementary concepts are important: the notion
of phase space and the other is concerned with the concept of attractors or basins of attraction (for
more detail see Waldrop,
representation
1994).
First, phase space can be thought
of as a geometric
of the universe of possibilities possessed by a system- in a sense, the allowable
territory within which it operates- an arena in which a phase portrait of the evolutionary history of
the system can be constructed.
Second, is their long-run behavior, which is manifest by a
particular attractor; a region of the phase space to which all points ultimately converge.
It is
effectively the 'signature' of the system.
With respect to human-modified
landscapes under discussion here, what we are faced
with is an empirical situation in which a number of different attractors are co-present.
In a sense
they inhabit an operational space constrained by non-linear causality on account of the multiple
periodicities represented by the wide variety of temporal rhythms, which define natural ecological
phenomena and their constant modification by human social groups who themselves are defined
by alternative periodicities (i.e., economics). Research in a number of fields has shown that nonlinear feedbacks can amplify these rhythms causing either catastrophic collapse (Holling, 1986) or
the emergence
qualitative state.
of spontaneous
structure (Allen, 1993), with the system evolving to a new
2.4.3
Landscape Socio-ecodynamics
To elaborate the process within the context of landscapes, imagine that the systems of
Figure 2.3 - values, knowledge, social organization, and technology - are made up of different
ways of valuing, knowing, organizing, and doing things. Similarly the landscape (environmental)
system consists of different types of species and other particular ecological factors which it starts
with before human contact (Figure 2.5). From a starting pre-human landscape geography (To) a
perturbation
(unstable
state) occurs whereby a particular
human social organization
arrives
randomly, allowing that landscape to co-evolve (T)) to a new characteristic look (pattern, process,
use) with relative
morphology,
stability.
Sauer (1925) originally referred to this process
where a landscape
prediction difficult.
environment
could take a multitude
as landscape
of pathways
making
In effect changes within anyone of the components from Figure 2.3 acts to
evolve the landscape development process conceptualized in Figure 2.4 and simplified in Figure
2.5. The process of experiments, discoveries, chance mutations, and introductions within each of
the systems (Figure 2.3) drives co-evolution across all of the systems simultaneously
creating bifurcation on the landscapes.
and thus
The landscape bifurcations described in Figures 2.3, 2.4,
and 2.5 helps us to understand how policy overriding economic, social and environmental systems
(Figure 2.6) can cause critical instabilities (bifurcation) and thus instigate a new co-evolutionary
pathway within a landscape. Policy is a fundamental determinant of the way natural resources are
exploited and/or conserved and how human systems are organized.
development
policies of the Ex-Apartheid
In South Africa the separate
State created spatial separation
pathways for local indigenous African versus colonial Europeans.
and development
The landscape character of the
created African tribal homeland system versus the Western industrial development
White South Africa are still clearly evident today (Fairbanks et al., 2000).
implications
reproduction
of this can be the asynchronous
and consumption
biodiversity conservation.
particular
patterns
which
rhythms between
challenges
model of
The commonly held
the natural world, societal
the sustainability
of landscape
Munasinghe and Cruz (1995) note that linking specific causes with
effects is especially difficult where many conditions are changing simultaneously.
However, it is usually possible to identify a small number of linkages affecting high priority
environmental concerns.
Through the process of co-evolution, the world's landscapes can be thought of as having
become a patchwork quilt along a gradient of loosely to strongly interconnected,
co-evolving
social and ecological systems. Within each landscape the ecological system evolved in response
to cultural pressures and tended to reflect the values, worldview, and social organization of local
peoples.
At the same time, the cultural system in each landscape evolved within the constraints
imposed by the ecosystem and hence tended to mirror the fertility, species composition, stability,
and management options presented by the ecosystem.
/
I/
'\ \ ~6'
\
\
/'~
/~
c- ---*---;..
\
/
\
T 2 co-evolution
~
T I co-evolution
~
T 2 co-evolution
T 2 co-evolution
!
/
--\~---!-----~/
\
\
/~-~
Starting Geography
T. landscape evolution
/
/
/
~
T 2 co-evolution
~
T 1 co-evolution
T 2 co-evoluti on
Policy
(politics)
Figure 2.6: The interaction among the major sectors affecting sustainable development.
political action oversees and drives decisions and actions taken in the other sectors.
Policy-
Therefore, each landscape takes on unique characteristics particular to the non-random biological
and cultural structuring occurring within the landscape mosaic. This reading of a landscape can
provide us with a valuable framework for examining developing countries. Developing nations
area able to illustrate co-evolution more clearly than technologically advanced nations (Norgaard,
1994). Information about the path a landscape or region has co-evolved along may let us interpret
future actions. Future landscape paths could be assessed in terms of objectives such as sustaining
biodiversity both in the unprotected landscape matrix and formally protected areas.
The real world is represented
altered environments.
by finely fragmented habitats interwoven with human-
This is especially true in southern African landscapes where nomadism,
colonialism and the ever rising human population have affected every part of the region in some
way, shape or form (Puzo, 1978). The vast majority of biotic inhabitants that occur across the
remnant pockets of the original environment are important.
The species that can survive in the
habitat fragments come together to determine the integrity and resilience of ecosystems to a range
of environmental perturbations.
In essence biodiversity loss has a direct impact on the ability of
interdependent ecological-economic
sustainability.
fragmented
Wessels
In South Africa, for example, the vast majority of the natural landscape is
by land-uses
conservation
systems to maintain functionality, and thus in a policy sense,
planners
et al., 2000).
(Fairbanks
et al., 2000), which presents
using species-based
African
reserve selection
conservation
analysis
logistical
problems
for
designs (Lombard et al., 1997;
frequently
still maintains
expectation that large pristine tracts of 'natural' habitat to support irreplaceable
the false
species will be
available in seas of poverty (Western, 1989; Adams and McShane, 1996).
Landscape ecology has been broadly defined as the study of the effect of landscape
pattern on ecological processes (Turner, 1989). In a clearer sense landscape ecology is the study
of how landscape structure affects (the processes that determine) the abundance and distribution
of organisms. The object of landscape ecology is not to describe landscapes, but to explain and
understand the processes that occur within them.
Certainly, the most challenging aspect is to
extend this discipline to the analysis of pattern in a socio-economic context, given the need to find
more sustainable forms of landscape management.
The application of the principles of landscape ecology in the formulation and solving of
problems is of interest here. Human influences in landscapes tend to eliminate gradual changes
and to produce abrupt boundaries, however the diversity of human cultural groups and their
subsequent
economic development
environmental
levels based on a combination
of policy, historical,
factors effect landscapes and biotic diversity in a variant of ways.
and
Landscape
metrics employed to quantitatively measure the spatial patterns of boundaries and patches within a
natural landscape (Turner, 1989) could be linked to socio-economic and cultural systems to assess
the health of ecological systems (O'Neill, 1999; O'Neill et al., 1999) for biodiversity conservation.
The ecological structure, function, and potential change of landscape mosaics need to be
understood within the socio-cultural and economic structures of a region to adequately address
sustainable conservation action.
The spatial arrangement of local ecosystem level components
and land-uses within a landscape within a region will have an affect on the areas ecological
integrity.
To understand an area's conservation potential one must understand an areas current
and future landscape function, but within human economic and social systems. Thus, landscapes
should be perceived as the tangible matrix of the total human ecosystem (sensu Naveh, 1997), and
therefore as concrete systems in their own right and not just as ecosystems on km-wide stretches.
This argument to consider the evolution of human-ecosystem
complex,
co-evolutionary
context governed by metastable
interactions
within a
states, means that the resulting
ecodynamics of pattern and process can be viewed from a hierarchical perspective (O'Neill et al.,
1986). A key concept is that ecosystem processes operate over a wide spectrum of rates, and
these can be assembled into discrete classes.
The structure imposed by these differential rates
allows the system to be decomposed into organizational levels, with each level being segregated
on the basis of response times (i.e., higher levels associated with slower rates, and lower ones by
more rapid rates).
Within such a scheme, ecosystem structure is viewed as a series of weakly
coupled sets within a hierarchy of process rates involving biotic interaction and abiotic factors.
The non-linear couplings in these processes are further complicated by human action, whether as
the result of un-coordinated
stochastic events or by a series of policy-directed
interventions
(Giampietro, 1994).
Scale is critical, for as spatially heterogeneous areas, landscapes may exhibit stability at
one spatial scale, but not at another. Thus, the scale at which observations are made profoundly
influences the research and analytical interpretation process (Turner, 1989; Wiens, 1989b). In this
case a variety of local and regional studies would be ultimately required to confidently provide
conservation planning and management strategies.
Analysis oflandscape
pattern makes use of measurements of the connectedness, diversity,
shape complexity, and size of land-cover patches to study ecological condition at local to regional
scales (Turner and Gardner, 1991). These metrics (O'Neill et al., 1988; Ritters et al., 1995) have
been used to assess landscape condition (Krummel et al., 1987; Wickham et al., 1999), infer
ecological
process
configuration
from pattern
(Milne,
1992; Fahrig,
1997), and show how landscape
can impose constraints on biological populations
1996; Flather and Sauer, 1996).
From a regional perspective,
(Pearson et al., 1993; Flather,
land-cover
patterns may be
considered as either forcing or constraint functions for sub-regional dynamics, or as integral parts
of strictly regional models (Allen and Starr, 1982). Information about land-cover patterns has
proven useful for both local and regional assessments of ecological condition (Vos and Opdam,
1993).
Landscape metrics are a set of tools that can be used to measure pattern, which can be
correlated to ecological processes, biodiversity persistence, and define 'spatial signatures' which
describe the co-evolutionary response oflandscapes
Therefore,
by using a monitoring
(O'Neill et aI., 1996; Wickham et aI., 1996).
framework to quantify spatial patterns and their changes
(O'Neill et aI., 1999) we can quantify their effect on ecological processes and then combine these
indicators with biodiversity
elements, socio-economic,
and cultural information
to provide a
integrated conservation solution.
Physical location, transportation
profitability of an economic activity.
costs, social climate and policy often determine the
In turn, economic activity is the primary determinant of
landscape pattern and change, and therefore the resiliency of ecosystem function. Co-evolution of
human-ecosystem
'signatures.'
dynamics develop positive feedback loops which enforces landscape pattern
This allows remotely sensed imagery, GIS, and landscape ecological metrics to be
combined into a powerful approach for interrogation and interpretation of the pattern, which can
then be back related to social, economic, and environmental
indicators.
For example, let us
assume that we wish to evaluate the status of the landscape pattern for several defined coevolutionary landscape regions.
We could ask how far the present landscapes deviates from an
ideal landscape for sustaining all hierarchical levels of species diversity with complete habitat
cover (high dominance) in large (un-fragmented) and complex patches.
We might also ask how
far the landscape deviates from a total state of ecosystem decay with many human land-use and
natural land-cover types (low dominance), in dissected (fragmented) and simple patches.
In
statistical
parlance,
abundance/distribution/local
landscape structure.
the 'responses'
economic,
the
'response'
variables
In
landscape
ecology
are
process variables, and the 'predictors' are variables that describe
However, in order to understand present co-evolution from past interactions
are variables
socio-cultural,
that describe landscape
and environmental
promising analytical approach to understanding
functioning (Whittaker,
structure,
indicators.
and the 'predictors'
Gradient
analysis
are the
may provide
a
the effects of multiple stressors on ecosystem
1967; McDonnell and Pickett, 1993) by integrating the complexity of
multiple stress effects across the landscape (McDonnell et aI., 1995).
The gradient approach
relies on the assumption that graduated spatial environmental patterns govern the structure and
functioning of ecological systems.
Changes in population, community, or ecosystem variables
along the gradient can then be related to the corresponding spatial variation in the environmental
and socio-economic variables, with specific statistical techniques dependent upon whether or not
environmental variation is ordered sequentially in time or space, and whether single or multiple
variables are being monitored.
In the case of system responses to multiple stressors, complex,
nonlinear gradients are apt to be present and ordination techniques may provide insight into the
Totally
Dysfunctional
"Leaky"
Fully
Functional
"Conserving"
Landscape Ecological
Gradient
•
•••
a.
[Resource capture]
Low dominance
High fragmentation
Simple patches
High class richness
High dominance
Large patches
Complex patches
High contagion
[Resource status]
[Habitat status]
IMedium
Dysfunctional
Landscape
I High
y
•
Functional
Landscape
Figure 2.7: Landscape functionality as: (a) a continuum from functional to dysfunctional, and in
relation to (b) resistance and resilience to disturbances (modified from Ludwig, 1999).
biotic responses
to these gradients
(ter Braak and Prentice,
1988; Jongman
et al., 1995).
Therefore, how well a landscape functions to conserve resources and maintain biodiversity could
be viewed as a continuum (Figure 2.7a).
Ludwig (1999) proposed a conceptual model that
deemed landscapes as "fully functional" when they conserve resources to maintain rich and
diverse environments that provide many habitats suitable for a high species richness. At the other
end of the spectrum, a landscape may be very dysfunctional where all resources 'leak' from the
system, resulting in a landscape with poor resources and no habitats suitable for species.
concept of stability (resistance and resilience) can then be applied to how disturbances
landscape functionality.
The
affect
Resistance refers to the ability of the system to remain unchanged when
disturbed, while resilience refers to the ability of the system to rapidly return to an assumed
equilibrium
state.
Using these definitions (Figure 2.7b), a landscape has low resistance if a
disturbance causes a highly functional system to become dysfunctional.
A landscape with high
resistance will only slightly shift down the continuum under the impact of the same disturbance.
Highly resilient landscapes will rapidly recover, for example, in a matter of months or a few
years, to a displacement
down the continuum caused by a disturbance.
Landscapes with low
resilience may take centuries to recover from this same disturbance. This conceptual model could
show promise in being able to assess various landscape environments and the drivers that have
'pushed' them into different 'states' or pathways (McIntyre and Hobbs, 1999).
Following on the argument developed so far, this chapter proposes an enhancement of the
theoretical framework for biodiversity conservation planning by integrating both anthropogenic
and ecosystem integrity goals into a decision framework guided by co-evolutionary
landscape ecology methods.
theory and
The basic principle encompasses a larger approach to biodiversity
protection, by protecting levels of biodiversity linked by process and spatial organization.
This is
the underlying concept for integrated approaches to the management ofland resources (e.g., Noss,
1990). The implementation oflandscape
level plans in routine environmental policy and planning
is complex, but if we understand that environmental
acknowledges
pluralistic
change is a co-evolutionary
systems then suitable frameworks
can be developed
process that
for protecting
biodiversity based on each regions particular issues rather than on a general model.
In studies of the causes and consequences of tropical deforestation in Rondonia, Brazil,
Southworth et al. (1991) and Dale et al. (1993; 1994) indirectly developed a co-evolutionary
model. The authors acknowledged land-use change as one of the major factors affecting global
environmental conditions and that to address the problem, spatially combined explicit ecological
information and socio-economic
countries.
factors.
This aspect is particularly needed within developing
In Figure 2.8, a framework is presented for developing a methodology that integrates
the idea of co-evolution
biodiversity
profile
by addressing the state of human social and economic welfare, the
and the landscape
ecological
attributes
of a defined region.
Ethical
stewardship of the environment requires that society monitor and assess environmental change at
the national scale with a view toward the conservation and wise management at the local scale
(O'Neill et aI., 1997; O'Neill et aI., 1999). Most social and economic indicators are measured at
regional levels, while some of the most important environmental and social changes occur at a
landscape scale (e.g., Forman, 1995). The landscape scale is important because political decisions
to manage natural resources are made at broad scales, such as catchments.
to change land cover may be made by individual
landowners,
cumulatively, as a change in spatial pattern on the landscape.
Decisions about how
but their impacts are seen
These decisions are usually also a
reflection of global, national and regional policy, economic or social situations that draw attention
to a hierarchical reading of these co-evolutionary
reporting
zones (e.g., political
environment-
reflecting
districts,
systems. Resulting data from 'representative'
catchments,
etc.) of economic,
the true state of society and nature-
hierarchically (Figure 2.8) within some defined multidimensional
For example, in sub-Saharan
socio-cultural
are recorded
and
and analyzed
data reduction method.
Africa women and children invest enormous energy in
obtaining domestic energy from fuelwood and herding cattle. Dasgupta (1993) has described the
complexities of interactions among population growth, poverty, and environmental deterioration.
Men are typically part of a migratory labor system whereby they leave the rural tribal areas for
temporary work on the mines and in industry.
Monies are sent back to their wives for food and
cattle purchase, which is equated as wealth accumulation (Hall, 1987). As the human and cattle
population grows in these areas grazing range is placed under greater pressure leading to land
degradation.
An examination of KwaZulu-Natal,
South Africa illustrates this pattern noted by
Dasgupta (1993) and Ehrlich et aI., (1995) whereas areas of low male to female population ratios
in developing countries have a higher percentage of degraded land (Figure 2.9).
In KwaZulu-
Natal this can be depicted as a systems model (Figure 2.10), which has been documented
historically
(Cole, 1960) and linked anthropologically
(Hall, 1987).
Therefore,
along with
economic geography models, culture should also be assigned a central role in any theory
Co-evolving Future
- Aspirations and expectations
- Globalization economics
- Landscape transformation and restoration
- Conservation, exploitations, extinctions
a
Regional Level
- Gross geographic product
- Work force per sector
- Service needs
a
- Landscape metrics
- Ecosystem functioning
- Biological diversity
Co-evolved Past
-
Cultural identity and experience
Local economic development
Landscape transformation
Conservation, exploitations, extinctions
Figure 2.8: An overview of the hierarchical
evolutionary dynamics.
indicator reading framework
for analysing co-
5
---
4
~
~
I.
2
~
Q
....
•..==
"'0
bJ)
3
1
Cl.l
"'0
"'0
=
....:l
=
0
~
....:l
bJ)
-1
-2
-3
0.7
0.8
0.9
1.0
1.1
1.2
MalelFemale Population Ratio (1996)
Figure 2.9: Linear regression relationship of male/female population ratio to percentage degraded
land per magisterial district in KwaZulu-Natal (N=52). Human population data from 1996 census
and land degradation assessment from the South African National Land-cover Database
(Fairbanks et aI., 2000).
1. Expansion
2. Contraction
1. MaleIFemale ratio
2. Birth/Death rates
o
O
1. Cattle grazing
2. Subsistence Agriculture
3. Commercial Agriculture
1. Degradation
2. Rehabilitation
Figure 2.10: Economic core and poor rural periphery systems model of landscape development
within rural African communities in South Africa.
purporting
to characterize
the process
of land-use
intensification
and landscape
pattern
development among rural African communities.
2.7
Co-evolutionary Implications for Sustainable Biodiversity Conservation
The primary implication of the foregoing discussion is the need for a conceptual retooling
and use of multiple analytical methods if we are to come to a more complete understanding of the
future development of human-ecosystem
framework
fundamentally
changes
our ideas with respect
interactions and long-term predictability.
an understanding
interactions on biodiversity conservation.
to evolution,
The proposed
human-ecosystem
This argument also draws attention to how much deeper
can be made of developing nations through transdisciplinary
research and
shared learning.
The implications of the role of non-linear phenomena in generating long-term dynamics is
a pre-requisite
for understanding
the evolutionary processes which structures landscapes and
subsequent biodiversity loss. This type of thinking effectively renders evolutionary models which
support linear, simple trajectories redundant (see Turner et al., 1996; Wear et al., 1996), stressing
the fact that human-ecosystem
usefully be conceptualized
dynamics within landscapes seen as long-term history can more
as a series of transformations
time. Essentially, if human-ecological
of structuring and restructuring
systems are prone to complex co-evolutionary
over
pathways
and the kind of 'structured disorder' associated with chaos, then this has significant consequences
for biodiversity conservation and general land management policy decisions (e.g., Holling, 1986).
In developing
countries
simple systematic
conservation
planning
may fall short in
informing conservation planners and policy makers as to the future persistence of ecosystems,
landscapes, and species populations.
A new model of co-evolutionary
social, cultural, economic, as well as environmental
landscapes incorporating
indicators (species, habitats, landscapes,
landscape metrics) is needed to understand and develop conservation management plans which
incorporate the goal of persistence (sustainable biodiversity).
Although preserving biodiversity
through formal protected areas is an important short-term step, it will not be sufficient to solve the
problem of biodiversity loss (Western, 1989; Shafer, 1994). Reserves are embedded within the
larger environment,
and most reserves alone cannot deal with ecological attributes that cover
larger scales (e.g., broad climate, global climate change). Thus, conservation efforts should firstly
be planned at the scale of the regional landscape to assess the available landscape matrix of
'natural' fragments.
Small reserves will lose their distinctive species if they are surrounded by a
hostile landscape (Askins, 1995; Baillie et al., 2000).
driven by interconnected
biodiversity conservation.
Reserves, as islands in a sea of change
economic and social systems, may not be a basis for sustainable
Moving toward a model of a co-evolving patchwork quilt of discursive communities
conceptually presents social systems as systems against a responsive environmental backdrop.
These landscapes will change over time through mergers and divisions as the social and
environmental systems co-evolve. The strategy is to use the available social, economic and
environmental data in an analytical framework that helps promote sustainable landscape and
regional social, economic and environmental systems.
Conservation International (1998) revealed that just seventeen nations collectively claim
more than two-thirds of all known species worldwide, making conservation efforts in these
'megadiversity' countries essential for the survival of Earth's natural heritage. Not surprisingly,
fifteen of the countries singled out are considered developing. These countries are also home to a
major portion of the planet's cultural diversity, perhaps even a larger percentage than for
biodiversity. Positive human welfare is directly related to sustainability of the environment and is
the critical link in the chain towards a comprehensive conservation (persistence) goal. It should
be apparent, that human welfare has to be met at the same time as biodiversity conservation, for
they are not mutually exclusive.
Landscape ecology has made a significant contribution to conservation biology (Noss,
1983; Noss, 1990; Hansson and Angelstam,
1991; Forman,
1995).
However, much of the
landscape ecological research that investigates biological conservation problems has not occurred
within appropriately
defined landscapes, rather relying on arbitrary ecoregion delimitations (as
discussed Host et al., 1996; Wright et al., 1998).
For planning purposes,
a representative
landscape approach to conservation could potentially be used as a spatial surrogate to ensure the
long-term maintenance
of biodiversity.
The maintenance
of processes that sustain ecosystem
structure and functioning is essential for achieving persistence goals for systems of conservation
areas (Baker, 1992; Noss, 1996).
If a landscape approach to conservation
effective, the landscape units need to be properly defined.
biology is to be
At present, the only ecologically
defined system that exists within South Africa is for the Kruger National Park (Gertenbach,
1983). This is understandable
considering
the relatively
recent international
emergence
of
landscape ecology as a discipline (Wiens, 1992), the importance placed on species systematics
and inventorying in southern Africa (Huntley, 1989), and the emphasis placed on poorly sampled
species databases for reserve selection (e.g., Rebelo and Siegfried, 1990; Lombard, 1995; Freitag
and van Jaarsveld, 1997). The first step in developing a successful landscape level conservation
plan is identifying and locating the landscapes of a region.
The
classification
goals
and
objectives
of environmental
of regions based on measurable
management
environmental
frequently
characteristics.
require
the
Delineation
of
ecological landscapes is useful in a variety of contexts, for example, in the assessment of the
regional representation
Franklin,
of conservation
areas (Margules et al., 1988; Bedward et al., 1992;
1993; Pressey et al., 1994), defining zones for sustainable
ecological management
(Forman, 1995), and as a framework for assessing the diversity of species and processes within
landscapes (Lapin and Barnes, 1995).
An ecological
framework
that can integrate
multiple
environmental
characteristics
diminishes problems of duplication among government land resource agencies, and it can assist
in the exchange of information and research results. Towards this end, the utility of ecoregional
classifications,
developed for the conterminous
United States (Omernik,
1987; Gallant et al.,
1995; Omernik, 1995) and Canada (Wiken, 1986), have been successfully demonstrated
(e.g.,
U.S. Environmental Protection Agency: Environmental Monitoring and Assessment Program).
There are two broad approaches
classification
to classifying
landscapes:
human landscape-based
approaches mainly applied in European countries (Blankson
and Green, 1991;
Green et al., 1996), and biophysical approaches (Christian and Stewart, 1953; De Agar et al.,
1995; Bailey, 1996; Bernert et al., 1997) which combine climate, soils, vegetation and landform
into observable
assessments
and definable land units (e.g., Omernik,
1987).
Methods vary from visual
using elements like scenery, to quantitative procedures,
which group areas with
similar values for a set of mapped variables (Benefield and Bunce, 1982; Blankson and Green,
1991; Host et al., 1996; Bernert et al., 1997). These methods are not completely objective, as
variables for consideration have to be chosen, but are less judgmental than visual methods.
We used the biophysical approach, because the aim was to identify natural landscapes
and then assess their conservation
amount of human-induced
status by examining both the degree of protection and the
transformation
that has occurred.
classification system for the province ofKwaZulu-Natal
This study presents a landscape
(South Africa) by using biophysical data
and a combination of principal component analysis, clustering and spatial overlay techniques.
preliminary
analysis
is also undertaken
to illustrate
the important
A
role that this kind of
information can and should play in identifying conservation worthy areas.
The variables used were those commonly used in the description of ecological regions
(Omernik, 1987; Omernik, 1995; Bailey, 1996). The set of variables was broad, and included
those describing
the physical
(topography,
landform,
geology
and climate)
and biological
environments (vegetation) and was integrated into a geographic information system (GIS). Only
the topography,
landform and climate variables were used in the classification
geologic and vegetation
proposed by Omernik,
maps were not used directly in the demarcation
1987; Bailey, 1996).
analysis, the
of landscapes
(as
Rather, they are used to derive a typology of
attributes within the landscapes that allows the landscapes to be described according to the
vegetation types and geological substrates found in each unit.
This adds considerably to the
conservation planning objective by not subjectively combining the unit boundaries of vegetation
and geology with landscapes to create arbitrary units (Host et al., 1996) and thus mask the
landscape heterogeneity into a coarser ecoregional unit (Wright et al., 1998).
A systematic approach was developed for delineating landscapes (Figure 3.1) within the
KwaZulu-Natal province that could be applied to any geographical region. To prevent landscapes
Primary Database
Development
- elevation
- climate surfaces
- vegetation
-geology
_
,
I
--~~,
---
Sampling
",
- stratified random
- 25 random points
1:50 000 map sheet
Multivariate
Analysis
- correlation analysis
-peA
- clustering
- boundary cleaning
Visual Analysis
Classification
Development
- two tiered landscape
- verbal landscape
definitions described
- spatial visualization of
principal component factor
on each axis.
Overlay Operation
- classified
- classified
- classified
- majority
elevation
topographic
growth days
class
Validation
Assessment
- National land-cover
accuracy assessment field
- error matrix
occurring along the KwaZulu-Natal border from being defined by arbitrary political boundaries,
the study area was extended across the borders using catchment boundaries (DW AF, 1996). This
overlap will also allow for easier edge-matching of future landscape classifications developed by
neighboring provinces.
The analysis was raster grid cell based. The analysis cell size was partly determined by
the largest cell size of the already rasterised data sets and a logical cell size for future integrative
work, in this instance 1 lan2•
projection
for analysis.
All data sets were converted to Lamberts Azimuthal Equal-Area
To reduce the amount of data to be analysed, a stratified random
sampling of data sets was conducted. The 165 South African Surveyor General 1:50 000 map
sheets covering KwaZulu-Natal
were used to stratify a random sample selection, with 25 cells
being chosen from each sheet (i.e., a total of 4675 samples).
Pearson correlation coefficients were used to examine multicolinearity
and thus minimise
the duplication of variable information, and make decisions with regard to variables being
recorded in the field.
Principal component analysis (PCA) was performed
variables, which allows the important descriptors
to be standardized
on the resulting
against each other for
interpretation into spatial objects (see Legendre and Legendre, 1998).
Pattern and cluster analysis was undertaken on the PCA results in ArcView GIS (ESRI,
1998) using bivariate map plots of the axes factor scores produced by the PCA analyses and then
applying a natural breaks clustering classification technique.
This method identifies breakpoints
by looking for groupings and patterns inherent in the data using Jenk's optimization,
which
minimizes the variation within each class (Jenks, 1963). Using these techniques the data sets
responsible for the greatest amount of variation, as identified by the PC A, were classified.
The
classified data sets were then subjected to class boundary cleaning by smoothing transitions
between classes. This procedure removes class border roughness which is caused by inaccuracies
in the coarse resolution data (ESRI, 1998).
Landscapes were constructed by combining the classified terrain and climatic data sets in
a stepwise manner using Arc/Info GRID GIS (ESRI, 1998), and smoothing the intermediate
derived data sets with a 3x3 grid cell neighborhood majority class filter.
This transformation
reassigned pixel values based on the most prevalent class membership within a 3x3 grid cell
moving window.
Scarpace
et al. (1981) found that majority
filtering
classification accuracy by reducing 'random' noise in classification results.
actually
increased
When applying this
method over large regions the errors average out, so the landscape estimates are probably quite
accurate even if the cell by cell estimates may be less accurate.
A validation
exercise was performed using the South African National Land-Cover
Database accuracy assessment points (Fairbanks and Thompson, 1996). The overall accuracy of
the landscape classification map was tested using 530 stratified random field locations.
Actual
class membership for the sample locations was assigned on majority area coverage of a class
within a cell. A combination of using the extra attributes collected in the field (e.g., topography,
position, and vegetation) per point and inspection of the fixed ground photography of the area
around a point was used to determine actual landscape class membership.
This helped to ensure
that the derived landscape types were recognisable ecological units for conservation analysis and
planning.
A crucial consideration in maximizing the protection of biodiversity is the assignment of
priorities for protection in the face of real-world constraints (Pressey et al., 1996). The concepts
of irreplaceability (Pressey et al., 1994) and vulnerability (Pressey et al., 1996) were developed to
explicitly define conservation value and priority for representative
irreplaceability
is a measure
of the likelihood
areas.
In its simplest form,
that an area will be needed to achieve a
conservation goal; vulnerability is a measure of the imminence or likelihood of the biodiversity in
an area being lost to current or impending threatening processes.
Thus, irreplaceability
is a
measure of conservation value whereas conservation priority is the value of an area combined
with some assessment of the urgency with which it should be conserved (Pressey, 1997). Areas
of high irreplaceability
and high vulnerability
are highest priorities
for conservation
action
(Pressey et al., 1996). Focusing conservation resources on such areas will maximize the extent to
which representation goals will be achieved on the ground.
To demonstrate the value the landscapes add to the analysis of conservation goals, by
helping identification
of conservation worthy regions, we conducted an analysis of the derived
landscapes with the South African National Land-Cover database (Fairbanks and Thompson,
1996; Fairbanks et al., 2000) and a protected area database for KwaZulu-Natal.
database contains spatial information
on natural land-cover and identifiable
The land-cover
human land-use
mapped from Landsat TM imagery at 1:250 000 scale (Fairbanks et al., 2000).
The land-use
classes are essentially a measure of transformation status in the context of threats to biodiversity.
The protected area database described the boundaries of provincial reserves, digitized from 1:50
000 maps.
The land-cover data was used to assess the vulnerability
of the landscapes to future
human transformation based on the diversity of land-uses in each landscape.
The rationale being
that landscape types with several land uses are more vulnerable to future transformation
areas of single land uses because of their unique and favorable environment
than
(e.g. available
positive water balance and heat units) to a variety of human development potential (this will,
however,
depend
on the available
irreplaceability
was
determined
transformation,
representation
land cover classes being transformed).
using
a linear
in protected
weighted
combination
areas, and rarity (measured
The level of
of the extent
of
as the relative areal
contribution of each class):
The classification
of the measures was derived using the natural breaks classification
technique (Jenks, 1963). The vulnerability and irreplaceability
as calculated from classifications
Conservation Services (KZNNCS).
scores were scaled from 0-100%
and weights (Table 3.1) as defined by KwaZulu-Natal
Nature
Table 3.1: Landscape rarity, transformation, and protection
frequency classification with accompanying importance ratings.
% of Total
classification
rules based on
% Transformed
Weights
% Protected
Weights
>50%
1
< 10%
1
(Rarity)
< 1.7%
1.7 - 5%
0.75
34 - 50%
0.75
10 - 25%
0.66
5 - 7.6%
0.5
18 - 34%
0.50
>25%
0.33
> 7.6%
0.25
< 18%
0.25
Median minimum rainfall for driest and wettest quarters, growth temperature,
mean
annual temperature, mean maximum temperature for January, and mean minimum temperature
for July were highly correlated (r> 0.50; p < 0.05) with elevation (Table 3.2) and were dropped
from further analysis.
temperature gradients.
Elevation alone is a good predictor of orographic
precipitation
and
Similarly, median annual precipitation was highly correlated with growth
days (r> 0.50; p < 0.05) and was dropped from further analysis (Table 3.2). Growth days have
been found to be a better predictor of water balance for determining the effectiveness of rainfall
for biomass production in southern Africa (Ellery et al., 1992; Fairbanks, 2000).
The peA results (Table 3.3) showed that the elevation model accounted for most of the
variation, and therefore the primary gradient for the region, on axis one (0.84), similarly for the
topographical landform index on axis two (0.975) and growth days on axis three (0.966). These
three variables were therefore used for construction
heterogeneity
variable
was dropped
of the landscapes
from any further analysis.
and the topographic
By using local a priori
knowledge, visual interpretation and examination of the ordering of the factor scores on each axis
with the clustering technique we determined elevation could be meaningfully classified into two
hierarchical levels of ten detailed and four coarse classes (Table 3.4). The topographic landform
index was retained at seven classes and lumped to two classes at a coarser level (Table 3.4). The
growth days index was reclassified into 30 and 60 day ranges to produce a six level and three
level hierarchical classification (Table 3.4).
Table 3.2:
Pearson correlation matrix for environmental variables used in landscape
classification (n = 4675). Correlations highlighted in bold violate the r > 0.50 multicolinearity
limit defined for this study. t
demsd
dem
tli
dm
demsd
1.0
dem
0.37
1.0
tli
0.03
0.19
1.0
dm
-0.13
-0.52
0.01
1.0
WIll
0.50
0.70
0.05
-0.04
1.0
WIll
mdp
gd
mdp
0.35
0.22
0.06
0.53
0.79
1.0
gd
0.36
0.31
0.05
0.49
0.78
0.91
1.0
gt
mat
maxj
minj
gt
-0.43
-0.94
-0.08
0.28
-0.74
-0.38
-0.56
1.0
mat
-0.39
-0.98
-0.05
0.43
-0.72
-0.27
-0.43
0.98
1.0
maxj
-0.43
-0.84
-0.10
0.17
-0.73
-0.45
-0.67
0.97
0.91
1.0
nun]
-0.26
-0.92
0.05
0.63
-0.55
-0.02
-0.12
0.82
0.91
0.67
1.0
tYariable names: topographic heterogeneity (demsd); elevation (dem); topographic landform index (tJi); driest quarter precipitation
(dm); wettest quarter precipiation (wm); median annual precipitation (mdp); growth days (gd); growth temperature (gt); mean annual
temperature (mat); mean maximum temperature January (maxj); mean minimum temperature July (minj).
Table 3.3: Factor weights, eigenvalues, and total variance explained derived by the PCA analysis
on the chosen topographic and climatic variables. Values in bold denote the significant variable
identified for each axis.
0.77
0.84
GD
0.21
TLI
0.06
Eigenvalue 1.34
Total VarianceExplained (%) 43.46
DEMSD
DEM
tYariable names: topographic
(tli); growth days (gd).
heterogeneity
The first data combination
-0.15
0.25
0.04
0.97
1.03
25.28
(demsd); elevation (dem); topographic
0.30
0.06
0.97
0.03
1.02
16.63
landform index
involved the overlaying of the detailed level I elevation
classification
with the level I topographical landform index classification
combinatorial
classes from the input data.
producing 20 unique
All combinations of classes potentially could have
yielded 70 unique classes, but in this case, only 20 unique elevation-landform
types were derived.
This combination was then overlaid with the level I growth days index. The combined data set
derived 104 classes out of a potential
120, but several classes were shown to be small and
spurious in nature (~ 3 grid cells). The majority class filter was processed over the data surface
and a final 97 class landscape map was produced.
KwaZulu-Natal
These 97 classes represent the landscapes of
at the highest level of detail by being derived from the level I classification
hierarchies of the input data. The 97 classes were then hierarchically collapsed to the coarser 24
class landscape level II classification for ease of use and illustration (Figure 3.2).
Table 3.4: Elevation,
hierarchies.
topographic
landform
index and growth
days index
classification
Elevation range (m) from peA axis I
0-162
162 - 352
352 - 558
558 - 754
754 - 948
948 - 1138
1138 - 1353
1353 - 1610
1610 - 1986
1986 - 3484
Coastal plain
Coastal hinterland
Lowlands
Mid-lowlands
Upper lowlands
Low highlands
Mid-highlands
Upper highlands
Low Afromontane/Escarpment
plateau
Upper Afromontane/Lesotho Alpine
Coastal
Coastal
Lowlands
Lowlands
Lowlands
Highlands
Highlands
Highlands
Afromontane
Afromontane
Level/flat
Valley
Foot slope
Mid-slope
Upper slope
Scarp
Ridge/crest
Undulating/flat
Undulating/flat
Mountainouslhilly
Mountainouslhilly
Mountainouslhilly
Mountainouslhilly
Mountainouslhilly
Dry
Moderately dry
Moderately moist
Moist
Wet
Very wet
Dry
Dry
Moist
Moist
Wet
Wet
Topographic landform index
Growth Days ranges (days)
60 - 90
90 - 120
120 - 150
150-180
180 - 210
210 - 247
The coarser Level II landscape classification
was analysed using conventional
matrices for predicted versus actual class membership at field checked locations.
error
Three summary
statistics, percent correctly classified (PCC), 95% confidence limits and the Kappa statistic, were
generated from the matrix for comparing the performance of the landscape model. PCC provides
an intuitive measure of classification
accuracy.
The Kappa statistic is a measure of overall
agreement based on discrete multivariate analysis described by Bishop et al. (1975), which has
been promoted for use in the remote sensing community (Congalton et al., 1983; Foody, 1992).
Overall the level II landscape classification accuracy is good at 86.8% PCC (83.8 - 89.7%
at 95% confidence),
considering the coarse data resolution, with predictable
confusions along
landscape borders and within areas where the coarse data were not able to describe local
structural anomalies.
The Kappa statistic implies that our classification is 85.3% better than the
accuracy that would result from a random class assignment.
of the same classification
This means that a high repeatability
results could be acquired by another knowledgeable
analyst using
30
0
~
//
N
I
(13) Hlghlands MountlUnousIhJllydry
lJmilsclII>clJl>es
o (I)CoasIIl_inouaIhiIlydry
•
•
CJ
o
(2)CoasIIl_yrnoist
(3) CoasIIl MoomainouaIhiIl(4) CoasIIl UllCllUlinWflatdry
(5)CoasIIlU~moist
(I") Highlands Motm1ainottVht11y moist
•
D
(15)HIghlandsMoun •• """"""lty ••••
(16) HIghiandoUndulating/flat dry
o
(18) H1gh1ondsUnduIaIing/flat••••
D (17) IIigh/ando UnduIaIing/flatmoist
III (6)CoasIIlU~_
o (7) Lowlallds Mcvn1IU>ouaibildry
•
~
o
o
1
(19)
Allo Alpine Moomta!nou9IhiIlydry
•
(21) Allo Alpine Moomta!nou9IhiIlymoist
(8) LowIaIldsMcvn1IU>ouaibil moist
•
(~LowIalldsMoun~y_
(10) LowIandoUndula1ing'flatdry
D
(21)AlloAlpineMoomta!nou9IhiIlywet
(22) AlloAlpine UnduIaIingllla'dry
•
(23)AlloAlpinel1DdulaliDgffla'moist
•
(2A)Allo Alpine l1DdulaIiDgffla.wet
(II)Lo_Undula1ing'flatmolst
(12) LowlandoUndulaling'flat wet
Figure 3.2: Landscape classification (Level II; 24 classes) of KwaZulu-Natal Province, South
Africa,
Landscape rarity, current transformation status, and current protection provided by
conservation authorities are presented in Table 3.5.
Figure 3.3 illustrates the current human-
induced transformation status on the level II landscapes. The majority of the transformation has
taken place in the coastal and highland regions. Figure 3.4 demonstrates the bias the provincial
protected area network managed by KZNNCS has in its protection of landscapes versus the
landscape vulnerability status.
In this case, the Maputaland coastal region and the Drakensberg
Table 3.5: Calculations of percent rarity, current transformation percentage and percent protected
in managed nature reserves. The legend for the landscape numbers is given in Figure 3.2.
Level II
% of Total
% Transformed
% Protected
1.2
24.2
0.2
25.2
29.9
62.5
50.0
21.2
30.0
39.3
34.2
52.9
66.1
18.6
25.1
33.1
30.8
40.3
56.2
34.6
12.5
2.5
11.9
12.5
8.2
4.2
1.0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
0.6
0.01
12.7
5.9
1.7
4.1
4.0
0.1
6.2
7.6
0.2
1.4
6.7
1.6
13.9
15.0
0.4
1.6
5.1
3.0
3.3
3.7
0.2
0.0
13.3
13.9
5.8
6.3
0.5
1.4
1.5
1.5
0.0
0.7
2.0
10.9
0.8
0.9
1.9
14.8
20.2
51.7
2.6
4.1
7.8
Escarpment are well conserved (areas with Malaria and high rocky areas), but the landscapes
denoting the lowlands and highlands
protected.
(highly valued agricultural
lands) are severely under
This illustrates a much noted paradox in conservation's history: pieces of land have
been put aside in an ad hoc manner, often on economically marginal land or to conserve a few
charismatic species (Pressey, 1994).
Irreplaceability
and vulnerability (Figure 3.5) reveal the landscapes with high values for
both as areas of high priority for conservation
action.
The majority
of these areas have
undulating/flat terrain with moist-wet climates in the coastal, lowland, and highland regions (e.g.
5, 6, 12, 17, and 18). These priority landscapes are dominated by mixed woodland and upland
grassland ecosystems (Table 3.6), which are habitats considered in serious threat to development
throughout South Africa (Fairbanks et aI., 2000). By using the modest IUCN protection rule of
10% minimum area and a hypothetical division of vulnerability status at 50% (see Figure 3.4),
only three landscape
vulnerability
types (4, 5, and 15) are minimally protected
(Figure 3.6).
coastal gradient,
with greater than 50%
In the case of landscape type five, which lies along a north-south
only the far northern
section receives
adequate
protection.
By using a
combination of analytical graphs and spatially plotting these results, landscapes like type five can
be
identified
by
their
skew
I III Original • Transfurmed I
1800000
--I
1600000
1400000
1200000
Ul
l!! 1000000
J!l
(,)
41
J:
800000
600000
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 2~
Landscape types
I
I
L-
..
Figure 3.3: Preliminary assessment of the level of transformation within the second level
landscapes relative to their areal coverage (see Figure 3.2 for number code descriptions).
60
21
?J. 50
"""'
'-"
'"0
Q)
~
u
~
40
Q)
8 30
P-.
CI:l
~
<:
- 20
20
~
0
Eo-<
19
5 4
[5
10
24
7
o
23
It
1
0
20
40
60
80
100
Vulnerability
Figure 3.4: Scatter plot of current protection status vs. vulnerability for each landscape type (see
Figure 3.2 for number code descriptions).
100
13
80
19
2
24
.•....
.•...•
10
8
15
.0
23
14
60
,l:l
ell
<U
U
ell
17
22
16
II
20
0..
40
21
~
>-<
20
0
0
20
40
60
80
100
Vulnerability
Figure 3.5: Preliminary scores for irreplaceability (conservation value) and vulnerability to
threatening processes for the landscapes. Landscape types in the upper right-hand comer are
conservation priorities (see Figure 3.2 for number code descriptions).
Table 3.6: The values represent the percentage of each level II landscape type that is comprised
of each functional vegetation type. Values in bold represent vegetation types with >10%
affiliated area with level II landscape types.
Level II
Forestt
Arid
Woodland
Moist
Woodland
Mixed
Woodland
I
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
0.5
1.7
47.3
0.0
0.0
62.7
0.0
0.0
32.0
5.5
0.0
33.3
1.7
0.0
2.1
2.2
1.3
1.3
1.2
0.3
31.7
0.0
26.2
76.9
88.0
0.0
5.4
26.9
37.0
0.00
7.2
10.3
1.0
12.5
0.9
1.9
3.6
0.2
1.2
27.6
0.0
0.7
3.3
0.0
1.1
8.0
0.1
0.4
2.4
0.1
0.4
2.2
0.1
0.8
1.8
fNote: Forest is a combination
2.6
0.0
0.1
0.7
0.0
0.0
0.0
5.3
0.0
5.3
9.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
of Montane and Coastal Forest.
26.9
43.6
37.6
38.4
44.6
49.2
41.7
22.4
7.9
66.1
13.9
1.0
0.0
1.5
0.0
0.0
0.9
0.0
Thicket
Upland
Grassland
Highland
Grassland
24.2
25.6
0.0
2.8
10.6
2.7
23.5
28.9
8.8
11.2
20.6
4.1
21.7
4.6
2.8
4.9
1.1
1.8
0.0
0.0
0.0
0.0
0.0
0.0
0.0
3.9
87.5
0.0
0.2
4.7
4.6
12.0
20.6
2.7
22.5
33.9
17.6
36.4
61.1
8.5
46.2
86.9
0.8
28.2
34.6
0.1
45.0
79.3
0.8
0.0
0.0
0.1
0.0
0.0
12.8
3.4
0.0
14.3
4.5
0.0
16.8
33.4
18.9
19.1
37.2
5.4
99.1
69.7
62.5
99.8
53.3
18.9
Vulnerability
vs. IUCN 10% Rule
High/Sufficient
•
High/Insufficient
II
Low/Insufficient
•
Low/Sufficient
Figure 3.6: Landscape types classified by a 50% vulnerability
proposed IUCN 10% target for minimum protection of habitats.
Landscape types (6, 8,10,11,12,14,16,17,18,
have been historically
status boundary and using the
and 22) represent the bulk of the province and
ignored by the conservation
authorities and targeted for development.
They primarily contain fertile habitats of mixed woodland and upland and highland grasslands
(Table 3.6). The almost total transfer of land in the formerly white areas of South Africa, from
government to private ownership, is possibly unique in the annals of European colonisation.
The
state by the mid 1930's had lost control over resources, which in countries such as Australia,
Canada or the USA were retained by the authorities because of their unsuitability for agriculture
(Christopher, 1982). The strong tradition of land ownership rather than leasehold in South Africa
and the absence of state interest in land through a leasehold system has developed a strong
demand for land and an attempt to make a living in areas often highly unsuitable for the purposes
of farming (Christopher,
1982; Schoeman and Scotney, 1987).
Demand for land has further
driven land prices to levels far in excess of its value as an agricultural commodity, and thus
confounded past and present conservation efforts.
In practice,
conservation
developing conservation
(1990) hierarchy
plans.
framework
managers
priority
assessment
data as an aid to
However, it would seem more reasonable
for identifying
combination of landscape priority-species
landscape
rely on speCIes distribution
important
to adopt the Noss
areas for conservation
or vegetation priority-species.
(from Figure 3.5) and a vegetation
based on a
An example using the
priority
assessment
(see
Appendix B for Reyers et al., in review) were conducted with the bird atlas database (Harrison et
al., 1997). Landscapes are ranked in order of importance based on dividing the graph (Figure
3.5) into four quadrants based on the 50% boundaries
on each axis and then defining the
following ranked values for the landscapes
quadrants
(Cartesian
read clockwise;
based on
suggestions from the C-Plan website, http://www.ozemail.com.au/~cplan/background_1.html):
II.
High conservation
values but not threatened,
maybe
consider
off reserve
management. (Rank 3)
IV.
These areas may contain features that are already represented
in reserves, but
which are still at risk. (Rank 2)
The priority vegetation type ranks were conducted in a similar manner (Appendix B),
with the KwaZulu-Natal
province containing four ranked vegetation types based on a national
level assessment (Appendix B). The spatial distributions of the landscape and vegetation priority
ranks are contained in Figure 3.7. Each ranked class on each map is used sequentially in turn to
define the search areas for rarity and richness-based
reserve selection algorithms (Rebelo and
Siegfried, 1992; Howard et al., 1998; Reyers et al., 2000). The results of using the hierarchy of
ranked
landscapes
and vegetation
to determine
complementary
sets of bird
species
for
conservation are provided in Figure 3.8.
Clearly, from the examples given, the goal of conservation is not only to ensure minimum
landscape, habitat and species protection, but also to represent geographic gradients and to enable
longer-term ecological and evolutionary processes to persist.
This is not in conflict with the
~
Ptiaity Rlrks
\eg:tation Pri:rity Rui"
•
I
•
•
2
II!
3
III
III
D
4
o
I
2
3
4
Figure 3.7: Priority ranks for landscapes and vegetation types as inclusion to rarity and richnessbased reserve selection algorithms.
importance of habitat loss for the immediate persistence of biodiversity, but long-term persistence
goals also need to be considered in designing and implementing reserve systems, especially in
response to global change.
This analysis represents one of the few times that a landscape or ecoregion classification has been
properly assessed for accuracy and fitness for use in the field, and thus evaluated for use in
systematic conservation planning.
Using indirect methods, Wright et al. (1994) and Host et al.
(1996) also assessed the value of larger ecoregiona1 units (e.g., Omernik, 1987) and a machine
driven ecosystem classification
conservation
with mixed success.
planning is questionable
The use of ecoregion classifications
for
given the very coarse scale of the units, the mixing of
'potential' and actual data sets (e.g., potential vegetation, climate zones, land-use pattern, soils,
etc.), and the reliance on boundaries drawn by a consensus of experts, which may not provide a
repeatable methodology.
Rather, a data driven and parsimonious approach based on ecologically
important structural and climatic variables derived at a larger landscape scale may allow for a
better understanding
type of landscape
edaphic
of the pattern and processes required for biodiversity preservation.
model can then be independently
assessed with potential vegetation
This
and
Figure 3.8: Selection order results for potential reserve networks based on either rarity or richness
procedures for birds. (a) and (b) Results are based on a hierarchical mask of ranked landscape
values based on four quadrants derived from 50% cutoff points in Figure 3.8; and (c) and (d)
results are based on a hierarchical mask of ranked priority vegetation types based on current
versus potential transformation (see Appendix B).
While chosen data layers and analytical methods are relatively objective, there are a
number of decisions that require some a priori understanding
of the landscapes under study.
There are also data processing questions, such as determining, a statistically appropriate number
of classification
levels, selecting important variables or generalizing
subjective, yet defendable decisions.
boundaries
that require
It is unrealistic to expect that the process of landscape
classification can be accomplished entirely by spatial and numeric analysis; human understanding
is also an important component (Host et aI., 1996).
However, by defining a computationally
repeatable methodology the knowledge of experts may be captured for future refinements within
a data driven model.
Terrain analysis is the quantitative
analysis of topographic
studying surface and near-surface processes.
wide range of landscape-scale
surfaces with the aim of
In short, terrain analysis provides the basis for a
environmental models, which are used to address both research
and management issues and objectives. It is widely recognised that landscape pattern analysis is
sensitive to the resolution (spatial scale) of the source data (Turner et aI., 1989). As the distance
between neighbouring
elevation samples increases, fine-scale features are lost and the surface
becomes more generalised.
However, when identifying landscapes there is a tendency to focus
on specific finer detailed terrain or ecosystem elements within a landscape rather than the broad
scale structures that truly define a landscape.
For this study, a landscape was not defined
traditionally as a mosaic where the mix of local ecosystems is repeated in similar pattern over a
kilometers-wide
define
area (Forman, 1995), but rather where the physical systems integrate together to
identifiable
environmental
patterns
over
a kilometers-wide
area.
Therefore,
the
database
of
layers defined at a resolution of 1km2 was considered appropriate for striking a
balance between regional and local ecosystem heterogeneity.
This study has shown that it is possible to produce an ecologically inclusive inventory of
regional landscapes, notwithstanding
complexity.
the extensive areas they occupy and their inherent spatial
Noss (1990) described landscapes as the upper level in a hierarchical framework that
extends upwards from genes-species-ecosystems
to describe the range of biological diversity.
The analytical framework presented here is an appropriate model for elucidating the landscape
level biodiversity
dilemmas faced by conservation
practitioners.
By proposing a top-down,
constraint based modelling and conservation assessment an approximation of the main processes
and structure maintaining
long-term biodiversity pattern can be used in more specific species
protection
plans.
and recovery
Biophysically
defined
landscapes
containing
vegetation types with edaphic drivers determine and drive co-evolution
mammal, reptile, bird and insect.
The products of interacting
elements
of
with other species of
organisms in a hierarchically
defined landscape environment are ecosystems.
The majority of the work on preserving biodiversity
and selecting priority areas for
conservation has concentrated on the lower level of the biodiversity hierarchy: species (Pressey
and Nicholls,
populations
1989; Rebelo
(Lamberson
and Siegfried,
et aI., 1992; Breininger
(especially vegetation assemblages;
1995) patterns.
to identifying
1990; Lombard,
1995; Pressey
et aI., 1996),
et aI., 1995; Doak, 1995), and community
Scott et aI., 1993; Strittholt and Boerner,
1995; Barbult,
Recently, criticism has been levelled at especially the species based approaches
priority
conservation
areas (Noss, 1983; Franklin,
1993; Scott et aI., 1993;
Barbault, 1995; Maddock and du Plessis, 1999; Maddock and Berm, 2000). However, due to the
hierarchical nature of biodiversity any approach, which only concentrates on one of the levels, is
flawed. There has been virtually no research on designing reserve systems intended for long-term
persistence
of biodiversity
representation
in the face of global change.
and retention of both biodiversity patterns as well as the processes that maintain
and generate these patterns.
Thus, more comprehensive and inclusive biodiversity protection can
be obtained by focussing on as many levels as possible.
irreplaceability
representation
Such a strategy must embody the
and
vulnerability
analysis,
are focus
and identification
maintenance and genesis of biodiversity.
areas
Landscape areas representing
for follow-up
of key processes
species
high
and ecosystem
that are responsible
for the
If the information is available, important constituent
ecosystems within these priority landscapes can be identified using the classification procedure
developed here. The dominance of mixed woodland and upland grassland vegetation functional
types within the priority landscapes identified in the preliminary analysis suggests the ecosystems
needing consideration, and gives significant insight into what conservation actions are needed on
the ground.
Hierarchy theory (O'Neill et aI., 1986) suggests that constraints operate downward in
complex hierarchies
such as ecosystems
aggregated
In recognising
levels).
(i.e., from the more aggregated
levels to the less
this, it has been suggested that using higher levels of
biodiversity alone to select priority areas for conservation is preferable, especially in areas with
inadequate region-wide biological data (Margules and Redhead,
assumption
that diversity and spatial heterogeneity
1995).
are intrinsically
This is based on the
linked (Diamond,
1988;
Hunter et al., 1988; Samways, 1990; Forman, 1995). If for instance landscapes were to be used
in this manner, it assumes that a predictable relationship (surrogacy) between diversity at the
landscape
level and lower
levels
exist.
Unfortunately,
little research
has tested
these
assumptions, but some do suggest (see Hamer and Harper, 1976; Burnett et al., 1998; Nichols et
al., 1998) that the upper levels of biodiversity (e.g., Noss, 1990) may act as effective surrogates
for biodiversity as a whole. However, this will vary between ecosystems and depend on levels of
disturbance.
Until such relationships
are adequately explained, the best practice for selecting
priority areas and preserving biodiversity will involve multiple levels of biodiversity (i.e., broader
classification such as landscapes, vegetation, geology in conjunction with species data and human
development induced threats) guided by the principles of retention of pattern and process.
A final issue that must be addressed
classification
system over time and space.
is the robustness
of the derived landscape
The landscape classification
based on both structural and climatic components.
system developed was
The structural data layers are expected to be
robust over time and space due to their slow geological evolution, but climate may present
resiliency problems for the current classification.
precipitation
Under a predicted climate change scenario for
in southern Africa (Joubert and Hewitson,
expected to change over space and in magnitude.
climatic
data sets become
classification
system.
Re-defining the classification
can therefore
retain the relevance
when newer
of the landscape
This is not in conflict with the objective of providing a classification
system for a functional
time.
available
1997) the growth days index can be
landscape, which is also expected to undergo evolutionary change over
However, there is a trade-off between too much data resolution
versus the expected
resilience of the classification system, which can be tested through sensitivity analysis.
The use of regional ecological classification systems is increasing (Bailey, 1996; Host et
al., 1996; Pressey,
1997).
This is a result of efforts by resource
and nature conservation
managers to replace political boundaries with ecologically based management units that better
reflect the spatial distributions of natural features. This is particularly true in water resource and
nature conservation
planning sectors, where landscape and regional ecology can be used to
spatially
natural
combine
management
processes
and human
(Davis and Stoms, 1996).
activities
to promote
Developing a landscape classification
sustainable
land
allows for this
often ignored level of biodiversity to be inventoried and considered in conjunction with speciesbased conservation prioritisation exercises.
were chosen, and requires some a priori knowledge of the focus region's landscapes. However,
the method is systematic and extensible to other areas.
Furthermore, the method provides
approaches for quantitatively classifying data, allows for quantitative understanding of the data
heterogeneity among the themes, and can be updated as better data becomes available or
environmental changes are documented.
By developing data layers for all the levels of biodiversity we can then provide a protocol
for developing a reserve system that will enable biodiversity to persist into the next millennium.
Rather than maximizing conservation of contemporary biodiversity patterns, a system should
conserve ecological and evolutionary processes essential for sustaining biodiversity. The use of
the landscapes-species hierarchy and the identification and role of processes in maintaining
biodiversity patterns will help conservation planners to formulate clear representation goals in
balance with human induced threat.
4.
Species and Environment Representation: Selecting Reserves for
the Retention of Avian Diversity
Considerable progress has been made in developing and testing practical protocols for
designing representative conservation area systems (for review see Margules and Pressey, 2000).
Historically,
opportunistic
methods have been used for assigning land with low potential for
economic and political conflict; or high potential for recreation
conservation,
which has resulted
in an inefficient
and tourism to biodiversity
and ultimately
more costly means of
conservation area allocation (see Pressey, 1994; Rodrigues et al., 1999).
This has lead to the
'minimum set' approach to conservation planning to identify whole systems of complementary
areas that collectively achieve some overall conservation goal in a more efficient manner (Pressey
et al., 1993).
Its prevailing conservation focus is to identify potential conservation areas that
represent the greatest number of features (e.g., species, vegetation types) at least once. However,
the extent to which conservation areas fulfill the role of securing a region's biodiversity depends
only partly on the goal of sampling biodiversity pattern.
also requires the representation
The long-term retention of biodiversity
of the processes that contribute to shaping and maintaining
biodiversity patterns.
Several
authors
have
emphasized
that
current
biodiversity
representation
within
conservation areas is not equivalent to the ultimate goal of maintaining biodiversity over the longterm (Cowling et al., 1999; Fairbanks and Benn, 2000; Margules and Pressey, 2000; Rodrigues et
al., 2000).
The representativeness
concept implies that a reserve, or system of reserves, should
contain biota that ideally represents the entire range of biological and environmental
within a given geographical
area (Margules and Usher, 1981; Kirkpatrick,
variation
1983; Austin and
Margules, 1986; Mckenzie et al., 1989). Fairbanks and Benn (2000), along with Margules and
Pressey (2000), agree, but also emphasize the maintenance of natural processes as an important
component
of conservation
area selection.
distribution
patterns change over time, the selection of conservation
turnover in species or environmental
Rodrigues
et al. (2000) argue that as species
areas that are robust to
diversity is a critical component
selection for ensuring the long-term maintenance
of biodiversity.
of conservation
area
Thus, in selecting nature
reserves, one should attempt to identify the major gradients of biotic and environmental variation
within habitat types of interest in the study area and, if possible, the environmental variables that
most closely correlate with the distribution and abundance patterns of relevant taxa (De Velice et
al., 1988).
Emphasis should not only be placed on the identification and conservation of biodiversity
pattern, but also the natural processes that control and maintain that pattern within the biodiversity
hierarchy (Noss, 1990; Balmford et aI., 1998). Conservation of ecosystem processes that sustain
ecosystem structure and function (Fairbanks and Benn, 2000), and evolutionary processes that
sustain lineages and generate diversity (Cowling et aI., 1999), are essential for achieving the longterm maintenance of biodiversity in conservation areas (Nicholls, 1998). However, as Margules
and Pressey (2000) point out, because conservation
area selection is often a spatial exercise,
protection of these natural processes is often based on their spatial surrogates rather then on the
processes
themselves.
Nevertheless,
by ensuring that conservation
areas are large or span
substantial environmental gradients it should be possible to accommodate, at least partially, many
of these natural processes (Noss, 1996).
Ordination
environmental
analyses
have illustrated
gradients responsible
tremendous
potential
for identifying
important
for biodiversity pattern (DeVelice et aI., 1988; Faith and
Norris, 1989; Saetersdal and Birks, 1993; Taggart, 1994). This analytical approach is used for
integrating multiple environmental
effects across a landscape (Bray and Curtis, 1957; Gauch,
1982; Jongman et aI., 1995). Ordination, whether direct or indirect, is particularly useful when
studying the relationships between species composition and environment (Jongman et aI., 1995).
Beta diversity
is concerned
with species
spatial turnover
along habitat
gradients
(Whittaker 1977). Beta diversity is important in determining regional species richness patterns,
yet little attention has been paid to this component of diversity in selecting conservation areas. If
conservation areas are selected only to represent numbers of species, they may not necessarily
continue to serve this purpose over a period of years (Margules et aI., 1994; Virolainen et aI.,
1999; Rodrigues et aI., 2000).
The present study addresses the issues of conserving
natural processes
and spatial
turnover of species diversity in an investigation conducted to assist the KwaZulu-Natal
Conservation
conservation
Service (South Africa).
areas in KwaZulu-Natal
The goal was to identify additional
potential
Nature
avian
Province, as an added component to their strategic plan
(Armstrong et aI., 2000) for the long-term maintenance of regional biodiversity.
To date, no study
has been carried out on the complete bird fauna of the province to assess its representativeness
or
relationships with environmental processes and features.
The primary analytical tool used was canonical correspondence
analysis (CCA; ter Braak
and Prentice, 1988), a widely used direct gradient analysis method (Palmer, 1993), and detrended
correspondence analysis (DCA), an indirect gradient analysis method (Gauch, 1982). The
program CANOCO, version 4.0 (ter Braak and Smilauer, 1998), was used to conduct all gradient
analyses. DCA and a hierarchical classifier were used to determine the avian species communities
within KwaZulu-Natal
(Legendre
and Legendre,
1998).
Environmental
data (e.g., the 13
environmental parameters found under topography and climate in Table 1.1) were entered with
the species data using stepwise CCA to investigate which environmental variables explained the
patterns in observed avian diversity (ter Braak and Smilauer, 1998). Variables are added to the
model in the order of greatest additional contribution to total variation explained, but only if they
were significant (P::;; 0.01), where significance was determined by a Monte Carlo permutation
test, and if adding the variable did not cause any variance inflation
Variables
with large inflation
factors are strongly multicolinear
factors to exceed 20.
with other variables
contribute little unique information to the model (ter Braak and Smilauer, 1998).
and
In order to
combine this information on species patterns and the related environmental gradients responsible
for those patterns into practical conservation planning techniques, I propose the use of spatial
autocorrelation analyses.
In the analysis of spatial association among many spatial observations, the tendency is to
assess spatial autocorrelation based on global statistics such as Moran's lor Geary's c (Cliff and
Ord, 1981).
A focus on local patterns of association
(local spatial clusters) prompted the
development of local indicators of spatial association (Anselin, 1995). This form of analysis was
used to identify areas with high levels of species and associated environmental gradient turnover.
The software packages Spacestat (An selin, 1999) and S-plus with the spatial statistics component
(Mathsoft, 1999) were used to conduct this part of the analysis.
Using Moran's I analysis, based on the information gained from the previous CCAs, local
spatial clusters of integrated species compositions and their associated environmental
gradients
were identified. A grid cell with a high positive Moran's I value is highly autocorrelated
similar to neighbouring
parameters.
grid cells in terms of avian species contained
or is
and environmental
A grid cell with a negative to low positive Moran's I value shows low levels of
autocorrelation
and is thus very different from surrounding
assemblage and the associated environmental variables.
grid cells in terms of species
Thus, those grid cells with low levels of
spatial autocorrelation are indicative of areas with high turnover in species composition as well as
strong environmental gradients.
An algorithm based on species rarity or richness (Rebelo and Siegfried, 1992; Howard et
aI., 1998; Reyers et al., 2000) for selecting a set of complementary reserves was initially run on
the birds species distribution data. However, such selection procedures do not successfully select
areas for the representation
of natural processes responsible for generating biodiversity patterns.
Furthermore, they do not target areas of high beta diversity, i.e. areas with a high turnover in
feature diversity.
I attempted to include steps in these algorithms that selected areas high in beta
species diversity and with associated environmental
lowest spatial autocorrelation
gradients by ranking the grid cells from
to highest and iteratively incorporating
representation using either species rarity and richness approaches.
the required species for
Moran's I values were used as
indicators of the importance of grid cells in terms of species and environmental
turnover. This
then made it possible to represent not only alpha diversity patterns (numbers of species within a
community), but also beta diversity patterns and sample the underlying environmental gradients
during the reserve selection procedure.
First, a grid cell was considered protected if ~ 25% of its area fell within protected areas.
The species found within these grid cells were removed from the analysis. Second, Moran's I
values of each grid cell were categorized and ranked into four groups: negative autocorrelation,
weak
positive
autocorrelation.
autocorrelation,
moderate
positive
autocorrelation,
and
strong
positive
Third, two analyses were completed, one based on complementary rarity and the
other on complementary
richness. The algorithm starts by selecting grid cells from the first
category of spatial autocorrelation
scans them for un-represented
(i.e. grid cells in the negative autocorrelation
species not removed in the first step.
proceeded in a stepwise fashion through all spatial autocorrelation
category) and
The algorithm then
categories until all species
were represented at least once. In this way two real-world reserve system outputs were developed
for comparison,
based either on species rarity or richness, but also incorporating
dissimilar species compositions
and previously
and different environmental
characteristics
areas with
from neighbouring
selected grid cells (high beta diversity). This beta diversity (BD) algorithm,
therefore, selects a network that not only represents all species in the area, but also bases its
selection on the spatial structure of the species assemblages and environmental
samples both biodiversity pattern and process in a representative manner.
gradients, i.e. it
Geographic patterns of hierarchically classified DCA scores are indicated in Figure 4.1
illustrating
the five avian communities
community
in the northeast,
identified
within the province.
the East Coast, the Drakensberg
forming a transition between the Drakensberg and Maputaland
southern
Midlands
community
at the southern
The Maputaland
Escarpment,
Central Zululand
communities
and the Central-
end of the province,
each contain unique
combinations of species. The most important bird species in each community, based on indicator
species analysis (Dufrene and Legendre, 1997) is provided in Table 4.1. Eigenvalues and gradient
lengths were moderately higher for DCA than for CCA for the first two axes (Table 4.2). This
fact together with the strong and significant correlations between the DCA for axis 1 and axis 2
with the explanatory variables (Table 4.3) suggested that much of the variation in avian diversity
distribution is related to the measured environmental variables.
The stepwise CCA reduced the
number of significant variables required to explain the variation in species turnover (Table 4.4).
Most (81 %) of the variation in bird species assemblages in KwaZulu-Natal was accounted for by
the explanatory
environmental
variables of elevation heterogeneity,
growth temperature, mean annual evapotranspiration,
mean growth days, mean
and seasonality of precipitation.
The CCA results are graphed as a biplot, in which arrow length and direction indicate the
correlation between the explanatory variable and the CCA axes, and smaller angles between
arrows indicate stronger correlations between variables (Figure 4.2). The dominant compositional
gradient (axis 1) reflected an altitudinal gradient, which was represented by the mean growth
temperature and the seasonality of precipitation, from the sub-tropical climate of the coast to the
temperate-afromontane
climate of the Drakensberg Escarpment.
Grid cells towards the higher
lying areas experienced higher seasonal variability in temperature and precipitation, whereas low
lying coastal regions experienced lower seasonal variability in temperature, higher temperatures,
and lower variability in precipitation. The seasonality of precipitation and elevation heterogeneity
are moderately correlated with each other, but reflected low inflation factors in the CCA analysis
therefore each was able to provide explanation for the turnover in species composition.
This
altitudinal gradient runs roughly east-west from the Maputaland coastal plain to the Drakensberg
Escarpment, reflecting the strong climatic influence of the Indian Ocean and the generally northsouth orientation of the Drakensberg Escarpment.
The second CCA axis was a gradient in growing season moisture stress, from the areas of
warm, dry growing
seasons
around Maputaland
and the Lebombo
Mountains,
which are
characterized by arid woodlands to areas of warm, wet growing seasons along the southern East
Avian Communities
E21
Maputaland
[]
East Coast
8
Drakensberg
[]
Central Zululand
~
Central-Southern
Midlands
Figure 4.1: Identified avian diversity assemblages derived from hierarchical classification of first
two axes of the detrended correspondence analysis results.
Coast (Figure 4.2).
Areas of low summer precipitation
and high annual evapotranspiration
included the interior valleys to the west of the Lebombo Mountains, especially northern Zululand,
and the White Mfolozi and Tugela River basins. The Central-southern Midlands represents areas
of higher summer precipitation with variable elevation owing to lower annual evapotranspiration
being able to support montane forests and upland grassland.
An analysis of the available vegetation habitat and human impact on the bird communities
of KwaZulu-Natal illustrates the conservation conflicts and habitats to be managed.
In Table 4.5
potential functional vegetation types (Fairbanks and Benn, 2000) that would have occurred today,
were it not for all the major human-made transformations, were combined with currently mapped
major land-use types (Fairbanks et al., 2000). The proportion of vegetation types for each avian
community (Figure
1.3) provides a general description of the habitat requirements.
The land-use
Table 4.1: Avian bioindicators in order of importance based on Dufrene and Legendre (1997)
indicator species value measure for each identified avian community assemblage.
Trigonoceps occipitalis
Eupodotis rujicrista
Torgos tracheliotus
Coracias naevia
Eremomela usticollis
Nectarinia neergaardi
Tockus erythrorhynchus
Terathopius ecaudatus
Cossypha heuglini
Eremomela icteropygialis
whiteheaded vulture
red crested korhaan
lappetfaced vulture
purple roller
burntnecked eremomela
Neergaard's sunbird
redbilled hornbill
bateleur
Heuglin's robin
yellowbellied eremomela
Aquila rapax
Cossypha humeralis
Cisticola chiniana
Merops pusillus
Hieraaetus spilogaster
Vidua paradisaea
Aquila wahlbergi
Tricholaema leucomelas
Turtur chalcospilos
Sylvietta rufescens
tawny eagle
African whitethroated robin
rattling cisticola
little bee-eater
African hawk eagle
paradise whydah
Wahlberg's eagle
pied barbet
greenspotted dove
longbilled crombec
Morus capensis
Sterna hirundo
Sterna bengalensis
Calidris alba
Sterna sandvicensis
Charadrius leschenaultii
Sterna albifrons
Sterna paradisaea
Sterna bergi
Larus dominicanus
cape gannet
common tern
lesser crested tern
sanderling
sandwich tern
sand plover
little tern
Arctic tern
swift tern
kelp gull
Central-Southern
Midlands
Hirundo atrocaerulea
Zoothera gurneyi
Serinus scotops
Poicephalus robustus
Tauraco corythaix
Ploceus bicolor
A nthus lineiventris
Seicercus rujicapillus
Nectarinia chalybea
Anthreptes collaris
blue swallow
orange ground thrush
forest canary
cape parrot
Knysna lourie
forest weaver
striped pipit
yellowthroated warbler
lesser doublecollared sunbird
collared sunbird
Eupodotis caerulescens
Hirundo spilodera
Chaetops aurantius
Francolinus africanus
Euplectes afer
Spreo bicolor
Gypaetus barbatus
Chersomanes albofasciata
Myrmecocichla formicivora
Amadina erythrocephala
blue korhaan
SA cliff swallow
orangebreasted rock jumper
greywing francolin
golden bishop
pied starling
bearded vulture
spikeheeled lark
southern anteating chat
redheaded finch
Table 4.2: Eigenvalues and gradient lengths (1 Standard Deviation) for the first two axes from
DCA and DCCA of all bird species for KwaZulu-Natal.
Eigenvalue
Gradient
length
DCA
DCCA
DCA
DCCA
0.21
0.19
0.09
0.08
1.96
2.45
1.51
1.23
Table 4.3: Spearman's rank correlation of explanatory factors with axis scores from DCA and
intraset correlation coefficients from CCA that included all explanatory variables. t
Axis 1
Axis 2
DCA
CCA
DCA
CCA
DEMMEAN
0.93
0.93
-0.05
-0.05
DEMSTD
0.59
0.59
-0.5
-0.023
GDMEAN
0.05
0.04
0.83
0.84
MAP
-0.03
-0.86
0.73
-0.21
GTMEAN
-0.88
-0.92
-0.21
-0.11
NGTMEAN
-0.91
-0.91
-0.11
-0.10
MAT
-0.91
-0.03
-0.09
0.74
HOTMNTHMN
-0.89
-0.89
-0.16
-0.17
MINMNTHMN
-0.93
-0.93
0.02
0.01
EVANNMN
-0.42
-0.42
-0.66
-0.69
PSEAS MN
0.85
0.86
0.06
0.04
TSEAS MN
0.53
0.54
-0.49
-0.52
MXSEAS
0.57
0.57
-0.40
-0.42
MN
t Sign reflects arbitrary selection of gradient direction by CANOCO.
Axis 1
Axis 2
DEMSTD
0.61
-0.001
GDMEAN
0.05
0.86
GTMEAN
-0.88
-0.21
EVANNMN
-0.44
-0.72
PSEAS MN
0.87
0.04
information provides an indication of the current transformation processes taking place within the
avian community assemblages.
The heterogeneous nature of the Central Zululand and Central-
southern Midlands vegetation structures and avian assemblages is apparent.
heterogeneity
found within the Central-southern
The environmental
Midlands community has also provided ample
development opportunities for humans, with 43% of the landscape having been transformed, and
most of the existing protected areas here are small. The small sizes of these protected areas, their
scattered locations, their progressive isolation through the loss of connecting habitats are cause for
Figure 4.2: Species-environment gradients identified from stepwise canonical correspondence
analysis with convex hulls of avian community biogeographic zones. GTMEAN - Annual mean
of the monthly mean temperature (0C) weighted by the monthly growth days; PSEAS _MN Precipitation seasonality from the difference between the January and July means; DEMSTD Elevation heterogeneity; GDMEAN - Number of days per annum on which sufficient water is
available for plant growth; and EV ANNMN - Total annual pan evapotranspiration (mm).
The analysis was performed on axis I and 2 of the CCA results (Figure 4.3a, b). The resultant
Moran's I axis values were then derived for each grid cell for the analysis of the speClesenvironment
autocorrelated
spatial structure in the reserve selection procedure.
clusters of similar species-environment
On axis 1 strong positive
compositions
northern coast, Maputaland coastal plain, and Drakensberg Escarpment.
were located along the
Negative autocorrelated
clusters were identified in the interior associated with the Central Zululand and Central-southern
OCA Axis 2 ~ran's LISA
OCA Axis 1 ~ran's LISA
o
"l
o
t'-Egaiveal1oo>rraaim
fI1
Fbsitive rn:xierae aliOOOrraciirn
II
II
•
FbStivestrorg atcxxmlabn
•
Fb91ive\'o'E9<
al1oo>rrBaim
Figure 4.3: Moran's I spatial autocorrelation
combined Moran's I axes 1 and 2.
o
o
t'-Egaiveal1oo>rTelaim
II
RJsitive
•
RlStivestrorg atcxxmlabn
t'-EgaivealJo<Dmjaim
RJ9tive\'o'E9<al1oo>rrBaim
RJ.tive rrroe:aeal1oo>rraaim
FtlstivestJOrg atcx:::are<.ton
RlStive\'o'E9<al1oo>rrBaim
rn:xiErcie al1O<Drraciioo
results: (a) CCA axis 1; (b) CCA axis 2; and (c)
Midlands avian communities. These grid cells represent dissimilar species-environment
compositions from their immediate neighbours and therefore represent areas of high species
turnover along the identified environmental gradients.
Moran's I analysis of the second axis identified strong positive autocorrelated clusters in
the arid woodland region of northern Zululand, Maputaland and the Lebombo Mountains and
along the southern East Coast.
Negative clusters were found in the Tugela and Mhlatuze river
basins, central Drakensberg Escarpment and northern East Coast.
Table 4.5: Percentage of functional vegetation and land-cover/land-use
community assemblage.
Maputalandt
East
Coastt
types per identified avian
Drakensberg
Central
Zululand
CentralSouthern
Midlands
Coastal Forest
0.6
1.8
0.0
0.0
0.0
Afromontane Forest
0.1
0.7
0.7
0.5
1.2
Arid Woodland
46.6
0.1
0.7
9.8
0.0
Moist Woodland
9.6
25.2
0.0
0.0
2.0
Mixed Woodland
7.6
4.9
18.0
31.9
13.3
Thicket
0.0
3.9
0.3
16.3
9.6
Upland Grassland
0.0
2.9
19.2
20.6
Highland Grassland
0.4
0.0
Wetlands
4.4
6.2
36.7
1.0
4.5
6.4
Bare
0.1
0.5
Degraded
11.8
Exotic plantation
0.1
3.3
0.5
0.5
0.5
0.0
7.5
5.6
9.6
10.0
1.8
8.8
3.4
5.8
16.1
Agriculture Dryland
16.5
22.1
10.5
13.7
19.5
Agriculture Irrigated
0.4
0.1
2.0
0.5
2.0
Urban
0.1
7.4
1.3
0.4
2.0
t Missing area measurements
from coast and Mozambique border.
Spearman's rank correlation analysis of the individual Moran's I axis values and combined
values revealed
heterogeneity
relationships
(Table 4.6).
between
the Moran's
I values and definitions
of land type
A combined model of landscape types (Figure 3.2) and functional
vegetation types (Figure 1.3) had the highest correlations with Moran's I values for axis 1 and
combined Moran's I values. There was no meaningful relationship for the axis 2 Moran's I results.
This relationship
depicts decreasing Moran's I values as the variety of landscape-functional
vegetation types increases, i.e. with increasing environmental heterogeneity
(Figure 4.4). This
implies that local bird diversity turnover is more strongly linked to landscape and vegetation
structure (e.g., MacArthur, 1964; Wiens, 1989a) within the Central Zululand and Central-southern
Midlands, than to broad climate. These areas appear to represent important transitionary regions
for birds, between the richer and more homogenous high grassland areas of the Drakensberg
Escarpment,
the Maputaland
arid woodlands
and East Coast
moist
woodlands.
These
heterogeneous areas may also be of significance as zoogeographical barriers to avian distributions
because of deeply incised river valley conditions (Figure 1.2a - also see Benson et al., 1962;
Clancey, 1994).
Microclimates,
diverse habitat assemblages, and geomorphology
all seem to
play important roles in maintaining and driving the unique bird assemblages and rapid species
turnovers across the province's interior regions (Figure 4.3c).
Table 4.6: Spearman's rank correlation coefficients of the Moran's I analysis and the diversity of
landscape definition types (see Table 1.1).
Axis 1
Axis 2
Combined
LAND
-0.52
-0.16
-0.49
LANDVEG
-0.62
-0.07
-0.54
LANDVEGF
-0.70
-0.04
-0.59
VEG
-0.37
0.11
-0.23
VEGF
-0.51
0.14
-0.32
LCLUTYPES
0.01
0.09
0.00
LCLULAND
-0.57
-0.05
-0.49
4
•
-
3
.~
'"
--
•
~
~C.I
'0
"C
2
blI
'~
"
c.
~
•••
1
•
•
•
• ••
• • • •
I
•
•
•
_. • • • •• •• • • • ••
: •
• • • ••
• • • ••
• • • • •• • I • I • •
•
I
••
•
I
• • •• • • • ••
•
I
•
•
I
•
• I• •
I
•
•
'"
~=
~
'"
0
~
•
••
~
'"
=
-;
•
•
•
- • ••• •
• • •
•
• •
••
f-
0
•
•
• ••
I
•
•
•• ••
Figure 4.4: Graph of CCA axis 1 Moran's I values relationship
vegetation functional types found within each grid cell.
I
•
• • •
to the variety of landscape-
The study area of 165 grid cells included 19 (11.4%) grid cells that were considered
protected, i.e. ::::25% protected.
These grid cells are located almost entirely within the central
Drakensberg Escarpment and Maputaland areas, and represented 529 (93%) of the 566 recorded
bird species. This illustrates that these larger existing reserves do contribute significantly towards
the goal of conserving all avian species. The rarity and richness-based complementary algorithms
selected 15 (10%) and 14 (9%) of the remaining
147 grid cells respectively
to represent the
remaining 37 species at least once (Figure 4.5a,c). To achieve the goal of adequately sampling all
species while also representing the identified environmental
algorithm for both the rarity and richness-based
gradients, the BD (beta diversity)
analyses needed 18 (12%) grid cells (Figure
4.5b,d).
Figure 4.6 illustrates the rates of species accumulation
richness-based
for the four algorithms. The
algorithm rapidly represented most species (> 90%) within 7% of the remaining
land area, with the rarity-based algorithm requiring only slightly more land (9%).
The rarity-
based algorithm also illustrates the break levels its search rules creates by looking for pockets of
rare species while constrained
previously selected grid cells.
by proximity rules to pick grid cells that are closer to the
The richness-based
BD algorithm initially selected species at a
slow rate but increased after the first 3.5% of the grid cells were selected and the rarity-based BD
algorithm shows the same breaks but chose more land area earlier.
The results outlined above assume that the protected areas that are already proclaimed are
adequate, and that the procedure used can only produce results that add to defining an all inclusive
representative reserve network.
Once the environmental gradients that are associated with birds
species turnover are identified it may be more appropriate to ask what would an "ideal" network
for total bird protection look like if the current protected areas were not assumed adequate. Figure
4.7 provides such a result, which might provide a more resilient and thus viable option for longterm retention of the provinces bird diversity.
quo of using straight species-based
For either algorithm, the contrasts with the status
complimentary
procedures versus incorporating
associated
environmental gradients are strikingly apparent.
The original rarity and richness-based algorithms were the most efficient representing all
species in the least amount ofland area possible. These algorithms obviously concentrated on the
areas of high species richness and rarity.
The algorithm rules for either approach (rarity or
richness) select grid cells in a locally optimal manner, based on the species database and grid cell
proximity, rather than selecting based on regional optima.
The grid cells picked for either the
rarity or richness-based algorithms are similar, except for the selection order, with most areas
Figure 4.5: Comparison of algorithm results: (a) species rarity-based algorithm; (b) species rarity
and beta diversity algorithm; (c) species richness-based algorithm; and (d) species richness and
beta diversity algorithm.
selected in the Maputaland, Drakensberg Escarpment and East Coast regions.
The BD algorithm
attempts to provide the algorithm rules with important environmental information about the entire
region using the ranked spatial autocorrelation
for both the rarity and richness-based
classes.
Although similar grid cells are selected
BD algorithms, the masking action of the ranked spatial
autocorrelation categories forces the algorithms in this region to search the interior of the province
(Figure 4.3c) first to locate grid cells containing the required species.
Four to five grid cells are
chosen from the southern areas of Central Zululand and northern areas of the Central-southern
Midlands depending on the algorithm emphasis (Figure 4.5b,d). The other significant differences
....,;::::;::::;;
"'C
~
c:
Q)
en
~
,....r-..,.... /
80
...,.../..
,...
0-
~
en
-/
/
60
/
/
/
Rarity
- Richness
--Rarity & SD
................
Richness & SD
i //
Q)
'(3
/1
Q)
0-
en
C>
c:
c:
40
E
20
//
I
'ro
./
(
.;
l
.
1/
~
~
o
rjJ
5
10
% land area required
Figure 4.6: Graph of algorithm efficiencies detailing species representation
area required. (BD = beta diversity).
among algorithm outputs are the selection orders.
versus percent land
In this case, the rarity-based BD algorithm
results are the most useful for conservation as it ranks the rarest birds, landscapes and natural
processes most important for immediate conservation action (Table 4.7).
Like most other systematic conservation procedures (see review Margules and Pressey, 2000),
this proposed procedure is useful for identifying conservation-worthy
and multivariate.
The framework of complementarity
efficient selection of un-represented
areas because it is flexible
analysis can contribute to assessing the
species for conservation.
The long-term retention of those
species should also be improved by extending this methodology to select by spatial changes in
environmental gradients and associated species.
Existing
Maputaland
protected
areas within the prOVInce are concentrated
region and the central Drakensberg
«
1000 ha) ineffective
reserves.
within the
Escarpment along the Lesotho border.
leaves the other avian communities identified largely un-represented
small
mostly
The traditional
or under-represented
complementarity-based
This
with
algorithms
emphasizing rarity or richness do little to correct this representation bias as they select additional
grid cells in the already sufficiently conserved areas, leaving Central Zululand and the Central-
Figure 4.7: Comparison of algorithm results based on an ideal network, i.e., not taking into
account current protected areas: (a) species rarity-based algorithm; (b) species rarity and beta
diversity algorithm; (c) species richness-based algorithm; and (d) species richness and beta
diversity algorithm.
southern Midlands avian communities largely unnoticed and under protected.
This is due mostly
to the fact that the Maputaland,
regions are highly
East Coast and Drakensberg
Escarpment
species rich, containing >90% of the avian species recorded for the province.
Thus, once these
Table 4.7: Species conservation status and representation selection order based on algorithm type.
Species name
Accipiter ovampensis
Botaurus stellaris
Bubalornis niger
Campethera notata
Chersomanes albofasciata
Ciconia abdimii
Circus macrourus
Crex egregia
Cryptolybia woodwardi
Cursorius rufus
Daption capense
Diomedea melanophris
Eupodotis afraoides
Falco rupicoloides
Falco vespertinus
Gallinula angulata
Glareola nordmanni
Glaucidium capense
Hirundo atrocaerulea
Larus fuscus
Macronectes giganteus
Mirafra apiata
Mirafra cheniana
Mirafra ruddi
Numenius arquata
Oceanites oceanicus
Pachycoccyx audeberti
Pinarocorys nigricans
Podiceps nigricollis
Poicephalus robustus
Prinia flavicans
Procellaria aequinoctia
Serinus atrogularis
Spermestes fringilloide
Spizocorys conirostris
Spizocorys fringillaris
Zoothera gurneyi
Common name
Ovambosparrowhawk
bittern
redbilled buffalo weaver
Knysna woodpecker
spikeheeled lark
Abdim's stork
pallid harrier
African crake
Woodward's barbet
Burchell's courser
pintado petrel
blackbrowed albatross
northern black korhaan
greater kestrel
western redfooted kestrel
lesser moorhen
blackwinged pratincole
barred owl
blue swallow
lesser blackbacked gull
southern giant petrel
clapper lark
melodious lark
Rudd's lark
curlew
Wilson's storm petrel
thickbilled cuckoo
dusky lark
blacknecked grebe
cape parrot
blackchested prinia
whitechinned petrel
blackthroated canary
pied mannikin
pinkbilled lark
Botha's lark
orange thrush
Conservation
status
rare
critically endangered
common resident
globally near threatened
near endemic, common
migrant visitor
globally near threatened
locally common
local endemic, vulnerable
vulnerable, southern Africa
common visitor
common visitor
common resident
common
common migrant
common
globally near threatened
rare
rare, threatened
uncommon
common visitor
near endemic
endemic, threatened
local endemic, critically endangered
common, vulnerable
common
rare
uncommon
common
endemic, endangered
near endemic, common
common
common
rare, indeterminate conservation
near endemic, local nomad
local endemic, endangered
vulnerable, southern Africa
Species
rarity
Species
rarity and
BD
Species
richness
10
12
6
13
1
11
5
8
9
2
8
8
2
1
1
9
3
4
15
8
8
2
3
I
8
8
7
4
1
13
1
8
1
14
1
1
15
14
17
9
18
I
15
10
4
16
7
13
13
7
1
I
5
11
8
3
13
13
7
2
6
13
13
12
8
6
3
6
13
1
18
4
6
3
12
14
9
6
I
13
8
2
7
4
2
2
4
1
1
3
5
10
3
2
2
4
5
1
2
2
11
7
1
3
1
2
1
6
1
1
3
Species
richness
andBD
areas are represented, almost the entire avian diversity within the province is represented and from
a species representation point of view there is no need for additional grid cells.
This makes the investigation and identification
well as, the environmental
of the species community structure, as
gradients associated with that structure an essential component of
conservation area selection procedures. By attempting to protect not only the biodiversity pattern
but also the processes responsible
guaranteeing the representation,
for that pattern, conservation
design may come closer to
as well as the long-term retention of regional biodiversity. The
grid cells selected by the BD algorithm, although similar to those selected by the traditional
algorithm, differ in that some grid cells fall within the under-represented
avian communities,
particularly the highly heterogeneous areas in the Central Zululand and northern Central-southern
Midlands communities.
Both variants of the BD algorithm are able to begin selection in the
16
18
13
9
1
17
II
4
10
8
6
6
8
1
1
5
12
14
2
6
6
8
3
7
6
6
15
10
7
2
7
6
1
9
4
7
2
Central-southern
Midlands and southern Central Zululand then move progressively to the higher
richness areas of the East Coast, Maputaland and northern Drakensberg Escarpment.
In addition to the under-representation
by the traditional reserve-selection
of the avian communities in the province's interior
procedures, it is obvious from the CCA analyses that these
procedures succeed in representing the extremes of the CCA species-environment
gradients. By
focussing on species representation alone, the low lying, moist, hot Maputaland region and high,
wet, cool Drakensberg Escarpment are well represented, but the climatically variable interior midaltitude areas with their unique species assemblages are excluded.
Spatial autocorrelation
analysis proved to be a valuable tool in the identification of areas
of high beta diversity, as opposed to employing simple measures of alpha diversity traditionally
used by reserve-selection
techniques.
Moran's I values for both the identified
altitudinal-
temperature environment gradient of axis 1 and the water balance environment gradient of axis 2
from the CCA analysis (Figure 4.2) enabled the identification
of areas high in beta diversity.
These areas highlighted by low Moran's I values contained very different species assemblages
from their neighboring grid cells, as well as different environmental
these assemblages.
variables associated with
By focussing on grid cells with low levels of spatial autocorrelation,
the BD
algorithm identified areas with highly dissimilar species, and environmental
compositions from
neighboring grid cells in the southern Zululand and northern Central Midlands.
The Tugela River
basin and Central-southern
Midlands are the transition zones for flora and fauna from the
Drakensberg Escarpment and coastal plains (Poynton, 1961) and these dominant river valleys may
represent barriers to avian dispersal (Benson et aI., 1962; Clancey, 1994). They also represent
areas of high species turnover along the identified environmental gradients.
The contrasting
selection orders (Figure 4.5) of the algorithms
illustrate the highly
dissimilar approaches and values assigned to each selected grid cell by the four procedures. The
richness method favours areas of high species richness (Drakensberg Escarpment, East Coast and
then Maputaland
regions)
and the rarity
method
favours
the Drakensberg,
Escarpment,
Maputaland and then East Coast regions. The BD method using richness places emphasis on the
interior regions, as it should, but must pick up the remainder of the required species from the
Drakensberg
Escarpment,
East Coast and Maputaland
regions.
The BD and rarity method
chooses a similar selection order for the interior but re-assigns selection order importance to grid
cells in the Drakensberg Escarpment and Maputaland region. The spatial autocorrelation
employed
here allows for the incorporation
of measures
of beta diversity
method
into what are
traditionally alpha diversity based reserve selection techniques. The results of the present study
illustrate the value of the inclusion of areas with high levels of alpha and beta diversity.
This
investigation also highlights that the avian communities of high protection (Drakensberg
Escarpment and East Coast) are also skewed in their representation along north-south geographic
gradients.
The central Drakensberg Escarpment is adequately protected in the south and the East
Coast protected areas lie in the north.
Both the rarity and richness algorithms for all scenarios
place emphasis on adequately protecting the full length of the Drakensberg
Escarpment and
strategic locations along the coastline.
However, as with any species-based
reserve selection algorithm, problems emanating
from error or particular areas in the available databases are immediately evident.
The grid cell
covering the city of Durban and its harbor contains eight species of Palaearctic
seabirds only
found there because of the fishing trawlers that they follow for food sources (Harrison et aI.,
1997) and the tidal mudflats. Several of these birds are near globally threatened and will require
the conservation authorities to develop appropriate management plans at Durban harbor, which
will not necessarily lead to the declaration of extra coastal reserves, but will require the extensive
restoration of the mudflats and mangroves (Allan et aI., 1999).
Biological
representativeness
should
be used as the first objective
in selecting
conservation worthy areas (Margules, 1986). To date complementary approaches to conservation
have focussed primarily on maximising the conservation of contemporary alpha diversity patterns
using measures of species, habitat richness or rarity (Margules and Pressey, 2000). The present
study shows that the use of principles such as complementarity
always produce adequate biologically meaningful results.
on species data alone does not
Although they represent the required
species efficiently, they do little to address the long-term retention of species diversity through the
conservation
of underlying natural processes and turnover patterns that support this diversity
pattern (Balmford et aI., 1998; Cowling et aI., 1999; Fairbanks and Benn, 2000; Rodrigues et aI.,
2000).
Climatic variables are generally important at coarser scales, whereas disturbance variables
(e.g., management or successional stages), geology, or biotic factors tend to be important at finer
scales.
Of course, decisions on which environmental
variables to include in direct gradient
analysis will largely depend on the scale of the study (Wiens, 1989b).
Nevertheless, by applying
techniques such as CCA it is possible to find what the important environmental variables are, if no
a priori knowledge exists about the possible predictor variables. In this study, the landscapes and
physiographic
basins contain climatic patterns, which interact to limit the species pool.
By
applying methods like CCA and spatial autocorrelation analysis, it is possible to consider all these
environmental
variables and their spatial arrangement in an integrated manner.
Future studies
will however, need to incorporate landscape connectivity (Forman, 1995; Wessels et aI., 2000)
Fairbanks et a1. (1996) presented evidence from Californian floral communities that the
end points of species-environment gradients, where the climate is overly cold, hot, or dry, were
more strongly affected by climate change and therefore more liable to species composition
change. A South African climate change study conducted on invertebrate and vertebrate taxa
estimated that 66% of all species found within the Kruger National Park would have a < 50%
chance of being found there after a doubling of CO2 levels (van Jaarsveld et al., 2000). It is
important to raise the problem of how to preserve communities in a continually changing
environment (White and Bratton, 1980), although fluctuations in natural communities over a
variety of temporal scales are generally accepted (Wiens, 1984). How climate change impacts on
current conservation, is an issue often discussed but rarely applied in conservation planning
(peters and Darling, 1985; Balmford et a1., 1998; Huntley, 1998). Climatic change will certainly
affect bird populations, though its precise effects are difficult to predict (Botkin et a1., 1991;
Furness and Greenwood, 1993). Therefore, although the BD algorithm is less land-use-efficient it
manages to spatially represent the under-represented species, avian communities, and the
identified environmental gradients in the two proposed conservation area networks. It could
therefore be a surrogate for representing potential changes in temporal species assemblages (e.g.,
Rodriguez et a1.,2000).
African conservation agencies are charged with the task of incorporating broader levels of
biodiversity in an integrated manner to maintain systems and services (Maddock and du Plessis,
1999). However, the budgets of public conservation organizations fall far short of being able to
fund the acquisition of all the new reserves the province will require to be truly representative of
the avian biodiversity pattern identified in this study.
Therefore, the development of a
biologically sound logic and methods for identifying conservation areas must not be limited to
identifying a reserve network. This study identified only broad conservation-worthy linkages
among existing protected areas. This is the first of several steps in demarcating areas that could
contribute to longer-term retention of avian diversity outside the formally protected areas
(Armstrong et a1., 2000). Implementation will need to ensure that landowners are amenable to
conservation and that identified areas remain untransformed. In the short-term emphasis should
be placed on identifying critical conservation areas for all the major taxonomic groups, which can
then be included in a comprehensive regional conservation plan, integrating formal reserves and
priority areas in the human-managed matrix.
5.
Human-Ecosystem Co-evolution: Analysis of Bird Diversity and
Structure with Human Land Transformation
But when I consider that the nobler animals have been exterminated
here - the cougar, panther, lynx, wolverine, wolf bear, moose, deer, the
beaver, the turkey, etc., etc. - I cannot but feel as if I lived in a tamed,
and as it were emasculated country ... I listen to a concert in which so
many parts are wanting ... for instance, thinking that I have here the
entire poem, and then to my chagrin, I hear that it is but an imperfect
copy that I possess and have read, that my ancestors have tom out
many of the fIrst leaves and grandest passages ...
Biodiversity is suffering losses at an accelerated rate due to human action and biologists
are increasing
their efforts to understand
this decline and develop appropriate
conservation
responses (Wilson, 1988; Lubchenco et aI., 1991; Soule, 1991; Dale et aI., 1994; Pimm et aI.,
1995; Margules and Pressey, 2000). Although there is little consensus about the most appropriate
response strategy (Mace et aI., 2000), systematic approaches such as complementary
networks,
species richness "hotspots" and gap analysis (Pressey et aI., 1993; Scott et aI., 1993; Williams et
aI., 1996; Mittermeier et aI., 1998; Reid, 1998; Schwartz, 1999; Myers et aI., 2000) all employ
species or community assemblage patterns derived from biological
systematically
interpreted
surveys.
These data are
in a spatially explicit manner to identify "ideal" or "real world"
conservation land-use plans.
An understanding
of the structure of human modification dynamics across a landscape
and its co-relation with species presence and abundance is required.
In contrast to the degree of
human influence on ecosystems, ecologists have concentrated their research on relatively pristine
areas (Cairns,
1988; Lubchenco
et aI., 1991; O'Neill
and Kahn, 2000) and have failed to
incorporate human beings and their institutions as explicit agents in the functioning of ecosystems
(McDonnell and Pickett, 1990; McDonnell et aI, 1993; McDonnell et aI., 1995, Breitburg et aI.,
1998; O'Neill and Kahn, 2000). This focus of research has led to a lack of information on how
land-use affects biological diversity in general and in particular within developing regions of the
world. Landscape ecology, however, and its methods have advanced the farthest in attempting to
understand landscape pattern shaped by humans as explanatory variables to biological pattern and
processes (Turner, 1989; Nevah and Lieberman,
1993; Forman, 1995; Brooker et aI., 1999).
Several studies have looked at the results of human induced changes through pattern development
(i.e., fragmentation)
and linked the pattern to biodiversity dynamics (e.g., Lynch et aI., 1984;
Quinn and Harrison, 1988; Burkey, 1989; Opdam, 1991; Farina, 1997; White et aI., 1997).
Identifying the factors controlling the distribution, abundance, and diversity of species in
ecological communities continues to be a central problem in ecology, with increasing emphasis on
human dimensions of change to explain the patterns, at least partially (Forman, 1995; Lubchenco
et aI., 1991; Turner, 1989; McDonnell and Pickett, 1993). Community structure is considered not
only a product of local physical condition and interactions among species, but also of regional
constraints such as climate, and of historical processes such as dispersal and speciation, migration,
and extinction (Menge and Olson, 1990; Latham and Ricklefs,
1993; Ricklefs and Schulter,
1993). Plant community ecologists have long devoted considerable effort to quantifying local to
landscape-scale
variation in vegetation with recent efforts to quantify broad scale determinants
(Denton and Barnes,
contrast,
1987; Ohmann and Spies, 1998; Fairbanks
animal community
ecologists
and McGwire,
2000).
have spent much effort in understanding
In
local and
increasingly more landscape-scale variation in species and communities (Wiens and Rottenberry,
1981; Opdam et aI., 1984; Cody, 1985; Maurer and Heywood, 1993; McGarigal and McComb,
1995; Villard and Maurer, 1996; Flather, 1996).
regional-scale
descriptions
to provide contexts for interpreting landscape differences (but see
Wiens, 1973; Wiens, 1974; Rotenberry
McComb,
1995; Flather,
Still there are few systematic, quantitative,
1996).
1978; Rotenberry
and Wiens, 1980; McGarigal and
Yet, the collation, examination,
and synthesis of species-
community data in regional analysis has been cited as a major research need for conservation
(Soule and Simberloff, 1986; Balmford and Gaston, 1999).
In an effort to understand avian community temporal dynamics, pattern and scale, this
chapter undertakes a quantitative, systematic analysis of avian species data at the South African
extent followed by a more detailed analysis for the KwaZulu-Natal province.
consider these in a hierarchical
landscape-scale
It is instructive to
fashion, from the broadest to the most localized.
factors are examined
as contributions
to regional
variation
Broad and
in community
composition, and the influences of physical environment, biotic factors, and human disturbance
processes are explored.
The study objectives were to identify and quantify environmental and
landscape pattern factors associated with regional gradients in avian species diversity between the
two survey periods outlined in Chapter
1.
This chapter addresses
questions about factors
controlling avian species assemblages by considering a broader region and a large data set from
two periods,
and uses contemporary
accomplish its task.
multivariate
statistical
and spatial analytical
tools to
Within a local area, the range of climatic conditions is usually small, and most sites fall
within an animal species' physiological tolerances.
Thus, shifts in species' relative abundances
are thought to be associated with local variation in topography, microclimate, vegetation, biotic
interactions, and human impact, as well as with stochastic disturbances that are highly variable
over time and space. Therefore, one can hypothesize that species variation explained by regional
climate decreases, and variation explained by local factors increases, with decreasing geographic
extent.
Different environmental
factors probably assume varying degrees of importance among
localities within a region, and landscape pattern or land cover proportion are likely to dominant in
importance in a similar manner. In KwaZu1u-Nata1, the hypothesis that topographic, temperature
and moisture factors assume the greatest importance for explaining variation in bird diversity at
coarser scales, and landscape pattern and land cover proportion explains the remaining variation
in diversity in a hierarchically scaled manner is explored. In Chapter 4 this concept was explored
on a reduced data set of species targeted for conservation (Table 1.3), with results showing that
the climatic factors and the landscape-vegetation
by limiting species ranges.
complex are interacting to define communities
This chapter continues the examination
in more detail to study
community changes and how they might affect the use of systematic conservation procedures
conducted in chapter four.
Biological
atlases routinely
collation and interpolation
caused
by differences
specialization
procedures
in sampling
1984; McArdle,
and community
appropriate
sampling designs, data
(Prendergast et al., 1993) to minimize sampling errors
effort,
and patchy habitat occupancies
species (Brown,
richness
strive to incorporate
sampling
duration,
the degree
flowing from metapopu1ation
1990; Hanski et al., 1993). Therefore,
assemblage
data derived from contemporary
of ecological
characteristics
of
in practice, species
biological
atlases are
considered adequate (Donald and Fuller, 1998), or at least, the best available basis from which to
conduct conservation planning.
The relationship
between distribution
and abundance
has important implications
for
species richness and conservation planning. The term distribution here is referred to as a species'
geographical
range, though recognizing
that, given a set of observations
with geographical
coordinates, a range boundary may be drawn in different ways (Gaston, 1990). Abundance refers
to the size of a "local" population found in an arbitrarily defined study area. Species richness is a
problematic indicator (Stoms, 1994).
Sampling effort, sampling period, size of sampling area,
and human influence can hugely effect the final species distribution and abundance for a taxon
(e.g., birds) found in a particular area.
inventorying
exercises
Where effort is recorded for species, however, for
over a short time frame, richness
correction
techniques
have been
suggested (e.g., Soberon and Llorente, 1993; Fagan and Karivera, 1997).
Since distribution and abundance may change over time, they should both be estimated
within a sufficiently short period. This is a potentially serious problem for faunal atlasing studies
and national biological surveys, on which estimates of distribution are often based, because such
studies typically accumulate
longer-term
observations
studies of distribution
over a long period.
The results from these larger
are used to develop simple indices of species richness or
diversity that gets used in reserve selection analyses to derive strategic decisions.
The unique
characteristics of atlas studies, however, may allow for the study of species assemblage responses
to degrees of human disturbance (disturbance hypothesis; sensu Connell, 1978; Burel, 1998).
This could allow for a more informed approach to conservation planning.
In addition to viewing human land transformation as a threat to conservation, this chapter
explores the phenomena as a shaping force of species richness or community assemblage patterns,
which can be derived from biological atlases. This chapter examines the interactions between bird
species richness, assemblage characteristics,
and the degree of land transformation
at the South
African (including Lesotho) spatial extent and then a more detailed analysis is conducted for
KwaZulu-Natal.
Three ecological measures were used to characterize
Burel, 1998) per quarter-degree
bird species assemblages
(e.g.,
grid cell: Species richness, Shannon diversity, and an adjusted
evenness measure to reduce sampling effort bias. Species richness (S) is simply the total number
of bird species recorded in a grid cell. Evenness (E) is a relative measure to quantify unequal
species representation
against a hypothetical assemblage in which all which species are equally
common (Krebs, 1989). Evenness was calculated using:
E=
H'-H' rom.
H~ax
-H~in
s
H'
= - LPi
logpi
i=1
where S is the total number of birds speCIes with proportional
(Ludwig and Reynolds,
abundances
PI' P2' ... ,p s
1988). In addition to providing information on presence/absence,
ADD data also provided a simple measure of abundance based on reporting rates.
the
A reporting
rate is the percentage of checklists on which a species was recorded relative to the total number of
checklists
for a grid cell.
This form of relative abundance
from the atlas data has been
quantitatively evaluated previously to correspond well to other field data and deemed fit for use in
general population
studies (Robertson et aI., 1995).
As a result, the relative abundance of a
species was estimated by the ratio of number of observations reported for species i over the total
number of observations
for all species in a grid square (Harrison and Martinez,
1995). The
variables H'rru.n and H'max are the minimum and maximum values of the index respectively.
H'rru.n
= log(N)
- (N-S+1)
N
The
log(N - S + 1)
where N is the total number of species presences observed in a grid square (Harrison et aI., 1997).
The evenness index (E) varies from 0, when a few species are dominant, to 1, when all species are
equally abundant. In general, the total number of species in the sample and sampling effort
(Magurran,
1988) can largely influence the Shannon diversity index and the evenness index.
However, since evenness is measured as a ratio with the number of species in both the numerator
and the denominator, it may effectively cancel the impact of the number of species in the sample,
which is a product of effort (Magurran, 1988; Ludwig and Reynolds, 1988; Krebs, 1989). In this
study, species richness is significantly positively correlated with the Shannon index (r
1844, P < 0.001) and the pre-correction
correlated (r
= 0.91, N =
relationship to evenness was significantly negatively
= -0.62, N = 1844,p < 0.001), but after correcting for sampling effort, the correlation
was negligible (r
= 0.09, N = 1844, P < 0.1).
Two levels of analysis, South Africa and KwaZulu-Natal,
influence of geographic extent on the analysis.
are reported to examine the
Two datasets are reported for the South African
study extent: original species richness (Figure 5.1a) and evenness (Figure 5.lb), and smoothed
measures (e.g., Prendergast
performing neighborhood
and Eversham,
1997; Williams and Gaston, 1998) generated by
averaging (Figure 5.lc, d). Smoothing of the richness and evenness
maps was carried out to further remove the vagaries of sampling and to make larger scale
variation more apparent. For each grid cell, species richness and evenness measure are calculated
94
as the median for the group including the core cell and the surrounding neighborhood
of eight
cells (i.e., equivalent to an area approximately 75 x 75 Ian). Simple median filtering has the effect
of reducing local differences in species richness and evenness. Given a relationship between two
variables, smoothing will tend to reduce the local variance about that relationship, reducing the
impact of outliers. This operation also changes the scale of analysis to provide a regional estimate
of biodiversity.
transformation
The original local scale human disturbance patterns of low and high intensity
were retained (Figure 5.2). The two data sets provide local scale and regional
scale estimates of measured biodiversity co-variation with human disturbance for South Africa.
The assessment of human disturbance on species richness and evenness was conducted at
the South African extent and then by biome (Figure 5.3) to control for expected variation in
species richness and evenness.
This is particularly
the case since high intensity land-use is
concentrated in only a few of the biomes (Figure 5.2b and 5.3). Not controlling for this effect
assumes that all biomes have the same mean species richness and evenness, which is certainly not
true. This was confirmed using analysis of covariance (ANCOV A). Spatially corrected Pearson
correlation coefficients (rs) are calculated between the data sets using the procedure developed by
Dutilleul (1993), which corrects for spatial autocorrelation (using Morans I; Cliff and Ord, 1981)
when calculating the significance (P) of the correlation coefficients.
The procedure does not
change the value of the coefficient but reduces the degrees of freedom and hence alters the p
significance.
This exploratory analysis accounts for the spatial dependence of the data and their
covariance patterns.
For KwaZulu-Natal,
the assessment was conducted on species richness,
Shannon diversity, and evenness measures for the province by life history bird assemblage
(Chapter 1; Table 1.3) from the ADD database.
It is clear that ecologists have a great need to incorporate spatial information into their
analyses. Spatial pattern reflects the underlying structure in the variation of a variable of interest,
and it often provides clues as to possible cause. The development of geostatistics by Matheron
(1963) provides a theoretical and methodological
framework for addressing problems unique to
spatial data. Variables that characterize a spatial property or process generally exhibit a localized
spatial dependence
(i.e., autocorrelation)
within a measurement
field, which satisfies some
reasonable level of stationarity at a more global scale. Geostatistical techniques are based on the
general regionalized
variable model (Cressie, 1993).
The model is a linear model of spatial
process that includes both an explanatory component and a random component.
Geostatistics
embody a set of methods applicable throughout the Earth sciences for investigating
variation and extent in a continuous random variable (Cressie, 1993; Burroughs, 1986).
spatial
a.
Provinces
Species
•
b.
Provinces
Evenness
richness
< 100
0.6125 - 0.7948
100-150
0.7948 - 0.8761
150 - 200
0.8761 - 0.905
•
200>
c.
d.
0.905 - 0974
Provinces
Provinces
Evenness
(filtered)
0.822 - 0.876
100-150
•
150 - 200
200 >
••
0.876 - 0.894
0.894 - 0.908
0.908 - 0.928
Figure 5.1: (a) Bird species richness across South Africa; (b) evenness structure of birds across
South Africa; (c) Bird species richness using smoothed data; and (d) evenness structure using
smoothed data.
Low intensity transformation
<10
•
10-30
•
~-5O
•
ED>
High intensity transformation
< 10
10- 30
30-50
b.
Figure 5.2: The separation between the transformation categories illustrates the spatial
heterogeneity found within South Africa, particularly highlighting the development differences
between: (a) low intensity transformation representing African ex-homeland areas; and (b) high
intensity transformation representing "White" developed South Africa.
Vegetatioo bianes
D
D
Savanna
lIVOOCIand
G"assland
••
Figure 5.3: The vegetation biomes of South Africa based on the map by Low and Rebelo (1996)
and from the original classification work by Rutherford and Westfall (1986).
The central tool of geostatistics is the empirical semivariance illustrated by the variogram.
Variograms can be computed to determine the strength and spatial scale of any pattern, and to
summarize the variation.
Biodiversity measures (e.g., species richness, Shannon diversity and
evenness) and human impact vary spatially and the same qualities apply. Semi-variograms may
be used to examine the spatial variation in species distribution data as a function of ecological and
sampling parameters.
Research has indicated direct ties between measured map components and
variogram form (Burroughs, 1986). The information content of a multi-species diversity map is a
function of both the complexity of the terrestrial area and the spatial and temporal resolution of
the sampling routine.
Multi-species
diversity map variance can be directly related to the
frequency of observation of a species relative to the spatial resolution of the sampling unit and
that this relationship is manifested in both the local and the overall variance of the diversity map
(see Woodcock and Strahler, 1987 for remote sensing parallels).
Recently Pearson and Carroll
(1999) conducted a study of the congruency of species richness scale and extent for comparison
between taxa using a semi-variogram
methodology.
In bird atlasing, analysis scale may vary
from that of individual birds, to communities of birds, to large-scale gradients in bird turnover
over tens or hundreds of kilometers.
The purpose in this chapter is to use semi-variograms
describe and analyse biodiversity measures and human transformation
to
data, and to suggest how
they could be used in a complementary way to predict biodiversity changes in the future.
Variograms compare the similarity between pairs of points a given distance and direction
apart (the lag), and expresses mathematically
the average rate of change of a property with
separating distance, which provides a measure of the form and scale of variation in a variable.
Empirically
derived semi-variograms
patterns in spatial autocorrelation.
increasing distance.
may be fitted with a model, which quantifies observed
Characteristically,
the semi variance tends to increase with
Figure 5.4 provides some examples of idealized semi-variograms
fitted by
the common spherical model, where a minimum nugget variance (c) is found between adjacent
samples and sample variance increases throughout a region of influence (a) beyond which an
asymptote, or sill (co + c), is reached.
The following
is summarized
from Webster
and Oliver (1990).
variogram can provide insight into the structure of the variation.
increases
with increasing
separating
distances.
Interpreting
the
In most instances, the variance
This corresponds
with more or less strong
correlation
or spatial dependence at the shortest distances, which weakens as the separation
increases.
Variograms often flatten when they reach a variance known as the sill variance; they
are bounded (Figure 5.4). Such flattened variograms are second order stationary and suggest that
there are patches
or zones with different levels of species richness
for instance, whereas
unbounded ones suggest continuous change over a region.
The distance at which the 'sill' is
reached, the 'range', marks the limit of spatial dependence.
The variogram often has a positive
intercept on the ordinate, the 'nugget variance'.
This part of the variation cannot be predicted.
Much of it derives from spatially dependent variation within the smallest sampling interval,
somewhat
less from measurement
error and purely random variation.
A completely
flat
variogram, 'pure nugget' , means that there is no spatial dependence in the data.
I
I
I
I
I
I
I
I
I
I
I
C -------------------~------I
nugget
variance
:
:
I
I
---------------~
range
Figure 5.4: Some examples of forms of variograms:
unbounded variogram; and (d) pure nugget variogram.
a
.
(a) and (b) bounded
variograms;
(c)
Empirical semi-variograms were generated for transformation class, species richness, and
evenness for the extent of each biome in South Africa. In KwaZulu-Natal,
generated
for species
richness,
Shannon
diversity,
evenness,
semi-variograms were
transformation
class,
road
disturbance and 1996 population density estimates using the extents of each bird community class
identified from ordination and clustering for each ADD life history bird assemblage.
coordinate of each grid cell was used as the spatial component for all analyses.
were fitted with a spherical model in order to allow quantitative comparisons.
The center
Semi-variograms
The spherical
model was chosen based on visual inspection of plots and because this model's
region of
influence parameter (a) provides a quantitative measure of autocorrelation distance. Other models
were tried, but found to be inadequate.
Visual assessments of the fit of the model to the data were
made and displayed for the national analysis, but due to the large number of estimates that were
required (~344) for the KwaZulu-Natal analysis, only a table of the range value based on the sill
is provided.
Gradient analysis may provide a promising analytical approach to understanding
effects of multiple stressors on ecosystem function (Whittaker,
the
1967; McDonnell and Pickett,
1993; Breitburg et aI., 1998) by integrating multiple stress effects across the landscape.
The
gradient approach relies on the assumption that graduated spatial environmental patterns govern
the structure
and function of ecological
systems.
Changes in population,
community,
or
ecosystem variables along the gradient can then be related to the corresponding spatial variation
in the environmental and socio-economic variables, with specific statistical techniques dependent
upon whether or not environmental variation is ordered sequentially in time or space, and whether
single or multiple variables are being considered.
In the case of system responses to multiple
stressors, complex, nonlinear gradients are apt to be present and ordination techniques may
provide insight into the biotic responses to these gradients (ter Braak and Prentice, 1988; Jongman
et aI., 1995).
Canonical correspondence analysis (CCA; ter Braak and Prentice, 1988), a direct gradient
analysis
method
correspondence
used
widely
in
community
(Palmer,
1993),
and
detrended
analysis (DCA), an indirect gradient analysis method (Gauch, 1982) was used as
the analytical tools.
CCA was chosen because the goal was to better understand environmental
factors associated with avian diversity patterns.
multidimensional
ecology
In CCA, sites and species are arranged in a
space, with the restriction that the ordination axes must be linear combinations
of the specified environmental variables. DCA (using 2nd order polynomial detrending) was used
before CCA to determine the dominant unconstrained avian diversity trends for the first two axes
of variation based only on the species-site matrix.
100
The two axes of variation were then
hierarchically
classified by Euclidean distance using Ward's linkage (Legendre and Legendre,
1998) to identify bird communities
within each life history bird assemblage.
The program
CANOCO, version 4.0 (ter Braak and Smilauer, 1998), was used to conduct all gradient analyses.
Environmental data (e.g., the 11 environmental parameters found under topography and climate in
chapter one, Table 1.1) were entered with the species data using stepwise CCA, with detrending
by 2nd order polynomials to avoid the arch effect. All other CANOCO defaults were used. All
CCA plot scores in this chapter are linear combinations.
In CCA and in this chapter, the fraction
of species variation explained by a set of explanatory variables (total variation explained, TVE) is
the sum of all constrained eigenvalues divided by the total variation (TV) in the species data (or
"total inertia," sensu ter Braak and Smilauer, 1998), which is the sum of all unconstrained
eigenvalues.
The TV is the ratio of the dispersion of the species scores to the dispersion of the
plot scores (ter Braak and Smilauer, 1998), a property of the species-by-plot data matrix.
Data
matrices with greater TV contain many species and little overlap of species occurrence among
plots, and thus higher beta diversities (similar to Whittakers, 1960 beta diversity measure,
Explanatory
variables
~w
=
are added to the model in the order of greatest additional
contribution
to total variation explained, but only if they were significant (p < 0.01), where
significance
was determined by a Monte Carlo permutation
test using 499 permutations
(Ra:
additional influence of variable on avian diversity is not significantly different from random), and
if adding the variable did not cause any variance inflation factors to exceed 20. Variables with
large inflation factors are strongly multi-collinear with other variables and contribute little unique
information to the model (ter Braak and Smilauer, 1998). These were excluded to improve the
interpretability
and parsimony
of the model.
Although several of the explanatory
variables
included in the stepwise models were still intercorrelated, CCA is robust to this multi-collinearity
(Palmer, 1993).
CCA was also performed with variance partitioning (ter Braak, 1988; Borcard et al.,
1992; 0kland and Eilertsen, 1994; Ohmann and Spies, 1998), using partial CCA, to quantify the
relative
contributions
of variable
subsets to explained
variation.
In partial CCA, species
variations associated with explanatory variables that are not of direct interest (i.e., covariables) are
partialled out, in order to examine a selected set of explanatory variables of interest.
The usual
explanatory variables are replaced by the residuals obtained by regressing each of the variables of
interest on the covariables.
Those regional-scale factors (climate and topography) found to be of
importance for each bird assemblage became covariables and then landscape pattern and land
cover proportion factors (chapter one, Table 1.5 and 1.6) were analyzed as local scale operators to
explain the residual variation.
In this way the CCA with variance partitioning
quantified the
relative contributions to explained variation of regional-scale environmental factors vs. landscape
scale spatial land pattern factors.
The CCA results are graphed as a biplot, in which arrow length and position of the
arrowhead indicate the correlation between the explanatory variable and the CCA axes, arrow
direction indicates how the variable is correlated with the CCA axes, and smaller angles between
arrows indicate stronger correlations
between variables (ter Braak and Smilauer,
1998).
A
comparison between CCA and DCA eigenvalue scores showed how well environmental variables
accounted
for variation
in the avian data and suggested
variables were overlooked.
whether
important
environmental
Intraset correlations and the change in eigenvalues for the first and
second axes were evaluated.
Community dynamics are an important area of analysis in ecology and for conservation
efforts. The two survey periods provided by the bird atlases (Cyrus and Robson, 1980; Harrison
et aI., 1997) used in this study allow the opportunity to assess the similarity between the two
sampling periods.
The case provided here for the detection of species association over time has
important ecological implications.
Some human land transformation
processes may result in
continued similar species distribution patterns or impart dissimilar patterns of change. However,
the detection of pattern does not provide a causal understanding of why such a pattern might exist.
Rather, pattern
detection
should ideally lead to the generation
of hypotheses
of possible
underlying causal factors. The study of two data sets involves two distinct components.
The first
is a statistical test (denoted by the X2 distribution) of the hypothesis that the two sampling areas
are similar between times or not at a predetermined probability level (a
measure
of the degree
or strength
characteristics of an association.
of the association.
These
=
0.005). The second is a
are regarded
as separate
The analysis used here relies on the properties of a contingency
matrix (Table 5.1), where two survey dates are compared for each of their sampling units. The
value d, which is joint absences, is usually disregarded in ecology when determining association
by species, but in this study, d is important.
Therefore, instead of using one of the common
binary indexes of association commonly used in ecology, Jaccard or Sorenson, the Kappa index
(K) has been shown to be more sensitive in assessing agreement using the entire matrix (Foody,
1992; Fielding and Bell, 1997).
The Kappa index (K) is typically used to adjust for expected chance occurrences on the
diagonal (Congalton, 1991; Foody, 1992):
K =
E.
Q
-p~
I.Pe
Table 5.1: Setup of a 2 x 2 contingency table used to compare species sampling surveys per
sampling unit. t
Survey 1
Survey 2
Species Present
Species Absent
Species Present
Species Absent
b
d
a+b/N
c+d/N
b+d/N
Pc
tN= a + b + C + d; Po = observed agreement = a + d / N
= Species present in sampling unit during both survey periods.
b = New species are present. Species may possess unique behavioral or physiological adaptations that have allowed it to
colonize locations outside previously recorded range, or sampling unit may have been poorly surveyed originally.
c = May reflect species loss due to human impacts (i.e., fragmentation) or recent survey of sampling unit may have been
poorly sampled.
d = Species absent from sampling unit during both survey periods.
a
Where Po is the observed proportion of agreement, and Pe is the proportion of agreement that may
be expected to occur by chance. Kappa can be calculated from the row and column marginals as:
where Pr is the row marginal error, and Pc is the column marginal error. The result is 0 < 1C < 1,
where zero represents agreement no better than chance, and 1.0 is perfectly agreeing survey data
with all elements on the diagonal.
This implies that a high kappa (> 0.8) is better than the
agreement that would result from a random survey assignment, or a high association between the
two surveys was acquired.
For most purposes values larger than 0.8 represent almost perfect
agreement, those below 0.4 signify poor agreement, and values between 0.4 and 0.8 represent
moderate to substantial agreement.
The analysis of covariance (ANCOV A) determined that there is significant covariation
between species richness and transformation classes while controlling by biome (low intensity, p
< 0.014; high intensity,p < 0.0001; total,p < 0.0001). While, the significance with evenness and
transformation
was not as great across all transformation
transformation (low intensity,p
classes, except for high intensity
= 0.185; high intensity,p < 0.001; total,p = 0.57). These results
confirm that overall correlations between land-use intensity and species richness/evenness
should
be controlled for expected variation in species richness and evenness with biome. Tables 5.2 and
5.3 present the spatially corrected Pearson correlation results considering all of South Africa and
broken down by biome, with values for p < 0.30 highlighted in bold.
As evidenced by the
ANCOV A the national result is largely driven by the differing levels of species richness/evenness
in different biomes, while the high intensity land-use is concentrated in only a few of the biomes.
At the national extent, species richness patterns are positively
correlated with high
intensity human disturbance (Table 5.2). This correlation is weaker for total disturbance, which
also incorporates non-significant
not considering
low intensity disturbance patterns, illustrating the effects when
spatial heterogeneity
in the analysis transformation
variance between the two levels of disturbance; r. = -0.001, N
artificial landscaping
processes (no spatial co-
= 889, P = 0.98). The increased
and an increased number of artificial water impoundments
contained in
several of the biomes across the country can explain this positive relationship between species
richness and human dominated areas, as artificial habitat is created for many species.
Or,
alternatively, the increase in species richness is a product of attracting weedy generalist taxa to
developed
landscapes.
negatively
correlated
In contrast with richness, the evenness of the bird assemblages
with high intensity
transformations
suggesting
is
that bird assemblage
evenness could be adversely affected by high intensity human disturbance.
Table 5.2: Spatially corrected Pearson correlation coefficients (r.) for comparisons of species
richness and evenness against transformation classes among South African grid cells (only cells
with records for all data sets are included). Richness and human disturbance data were square
root and log-transformed before analysis to improve normality.
Analysis level
South Africa (N= 1046, 1405, 1562)
Richness
Evenness
Woodland (N= 521, 499, 582)
Richness
Evenness
Grassland (N = 275, 464, 486)
Richness
Evenness
Shrub Steppe (N = 139, 235, 280)
Richness
Evenness
Succulent Desert (N = 43, 92, 98)
Richness
Evenness
Fynbos (N= 73,115,116)
Richness
Evenness
High intensity
transformation
p
rs
Low intensity
transformation
rs
p
Total disturbance
rs
p
0.004
0.006
0.98
0.90
0.49
-0.16
0.002
0.009
0.38
-0.10
0.07
0.11
0.06
0.13
0.67
0.03
0.43
-0.15
0.003
0.03
0.29
0.04
0.16
0.58
-0.41
-0.03
0.02
0.66
0.39
-0.05
0.03
0.55
0.04
-0.07
0.77
0.26
0.28
-0.003
0.03
0.97
0.41
-0.05
0.009
0.63
0.36
-0.05
0.002
0.57
-0.17
-0.12
0.32
0.50
0.44
0.11
0.000
0.27
0.40
0.13
0.002
0.20
-0.06
0.66
0.85
0.39
-0.18
0.004
0.13
0.37
-0.17
0.006
0.12
-O.oz
Table 5.3: Spatially corrected Pearson correlation coefficients (rs) for comparisons of smoothed
species richness and evenness against transformation classes among South African grid cells (only
cells with records for all data sets are included). Richness and human disturbance data were
square root and log-transformed before analysis to improve normality.
Analysis level
South Africa (N= 1046, 1405, 1562)
Richness
Evenness
Woodland (N= 516, 495, 578)
Richness
Evenness
Grassland (N = 275, 464, 486)
Richness
Evenness
Shrub Steppe (N = 139, 235, 280)
Richness
Evenness
Succulent Desert (N = 43,93,99)
Richness
Evenness
Fynbos (N = 73, 118, 119)
Richness
Evenness
High intensity
transformation
rs
p
Low intensity
transformation
rs
p
Total disturbance
p
rs
0.43
-0.18
0.08
0.08
0.07
-0.04
0.69
0.60
0.51
-0.23
0.005
0.02
0.13
0.09
0.42
0.33
0.47
-0.17
0.009
0.09
0.35
0.008
0.14
0.94
-0.40
-0.11
0.05
0.33
0.42
-0.08
0.05
0.56
0.03
-0.14
0.82
0.17
0.35
0.03
0.04
0.75
0.39
-0.12
0.06
0.37
0.35
-0.Q7
0.02
0.55
-0.04
0.15
0.79
0.43
0.37
0.21
0.003
0.16
0.39
0.27
0.005
0.07
-0.12
-0.12
0.48
0.48
0.47
-0.22
0.003
0.27
0.46
-0.20
0.003
0.27
Vegetation biome analyses highlight bird-vegetation sensitivities (Table 5.2).
Low
intensity disturbance tends to be correlated with bird richness in grassland and shrub steppe
biomes. Low intensity disturbance in grasslands has an inversely negative effect on bird species
richness, but a positive effect in shrub steppe. This could assert that heavily grazed or near to
barren grassland ranges promote reduced richness in grassland related birds, while heavy grazing
in shrub steppe regions opens up the shrub cover, which seems to increase bird richness. Only
woodland bird assemblages are affected by low intensity disturbance, however, in a positive trend
(Table 5.2). This possibly reflects the opening up of woodland areas through heavy grazing and
fuelwood removal, which in turn creates grass patch mosaics favoring grassland birds, lowering
the dominance of woodland species (Harrison et al., 1997). High intensity disturbance tends to
promote increased species richness across all biomes, with a slight emphasis on drier biomes of
woodland, shrub steppe, and succulent desert (Cowling et al., 1997). The woodland and fynbos
biomes have bird assemblage evennesses that reflect an inverse relationship to high intensity
transformation. While for the succulent desert biome the bird assemblage evenness has a positive
effect from high intensity transformation.
The tendency towards single species dominance
reflected in the woodland and fynbos biomes seems to reflect the total removal of vegetation
structure from these areas for replacement by low stature annual and permanent agricultural
landscapes. The birds of these biomes are sensitive to land transformations, with generalist
species tending to dominate (Harrison et al., 1997). The succulent desert biome has low species
richness (Figure 5.1a), and typically, only a few of the species tend to dominate within these
resource strained environments (e.g., Rottenberry and Wiens, 1980; Wiens and Rottenbery, 1981;
Harrison et aI., 1997). The slight positive influence in bird assemblage evenness may reflect the
increase number of species brought about by water impoundments, wells, and tree cover around
human habitation.
Total disturbance in all biomes, with the exception of the grassland biome,
shows increased
species richness.
Transformation
relationship
with woodland
biome bird
assemblage evenness is not significant, and significance is reduced for grassland, succulent desert
and fynbos biome birds (Table 5.2).
The application
analytical
of the data smoothing procedure improves the correlations
extents (Table 5.3).
correlation
between
woodland
One exception
includes the loss of statistically
bird evenness and low intensity transformation.
across all
significant
The other
relationships have mostly reduced significance in p against low intensity disturbance, while p is
increased
for high intensity transformation
and richness but reduced
for evenness.
These
exceptions may reflect instances where the smoothing of biological richness and evenness data
reduced their spatial overlap with transformed areas.
Empirical semi-variograms were generated for each measurement by biome at a distance
resolution of 22.5lan (distance from grid cell midpoint to neighboring grid cell midpoint).
Range
values representing the autocorrelation pattern for each variable by biome are provided in Table
5.4 with the modeled semi-variogram shapes shown in Figure 5.5 (Shannon diversity plots are not
provided).
The richness and Shannon diversity spatial extents tended to provide the same results
as would be expected for the variables because of their high correlation.
this relationship
In grassland and fynbos
does not hold with the Shannon diversity measure reaching a sill, where the
variance measurements remains stable at a range of less than 112.5 lan. This may show that in
the case of these two biomes the Shannon diversity measure is more sensitive to the overall
erosion in bird diversity than species richness especially when compared to the semi-variogram
results for high intensity transformation (HI) and total transformation (TT). The spatial extent of
Shannon diversity for grassland is nested within the influence extent for HI and similar in spatial
extent to IT.
The spatial extents of each measurement with respect to the degree of nestedness of the
spatial phenomena
are important indicators.
desert bird assemblages
Savanna, grassland, shrub steppe, and succulent
have extents greater than the human transformation
processes.
The
fynbos bird spatial variation (richness and evenness) is completely dominated by the human
transformation extent.
Grassland evenness structure is also nested within the human
106
70
60
1000
900
Vi
Ul
CI> 800
Vi
Ul
CI>
c: 55
c:
CI>
>
c:
.<::
CI>
0
c:
'">,
'"
'C
..;
'f
CI>
C/)
500
>,
CI>
C/)
10
a
~
~
c:
200
400
'">,
'f
a
700
350
40
1050
Distance (km)
60
70
//
50
~
~ 40
.~
""
/
~ 30
/
""
Vi
Ul
"
""
20
~
~
c:
/
............•.... ....... ......•...... ............•.......•.......
'
'
0
200
..-,
. MPER
TPER
'
10
a
CI>
c: 60
c:
>
0
/
-.,"
Vi
Ul
500
CI>
.<::
c:
/
/
.•..................
b.
400
~TTOT
CI>
400
~
300
c:
.~
'f 200
'f
>,
'"
CI>
C/)
600
130
260
390
520
30
650
75
Distance (km)
130
60
~ 600
50
CI>
c:
.<::
a:~
0
c: 40
,."
.~
'"
CI>
//
//
'f
• MPER
.........
, TPER
",,/,,/"/
CI>
C/)
~TTOT
''
a
0
200
c:
c: 110
CI>
>
400
~ 100
0
CI>
.~
'"
'::6
0
c:
400
>,
'f
'f
600
130
260
390
520
60 0
650
:g
c: 60
c:
CI>
>
~
~
.~
'"
fl
>,
. MPER
'f
CI>
C/)
~i~6~
c:
'f'">,ij
'f
fl
200
c:
.~
>,
CI>
C/)
00
200
Distance (km)
200
400
30
0
600
90
:ll
60
.~ 50
>, 40
.~ 30
'"
.;~~~~
C/)
~
20
10
TTOT
Ul
200
CI>
c:
c:
!!S 150
~
~
c:
flc:
g? 30
'">,
"~ 100
"~ 25
'"
>,
"f
"f 50
CI>
CI>
C/)
C/)
125
250
375
Distance (km)
500
625
600
Vi
.<::
c:
.!.!
400
35
250
Vi
70
200
Distance (km)
Distance (km)
80
0
50
'"
'f 40
100
CI>
C/)
150
650
CI>
300
0
c:
520
70
Vi
Ul
CI>
390
Vi
Ul
.<::
0
260
130
Distance (km)
400
100
70
Distance (km)
80
50
80
CI>
C/)
Distance (km)
d.
90
'">,
CI>
C/)
10 .,,;;~."
"...... ...''''' ... ''' .... '''' .... .... ,.
"
500
c:
//
>,
Vi
Ul 120
CI>
0
a
50
25
0
Distance (km)
CI>
e.
40
>,
CI>
C/)
70
flc:
50
CI>
0
.~
'"
Distance (km)
c.
1050
700
350
0
Distance (km)
Distance (km)
CI>
C/)
45
CI>
C/)
200
600
50
.~
400
100
a
~
~
c:
600
'f 300
~TTOT
a.
700
.~
'"
~~~~
20
0
a
0
100
200
Distance (km)
300
100
200
300
Distance (km)
Figure 5.5: Model semi-variograms of transformation level, bird richness and community
evenness in South African biomes: (a) savanna woodland; (b) grassland; (c) shrub steppe; (d)
succulent desert; and (e) fynbos. (MPER - Low intensity transformation; TPER - high
intensity transformation; and TTOT - total transformation).
Table 5.4: Spherical model estimates of range (km) lag value derived from the sill when a stable
variance is reached for South African biomes: species richness (S), Shannon diversity (H'),
evenness (E), low intensity transformation (Ll), high intensity transformation (HI), and total
transformation (IT).
S
H'
E
LI
HI
TT
Woodland
1260.0
1260.0
1057.5
45.0
247.5
832.5
Grassland
517.5
112.5
22.5
22.5
472.5
112.5
Shrub Steppe
832.5
832.5
652.5
22.5
562.5
540.0
Succulent Desert
630.0
630.0
630.0
45.0
22.5
22.5
Fynbos
202.5
112.5
337.5
652.5
652.5
652.5
Biome
transformation extent, but the spatial extent of richness is still just greater than the transformation
indices. These geographic-scale results may provide evidence for non-randomly distributed areas
of decline throughout the ranges of birds species as measured by the evenness index, with or
without range contractions.
Dataset results are denoted by the following: CR (Cyrus and Robson, 1980) and ADD
(Harrison et al., 1997).
using presence/absence
Two sets of results comparing the CR and ADD bird census data sets
data are provided, each detailing the life history bird assemblages (chapter
one, Table 1.3). Analyses of the ecological habitat bird assemblages are only conducted from the
ADD survey as they use the relative abundance values of the bird species to drive the ordination
analyses.
For the CR bird assemblages representing
human, eigenvalues
all birds, summer, winter, passerine, breeding, and
and gradient lengths were higher for detrended correspondence
(DCA) than for detrended canonical correspondence
analysis
analysis (DCCA) for the first two axes
(Table 5.5), indicating that a portion of the species variation was not accounted for by the
environmental
variables identified in DCCA.
lengths were derived for the non-passerine,
Moderate to weaker eigenvalues
non-breeding,
and gradient
and non-human bird assemblages.
Gradient lengths derived from DCA for the first axis showed some large species turnover pattern
for all birds, non-passerine,
non-breeding,
and non-human bird assemblages.
The human bird
assemblage had the lowest species gradients. Strong correlations between the DCA for axis 1 and
the explanatory variables suggested that much of the variation in avian diversity is related to the
measured explanatory variables.
Geographic patterns of DCA scores also were quite similar to
DCCA for the first two axes.
The first two axes of significance for DCA were hierarchically
clustered to identify avian community groups for each bird assemblage (Figure 5.6), except in the
case of non-breeding and non-human bird assemblages that had three significant axes.
ADU bird assemblage eigenvalues were also near similar between DCA and DCCA for
the first two axes (Table 5.6) with each bird assemblage showing only a small portion of its
species variation not being accounted for by the explanatory variables identified in DCCA. The
weakest fit was obtained for the non-breeding bird assemblage.
Gradient lengths derived from
DCA for the first axis showed some large species turnover pattern for non-breeding
human influenced bird assemblages.
gradients.
and non-
The human influenced assemblage had the lowest species
In comparison with the CR data all the ADU data sets had lower eigenvalues, TV and
gradient lengths.
Strong correlations between the DCA for axis 1 and the explanatory variables
suggested, however, that much of the variation in avian diversity is related to the measured
explanatory variables.
Geographic patterns of DCA scores were also quite similar to the DCCAs
for the first two axes. The first two axes of significance for DCA were hierarchically clustered to
develop avian communities for each bird assemblage (Figure 5.7).
Results from stepwise CCAs varied across CR bird assemblages and numbers of species
(Table 5.5). Total variation (TV) increased and total variation explained (TVE) decreased, with
increasing gradient length (Table 5.5).
TV, representing beta diversity, was highest for non-
breeding, non-human influenced, winter, and non-passerines.
influenced
(76%), non-breeding
TVE was highest for non-human
(64%), and winter (56%) bird assemblages.
ADU bird
assemblage details are in Table 5.6. TV for ADU bird assemblages was highest for non-breeding
and non-human influenced; however, all bird assemblages from this census period had lower TVs
than for the CR data sets. TVE was highest for non-human influenced (74%) and non-breeding
(51%) bird assemblages.
Unfortunately,
the statistical significance of TVE differences among
CCA models cannot be tested (ter Braak and Smilauer,
1998), and thus interpretations
of
differences are somewhat subjective.
Results of stepwise CCAs for each bird assemblage and census period are presented
below.
between
The analyses of the bird assemblages revealed considerable
environment
differences
underscored
and compositional
the importance
gradients.
variation in associations
These within-region
of ecological modelling
locations' particular biota, physical environment, and history.
approaches
bird assemblage
that consider a
For each bird group, species with
highest and lowest scores on stepwise CCA axes are show in Appendix B.
Table 5.5: Eigenvalues and gradient lengths (1 Standard Deviation) for the first two axes from
DCA and CCA of all bird species groups in KwaZulu-Natal from the Cyrus and Robson (19701979) survey.
Eigenvalue
Bird
assemblage
All
Species
(no.)
TV
TVEt
614
1.974
0.398
Summer
576
1.864
0.491
Winter
591
2.642
0.558
Passerine
260
1.469
0.422
Non-passerine
354
2.295
0.426
Breeding
506
1.538
0.444
Non-breeding
108
4.146
0.640
Human
334
1.154
0.285
Non-Human
280
3.625
0.763
Gradient
length
Axis
DCA
DCCA
DCA
1
2
1
2
1
2
1
2
1
2
1
2
1
2
3
1
2
1
2
3
0.28
0.12
0.27
0.12
0.31
0.15
0.26
0.10
0.35
0.15
0.25
0.10
0.69
0.32
0.14
0.17
0.07
0.49
0.25
0.13
0.23
0.10
0.25
0.10
0.28
0.12
0.24
0.09
0.24
0.13
0.23
0.09
0.33
0.24
0.06
0.15
0.05
0.42
0.22
0.04
2.514
1.742
2.147
1.704
2.384
2.129
2.150
1.481
2.632
1.987
2.111
1.567
3.039
2.821
2.423
1.727
1.426
3.780
2.296
1.835
t Decimal fraction of TV
Table 5.6: Eigenvalues and gradient lengths (1 Standard Deviation) for the first two axes from
DCA and CCA of all bird species groups in KwaZulu-Natal from ADD Bird Atlas (1987-1992)
survey.
Eigenvalue
Bird
assemblage
All
Species
(no.)
600
1.326
0.391
Summer
595
1.501
0.420
Winter
558
1.777
0.464
Passerine
256
1.142
0.358
Non-passerine
344
1.451
0.380
Breeding
501
1.155
0.366
Non-breeding
99
2.567
0.511
Human
335
0.847
0.227
Non-human
265
2.467
0.737
Woodland~
135
0.821
0.391
Forest~
91
0.888
0.412
Thicket~
38
0.951
0.422
Grassland~
137
0.898
0.362
TV
TVEt
Gradient
length
Axis
DCA
DCCA
DCA
1
2
1
2
1
2
1
2
1
2
1
2
I
2
I
2
I
2
1
2
I
2
1
2
1
2
0.21
0.10
0.22
0.10
0.25
0.11
0.23
0.09
0.20
0.09
0.21
0.08
0.34
0.12
0.12
0.06
0.40
0.19
0.22
0.10
0.31
0.08
0.28
0.16
0.21
0.10
0.20
0.08
0.21
0.08
0.23
0.09
0.21
0.07
0.18
0.08
0.19
0.07
0.27
0.08
0.11
0.04
0.38
O.
0.20
0.07
0.29
0.07
0.25
0.13
0.19
0.09
1.958
1.587
2.022
1.736
2.302
1.861
2.204
1.268
1.752
1.582
1.954
1.359
2.691
2.028
1.357
1.332
3.587
2.242
1.981
1.859
2.531
1.792
2.020
2.195
2.091
1.571
Decimal fraction of TV.
~The following number of samples were used: woodland, 71; forest, 114; thicket, 162; and grassland, 162.
Results are derived from relative abundance values ofthe birds instead of presence/absence.
f
110
Drakcllsbcrg
Drakensberg
Midlands
Midlands
Passerine
Nn-passerine
Mip>ollnd
MipuUlbnd
Mlputaland
East (nast
East Coast
East Coast
[)-dlcell5belg Escarp
,,",.-.I;
O"akensberg
Escarp
Catr.i1Z<Wllnd
CentralZ11Juland
Mdlands
Nm-breeding
II
II
o
o
Woodland
East Coast
Wet grassland
Dry Grassbnd
Figure 5.6: Assemblage classifications derived from ordination analysis of the CR life history bird
datasets.
&mrrer
MiJll~b.nd
MiJll~hnd
Blst O>ast
East Coast
Q-dkensbcrg
Q-dkcnsbcrg
CatrdlZubland
CamlZuhbnd
M:!Jallds
M:!JaRl;
Drakcllsbcrg
Mdlands
Passerre
•
E!!!
D
o
o
~
M<po"land
M<po"land
Fost Cl>aSl
East Coast
frili",belg
Ihb:Ribcrg
Drakensberg
C<waIZWbnd
M:!JaRl;
Hmm i1fu::nce
Wctgrasslands
Dry grasslands
C<waIZWland
Midlands
MlaRl;
NnhllTllllilllum:e
•
M<po"land
•
M<po"land
E:!ll
Fosl Cl>aSl
mm
o
o
o
[I
East Coast
o
o
o
Dakcnsbcrg
iigh grassBIDs
CcntTdlZlWtand
eoo.taIlilUcrl~x1
M:!JaRl;
Higlm:ls
frdkcnsbcrg
Figure 5.7: Assemblage classifications derived from ordination analysis of the ADD life history
bird datasets.
All birds.-The
dominant compositional gradient (CCA axis 1) reflected a gradient in elevation
and topographic
heterogeneity
from the humid Maputaland
plain and northern coast to the
temperate montane climate of the Drakensberg (Figures 5.8a and 5.9a) with a TVE of29% (Table
5.7). Grid cells with the lowest scores on axis 1 were at higher elevations and experienced higher
seasonal variability
rainfall.
in temperature,
colder maximum and minimum temperatures,
and higher
These plots were concentrated along the length of the Drakensberg Escarpment (Figure
5.8a) within the high grassland zone.
elevations and experienced
minimum
temperatures,
Grid cells with high score on axis 1 were at lower
lower seasonal variability in temperature,
and higher evapotranspiration.
warmer maximum and
The highest grid cell scores were
concentrated along the coast from Durban north and encompassing the Maputaland plain, with the
highest score situated over Ndumo nature reserve (Figure 5.8a).
High-scoring
grid cells fell
largely within the moist and arid woodland zones (Chapter 1, Figure 1.3). With minor exceptions,
the axis 1 gradient was longitudinal from the coast to the Drakensberg escarpment, reflecting the
strong climatic influence of the Indian Ocean, the pronounced sharp rise in elevation from the
coast, and the generally north-south orientation of the Drakensberg escarpment.
The second CCA
axis was a gradient in growing season moisture stress, from areas of warm, dry growing seasons
to areas of humid, wet growing seasons along the coast (Figures 5.8a and 5.9a) with a TVE of
12%. Areas of low summer precipitation, high evapotranspiration
and high summer temperature
included the interior valleys on the western side of the Lebomo Mountains, especially the Pongola
and Mfolozi
River valleys, and the Tugela River valley.
Lowest grid cell scores were
concentrated in these areas, and tended to be in arid and mixed woodland and thicket (Chapter 1,
Figure 1.3). Highest grid cell scores on axis 2 were situated at coastal river mouths, wetlands and
bays.
These areas included Durban Bay, Richards Bay, Kosi Bay, St. Lucia wetland, Mfolozi
River mouth, Tugela River mouth and Mvoti River mouth (Chapter 1, Figure 1.2). The species at
these points reflect the coastal wetland environment (Appendix B).
The ADD data set had a similar pattern and gradient on axis 1 (Figure 5.10a and 5.11a)
with the addition of temperature variables and the seasonality in precipitation
(TVE of 27.5%;
Table 5.8). The spatial pattern of the low scores along the Drakensberg Escarpment were not as
wide, but rather shrunk along the escarpment edge compared to the CR data set. High values
covered the Maputaland plain and coast as far down as Durban Bay.
The second axis was also
similar to the CR result and identified variables, however, with the addition of seasonal variability
in evapotranspiration
(TVE of 10.5%). High values however were calculated down the coast to
cover the region from Richards Bay to the southern border of the province.
Table 5.7: Increases in total variation explained (TVE) by explanatory variables in stepwise
canonical correspondence analysis of CR bird species, by group type; the three greatest
contributors to TVE in each group type are show in boldface. t
Additional variation explained (proportion ofTVE)
Variable
All
Summer
Winter
Passerine
Nonpasserine
Breeding
Nonbreeding
Human
Nonhuman
0.23
0.05
0.23
0.04
0.26
0.05
0.22
0.03
0.22
0.05
0.22
0.04
0.27
0.13
0.14
0.03
0.42
0.08
Topography
DEMMEAN
DEMSTD
Climate
GDMEAN
0.10
0.09
0.11
0.08
0.11
0.08
t
§
0.21
MAP
t
§
§
§
t
§
t
0.02
§
GTMEAN
t
t
t
§
t
0.03
0.23
§
t
NGTMEAN
t
t
t
t
t
§
t
§
t
MAT
t
t
t
t
t
§
t
t
t
HOTMNTHMN
t
t
t
t
t
§
§
t
t
MINMNTHMN
t
§
t
§
t
§
t
t
t
EV ANNMN
0.03
0.02
0.03
0.02
0.04
0.02
t
0.05
0.06
PSEAS_MN
t
0.01
t
0.01
t
§
t
0.01
t
TSEAS_MN
t
0.04
0.03
0.04
t
0.02
t
0.01
t
MXSEAS_MN
t
0.03
0.04
0.02
t
0.02
t
0.01
t
EVSEAS MN
t
0.01
0.02
t
t
0.01
t
0.01
t
f Increase in TVE is additional species variation explained by adding the variable after previously selected variables already are
included, expressed as a proportion of TVE, and thus reflects selection order. Values are for variables included by forward
selection (P < 0.0 I, where significance was determined by a 499 iteration Monte Carlo permutation test, Ho: additional influence
of variable on vegetation is not significantly different from random), and where adding the variable did not result in inflation
factors> 20.
t Variable was not significant in the stepwise procedure.
§ Variable was significant in the stepwise procedure but excluded because ofmulti-collinearity.
Table 5.8: Increases in total variation explained (TVE) by explanatory variables in stepwise
canonical correspondence analysis of ADD bird species, by functional type; the three greatest
contributors to TVE in each group type are show in boldface. t
Additional variation explained (proportion ofTVE)
Variable
All
Summer
Winter
Passerine
Nonpasserine
Breeding
Nonbreeding
Human
Nonhuman
0.19
0.02
0.20
0.03
0.22
0.03
§
0.02
0.17
0.02
§
0.02
0.19
0.03
0.11
0.02
0.37
0.03
Topography
DEMMEAN
DEMSTD
Climate
GDMEAN
0.07
0.07
0.08
0.03
0.07
0.03
0.13
0.01
0.15
MAP
t
§
§
§
§
§
t
§
§
GTMEAN
§
t
§
§
§
§
t
§
§
NGTMEAN
t
t
§
§
t
§
t
§
t
MAT
t
§
§
§
t
§
t
§
§
HOTMNTHMN
t
§
§
§
t
§
t
§
t
MINMNTHMN
§
§
§
0.20
§
0.18
§
§
§
EV ANNMN
0.03
0.03
0.02
0.07
0.03
0.06
0.04
0.04
0.06
PSEAS_MN
0.01
0.01
0.02
0.02
0.01
t
t
0.01
0.03
TSEAS_MN
0.03
0.03
0.04
§
0.02
0.02
0.02
0.01
0.05
MXSEAS_MN
0.02
0.02
0.04
0.02
0.03
0.02
0.05
0.01
0.05
EVSEAS MN
0.01
0.01
t
t
0.01
0.01
0.04
0.01
t
t Increase in TVE is additional species variation explained by adding the variable after previously selected variables already are
included, expressed as a proportion of TVE, and thus reflects selection order. Values are for variables included by forward
selection (P < 0.01, where significance was determined by a 499 iteration Monte Carlo permutation test, Ho: additional influence
of variable on vegetation is not significantly different from random), and where adding the variable did not result in inflation
factors> 20.
t Variable was not significant in the stepwise procedure.
§ Variable was significant in the stepwise procedure but excluded because ofmulti-collinearity.
Figure 5.8: Patterns of variation in the first two axes of variation derived from detrended
correspondence analysis (DCA) for each CR life history bird group, KwaZulu-Natal: (a) all birds;
(b) summer; (c) winter; (d) passerine; (e) non-passerine; (f) breeding; (g) non-breeding; (h)
human; and (i) non-human. (Figure continued on next page).
Figure 5.8: Continued.
Summer.- The first axis on the CR data set was strongly correlated with elevation and
seasonality
in precipitation
and moderately
correlated
seasonality in temperature and evapotranspiration
with topographic
heterogeneity
and
(Figure 5.9b) with a TVE of 32% (Table 5.7).
Low grid cell scores were on cooler, high-elevation sites along the Drakensberg Escarpment, and
high scores were on warmer, low-elevation sites along the coast from Durban in the south to the
Maputaland plain in the north (Figure 5.8b). The second axis was a gradient in growing season
moisture stress and low variability in temperatures, from areas of warm, dry growing seasons to
areas of humid, wet growing seasons along the coast (Figures 5.8b and 5.9b) with TVE of 13%.
Low scores were in areas with arid and mixed woodland representing hot, moist summers along
the Lebombo Mountains and the Tugela, Buffalo and White Umfolozi River valleys, and high
scores were situated at coastal river mouths, wetlands, bays and south coast. These areas included
Durban Bay, Richards Bay, Kosi Bay, St. Lucia wetland, Mfolozi River, Tugela River, Mkomazi
River, and Mzimkhulu River mouths.
Figure 5.9: Biplots from canonical correspondence analysis of life history bird assemblages. All
axes have been resealed to range from -1.0 to 1.0. Axes for explanatory environmental variables
that were not significant or that had very low correlations with the canonical axes are not shown.
The first axis results on the ADD data set correlate to similar variables (Figures 5.l0b and
5.11b) with slight improvements in the relationships, but a lower TVE of27.5%
(Table 5.8). An
arbitrary flip in scores by CANOCO derived low scores in the Maputaland plain and northern
coast and high scoring grid cells in the Drakensberg Escarpment.
The score calculations (between
negative and positive values) are known to be calculated arbitrarily in CANOCO and have no
influence on the results or for comparisons between data sets (ter Braak and Smilauer, 1998).
Axis 2 was also related to water balance and seasonal variability in evapotranspiration
with a
similar TVE of 12.5%. Low and high scoring grid cells were in similar areas, but with reduced
emphasis on high scoring cells along the north coast.
Figure 5.10: Patterns of variation in the first two axes of variation derived from detrended
correspondence analysis (DCA) for each ADD life history bird group.
MXSEAS_MN
TSEAS_MN
Winter
EVSEAS_MN
EVANNMN
EVANNMN
EVSEAS_MN
Figure 5.11: Biplots from canonical correspondence analysis of life history bird assemblages. All
axes have been resealed to range from -1.0 to 1.0. Axes for explanatory environmental variables
that were not significant or that had very low correlations with the canonical axes are not shown.
120
EVANNMN
EVSEAS_MN
Winter.-The first axis for the winter bird assemblage followed a similar gradient as for
summer birds, an elevation gradient but with emphasis on seasonality in temperature (Figure 5.8c
and 5.9c) with TVE of 29.5% (Table 5.7). The second axis was also similar to summer birds,
where water balance and evapotranspiration
represented the gradient, but with the addition of
topographic heterogeneity and a TVE of 25%. The low scoring grid cells were in similar areas,
however the high scoring grid cells were largely on the south coast and confined to Durban and
Richards Bays and the Tugela River mouth.
The winter ADD data set of birds had arbitrary
compositional
sconng
flips on both axes of
variation (Figure 5.1 Oc). Axis 1 was similar in pattern and variable selection
(Figure 5.11 c) with the added variable of seasonal variability in precipitation and a higher TVE of
33% (Table 5.8).
Axis two had a much lower TVE of 12% with similar water balance and
temperature variables selected to explain the variation.
Passerine.- The first axis gradient for the CR passerine bird assemblage was strongly
correlated
with elevation
and seasonality
in precipitation,
and moderately
correlated
with
topographic heterogeneity and maximum temperature (Figure 5.8d and 5.9d) with TVE of 25%
(Table 5.7). Low scores were located in the high-elevation zone of the Drakensberg Escarpment,
and high scores were located tightly covering the Maputaland plain and far north coast. Axis 2
was strongly associated with water balance, evapotranspiration
and seasonality in temperature
(TVE of 11%), with low scoring grid cells in the arid and mixed woodlands along the Pongola,
Black and White Umfolozi, Buffalo and Tugela River valleys.
High scoring plots were located
either along the coast as far north as the St. Lucia wetland complex extending to the southern
coast.
Several of the grid cells were located in the southern interior comprising afromontane
forest areas and high Drakensberg Escarpment grassland.
ADU axis 1 compositional
minimum temperature
variation was similar (Figure 5.1 Od and 5.11 d), but with
and seasonality in precipitation
slightly lower at 25% (Table 5.8).
explaining the axis, however TVE was
The low and high scores were similar in location but low
scores were tighter against the Drakensberg Escarpment.
The second axis of variation had a flip
in scores, but had similar variables explain the variation with a TVE of 11%. Low scoring grid
cells were more clustered on the south coast than for the CR data set.
Non-passerine.- The dominant compositional gradient within the CR non-passerine bird
assemblage was strongly associated with elevation and topographic heterogeneity
(Figure 5.8e
and 5.ge) with a TVE of 27% (Table 5.7). The pattern for axis 1 and environmental variables
were similar to that derived using all birds. Axis 2 also related to similar variables, water balance
and evapotranspiration,
and elucidated similar pattern with a TVE of 15%. Exceptions to this
pattern, are the confinement of high scores to only Durban and Richards Bays and the Tugela
River mouth.
The non-passerine ADU data set required several more variables to explain the first axis
of variation (Figure 5.11e). These were mostly variables describing seasonality in temperature
and precipitation
and TVE was mostly similar at 25% (Table 5.8). The low scoring grid cells
were in similar areas along the Drakensberg Escarpment and high grasslands, but the high scoring
areas were located along the coast from Kosi Bay to Durban Bay with clustering around the St.
Lucia wetland (Figure 5.10e). Axis 2 contained similar variables with a slightly smaller TVE of
11 %. Low scores were more northerly in the arid woodland and reduced in the mixed woodland
and Tugela River valley. High scoring grid cells were confined to the coast from Durban Bay
southwards.
Breeding.-For breeding birds, axis 1 followed a gradient similar to that of the summer
bird groups but with emphasis on temperature (Figure 5.8f and 5.9f): elevation, summer
temperature, maximum temperature, and seasonal temperature variability contributed 34% of
TVE (Table 5.7). Distribution oflow and high grid cell scores was similar to that of the summer
bird group. Axis 2 represented a water balance and evapotranspiration gradient related to the
seasonality in temperature variability with a TVE of 10%. Grid cell scores and pattern were
similar to axis 2 of the summer bird assemblage, but with greater emphasis on the differences
between bird compositions of the interior arid and mixed woodlands and the coast.
The ADD data set of birds had arbitrary scoring flips on both axes of compositional
variation (Figure 5.l0f and 5.l1f). Axis 1 was similar in pattern, but temperature variables largely
explained this axis, but with a much lower TVE of 22% (Table 5.8). Axis two had a slightly
higher TVE of 12% with similar water balance and seasonal temperature variability to explain the
variation.
Non-breeding.- The non-breeding bird assemblage was much more variable than the
other groups (Table 5.5). The gradients were not as well defined (Figures 5.8g and 5.9g), with
axis 1 only moderately associated to elevation and summer temperature (TVE of 36%; Table 5.7).
The low scoring cells were located along the Drakensberg Escarpment and in un-developed
grassland areas within the Midlands and Central Zululand areas, and high scores were located in
the Maputaland plain, Pongola River valley and the central coastal and Pietermaritzburg-Durban
economic corridor. This pattern alignment with human dominated areas appears to relate to the
summer visiting behaviour of Eurasian birds.
The second axis associated moderately with
elevation and topographic heterogeneity and a TVE of 14%. Low scores were undifferentiated,
and high scores located along the coast. Axis 3 was less interpretable than the first two axes.
Topographic heterogeneity and temperature were the correlates. Low scoring grid cells were on
the river valleys and high scoring grid cells were in the Drakensberg Escarpment and foothills.
The non-breeding birds in the ADD data set were much less variable, yielding only two
significant axes of variation (Figure 5.1Og). Axis 1 required many more variables to explain the
compositional variation (Figure 5.11g) with inclusions of topographic heterogeneity, and seasonal
variabilities in temperature. The TVE was, however, higher at 36% (Table 5.8). Pattern was
more interpretable with low scoring grid cells along the entire coast and high scores along the
Tugela, Buffalo, Mzimkhulu, and Mooi River valleys. Axis 2 was associated with water balance
and evapotranspiration
undifferentiated,
variables, but with a much lower TVE of 14%.
The low scores were
while the high scoring areas were in the arid woodland zone.
Human.-Axis 1 (Figures 5.8h and 5.9h) was correlated with elevation, seasonality in
precipitation, topographic heterogeneity, and temperature (TVE of 19.5%; Table 5.7). Low scores
were situated along the Drakensberg Escarpment and high scores were in Maputaland and along
the coast.
The second axis associated with evapotranspiration,
mean annual precipitation,
and
seasonal temperature variability (TVE of 7.5%). Low scores were situated along the south coast
and Midlands and high scores were in the arid and mixed woodland regions of the Zulu homeland.
The ADD data set for axis 1 was similar in pattern and explanatory variables (Figures
5.10h and 5.11h) but with a slightly lower TVE of 15.5% (Table 5.8).
Axis 2 also required
similar explanatory variables with a similar TVE of 6.5%, but the spatial pattern in allocated
scores was different.
Low scoring grid cells were located along the coast south of Sodwana Bay,
the Midlands within the Pietermartizburg-Durban
Newcastle.
economic corridor, and along the main roads to
The high scoring grid cells reflected these changes by becoming more constricted to
the arid and mixed woodlands of Central Zululand, with a reduced presence in the Maputaland
plain.
Non-human.- The non-human influenced bird assemblage was much more variable and
elucidated three axes of variation (Figure 5.8i). The major compositional gradient on axis 1 was
elevation and topographic heterogeneity (Figure 5.9i) with a large TVE of 50% (Table 5.7). Low
scoring grid cells were located in the Drakensberg
Escarpment
and high scoring areas were
situated along the coast from Durban north. These areas tended to be characterized by bays, river
mouths, or large wetlands.
Axis 2 was associated with water balance and evapotranspiration
and
a TVE of 27%. Low scores were in the arid and mixed woodland and thicket regions and high
scores were on the coast at Durban and Richards Bays and the Tugela and Mfolozi River mouths.
Axis 3 was moderately related to topographic heterogeneity
scores throughout
the Midlands
and evapotranspiration
and high scores in the southern Drakensberg
northern Drakensberg Escarpment near Newcastle and Wakkerstroom,
with low
Escarpment,
and along the coast from
Richards Bay north.
The non-human influenced birds in the ADD data set were much less variable, yielding
only two significant axes of variation (Figure 5.1 Oi). Axis 1 required many more variables to
explain the compositional variation with inclusions of seasonal variabilities in precipitation and
temperature (Figure 5.1li).
The TVE was, however, similar at 50.5% (Table 5.8). The spatial
pattern was more interpretable with low scoring grid cells along the coast from Durban Bay to the
Maputaland plain, and high scores were located along the Drakensberg Escarpment.
Axis 2 was
associated with water balance and evapotranspiration variables, with a slightly lower TVE of
23.5%. The low scores were in a much more confined area in the arid woodland and along the
thickets in the river valleys. High scoring grid cells were situated along the coast from Richards
Bay south, with one exception at Sodwana Bay.
The relative contributions of explanatory variables to TVE in stepwise and partial CCA
were influenced by location and illustrated the nested scales in processes, coarse to fine scale, that
are required to explain variation within the life history species assemblages (Table 5.9). In partial
CCA conducted on the ADU life history data sets, regional factors (climate and topography)
accounted for more of the TVE (22-74%) than the landscape factors (8.6-32%) for all the bird
assemblages (Table 5.9). Landscape factors contributed less to TVE for the human influenced
(8.56%; Figure 5.12h), passerine (10.2%; Figure 5.l2d) and breeding (11.1%; Figure 5.l2f) bird
assemblages. Contributions oflandscape factors to TVE were greater for the non-breeding (32%;
Figure 5.12g), non-human influenced (24.7%; Figure 5.12i), non-passerine (19.5%; Figure 5.12e),
summer (18.4%; Figure 5.12b), all birds (15.4%; Figure 5.12a), and winter (14%; Figure 5.l2c)
bird assemblages. Passerine and breeding birds were the only assemblages not related to the
richness and density of LCLU classes. The life history bird assemblages were mostly related to
LCLU proportions than to the landscape mosaic pattern metrics. Passerine, breeding and nonhuman bird assemblages were related to the amount of woodland and forest coverage. While all,
summer, winter, breeding and human bird assemblages were related to the amount of grassland
coverage and extent of subsistence agriculture and degraded lands. Passerine, non-breeding, and
human bird assemblage variations had the greatest relationships with human built landscapes,
which included extent urbanized and road disturbance.
In order to resolve finer resolution in the power of landscape metrics to explain avian
diversity patterns analysis was conducted on birds grouped by associated primary vegetation
habitat (Table 1.3). These ecological habitat bird groups were ordinated using the relative
abundance of each bird rather than presence/absence in order to understand the land-cover class
patch characteristics relationship to bird population variation. This analysis was conducted to
overcome any confusions associated with landscape mosaic measurements, which include all
LCLU classes, and to come closer to an ecological explanation of bird reactions to landscape
pattern that would be masked using life history assemblages. Figure 5.13 illustrates the dominant
gradients derived from DCA analysis on the relative abundance of the birds group by primary
associated habitat classes. Table 5.10 presents CCA results of the bird habitat groups in relation to
the environmental variables chosen in the stepwise CCA to explain the ecological habitat birds.
Table 5.9: Proportion of total variation explained (TVE) by landscape variables while constrained
by the topography and climate variables chosen for each group type in partial canonical
correspondence analyses (CCAs) of ADD bird species; the three greatest landscape contributors
to remaining TVE after constraining by the topography and climate variables in each group type
are show in boldface. t
Additional variation explained (proportion ofTVE explained after constraining
NonNonSummer
Winter
Passerine
Breeding
breeding
passerine
1.081
1.313
1.071
0.934
2.056
0.784
0.790
All
Partial CCA TV
Variable
Landcover
POPTOT96
POPDEN96
FOR]ER
GRS]ER
WET]ER
LOWI]ER
PLNT_PER
DRY]ER
IRR]ER
URB]ER
M]ER
T]ER
T_TOTAL
ROAD_INDEX
§
0.01
t
t
0.01
0.01
0.015
0.014
0.02
t
t
t
t
t
0.02
t
t
t
0.014
0.01
0.015
0.013
0.01
t
t
t
0.011
0.013
t
t
0.01
§
§
§
§
0.012
t
t
t
t
t
t
0.02
t
t
t
0.01
0.014
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
§
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
0.012
0.01
0.012
0.03
0.Ql5
0.014
0.03
0.01
§
§
§
t
t
t
§
0.02
§
§
§
0.02
0.012
t
0.01
0.012
0.01
0.013
0.02
t
0.012
t
t
t
t
t
0.011
t
0.02
t
t
t
t
t
t
0.02
0.011
0.007
0.012
0.011
t
t
t
t
t
t
t
0.02
t
t
0.02
0.03
0.02
0.02
0.02
by covariables)
0.619
Nonhuman
1.730
§
0.005
t
t
t
0.03
Human
0.01
0.006
0.013
0.008
t
0.023
t
0.022
t
t
t
t
t
t
t
0.05
0.015
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
0.02
t
t
Patchiness
LPI
NP
PO
MPS
PSSO
CI
§
§
§
§
t
t
t
t
t
t
t
0.008
0.01
t
t
t
t
t
t
t
t
t
t
0.01
t
t
t
0.02
t
t
Shape
MSI
AWMSI
FD
MPFD
AWMPFD
t
t
t
§
0.01
t
§
0.01
0.011
t
0.007
0.007
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
t
0.012
0.008
t
t
t
t
0.006
t
t
t
t
0.014
0.04
0.01
t
0.03
0.11
0.01
0.007
0.02
0.06
§
§
§
§
t
t
t
t
t
t
t
t
t
t
t
t
0.006
0.03
t
t
Interior
MCAPP
PCASD
MAPDC
DCASD
DCACV
t
t
t
t
0.01
0.Ql5
0.022
t
t
Isolation
MNND
NNSD
MPI
11
t
t
t
t
Richness
CR
CRD
0.04
t
Heterogeneity
SHDI
SDI
MSDI
t
t
t
§
t
Evenness
SHEI
§
§
0.013
§
t
t
0.Ql5
SEI
0.01
§
0.01
0.01
t
MSEI
§
§
§
t
t
t
t Same rules as for Table 5.8.
I Variable was not significant in the stepwise procedure.
§ Variable was significant in the stepwise procedure but excluded because ofmulti-collinearity.
126
CRD
d.
C.
Axis 2
Road_index
Axis 1
Foryer
Myer
Axis 1
SOl
II
Plntyer
Winter
Passerine
Figure 5.12: Bip10ts from canonical correspondence analysis of life history bird assemblages. All
axes have been resealed to range from -1.0 to 1.0. Axes for explanatory landscape variables that
were not significant or that had very low correlations with the canonical axes are not shown.
Axis 1
CRD
Figure 5.13: Patterns of variation in the first two axes of variation derived from detrended
correspondence analysis (DCA) for each ADD ecological habitat bird group. Areas with no
coverage of the respective vegetation class are depicted in white.
Table 5.10: Increases in total variation explained (TVE) by explanatory variables in stepwise
canonical correspondence analysis of ADD bird species, by ecological type; the three greatest
contributors to TVE in each group type are show in boldface. t
Variable
Woodland
Forest
Thicket
Grassland
§
§
0.02
0.13
0.02
0.19
0.03
Topography
DEMMEAN
DEMSTD
t
Climate
GDMEAN
0.03
0.04
0.04
0.02
MAP
t
§
§
§
GTMEAN
0.20
§
§
§
NGTMEAN
§
§
§
§
MAT
t
§
§
§
HOTMNTHMN
§
§
§
§
MINMNTHMN
§
0.28
§
§
EVANNMN
0.10
0.07
0.21
0.08
PSEAS_MN
0.02
t
0.02
0.02
TSEAS_MN
§
t
§
§
MXSEAS_MN
§
t
0.01
0.02
EVSEAS MN
0.02
t
t
0.01
t Increase in TVE is additional species variation explained by adding the variable after previously selected variables already are
included, expressed as a proportion of TVE, and thus reflects selection order. Values are for variables included by forward
selection (P < 0.01, where significance was determined by a 499 iteration Monte Carlo permutation test, Ho: additional influence
of variable on vegetation is not significantly different from random), and where adding the variable did not result in inflation
factors > 20.
I Variable was not significant in the stepwise procedure.
§ Variable was significant in the stepwise procedure but excluded because ofmulti-collinearity.
Woodland bird variation was related to temperature and precipitation
seasonality on the
first axis, with higher temperatures in Maputaland and stronger seasonality in precipitation in the
Tugela and Buffalo River basins. The second axis gradient contrasted the northern coast versus
the inland low-lying Tugela, Pongola, Mkuze and Mhalutze
River valleys.
illustrated an aridity gradient represented by growth days and evapotranspiration.
The variables
The forest
bird's first axis gradient depicted a trend from the Drakensberg escarpment and southern midlands
to the flat and warmer coastal plains of Maputaland, appropriately
elevation heterogeneity and
mean minimum temperature of the coldest month described the axis. The second axis of variation
described the lower moisture regimes in the interior of the province and Maputaland from the wet
southern coast and Drakensberg escarpment, with evapotranspiration
and growth days explaining
the pattern. The variation in thicket birds illustrated a trend from Maputaland and Zululand to the
Drakensberg escarpment, which was explained by evapotranspiration
was related to elevation, elevation heterogeneity,
and growth days. Axis two
seasonality in maximum temperature
and a
further contribution by growth days. The variation in grassland birds was the most complicated to
explain with the available environmental variables, which lead to seven variables being chosen.
Axis one depicted a gradient in elevation and seasonality in maximum temperature from the coast
and Maputaland to the Drakensberg escarpment.
Axis two was mostly represented by a moisture
gradient between Tugela and Buffalo River basin and Maputaland grassland birds and birds in the
Midlands, south coast and high central Drakensberg escarpment.
130
GTMEAN
Axis 1
PSEAS_MN
Axis 1
DEMsm
MINMNTIIMN
EVANNMN
EVSEAS_MN
Woodland
a.
Figure 5.14:
assemblages.
environmental
canonical axes
b.
GDMEAN
Forest
Biplots from canonical correspondence analysis of ecological habitat bird
All axes have been resealed to range from -1.0 to 1.0. Axes for explanatory
variables that were not significant or that had very low correlations with the
are not shown.
In partial CCA conducted
(climate and topography)
on the ecological habitat bird data sets, regional factors
accounted
for the following
in TVE: Woodland
(47.6%);
forest
(46.4%); thicket (44.3%); and grassland (40.3%), with forest requiring the least amount of
explanatory variables (Figure 5.14). The landscape class patch variables contributed to explaining
a further 24.5% - 31.4% amongst the ecological habitat bird assemblages (Table 5.11), overall this
is significantly higher than for the life history bird assemblages.
The variable set included not
only class patch pattern characteristics by vegetation type but also the proportion of human landuse and transformation impact indicators. Woodland and grassland bird variation responded more
131
Table 5.11: Proportion of total vanatIOn explained (TVE) by landscape variables while
constrained by the topography and climate variables chosen for each group type in partial
canonical correspondence analyses (CCAs) of ADD bird species; the three greatest landscape
contributors to remaining TVE after constraining by the topography and climate variables in each
group type are show in boldface. t
Partial CCA TV
Additional variation explained (proportion of TVE explained after constraining
Woodland
Forest
Thicket
0.430
0.477
0.529
by covariables)
Grassland
0.536
Variable
Landcover
POPTOT96
POPDEN96
LOWI]ER
PLNT]ER
DRY]ER
IRR]ER
URB]ER
M]ER
T]ER
T_TOTAL
ROAD_INDEX
t
t
0.015
t
t
t
t
t
§
§
t
0.02
t
0.01
0.04
0.01
t
t
t
t
t
0.02
0.01
0.02
0.02
0.01
0.02
t
t
t
t
§
0.01
t
t
t
t
0.01
0.01
0.03
0.03
t
t
t
t
0.01
0.01
0.01
0.01
t
t
t
t
t
t
t
t
t
t
t
t
Patchiness
%LAND
LPI
NP
PD
MPS
PSSD
PSCV
t
§
§
t
t
t
t
t
0.01
0.02
t
Shape
MSI
AWMSI
MPFD
AWMPFD
t
0.01
t
§
§
§
§
0.01
t
t
t
t
t
t
t
t
t
t
0.01
§
t
t
0.01
t
Interior
CADI
TCA
NCA
CAD
MCAPP
PCASD
PCACV
MAPDC
DCASD
DCACV
TCA%
MCA%
§
0.02
t
t
t
t
t
t
t
t
0.01
t
t
t
t
0.01
t
t
t
t
t
t
t
t
t
0.01
0.04
0.01
t
0.01
t
0.01
0.015
t
t
0.01
t
t
t
t
Isolation
t
t
t
MNND
t
t
t
NNSD
t
t
t
NNCV
0.01
0.01
t
MPI
t
t
t
II
t Same rules as for Table 5.8.
I Variable was not significant in the stepwise procedure.
§ Variable was significant in the stepwise procedure but excluded because ofmulti-collinearity.
t
t
t
t
0.01
Figure 5.15: Biplots from canonical correspondence analysis of ecological habitat bird
assemblages. All axes have been resealed to range from -1.0 to 1.0. Axes for explanatory landcover class type variables that were not significant or that had very low correlations with the
canonical axes are not shown.
to the pattern metrics than to the human transformation
reacted mostly to variations in human transformation
indicators.
indicators.
The forest and thicket birds
Overall the metrics explained
31.4% in woodland (Figure 5.15a), 28.3% in forest (Figure 5.15b), 24.5% in thicket (Figure
5.15c), and 27% in grassland (Figure 5.15d) bird variations.
Grassland birds required the most
variables to fit the remaining TV. The trend in the results suggests that as vegetation becomes
more 2D in structure, i.e. trees to grass, the pattern metrics explain less of the local variations in
birds.
Forest, thicket, and grassland bird variation
was related
strongly to the extent of
urbanization in the habitat class, and exotic tree plantations affected forest birds more strongly
than grassland birds.
Woodland birds were affected by total woodland removal, along with the
number and density of the available woodland patches. Thicket bird variation was related to road
disturbance and the total amount of thicket patch core area available.
also related to the variability
in patch size distribution
Grassland bird variation
of grassland patches and the shape
complexity of the available patches which was related to the amount of exotic plantations in the
area.
5.5.2.2 Correlation Results
These results only pertain to the ADD data sets for which biodiversity
assemblage structure could be calculated.
measures of
Analysis by identify assemblage class within the life
history bird assemblages, as derived from the ordination analysis, provides an ecological basis for
the analysis from which landscape level descriptions
can be made.
Table 5.12 provides a
breakdown of the biodiversity measure relationships to human impact for each assemblage class
identified within a bird assemblage group.
Midlands
communities
In almost all cases the identified East Coast and
of each of the bird assemblages
is tending towards
single species
dominance with a negative correlation with high intensity transformation.
The correlations are
even higher when compared to the road disturbance index.
include the negative
correlation
between
high intensity
community
of passerines,
transformation
Maputaland
Exceptions
and evenness
community of non-human
grasslands community of human influenced birds.
for the Central
influenced
Zululand
birds, and high
The negative trends in evenness occur once
grid cells in a community group record high intensity transformations
of greater than 60%, which
has been noted as a fragmentation transition point in other analyses (Andren, 1994; With and
Crist, 1995; Bascompte and Sokal, 1996). High intensity transformation
and road disturbance
effects would appear to be the most important general indicators to monitor and assess the fate of
bird assemblage structure.
Again, the assessment of these relationships at the community extent
has species richness positively correlated with high intensity transformation
groups and assemblage groups.
human influenced
for many of the bird
Exceptions to this relationship occur among the passerine and
bird assemblages.
Another notable exception
is the Drakensberg
Escarp
community of non-human influenced birds, which is shown to be negatively associated with high
intensity transformation
sensitive environmentally
and road disturbance.
This alpine bird community type is not only
but the majority of the birds are negatively
influenced by human
disturbance.
The relationship
of low intensity transformation
to species richness
documents
the
negative effects of habitat degradation to overall species richness in all bird assemblages and the
majority of assemblage classes were the phenomena
is present.
Reducing habitat quality is
detrimental to interior species and because it adds no new novel land-use types through total
transformation
there is no immigration of edge dwelling and generalist species. This problem
leads to a general reduction in sensitive species and a therefore a general evenness in bird
assemblage for the areas were there was natural tendency towards single species dominance, i.e.,
134
Table 5.12: Pearson correlation coefficients for comparison of species richness (SR), Shannon
diversity (H'), and evenness (E) against transformation and disturbance variables among
classified groupings of ADD birds derived from ordination (DCA) and hierarchical classification.
Human induced transformation data were square root-transformed before analysis to improve
normality. t
All birds
Maputaland
East Coast
Drakensberg
Central Zululand
Midlands
Summer
Maputaland
East Coast
Drakensberg Escarp
Central Zululand
Midlands
Winter
Maputaland
East Coast
Drakensberg
Midlands-Zululand
Passerine
Maputaland
Drakensberg
Central Zululand
East Coast
Midlands
Non-passerine
Maputaland
East Coast
Drakensberg
Midlands
Breeding
Maputaland
East Coast
Drakensberg
Central Zululand
Midlands
Non-breeding
Woodland
Wet Grasslands
East Coast
Dry Grasslands
Human
Maputaland
Drakensberg Escarp
High Grasslands
Central Zululand
East Coast
-0.44'
-0.47'
§
§
§
§
§
-0.68'"
0.31;
§
0.29;
§
§
§
§
0.27;
0.61'
§
§
§
-0.43;
0.37"
0.42'
-0.36"
0.38;
-0.64'"
-0.28'
-0.34;
-0.56"
-0.56"
§
§
§
§
-0.38"
-0.73'"
-0.30'
-0.41'
-0.44'
-0.36;
§
§
§
0.36;
-0.35"
-0.65'"
-0.26'
-0.48'"
§
§
§
§
0.61'"
0.48'"
0.42'
§
§
§
§
0.38'
§
0.32;
-0.42"
-0.32;
-0.74'"
§
0.48'
§
§
§
-0.41"
-0.74'"
§
§
§
-0.43'
0.36;
§
§
§
§
0.53"
§
§
§
0.40'
-0.44'
0.46'
§
§
§
§
0.57"
-0.41'
§
§
§
§
0.61"
§
§
-0.53"
0.41'
§
§
§
§
§
§
0.50'"
0.34'
§
§
§
§
0.38'
0.39'
-0.33'
§
§
-0.37'
0.38'
§
§
§
-0.48"
-0.28;
-0.72'"
-0.42'
0.44'
§
§
§
§
-0.73'"
§
-0.37'
§
-0.43'
§
§
§
§
§
§
§
§
§
-0.54"
0.37'"
0.54"
0.36'"
0.33;
§
§
0.35;
0.36'"
0.56'"
-0.40'"
-0.62'"
-0.52"
§
§
-0.39'
-0.39'
§
§
§
§
§
0.35;
0.34'
-0.40'
-0.70'"
§
§
§
§
§
§
§
§
§
§
-0.37'
0.32;
0.61'"
-0.37'
-0.41'
-0.27'
§
§
§
§
0.33"
§
-0.44'
-0.43"
-0.35;
§
§
§
§
§
0.39'
0.36'
§
§
§
§
§
§
§
§
§
-0.52'"
§
-0.34;
-0.31'
-0.35'
-0.34'
§
§
0.40"
§
§
0.71'"
§
0.40'
§
§
0.60'
Non-human
Maputaland
-0.45;
-0.43'
§
§
§
East Coast
§
§
§
§
§
Drakensberg Escarp
§
§
0.50'
-0.46'
-0.52'
Highlands
§
§
§
§
§
Midlands
-0.47"
§
§
0.60'"
§
Coastal Hinterland
-0.53"
-0.38'
§
0.31;
§
f Significance is denoted by the following: § not significant, t p < 0.1, * P < 0.05,
§
§
§
§
0.42"
0.38'
-0.33'
§
§
0.37'
0.43'
§
0.36'
0.42'
§
0.37"
0.42'
0.25;
-0.29'
§
§
§
§
§
0.32;
0.57'
§
§
-0.36;
0.35"
§
§
-0.74'"
§
§
§
0.44"
§
§
§
§
§
-0.49"
§
§
§
§
§
§
0.29'
§
§
§
§
§
§
§
-0.42;
§
§
§
§
0.43'
§
§
0.35;
§
** P < 0.01, *** P < 0.001
§
§
§
§
-0.68"
§
§
-0.48'
§
§
§
Drakensberg and Midlands. The East Coast bird assemblages for non-passerine and non-breeding
birds have the only negative relationships of evenness to low intensity transformation.
Empirical semi-variograms were generated for each biodiversity measurement and human
impact measurement, while also including a road disturbance index (Stoms, 2000; Reyers et aI., in
review) and the 1996 human population density. The analysis results are only for the ADD data
sets, documenting the influence of extent on the variables from the identified assemblage class per
bird life history group as derived from the ordination analysis.
The same distance resolution of
22.5 km between grid cell centers was used to calculate the variograms.
Range values where a
flat sill is reached, which represents the autocorrelation pattern for each variable by assemblage
class within a life history bird assemblage are provided in Table 5.13.
The semi-variogram
estimates for most of the measures are bounded.
As can be seen
from the table, ranges with small local distances describe small auto correlated patches in the
distribution
of low and high values of the measures.
Large distances describe large uniform
patches of uniform autocorrelated low or high values. Five situations had flat variograms (pure
nugget) that depict those variables as random processes across the extents defined in the model.
Focusing on all, summer and winter life history groups in Table 5.13 the winter groups measures
and assemblage class extents tend to yield fine grained pattern, while the summer group generally
has larger coarse grained pattern.
Considering all the birds in the province further increases the
coarse grained pattern in each of the assemblage classes.
The trend in spatial pattern for the
biodiversity measures in the other bird assemblages shows all but non-passerine bird assemblages
as being generally coarse grained. A comparison between the spatial extents of evenness and high
intensity transformation
for the assemblage classes with negative relationships
decreased correlation strength when the high intensity transformation
than the evenness pattern.
disturbance
index.
shows a trend in
pattern is larger grained
This pattern is also similar with comparisons
against the road
Any general trends in the relationship between species richness and high
intensity transformation
could not be discerned as changes were confounded by the changing
extents among assemblage classes by bird assemblage group.
Kappa index values and X2 tests were computed for each grid cell and bird assemblage
between the two survey periods (1970-1979 and 1987-1992).
2
Figure 5.16 provides a breakdown
of the classified Kappa index values with indications of X tests that were not rejected (Ra: grid
cells from two time period are dissimilar).
Table 5.13: Semi-variogram derived distances (kilometers) of spatial dependence for species
richness (SR), Shannon diversity (H'), evenness (E), low intensity transformation (Ll), high
intensity transformation (HI), total transformation (IT), road disturbance index (RI), and 1996
population density (PD96) among classified groupings of ADD birds. t
Community
Group
HI
TT
157.5
22.5
337.5
22.5
112.5
90.0
270.0
45.0
225.0
180.0
157.5
270.0
337.5
112.5
180.0
67.5
337.5
22.5
225.0
22.5
157.5
22.5
90.0
180.0
112.5
112.5
405.0
180.0
225.0
135.0
22.5
22.5
337.5
22.5
112.5
22.5
202.5
247.5
225.0
135.0
22.5
270
337.5
112.5
202.5
22.5
405.0
22.5
225.0
22.5
157.5
22.5
90.0
180.0
112.5
45.0
405.0
337.5
45.0
22.5
337.5
337.5
22.5
22.5
22.5
337.5
67.5
90.0
247.5
247.5
22.5
112.5
270.0
45.0
337.5
22.5
292.5
22.5
337.5
157.5
22.5
22.5
22.5
22.5
157.5
112.5
22.5
112.5
112.5
157.5
225.0
22.5
22.5
180.0
112.5
225.0
45.0
135.0
22.5
337.5
67.5
22.5
112.5
180.0
247.5
225.0
22.5
22.5
180.0
247.5
112.5
135.0
225.0
180.0
22.5
225.0
157.5
225.0
157.5
90.0
112.5
22.5
225.0
67.5
22.5
67.5
135.0
22.5
22.5
67.5
112.5
202.5
405.0
337.5
112.5
22.5
22.5
292.5
22.5
202.5
247.5
90.0
45.0
270.0
292.5
157.5
202.5
292.5
337.5
202.5
90.0
22.5
157.5
22.5
225.0
405.0
337.5
202.5
22.5
225.0
202.5
135
225.0
22.5
22.5
202.5
135.0
22.5
22.5
337.5
90.0
22.5
180.0
247.5
22.5
157.5
157.5
180.0
270.0
337.5
22.5
22.5
180.0
292.5
22.5
202.5
202.5
225.5
22.5
90.0
22.5
22.5
67.5
67.5
22.5
202.5
247.5
292.5
22.5
157.5
202.5
22.5
247.5
202.5
225.0
22.5
270.0
315.0
22.5
247.5
22.5
315.0
202.5
270.0
22.5
292.5
22.5
292.5
157.5
270.0
22.5
270.0
180.0
22.5
157.5
135.0
157.5
292.5
247.5
202.5
180.0
22.5
202.5
67.5
22.5
315.0
315.0
202.5
22.5
247.5
247.5
22.5
22.5
135.0
247.5
315.0
112.5
202.5
180.0
247.5
22.5
180.0
22.5
202.5
247.5
67.5
135.0
202.5
22.5
11
SR
H'
45.0
360.0
337.5
112.5
135.0
112.5
22.5
337.5
225.0
22.5
157.5
405.0
292.5
225.0
235.0
45.0
405.0
337.5
135.0
135.0
22.5
22.5
315.0
180.0
135.0
45.0
382.5
157.5
67.5
E
R1
PD96
All birds
Maputaland
East Coast
Drakensberg
Central Zululand
Midlands
Summer
Maputaland
East Coast
Drakensberg Escarp
Central Zululand
Midlands
Winter
Maputaland
East Coast
Drakensberg
Midlands-Zululand
Passerine
Maputaland
Drakensberg
Central Zululand
East Coast
Midlands
Non-passerine
Maputaland
East Coast
Drakensberg
Midlands
t
Breeding
Maputaland
East Coast
Drakensberg
Central Zululand
Midlands
t
t
Non-breeding
Woodland
Wet Grasslands
East Coast
Dry Grasslands
t
Human
Maputaland
Drakensberg Escarp
High Grasslands
Central Zulu land
East Coast
t
202.5
157.5
292.5
Non-human
157.5
Maputaland
157.5
67.5
45.0
22.5
90.0
45.0
157.5
22.5
East Coast
337.5
22.5
247.5
270.0
247.5
22.5
22.5
135.0
247.5
Drakensberg Escarp
180.0
157.5
337.5
337.5
337.5
337.5
22.5
Highlands
270.0
157.5
157.5
157.5
22.5
270.0
202.5
270.0
22.5
Midlands
202.5
202.5
157.5
22.5
22.5
157.5
202.5
Coastal Hinterland
180.0
247.5
22.5
337.5
337.5
22.5
180.0
f Semi-variogram models have been fit using a spherical function.
~Variable represented a completely flat variogram (pure nugget), meaning no spatial dependence in the data, i.e.
no discernible pattern, and that deriving a semi-variogram is inappropriate.
137
Kappa classes
~
OJi Square not rejected
~
Protected areas
D
••
<0.4
0.4 - 0.8
>0.8
Figure 5.16: Kappa coefficient maps of each comparison between CR and ADD surveys and life
history bird assemblages.
The summer, winter and non-human influenced bird assemblages have:::: 50% of their grid cells
with Kappa values < 0.4, which equates to agreements no better than chance. The only grid cell
to have almost perfect agreement (> 0.8) with all, passerine, breeding and human influenced birds
covered the Ndumo and Tembe nature reserves on the Mozambique border in Maputaland.
This
grid cell is remote, > 50% protected and the lands on the Mozambique side were in a near pristine
condition with little to no human habitation during the bird census periods.
Areas of moderate to
substantial agreement for all bird assemblages tend to contain protected areas, are in remote
regions, or have continued
with substantial
levels of either low intensity or high intensity
development since prior to 1970. It is possible that bird inventories had either already changed
prior to 1970 to a stable inventory in the previously developed areas (coastal areas in and around
Durban and Pietermaritzburg
since 1911) or no changes in "natural" areas up to 1992. Many of
the high intensity developed areas were that way before at least 1970 (Midlands, Durban and
coastal regions), while the low intensity developed areas could have expanded between census
periods due to the demarcation of the KwaZulu homeland areas during the mid-1970s.
areas are consistently represented with dissimilar bird inventories
«
Several
0.4) for all the bird groups.
These areas were also the ones targeted for development between the two governments (Republic
of South Africa and KwaZulu government) between the census periods (Fair, 1975; South African
Government, 1975; Thorington-Smith
et al., 1978). These areas consist of the coastal and coastal
hinterland regions south of Umfolozi game reserve, the areas south of Ndumo and Tembe nature
reserves on the Maputaland plains, grid cells to the west ofPongola dam along the Pongola River,
grid cells in the Newcastle, Vryheid and Utrecht areas, grid cells in the Bergville area and down
along the Tugela River valley, and grid cells along the Pietermartizburg-Durban
corridor fanning out north and south along the coastal hinterland.
economic
Appendix D provides a list of
the birds found in the CR survey but not recorded in the ADU survey, and birds found in the ADU
survey but not found during the CR survey.
There is a growing recognition that biogeographic-scale
comprehensive
impractical
studies are needed to gain a more
assessment of faunal response to anthropogenic disturbance.
for individual investigators
Because it is often
to collect the necessary biogeographic
data (Kodric-
Brown and Brown, 1993), one must rely on the independent efforts of others to develop relations.
The use of comprehensive
biological
atlases, developed
with consistent
methodologies
in
surveying can help in this task. In this study, consistent methodologies were applied to examine
two temporally distinct avian atlases, landscape structure and environmental variables.
The two
spatial extents of analysis provided useful insight into the avian assemblages relation to land-use
intensity and pattern with spatial extent.
While the two time periods provide the landscape
dynamics for a selected region.
The ADU data supporting
the spatial correlation
and vanogram
analyses
can be
considered both comprehensive and reliable. While for South Africa the average number of bird
checklists having been compiled for each grid was 30, KwaZulu-Natal
Province was one of the
most thoroughly surveyed regions during the atlassing effort, with an average of 105 bird checklists having been compiled per grid cell in the province.
subjected
to a rigorous
ornithologists.
vetting procedure
The distributional
by both experienced
amateur
data were also
and professional
A quantitative testing against independent and more sensitive survey techniques
(Allan, 1994; Robertson et al., 1995), also confirmed its reliability in reflecting relative abundance
in terms of reporting rates. However, Harrison et al. (1997) do note that there were still problems
in identification between species, consistency of species probability of being sighted in different
habitats and the technique used in this chapter to dampen the effects of bias in recording effort
cannot help in some areas of South Africa were reporting rate was less than 10 check-lists.
Some
may even argue that using cumulative data from a five year period is quite risky for vertebrate
populations.
Nevertheless, these challenges in using biological atlases should not deter from their
importance for recording distributions and relative changes in bird assemblages, as their survey
frameworks have the added advantages of representing less common species due to the longer
survey period (Preston, 1948), which also has the added effect of not having to worry about high
frequency spatial and temporal effects (Preston, 1960). In this case the averaging effects over any
high frequency population
changes due to the longer survey period should provide a clearer
pattern of true compositional
and structural changes due to large-scale land-use developments.
However, the limitations noted will temper the results and discussion accordingly.
South Africa and KwaZulu-Natal correlation analyses.-Areas
of high intensity human
disturbance across South Africa and for each biome demonstrate the fragmentation process in
developing new landscape mosaics and therefore changes in bird assemblages.
Bird species
richness increased, but the bird assemblage evenness across South Africa, and for the woodland
and fynbos biomes showed a trend towards single species dominance.
In contrast, the succulent
desert biome showed increases in species richness and bird assemblage evenness.
Each of these
responding biomes has areas comprising urban residential, industrial, and commercial agriculture
land-uses (Fairbanks
et al., 2000), with fynbos, woodland, and then succulent desert ranked
accordingly for the relative importance of these land-uses covering each biome. Urban, industrial
and commercial agriculture forms ofland-use lead to structural habitat changes (i.e., tree planting,
water impoundments, infrastructure) that appear to positively influence the number of bird species
found, but would also promote selected species dominance in those species adaptable to the new
environments.
This trend towards single species dominance may not require reduced sightings of
rare species, but may simply reflect increasing populations
of species that thrive in modified
human landscapes due to their specific life-history requirements.
to be intimately tied to the evolving natural-urbanizing
These patterns, however, appear
landscape.
In contrast low intensity
disturbance has a negative effect on the species richness of birds in the grassland biome, but a
positive effect on birds in the shrub steppe. As opposed to degraded woody biomes where some
level of tree or shrub structure remains, degraded grassland is highly detrimental to bird richness.
In the shrub steppe biome the opening up of shrub patches from heavy grazing, which would
allow for an increase in mosaic diversity (i.e., shrub, grass, and bare) appears to have a positive
effect on bird richness.
The woodland biome is typically over grazed and coppiced for fuelwood
in South Africa (Fairbanks, 2000), which appears to cause a trend towards dominance in the
assemblage structure.
The South African and biome extent analysis shows that species distributions
and
assemblage patterns change between natural and highly transformed areas. First, species richness
increases and peaks in areas with optimal environmental
proportion
of highly transformed
land.
levels containing
a high
Secondly, the relative evenness of bird assemblages'
decrease from natural to more highly transformed
transformation
resource
areas.
This suggests that high intensity
may bring in novel resources for birds not normally found in areas, but that this
does not directly translate into evenly structured communities.
intensity transformation
It appears that some species (high
exploiters) are adept at exploiting these changes and reach their highest
densities with a tendency for assemblage structure to drift towards single species dominance.
The correlation analysis amongst biodiversity measures and land transformation values by
identified assemblage class in KwaZulu-Natal provided further support for the preponderance of
species assemblages
with negative relationships
to human disturbance
patterns.
Also, since
assemblage structure and total bird species richness summarizes a composite response of the
habitat needs of individual species (Hansen and Urban, 1992), the performance of separate spatial
correlation analysis for the main life history groups of birds allows to increase the predictive
power of the model. Two community assemblages identified in several of the bird assemblages
show this relationship: The East Coast assemblage is tending towards single species dominance in
all, summer, winter and non-passerine
bird assemblages, and the Midlands assemblage in all,
summer, passerine, breeding, and human bird assemblages (Table 5.11). Two other communities
were identified once for two bird groups: Central Zululand for passerines and Maputaland for
non-human influenced birds. The non-human influenced bird assemblage structure in Maputaland
could be a serious problem.
These birds are neutral or negatively influenced by human activities.
The Maputaland region represents South Africa's only link to the Afrotropical avifauna (Harrison
et al., 1997) and therefore a greater share of the sensitive species in the region.
The results in
Table 5.11 show a tendency towards reduced species richness for low intensity disturbance and
towards
single species
transformed.
dominance
under high intensity
transformation
for areas < 30%
Another important aspect is the affect road disturbance has on the biodiversity
measures by bird assemblage.
Generally there is a similar positive relationship of roads to species
richness as there is for high intensity transformation,
and a negative relationship
to evenness.
However, these results should be interpreted cautiously as the species inventorying may be partly
affected by road bias (e.g., Freitag et al., 1998), even though surveying procedures
were
developed to avoid this bias during the development of both atlases. Therefore, there is no reason
to expect potential errors or biases to be systematically or non-randomly distributed throughout
the country.
Changing the spatial extent and scale of the correlation analysis in a meaningful manner
between national, vegetation biome, and ecologically identified sub-regional assemblages brought
to bear important relationships
regarding pattern and process.
become a paradigm in many fields of ecology.
The importance of space has
Studies in community ecology, biodiversity,
landscape ecology, and biogeography all rely on the analysis of spatial patterns.
It is well known
that the spatial pattern that we observe depends on the scale of the study (Wiens, 1981; Wiens,
1989b), which is conveniently
summarized by its grain or resolution and by its spatial extent.
While the effect of grain size has been formally explored in many studies (Turner et al., 1989), the
effect of spatial extent has not always been thoroughly considered.
However, this study illustrates
the crucial role spatial extent can play for prominent operational definitions of richness, diversity,
and community structure.
biodiversity
measures
transformation
activity.
The biome level results for birds incorporate a lot of variation in the
in association
with the spatially
structured
response
of the human
Therefore, the results at this level of analysis are not as strong as the
community level results for KwaZulu-Natal.
This is probably because the bird communities, as
shown in the ordination analysis, are being shaped more by the human developed land-covers at
this scale rather than general vegetation
characteristics
(i.e., grassland or woodland).
The
identified communities would seem to provide a more appropriate measure of each bird groups
relationship with the degree of human disturbance in a particular area. However, as identified for
the national level analysis, the patterns of the birds in each group do not seem to be consistent
with the intermediate disturbance hypothesis for human land-use as proposed by McDonnell et al.
(1993). Therefore the results at the community extent of analysis provides little support for this
suggested pattern, except for possibly the Drakensberg
influenced birds.
Escarp community of the non-human
The smaller extent of analysis did provide some empirical support for the
theoretical results found from percolation theory, which has been used to describe multi-species
persistence
in simulated fragmented landscapes (Gardner et al., 1987; Gardner et al., 1989;
Pearson, 1993; Bascompte and Sole, 1996). In all of the negative associations for evenness versus
high intensity transformation the regression line was pulled down once sampling sites within the
population were> 60% transformed. The positive relationship between species richness and high
intensity transformation occurred when grid cells were> 40% transformed.
This pattern for South African birds is not consistent with the intermediate (natural)
disturbance hypothesis (Connell, 1978) or with its extension (McDonnell and Pickett, 1990;
McDonnell et al., 1993) to include human land-use. It has been suggested that species richness
should peak at intermediate levels of human landscape development because biotic limitations are
high in natural landscapes and physical limitations high in highly transformed landscapes
(McDonnell et al., 1993). These results, at the coarse grid cell scale of analysis, provide little
support for this suggested pattern in South African birds, and may instead provide some support
for declining assemblage evenness regimes among cumulative local scale bird communities
within a grid cell in response to high intensity transformation. The correlation slope increase in
species richness appears to be significant once greater than 40% of the land-cover within a grid
cell is transformed to high intensity land-uses. This situation would allow for both new human
adaptable exploiters and the remnant natural vegetation related birds to co-exist, but with
increasing high intensity land transformation the transformation exploiters appear to become
dominant in the landscape.
However, in the absence of land-use time series, interpretation is
difficult since the rate of land-use change may be high in the woodland, grassland, and fynbos
biomes (e.g., Fairbanks et al., 2000).
The use of the index of relative abundance tells us something of the changes of numerical
relationship between two or more species, but nothing on the true abundance of each of them. If
species #2 becomes much more numerous, it does not imply that species #1 is less abundant than
before but only that species #2 has become more abundant. Therefore, an uneven assemblage in a
high transformation area is not necessarily an impoverished community, but rather one that has a
few species that are overly abundant in comparison to the total assemblage within an area. The
potential causes for the correlations between species richness and high land-use intensity could be
a function of time since transformation (i.e., recently transformed areas being richer than
previously transformed areas), apparent increase due to better sampling, or to compositional or
habitat differences in highly transformed versus un-transformed local scale communities that alter
conspicuousness between land-use types. The apparent increase in species richness and decrease
in assemblage evenness for some of the biomes may be due to maintenance of species richness in
untransformed fragments in highly transformed grid cells combined with species that exploit
transformed habitats well in the transformed areas (Quinn and Harrison, 1988). The pattern
explored could be highly dependent on the scale and homogenizing character of grid cells on local
scale bird community attributes (Wiens, 1981; Wiens, 1989a).
South Africa and KwaZulu-Natal geostatistical analyses.--Variograms also provide
valuable information on the ability to sample and characterize bird assemblages at the defined
grid cell resolution. The semi-variograms provide information on the spatial autocorrelation of
the bird assemblage measures within each biome. Distances less than the sill derived range values
are spatially autocorrelated and thus provide information about the processes and structures of the
bird assemblages. The evenness assemblage measure derived for the grassland biome shows that
to correctly characterize the bird assemblage structure in grassland the sampling resolution would
need to be smaller than the current grid cells used for biological atlasing in South Africa. This
result for grassland bird fauna may mean that characterizing birds using fine scale survey data
may be required to completely understand their distributions and abundance, especially in
response to land-use impact. This is important as the grassland bird species are tending towards
the same problems as the fynbos species (Table 5.2), thus the current analysis may be limited
because of scale. Nevertheless, the measures for all the other biomes confirm that the grid cells
used for sampling are a small enough grain size to characterize the bird groups.
The variogram results for the KwaZulu-Natal provide more power in explaining the grain
pattern of the life history bird assemblages within ecologically defined zones. An important
aspect of the analysis points out the ecological extent of these bird assemblages against the landuse development within the assemblage classes. Most of the bird assemblages generally have a
coarse autocorrelated measurement similarity across the zones, but this is not the case for species
richness in non-passerine and non-breeding birds which tends to show fine scale detail. The nonpasserines include the raptors, kingfishers, cranes, rollers, etc. These birds tend to have very
specialized habitat requirements (i.e., river frontage, wetlands, cliffs) and are not as abundant or
gregarious as the passerines. The non-breeding birds are Afrotropical and Eurasian visitors that
also have special habitat requirements (e.g., blue swallow in mist belt grassland; European
swallow in areas providing cliffs). Therefore, both assemblage types will tend to congregate at
very specific locations producing fine scale species richness patterns. The results of the
geostatistical analysis confirm the special landscape management nature that these birds would
require for population persistence. The evenness index generally illustrated coarse grain pattern,
except for several assemblage classes within the non-human influenced bird assemblage. The fine
grained pattern in evenness structure occurs in assemblage classes that are dominated by high
intensity transformation, which could reflect the fine scale abruptness in assemblage structure
these birds reflect owing to their mainly negative relationship to human activity.
KwaZulu-Natal
bird ordination and survey association analysis.-Qrdination
analysis of
KwaZulu-Natal Province birds provided insight into the unique bird assemblage structures that
have evolved within the region. Comparisons between the two survey periods highlighted several
important changes in bird community structure over the 10-year period between surveys. Many
of the shifts, contractions and splits in the assemblages are primarily associated with broad scale
climate regimes and secondly to finer scale habitat patchiness and human land-use impacts. In
particular the creation of woody patches of alien trees in the grasslands, the addition of water
impoundment's, dryland agriculture, and urbanization have all played a part in changing the
geography of bird distributions. This was reflected in the general decrease of the gradient lengths
for each species assemblage, with the more recent ADD survey reflecting the homogenising
affects land-cover changes are having on bird diversity. Comparisons between Figures 5.6 and
5.7 show several general trends: thinning of the East Coast bird zone around Richards Bay,
compaction and stretching of the Maputaland zone southwards, shrinking of the Drakensberg
zone, and a general split in the Midlands zone from the CR period into a Midlands and Central
Zululand zones during the ADD survey. Several hypotheses could be driving these changes. First,
rainfall varies above or below the mean in approximately lO-year cycles (Dyer and Tyson, 1977)
in the region, with the CR survey coinciding in with a pluvial cycle and the ADD survey
coincided with a dry and in some areas drought cycle. Therefore, species responding to this wet
cycle could have moved into KwaZulu-Natal down along the Mozambique coast, or some species
from the dry interior of South Africa could have moved in during the ADD period. Second, an
increase in land transformation within the economically active areas along the coast and Midlands
could have added more artificial habitats for transformation exploiters to distribute against
(Figures 5.6h and 5.7h), or alternatively sensitive species could have withdrawn from those areas
into the underdeveloped Zululand region (e.g., Winterbottom, 1962). Finally, there is still the
possibility that despite the thorough surveys conducted in both periods, a majority of the species
could still reflect biased distribution estimates.
Elevation contributed more to total variation explained (TVE) than any other variable
(Tables 5.7 and 5.8). However, elevation is a complex-gradient (sensu Whittaker, 1960; 1965)
that co-varies with a host of historical and environmental factors such as vegetation and climate.
Therefore, in many of the datasets the elevation gradient was collinear with climate variables,
such as minimum temperature.
Nevertheless, the strong association between elevation and
species composition observed in both sampling periods is consistent with other studies (e.g.,
Liversidge, 1962; Brown and Barnes, 1984).
One could debate excluding elevation from the analyses because it measures spatial
position and only indirectly reflects physical environment. Excluding DEMMEAN from stepwise
CCAs, however, did not appreciably affect TVE, probably because of multicolinearity with other
climate measures, nor the relations among species, grid cells, and explanatory variables. This
robustness to multicolinearity among explanatory variables is a strength ofCCA (Palmer, 1993).
Therefore, DEMMEAN was retained in the analyses because of its value in interpreting results.
Results for the gradient analysis supported the hypothesis of primary importance of
macroclimate and the secondary role of land-cover proportion and pattern in controlling regional
compositional gradients in the bird diversity of KwaZulu-NataI. These findings can only be
compared indirectly with previous studies because of differences in methods. Similar type studies
in North America used plot sampling or linear sampling along predefined road transects, and
examined relative contributions of regional climatic measures and plot based vegetation pattern
statistics or coarse-scale land-cover (Flather, 1996; Wiens and Rotenberry, 1980; Rotenberry,
1978; Wiens, 1973; McGarigal and McComb, 1995; Flather, 1996; 0' Connor et aI., 1996).
Evapotranspiration, seasonal variability and extremes in climate were more important in
explaining species gradients than were mean annual climatic conditions. The two dominant
species gradients were associated with elevation, which integrates temperature extremes, seasonal
variability in moisture and temperature related to the Indian Ocean and escarpment
(continentality), and climatic gradients that integrate elements of both temperature and moisture.
The importance of continentality in this study was consistent with Rotenberry's (1978) study that
spanned a west-east gradient from Oregon to Colorado, although this present study spanned a
much narrower latitudinal range (see also Cody, 1981; Cody, 1985; Cody, 1993). Species
gradients for all groups and survey periods were much more strongly associated with minimum
temperature during winter and moisture stress during summer. This association was consistent
with observed correspondence between growing season precipitation and temperature gradients
and latitudinal vegetation gradients in the grassland (Ellery et aI., 1995) and woodland (Fairbanks,
2000) biomes of South Africa.
Topographic heterogeneity contributed slightly more to TVE for the non-breeding and
non-human influenced bird groups in the CR datasets, but had a smaller contribution across all the
ADD datasets (Tables 5.7 and 5.8).
The CR datasets all responded more significantly to
topographic heterogeneity than to evapotranspiration and seasonal variability's in temperature and
precipitation.
In the ADD datasets, only the gradients in human influenced birds showed
significant relationships with topographic heterogeneity. In western KwaZulu-Natal, moisture is
substantial and temperatures are colder and more variable, topography is deeply dissected from
the rivers flowing off the Drakensberg Escarpment, and topographic effects would be more
pronounced (Fairbanks and Benn, 2000).
Analysis of the birds grouped by ecological habitat associations and using their relative
abundance values to examine relationships
illustrated that the effects of elevation were only
substantial for the thicket and grassland bird assemblages, and elevation heterogeneity was only
significant for the grassland bird assemblage.
These two vegetation types were found to extend
across the entire province and thus reflect the strong gradient between coastal and Drakensberg
Escarpment
grassland and thicket type bird variation.
evapotranspiration
Climate variables of temperature
explained most of the variation in bird variation.
Structural
and
changes in
woodland types between the arid woodlands of Maputaland
and the mixed woodland of the
Tugela
by growing
and Buffalo
(Fairbanks, 2000).
River basins
are largely determined
season temperature
Growing season moisture stress accounted for the variation in forest and
thicket associated birds, with minimum monthly temperature explaining the differences in bird
community
type in forest. This reflects the differences
in forest structure and composition
between the coast and afromontane forests of the Midlands and canyons along the Drakensberg
Escarpment (Everard et al., 1995).
The findings on the influence
of landscape
structure variables
on regional specIes
gradients were not very conclusive with the life history bird assemblages and varied widely in
importance
with regard to remaining
TVE (Table 5.9).
Under partial CCA the landscape
variables explained between 8.6-32% of TVE for the life history bird assemblages, with the most
sensitive bird assemblages more significantly influenced.
and non-passerine
bird assemblage
Non-breeding, non-human influenced
gradients were associated
the strongest with landscape
variables. For all the bird assemblages, however, the proportion ofland-cover
was more important
to TVE under the stepwise CCA than landscape mosaic pattern variables.
There were several
exceptions with class richness density having significance for all, summer, winter, non-passerine,
non-breeding,
and non-human influenced bird assemblages. LCLU evenness was important for
the non-human influenced birds and the variability in patch size with distance was important for
explaining
the passerine
assemblage.
Contrary
to the idea that landscape
structure
(i.e.,
configuration) plays a dominant role in the regulation of wildlife populations (e.g., Turner, 1989),
these results suggest that LCLU proportion (i.e., habitat and land-use area) in southern African
landscapes explain most of the TVE. Of the 9 life history bird assemblages studied, landscape
variables typically explained less than half of the variation in presence/absence
distributions
among grid cells. Grassland, forest, subsistence agriculture, low intensity transformation, urban,
and road disturbance
percentages
contributed
significantly
to the partial CCAs.
In contrast,
passerine, breeding and human influenced bird assemblage gradients were conspicuous by their
"insensitivity"
CCAs.
to variation in landscape structure with only 8.6-11.1 % of TVE from the partial
The remaining data sets of all, summer and winter bird assemblages
had moderate
explanation and really only represent more homogeneous combinations of the other stratified bird
147
assemblages.
The moderate performance of the landscape structural variables may be explained
by the CCA partialling procedure.
without the confounding
Though partialling out covariables allows one to test other data
effects of other unwanted pattern.
Those co-variables
may have a
confounding relationship with the new variables of interest and therefore reduce the explanatory
power of the new variables, once they are removed.
Table 5.14 presents the relationship of the
primary environmental variables to the landscape structure variables.
collinear effects between DEMMEAN, MINMNTHMN
variables of interest.
There are an abundance of
and MXSEAS_MN
and the landscape
However, what is more striking in this table, is the strong presence of
collinearity in bird species gradients (identified by environmental variables) and historical human
development (identified by land-cover variables) within the province.
The two biotic processes
are competing and evolving together along the same environmental
gradients.
These results
would suggest that policy makers and managers should consider emphasizing the landscape-avian
diversity-human
development relationship
for defining and implementing
conservation
policy.
Avian community structure seems to react well to changes in the landscape brought about by
human
development
and therefore
could be good monitoring
agents
(sensu Furness
and
Greenwood, 1993; Dufrene and Legendre, 1997).
These results also suggest that temperate migrants, non-passerines
and species deemed
neutral or negatively affected by humans are related to landscape structure in a way that is unique
when compared to species from other functional groups. Past studies from North America have
noted differences among migratory habitat categories with respect to population trends (Robbins
et al., 1989b).
Explanations
for these differences
include:
differential
susceptibility
of
Neotropical migrants to forest fragmentation (Robbins et al., 1989a); differential susceptibility of
permanent residents, and to a lesser degree, tropical migrants, to severe weather (Robbins et al.,
1989b); and the broader environmental tolerances expected in permanent residents (O'Connor,
1992). The patterns of association between the compositional gradients in bird assemblages and
landscape structure (Table 5.9) seem to be consistent with these explanations.
Flather and Sauer
(1996) uncovered similar associations to landscape structure for Neotropical migrants in North
America.
Analysis of the birds assembled by their association with ecological habitat, using relative
abundance and patch level statistics of their habitat sharpened the relationships,
structure explaining roughly half of the variation in abundance among landscapes.
with landscape
Examination
by habitat association brought forth clearer results that tied in well with biome level results and
past studies (e.g., Armstrong et al., 1998; Allan et al., 1997). The variation in woodland bird
species related closely to gradients in high intensity land transformation,
which related to
fragmentation represented by the number of patches and the patch density. In relation to the other
habitat assemblages studied it would appear that woodland bird variations at the landscape level
are highly related to structural relations brought forth at the landscape level from human impact.
This was noted as a factor in changing species richness and structure at the woodland biome level
of examination.
These results are similar to other bird-forestry
studies conducted in North
America (e.g., Askins et aI., 1987; Derleth et aI., 1989; McGarigal and McComb, 1995). Though
always small in extent to begin with (e.g., Midgley et aI., 1997), birds of indigenous forest habitat
are being strongly affected by the amount of surrounding exotic tree plantations and urbanization.
The size and number of patches of forest play an important part in the variation in bird species,
but the surrounding
human land-use matrix is overshadowing
149
them.
Thicket birds are also
responding to increased levels of urbanization, which is being caused by actual urban extent and
the dissection of habitat caused by roads, this leads the thicket birds to be sensitive to the amount
of core area in thicket. Grassland bird variation is being influenced by urbanization, planations,
and dryland agriculture, which creates gradients in patch size variability and increased flexibility
in grassland patch shape. Grassland birds are the most sensitive functional bird group in South
Africa, and respond strongly to changes in the continuous cover in grasslands required for their
maintenance (Allan et a1., 1997).
No single landscape descriptor is likely to consistently explain variation in assemblage
composition across sets of landscapes. A descriptors explanatory power is probably a function of
its range of variation, which typically increases with the area encompassed (Wiens, 1989b), as
well as the biota's sensitivity to change in the descriptors. For example, elevation and elevation
heterogeneity were dominant explanatory variables for bird variation recorded in the 1970s, but
for the ADU birds they had a reduction in significance or were replaced with temperature related
variables. Changing landscape structure (i.e., increased exotic trees in grasslands, development of
dams, urbanization, etc.) may largely be changing the majority of bird distributions altering biotic
pattern that were at one time closely related to broad environmental gradients.
Although
landscape structure was demonstrably strongly related to several species assemblage types, one
cannot conclude that it was a dominant factor given the large amount of unexplained variation.
Therefore, the limited evidence gathered so far for this region containing woodland, grassland,
forest and thicket suggests that one should not blindly accept the landscape structure hypothesis
described in the introduction. Habitat configuration and subdivision undoubtedly playa role in
regulating population dynamics, but the magnitude and nature of this role may vary
geographically and over time in relation to changes in regional habitat conditions and other
factors, and probably varies among species in relation to habitat selectivity, vagility, and scale.
The inclusion of the association analysis was able to further highlight several areas of
concern that may have driven the changes found in the ordination analysis. Unfortunately, with
the rather rapid development processes being conducted currently within post-apartheid South
Africa, the comparisons between the 1970s and early 1990s would seem woefully out of date.
Nevertheless, the use of the kappa index value and X2 test provided important results that could
address precautions to future development options in the province (i.e., Lebombo Spatial
Development Initiative). The areas that had changed bird inventories were generally targeted for
development in the former KwaZulu homeland and Natal province (Thorington-Smith et a1.,
1978; Fair, 1975; South African government, 1975).
In the Maputaland region the changes were quite substantial and may in fact demonstrate
how sensitive the birds of the region are to low levels of human disturbance. The cells with < 0.4
150
kappa values and
"l values not rejected were in areas developed since
1980. These include the
Makhatini flats agricultural development region (sugarcane and cotton), which began large-scale
farming in 1984 and had > 30 kms of irrigation canals installed and substantial bush clearing
concluded by 1987. Also during the 1980s the road from Jozini to Kosi Bay was paved for
greater traffic mobility. At the same time, the Tembe elephant reserve was created next to Ndumo
in 1983, which seemed to have helped that grid cell area retain a similar bird inventory over the
10 year period. Probably the greatest change, however, to the area that precipitated the changes
associated between the two bird surveys is the increase in human population. As recorded from
the combined magisterial districts of Ubombo and Ingwavuma, the human population expansion
occurred as follows: 1970- 108964; 1980- 148453; 1991- 260948; and 1996- 304222. Therefore,
between 1979 and 1992 (between survey periods), there was a 75% increase in the human
population within the Maputaland region.
Other areas of significant change include the Richards Bay development, where the
harbor was partially completed by 1977 and fully completed in the 1980s with increased shipping,
industrialisation and surrounding exotic forest plantations.
The Newcastle, Vryheid, and
Ladysmith areas were denoted by the government in 1975 to be development nodes (Fair, 1975;
South African government, 1975), and therefore during the 1980s saw industrial expansion
through mining, smelting and irrigated agriculture expansion.
The scope of this study was restricted in several ways.
These limitations identify
additional research required before management recommendations can be suggested. First, the
scale of the investigation placed lower and upper limits of resolution on the ability to detect
habitat configurations and assess bird-habitat relationships (Wiens, 1989a,b). The extent of the
landscapes was roughly 62500 ha; this defined the upper limit of resolution. Populations of the
species undoubtedly are subjected to demographic influences operating over larger areas. These
areas should also be related to more ecological defined types, rather than the use of an arbitrary
grid cell. The use of the grid cell more than anything makes this study a test against a random
sample structure and therefore any relationships found would be more than likely stronger using
another sampling system (e.g., landscapes, catchments, vegetation types, etc.). The lower limit
has been set by the resolution of the landscape structure variables, 1ha based on 100m2 raster.
Patchiness occurs at many scales and patches can be defined in hierarchical fashion at
progressively finer and finer scales (Kotliar and Wiens, 1990). Because landscape metrics are not
invariant to scale (Turner et al., 1989), changing the minimum patch size would have significant
effects on measures of landscape structure for specific patch types.
Second, the analysis presented is largely limited to a single scale.
As the studies by
Wiens and Rotenberry on shrub steppe birds (Wiens and Rotenberry, 1981; Wiens and Rotenberry,
1985; Rotenberry,
1986; Wiens et a1., 1987) demonstrate, habitat selection occurs at multiple
scales, and habitats association often very among scales of investigation or analysis. This study is
.
unable to infer about habitat associations at finer or even coarser scales .
Third, the classified habitat is based on structural characteristics
mapped from satellite imagery.
organism-centered
that could be reliably
The broad vegetation definitions may not be important from an
or even community-centered
perspective.
Moreover, vegetation patches were
somewhat arbitrarily and subjectively made discrete during the cover mapping and digitization
process (Fairbanks et a1., 2000). Vegetation structural differences were in many cases not abrupt
(i.e., grassland-thicket
or thicket-woodland transition).
over-simplified representations
Thus, the final patch mosaics represented
of the actual spatial heterogeneity present in the grid cells.
All three limitations were originally pointed out by McGarigal and McComb (1995), and
are inherent to all landscape ecological investigations as they deal with the issue of "measured
heterogeneity" vs. "functional heterogeneity" (Kolasa and Rollo, 1991). Measured heterogeneity
mayor may not correspond to something functionally meaningful to a particular species or group
of species (Wiens, 1989a; Wiens 1989b). Thus, results based on measured heterogeneity may
lead to erroneous conclusions.
This study had no a priori knowledge of what the functionally
relevant scale would be, so even though the scale was set based on available data the bird
assemblages were designed to make-up for any scale limited functional configurations and were
meaningful from a land management perspective.
This study was the first systematic quantification,
synthesis, and mapping of avian-
environment gradients across a large, contiguous region of southern Africa based on two periods
of biological atlas data. The goals of faunal atlases are sound and there is a need for continued
national biological surveys of taxa (Balmford and Gaston, 1999). However, the uncritical use of
these data to determine conservation priorities (Scott et a1., 1993; Williams et a1., 1996), as well as
to understand the environmental mechanisms behind biodiversity patterns (O'Connor et a1., 1996;
Bohning-Gaese,
1997) could prove problematic.
Distribution and abundance may change over
time; therefore, ideally their estimate should be conducted within a sufficiently short period. This
is a potentially serious problem with biological atlas surveys, on which estimates of distribution
are often based, because such studies are forced to accumulate observations across several years.
The results are frequently used to develop "hot spot" indices, provide data for reserve selection
analyses or are used as inputs into macroecology studies. Biological atlases effectively show the
152
compound errors in richness over time diverging from ecologically meaningful community
assemblage indices. Thus, that contemporary biological atlases may already present transformed
distribution and diversity patterns can easily be overlooked.
Atlasing projects, however, do provide a unique opportunity to explore at a macro-scale,
contemporary species distribution-abundance relationships with human disturbance, and
especially how human disturbance may be providing contradicting effects in commonly used
ecological indices. While this analysis underscores the value of sound species-related distribution
data and emphasizes the necessity for survey research in conservation biology (Haila and
Margules, 1996), equal emphasis must be placed on the role of human influence in shaping extant
biological communities. Although largely transformed areas may currently harbor many species,
these areas may not be able to sustain natural ecological processes and viable populations
(Baudry, 1993; Hobbs, 1993; Freemark, 1995).
Understanding the influence of landscape structure on spatial and temporal patterns of
species is an important component of developing prescriptive management recommendations to
conserve biological resources. Wiens (1992), and Flather and Sauer (1996), however, note that
much of what has been reported in the literature is of a descriptive nature, documenting patterns
of association such as has been reported here. There is a great danger from inferring causation
from correlations, especially as associated to the scale of this study. However, Flather and Sauer
(1996) also note that careful interpretation and analysis of extant data sources serves an important
heuristic function (Carpenter, 1990), can lead to insights into the factors affecting patterns in the
distribution and abundance of species (Brown, 1984), and can provide a regional context for
interpreting and guiding future local studies (Wiens, 1992).
This chapters fmdings support a conceptual model of multi-scaled controls on bird
distribution, and the related notion that local community structure is the result of both regional
and local scale abioticlbiotic processes, landscape structure and human action.
Broad-scale
climate and topography were the primary controls to the differences in bird composition within
the KwaZulu-Natal province.
The study also showed that the broad scale environmental
relationships are not stable over time and bird assemblage changes may be related to longer
climatic cycles and land-cover change processes. The use of ordination analysis provided a good
exploratory tool for untangling this ecological complexity. Even though the landscape measures
explained a portion of the variation in bird variation, this analysis demonstra~es the potential
limitations of using a simple pattern association with presence/absence or relative abundance data.
It is suggested that investigators use several analytical approaches and use the consistency in
results among approaches to gauge confidence in the conclusions. This study tried to adhere to
this approach using spatial correlation, geostatistics, ordinations, and temporal association
153
analysis.
Each provided key pieces to understanding
the changes in bird distributions
and
structure.
Despite strong ecological contrasts within the KwaZulu-Natal
presented here succeeded in synthesizing species-environment
relations.
province, the analyses
These findings suggest
that apparent conflicts among local bird studies can be explained by real ecological differences
among places.
Indeed, the results in this chapter provide a broader context for considering
gradient and classification
studies conducted at a finer, local scale within a landscape.
This
analysis also places crucial questions on the roles of establishing isolated nature reserves and their
ability to persevere bird community persistence.
regarding bird populations
In Baillie et al. (2000) several theoretical points
were examined and are supported by this chapter's result.
These
include: (1) Habitat deterioration may not only lead to population declines within that habitat but
also in adjacent habitats of good quality; and (2) if dispersal is an important process then
protecting only isolated areas may be insufficient to maintain the populations within them, which
provides
support to the landscape
ecology principles
involving
conservation
of the wider
landscape through connectivity.
Macro-economic
policies fueled by globalization
and human population
growth are
rapidly changing the landscapes of southern Africa, which in turn affect wildlife populations (and
consequently
the results of longer term surveys). Consequently,
we cannot just identify and
manage for "species diversity", as measured by richness and diversity indices, alone as they
ignore species assemblage structures (Soule and Simberloff,
detrimental
effects of development
1986).
on bird species populations
Results suggest that the
may cause the erosion of
persistence of South African bird assemblages, while biological atlasing may artificially inflate
the species richness recorded in a study region.
6.
Analyzing Human Factors that Affect Biodiversity Conservation:
the Co-evolutionary Model
If everything occurred at the same time, there would be no
development. If everything existed in the same place, there would be
no particularity. Only space makes possible the particular, which then
unfolds in time ... to let this space-conditioned
particularity grow
without letting the whole run wild-that is political art.
Landscape
ecologists treat human factors as the primary driving force for landscape
change and subsequent biodiversity loss (Forman and Godron, 1986; Soule, 1991; Dale et aI.,
1994; Forman, 1995; Forester and Machlis, 1996; Chapin et aI., 2000; O'Neill and Kahn, 2000).
Nevertheless,
there are few empirical
studies or methodologies
describing
human actions
developing across a region and their affect on biodiversity (but see Dale et aI., 1994; Farina, 1997;
White et aI., 1997; Abbitt et aI., 2000). The study of human spatial development remains within
schools of geography and in particular human and economic geography (Thoman et aI., 1962;
Haggett et aI., 1977; Chapman, 1979; Bradford and Kent, 1986; Healey and Ilbery, 1990).
It is important for conservation planning purposes to understand the interactions between
landscapes and the cultural and social forces, which have shaped them in the past and are driving
them at present (Nassauer, 1992; Norgaard, 1994; Forman, 1995; Zimmerer and Young, 1998;
Farina,
2000; sensu Chapter
development,
2).
Agricultural
development,
can be thought of as a co-evolutionary
which
can lead onto urban
process between a social system and an
ecosystem (as discussed in Chapter 2). Human agricultural activities modify the ecosystem while
the ecosystem's responses can determine the nature of individual actions and social organization.
When these sequential adaptations of one system to the other are complementary and beneficial to
humans,
either
development
Alonso,
fortuitously
design,
agricultural
and subsequently
emerges (Von Thiinen, 1826; Weber, 1909; Hagerstrand,
1974; Norgaard,
development
or by strategic
1981; 1984).
urban
1956; Friedmann and
In modem terms, this process is referred to as the
of a space economy, which is the direct product of culture, personality,
and
environment within a political economy.
This study builds on the previous chapters and the conceptual framework outlined in
Chapter 2 to develop and examine an inter-disciplinary
economic,
framework.
environmental,
and landscape
pattern
model of biodiversity threat using socioindicator
datasets
within
The results highlight the importance of identifying relationships
a geographic
between human
social systems and biodiversity conservation strategies and the potential offered by interdisciplinary research for exploring pathways to sustainable biodiversity conservation.
Space economies represent open systems, as they exchange materials, energIes, or
information with their environments (e.g., von Bertalanffy, 1968). Since a space economy may be
viewed as a system of interrelated and interconnected parts, it should be possible to uncover a
degree of spatial order within its structure. Dacey (1964) observed that it is unlikely that
geographic distributions, particularly those determined by human decisions, are random, and thus
most geographic patterns reflect some system or order. Forman and Godron (1986), and Forman
(1995) acknowledged this logic in the development oflandscape ecology theory. The search for
order not only relies on observation, measurement, and description, but also demands the study of
system behavior and processes responsible for evolving emergent patterns (see Chapter 2 for
review).
Friedmann (1972), and Friedmann and Alonso (1974) recognized that human activities
are distributed in particular rhythms and patterns within a space economy, which are the results of
interdependencies that shape the economic space. Berry (1970) formalized a general framework
for economic space relationships, which highlights these interdependencies, based on the
following three components:
1. A national pattern of heartland and hinterland;
2. An urban hierarchy; and
3. Gradients of urban influence on their surrounding dependent regions.
This framework allows for the identification of a national core-periphery structure. Previous
studies by Fair (1965), Board et al. (1970), and Browett and Fair (1974) confirmed that South
Africa had developed towards this norm. Whereas, regional inequalities are inherent in the spatial
structure because growth does not happen everywhere and at once; it is concentrated in points or
development nodes, of variable intensity and spreads along diverse transportation and
communication networks (Hansen, 1967).
The work of Myrdal (1958) first recognized that the dominant factor responsible for the
persistence of the core-periphery structure appears to be the process of cumulative and circular
causation. The existence of external economies, economies of scale, and agglomeration in core
areas, compounded by the provision of transportation networks, serve to enhance and capitalize
upon existing advantages of relative locations (e.g., along coasts, or major navigateable rivers).
This has meant that the balance of forces have consistently led to development concentration in
core areas and a cycle of poverty in the periphery (Browett and Fair, 1974). In addition, spatial
political policy, such as the development of the former homeland system in South Africa had
further accentuated the poverty periphery (Fair and Schmidt, 1974; Christopher, 1982).
Moreover, once initiated, the core-periphery structure is perpetuated by a compelling inertia (see
Rogerson, 1975) and, as noted by Richardson (1973), the existing spatial distribution of
population and economic activity in a region in turn drives patterns of regional growth. Fair
(1976) presented evidence that South Africa had reached the stage where the spatial economy was
past the period dominated by a single national center and associated periphery, with clear signs of
a multinuclear network of regional economic core centers (e.g., Cape Town, Durban,
Johannesburg, Pretoria), minor metropolitan centers, regional market towns, and peripheral
country towns (see Chapter 1, Figure 1.1b for a settlement hierarchy structure in KwaZulu-Natal
Province).
The identification of homogeneous geographical regions and their interpretation through
environmental or socio-economic variables has always been an important topic of biogeography
and regional geography studies. The primary purpose of this chapter is to identify and describe
the socio-economic-environmental and landscape mosaic patterns of KwaZulu-Natal, in support
of the conservation pattern studies conducted in Chapters 3, 4 and 5. This line of investigation
accepts challenges posed from within the landscape ecology community (O'Neill, 1999), to apply
and integrate theories of economic geography within landscape ecological analysis to biodiversity
conservation problems. The geographical structure of KwaZulu-Natal is explored by examining
the covariance between socio-economic-environmental, LCLU, and landscape mosaic pattern
indicators. The exploratory use of principal component analysis and pattern recognition
techniques are employed to investigate the spatial significance of socio-economic and
environmental relationships in KwaZulu-Natal Province. First, the structure and spatial patterns
derived from a multivariate analysis of available socio-economic-environmental variables for
KwaZulu-Natal are examined. Secondly, these patterns are related to LCLU patterns derived
from landscape metric analysis. Thirdly, the implications of the distribution of socio-economic
resources and needs in the province, as well as, priority avian conservation areas and required
habitat are assessed collectively.
The study relies on the fact that socio-economic activity is the
primary determinant of landscape pattern and change, and in turn drives biological community
responses (see Chapter 5).
Development needs and tensions with regard to environmental
preservation are identified in order to develop a socio-economic and ecologically sound strategic
conservation plan for the bird diversity of KwaZulu-Natal.
For each of the 1996 magisterial districts in KwaZu1u-Natal (Figure 1.10) three separate
datasets were first used to identify homogeneous regions: LCLU, socio-economic-environmenta1
indicators,
and landscape
correspondence
mosaic pattern indices.
This was done either using detrended
analysis or principal component analysis depending on the type and structure of
the dataset, to derive axes of variation in data space.
Hierarchical
and k-means classification
strategies were then applied to derive homogeneous clusters based on the type of results and their
variance characteristics.
The multivariate analysis is aimed at uncovering the most important
underlying dimensions from the relationships between a range of socio-economic-environmental,
land-cover, and landscape pattern data.
Results are loosely compared with results from other
multivariate studies in Africa, notably those in Ghana, Kenya, Tanzania, Swaziland, and Nigeria
(Forde, 1968; Soja, 1968; Gould, 1970; Lea, 1972; Weinand, 1973). The results are explained
and then inserted
into a pattern recognition
development regions.
procedure
to derive rules that describe
the
The variations within the datasets and regional rules are then discussed in
relation to the conservation planning results obtained in Chapters 3, 4 and 5.
The success of the analyses depends upon the selection of a suitable cross-section of
variables that would enable conclusions
patterning in the province.
to be drawn about the structure and geographical
The socio-economic-environmental
variables chosen were made up of
eighty-four variables, under six main subject groups (Appendix A):
1. Population characteristics (11 variables);
These
2.
Social characteristics (18 variables);
3.
Economic characteristics (23 variables);
4.
Development needs (16 variables);
5.
Physical characteristics of economic importance (9 variables); and
6.
Environmental characteristics (7 variables).
variables
transformations,
were checked
for normality
and transformed
where required, before principal component analysis.
usmg
log or square root
The landscape pattern indicators used in the analysis are those found in Chapter 1, Table
1.5. The last data set represented the class type and proportion of LCLU (Chapter 1, Table 1.3)
found within each magisterial district.
The original LCLU map was used to extract presence/absence
each class type per magisterial
district.
transformation
before analysis.
magisterial
into a percentage
districts
correspondence
and
twenty-nine
The hectare
LCLU
and areal abundance of
measurement
was standardized
The data matrix consisted
classes
comprising
753
by
of fifty-two
occurrences.
A
analysis was used to describe the interactions among characters and bring out the
main covariance relationships.
The model assumes a relationship between the environment and
the LCLU class occurrence as a unimodal response to the environmental condition (Gauch, 1982).
It is assumed that LCLU classes across the province are related to environmental gradients, which
controls their presence and abundance.
category.
For example, consider the exotic plantation land-use
This type of land-use is strongly related to the water balance of a region (Fairbanks,
1995; Fairbanks and Smith, 1996). Therefore, this type of land-use will only be developed where
there is sufficient moisture for economic production and the area size of the development will
depend on the amount of land with the required moisture regime.
Of course, this simple model
ignores government development policy and land ownership issues, which will skew the spatial
development of land-uses within a region.
Other classes of land-use generally follow the same
patterns, except for historical colonization patterns along coasts.
A detrended correspondence
anlaysis (DCA), as outlined in Chapters 4 and 5, was applied to the dataset to avoid the horseshoe
effect and to obtain linear representations
Simlauer,
1998).
Legendre,
1998) using Wards minimization
corresponding
A hierarchical
for linear gradients (Gauch, 1982; ter Braak and
linkage agglomerative
clustering
algorithm
was then used to develop
to the various clustering levels found from the ordination.
(Legendre
and
a series of maps
The hierarchical
clustering procedure was adopted because it is assumed that recorded LCLU variables have a
nearest-neighbor
relationship when measured by the magisterial districts due to past development
policy and known general patterns in human spatial development
(sensu Bradford and Kent,
1986).
The 84 socio-economic-environmental
and 32 landscape mosaic pattern indicators were
subjected to a principal components analysis (PCA). PCA is a multivariate procedure designed to
reduce a large number of variables to smaller set of "factors" that account for most of the variance
among the original variables.
correlation matrix.
Factors are typically extracted by applying PCA to a standardized
A table of factor loadings shows which variable are grouped together on
which common factors, and the degree of correlation between individual variables and the factors.
The factors are interpreted as axes in state space, and the meanings of the axes are inferred from
the variables that are most correlated with them. Highly correlated variables are said to "load
heavily" on that factor. Factors can be rotated in an attempt to account for additional variance.
Changes in sample area and landscape mosaic indices were highly confounded in the 52
magisterial districts; that is magisterial district area and landscape mosaic indices covaried in a
somewhat predictable manner. Following McGarigal and McComb (1995), regression analysis
was used to remove any significant empirical relationship between magisterial district area and
landscape mosaic indices. Magisterial district area was regressed on each of the configuration
indices using general linear models. Based on an analysis of the residuals, appropriate dependent
variable transformations (log or square root) were conducted to ensure that regression
assumptions were met. Models were constructed for each landscape mosaic index separately,
trying for the most logical model exhibiting the largest significant R2, and best residual
distribution. Using this process, the 32 original pattern metrics were transformed into 32 new
residual metrics representing variation in magisterial districts independent of their area.
Factor scores were calculated for each magisterial district by each dataset. The factor
scores from the socio-economic-environmental indicators dataset were then used to group the
magisterial districts using first the same hierarchical procedure described earlier and then a kmeans method (Legendre and Legendre, 1998). The k-means method produced k groups (the
value of k is decided by the user) after an iterative procedure of object reallocation; the procedure
stops when the overall sum of squares, which is the sum of the within-group sum of squares, has
reached a minimum. The clusters obtained by the hierarchical clustering were used as initial
configurations to the k-means algorithm in order to develop a parsimonious set of clusters, since
the k-means algorithm relies on the user to decide the number of classes to obtain. The k-means
approach was deemed appropriate over the hierarchical approach because the socio-economic
indicators were not necessarily hierarchical in nature with reference to the magisterial districts.
The factor scores that were calculated from the landscape pattern indices by magisterial
district were grouped together using the hierarchical agglomerative procedure with Ward's
minimization. A hierarchical method was considered the most appropriate to cluster the districts
based on the same reasoning for the LCLU dataset. Both datasets measure patterns that have
strong geographic contiguity between magisterial districts, which makes them appropriate for
hierarchical clustering.
Systems for inducing concept descriptions from examples are valuable tools for assisting
in the task of knowledge acquisition for expert systems. Such systems include the class of neural
networks and others that produce rules to describe a problem set. Neural networks are popular,
but hampered by the fact that they are difficult to use, resemble "black box" thinking, assume no
noise in the domain, search for a concept description that classifies training data perfectly, and the
rules developed are not easily deciphered for further investigation. Instead, a system able to
handle noisy data and generate interpretable rules is required. In particular, mechanisms for
avoiding the overfitting of the induced concept description to the data are needed, requiring
relaxation of the constraint that the induced description must classify the data perfectly.
A rule induction program used for data mining purposes, CN2 (Clark and Niblett, 1989;
Clark and Boswell, 1991), was used to develop if-then rules to describe each of the classified
datasets against the other indicator variable data sets. CN2 uses a beam search to find a rule, or
rule set, that best describes each class. Each rule set, referred to as a complex, consists of a
conjunction of conditions. Complexes are built using the beam search over the space of all
possible complexes (conjunctions of conditions). The best complexes for a class are found using
the efficient Laplacian error estimate as a search heuristic (Clark and Boswell, 1991).
The CN2 program was used to identify if-then rules from the variable sets in order to
explore the tension between variables identified for the regions in relation to required avian
conservation areas. These rule sets provide good indicators of issues that define a region and may
need to be addressed, and they can be interpreted with the PCA results for a comprehensive view
of the co-evolutionary links within the province. For this analysis the landscape pattern indicators
were used to develop descriptive rules to describe the geographic regions developed from the
PCA of the socio-economic-environmental indicators.
In addition, the socio-economic-
environmental indicators were used to develop descriptive rules to describe the regions developed
from the landscape pattern indicators. The regions grouped using the DCA results of the LCLU
had rules developed from both the socio-economic-environmental and landscape pattern
indicators.
Implications for avian conservation are assessed by using the "ideal" reserve network
developed in Chapter 4 (Figure 4.7) to identify magisterial districts required for further
conservation assessment. Class patch metrics were developed for each of the major vegetation
types found within a magisterial district (woodland, forest, thicket, or grassland) and then a simple
habitat index was developed to depict habitat importance and quality (e.g., White et aI., 1997). It
is calculated by taking the total area of each vegetation class and dividing by the number of
patches of that class found within each district.
To avoid the confounding nature of vegetation
area and number of patches being affected by the variable size of a magisterial district, an area
weight based on the district size in relation to the total area of the province was multiplied against
both variables in the numerator and denominator.
This transformation removed area size effects
from the index. This habitat index provides a simple measure of the size and fragmentation level
of each vegetation type.
These vegetation class indices can then be summed together for each
district and divided by the number of vegetation classes present within each district. The resulting
value is an index measure of habitat connectivity
amongst all the vegetation
types.
These
measures were only calculated for "undisturbed" vegetation types, and therefore leaving out the
lower quality degraded classes of each vegetation type (Chapter 1, Table 1.4). This approach
provides an alternative look at the landscape patterns in the magisterial districts by focussing only
on the available vegetation habitat rather than the total landscape mosaic pattern (section 6.2.2),
which measures the total pattern ofland-use and land-cover.
The detrended correspondence analysis (DCA) results using a 2nd order polynomial fit on
the LCLU data yielded strong eigenvalues on the first two axes with acceptable strength on the
third axes to be kept for further analysis (Table 6.1). The first axis's gradient length was quite
large, confirming the difference in land-uses and land-cover across the province from the coast to
the Drakensberg Escarpment.
Gradient lengths greater than three standard deviations represent
almost complete differences in features found on the opposite ends of the gradient (Table 6.1).
The DCA results are graphed as two biplots, one with the patterns of the magisterial
districts in data space and the other displaying the LCLU classes responsible
for that pattern
(Figure 6.1a). The first axis clearly separates out the urban-industrial regions from the remainder
of the province (Figure 6.1b).
The second axis of variation separates out the ex-KwaZulu
homeland areas from the White commercial farming regions based on agriculture, plantations, and
dominant vegetation type.
Using rules of thumb developed by Johnston (1980), it was decided to limit the number of
axes calculated to only those with an eigenvalue greater than or equal to one using the principal
components method applied to the standardized correlation matrix, followed by an orthogonal
(varimax) rotation of axes. Twelve factors were extracted for the socio-economic-environmental
data explaining 91 % of the variation among indicators in the dataset (Table 6.1).
Table 6.1: Eigenvalues and cumulative proportion of variance explained by principal component
analysis for socio-economic-environmental
indicators and landscape pattern indicators, and
eigenvalues and gradient length for detrended correspondence analysis of LCLU.
Factor
PCA
Eigenvalue
Cumulative
Variance
Soci-economoicenvironmental
1
2
3
4
5
6
7
8
9
10
11
12
41.49
10.94
4.70
3.64
3.08
2.60
2.51
1.95
1.83
1.34
1.13
1.01
.40
.54
.60
.67
.71
.74
.76
.79
.83
.85
.87
.91
Landscape mosaic
pattern
1
2
3
4
5
17.56
7.80
3.52
1.52
1.44
.45
.66
.77
.82
.88
Axis
DCA
Eigenvalue
Gradient
length
1
2
3
0.56
0.29
0.13
3.4
1.6
2.3
o
T
Y
G
o
E
22
4
~212 2~
2
31
,.
4
K
N
:§
P
U
H
3 3
3
M
3
3
II
I
II
A - Woodland
B _ Forest
C - Thicket
D - Shrubland
E - Grassland
F - Pasture
G - Plantations
H - Waterbodies
1- Wetlands
J - Bare
K - Erosion
L - Degraded woodland
M - Degraded thicket
N - Degraded shrubland
o - Degraded grassland
P - Irrigated ago temp.
Q - Dryland ago temp.
R - Irrigated ago perm.
S - Dryland ago perm.
T - Sugarcane
U - Dryland subsistence
V - Urban-residential
W - Small holding woodland
X - Small holding thicket
Y - Small holding shrubland
Z - Small holding grassland
AA - Urban-commercial
BB - Urban-industrial
CC-Mines
Figure 6.1: Detrended correspondence analysis biplots: (a) two axes of magisterial district data
space (numbers match Figure 6.4a); and (b) two axes of feature variable data space.
The large and general factor one is defined by population, employment, housing, and land
transformation indicators (Table 6.2). The highest loadings are on population total (POPTOTAL,
0.97), male/female population
(MALE96, 0.97; FEMALE96,
0.96), various age classes (0.81-
0.96), pupils in school (IN_SCHL, 0.96), number of children (under 14 years old) working
(CHLDNWRK,
0.94),
unemployment
(UNMPLOY,
0.96),
total people
living
(POVERTY, 0.95; NO_APP, 0.96; IND_NEC, 0.91) construction (CONSTRUC,
in poverty
0.90), modem
houses (HOUSE, 0.88), and the percentage of untransformed land (UT_PER, -0.60). This latter
moderate inverse association confirms the fact that a high percentage of untransformed
not located within or near economically active or urban-industrial
core areas.
lands is
The other strong
correlations relate to commercial and industrial enterprise, government employment, and housing
related to South African urban areas (i.e., which includes townhouses,
shack dwellings, and
worker hostels).
Factor one would thus appear to indicate that the greatest percentage of the total variance
among the eighty-four variables is explained by large populations covering all age groups, high
numbers of children in school, high child labor, high unemployment, high poverty, employment
associated with a large urban metropolitan
area, and modem housing associated with South
African urban areas. Figure 6.2a illustrates the basic pattern of geographic
variation for the
variables associated with this factor. This emphasizes the findings of similar research conducted
in Swaziland (Lea, 1972), namely that the distribution of the working population and skilled
I
employment sector give a measure of the distribution of modernization.
The axis is fundamental
to the understanding of the distribution of social and economic space in KwaZulu-Natal.
The second factor is primarily associated with high loadings on access to a flush toilet
(FTOILET,
0.87), possession
of basic household
items (BASIC, 0.85), refuse management
services (REFUSE, 0.84), access to safe water (PROXH20, 0.81), satisfaction with government
services (SAT_SERVICE,
access to electricity
0.81), satisfaction with the general environment (SAT_ENV, 0.81),
(ELECTRIC,
0.78), and poverty ratio (POV _RAT, -0.73).
A moderate
inverse association with the ratio of household dependents (-0.57), highlights the lower quality of
life the African rural areas have when compared to the urban conditions in the former Whites only
areas. Other moderate associations are satisfaction with housing (SAT_HOUSE, 0.66), functional
literacy
(F_LITERACY,
(A_LITERACY,
0.60),
male to female
ratio
(RATIO_MF,
0.58),
adult
literacy
0.57), satisfaction with economic situation (SAT_ECON, 0.56), and satisfaction
with life in general (SAT_LIFE, 0.53).
This group of variables is related to the availability of
basic "quality of life" amenities urbanize areas versus the stark realities of poverty and lack of the
most basic services in parts of the former KwaZulu homeland districts (Figure 6.2b).
The highest loading variables on this factor were around supporting development goals
within the province, including addressing people's basic needs (BNEEDS, 0.95), upgrading of
infrastructure
(upGRADE,
0.95), improving the general development
situation (DEVELOP,
0.94), and improving administrative dependability and equity (DEPEND, 0.93).
The other two
variables associated with the development group included both measures of literacy as in factor
two. It would appear that the acknowledgement
coincide with the basic attainment
of supporting development needs and goals must
of education to contribute
to community
and economic
development. The pattern of high scores for this factor corresponds to highly underdeveloped
rural districts, as well as semi-urban and urban economic areas (Figure 6.2c).
This axis represents a gradient in the availability of government
services.
The high
loadings that define the pattern are total available hospital beds (BEDSTOT, 0.86), total number
of post offices (TOTPOSTOF,
the
total
number
of
0.85), total number of police stations (TOTPOLSTA,
people
living
in
flats/apartments
(FLAT,
0.77), and
0.78).
Table 6.2: Factor loadings from principal component analysis with varimax rotation for the socioeconomic-environmental indicators based on the 1996 magisterial districts. t (Table continued
next page).
Factor
Indicator I
1
2
3
4
5
6
-0.07
-0.07
PARK]ER
-0.12
-0.13
0.09
-0.08
-0.34
-0.45
UT]ER
-0.06
0.04
-0.18
-0.60
-0.31
M]ER
0.15
-0.36
0.00
-0.15
-0.05
0.49
0.45
0.47
T_PER
0.19
0.18
0.32
POPTOTAL
0.16
-0.04
0.11
0.06
0.97
0.06
-0.D2
POPDEN
0.36
0.23
0.00
-0.12
0.86
0.07
MALE96
0.97
0.18
-0.03
0.1I
0.07
0.15
0.10
FEMALE96
-0.04
0.96
0.05
0.05
0.12
RATIO_MF
0.10
0.58
-0.03
0.05
0.27
0.07
-0.07
AGE_0_4
0.95
0.06
0.04
0.04
0.07
AGE_0_5
-0.09
-0.04
0.05
0.93
0.04
-0.10
AGE_5_14
0.93
0.06
-0.03
0.05
0.01
0.07
AGE_15_44
0.96
0.20
-0.02
0.1I
0.09
0.07
AG_15_64
0.21
-0.01
0.14
0.96
0.08
AG_65_99
0.81
0.1I
-0.02
0.48
0.01
-0.01
NO_SCHL
-0.09
-0.1 I
0.1I
0.76
-0.13
-0.05
0.07
IN_SCHL
0.19
-0.05
0.04
0.08
0.96
0.48
NO_DEGRE
0.78
0.26
0.08
-0.05
0.03
YES_DEGR
0.27
0.05
0.38
-0.01
0.12
0.85
CHLDNWRK
-0.09
0.94
0.07
-om 0.05 0.03
EMPLOYED
0.09
0.89
0.29
0.05
0.28
0.08
0.03
UNMPLOY
0.96
0.09
-0.06
-0.08
0.12
-0.57
DEP_RAT
-0.12
0.03
-0.08
-0.32
-0.03
-0.07
POVERTY
0.11
0.01
0.08
0.95
0.03
0.39
ABV]OVR
0.85
0.27
0.06
-0.02
0.10
POV_RAT
-0.27
0.08
-0.09
-0.29
-0.73
-0.10
RR_INDST
-0.03
0.10
0.10
0.04
0.94
-0.04
0.47
NR_INDST
-0.05
-0.09
-0.06
-0.18
0.02
0.32
MANUFAC
0.04
0.22
0.06
0.11
0.86
0.07
ENERGY
0.31
0.01
0.87
0.23
0.02
CONSTRUC
0.10
0.90
0.25
0.06
0.21
0.04
0.34
TRADE
0.88
0.25
0.06
0.03
0.06
TRAN_COM
0.88
0.27
0.04
0.30
0.00
0.15
BUS_SERV
0.79
0.23
0.08
0.49
-0.05
0.09
-0.D2
0.31
0.11
SOC_SERV
0.27
0.88
0.04
0.07
PRIVATE
0.04
0.89
0.27
0.16
0.06
EXT_ORG
0.70
0.10
0.04
0.36
0.08
-0.04
REP]ORG
0.80
0.23
0.05
0.38
-0.07
0.02
0.30
0.07
IND_NEC
0.91
0.04
0.21
0.06
NO_APP
0.96
0.09
-0.08
0.01
0.05
0.05
0.37
NA_INST
0.08
-0.05
0.65
0.25
0.05
0.36
HOUSE
0.88
0.02
0.17
0.1I
0.05
-0.20
TRADHOME
0.21
-0.39
-0.18
-0.13
0.06
0.11
FLAT
0.08
-0.05
0.55
0.78
0.01
0.17
TOWN
0.23
-0.02
0.85
0.04
0.01
0.14
-0.07
RETIRE
0.62
0.21
0.19
0.05
0.07
ROOM
0.34
0.30
0.00
0.29
0.73
SHCK_BCK
0.28
-0.10
0.01
0.83
0.05
0.00
0.30
SHCK_EW
0.84
0.14
0.07
0.00
0.09
FLATLET
0.70
0.22
0.10
0.09
0.23
-0.01
CARAVAN
0.05
0.21
0.19
-0.17
0.65
0.17
0.46
HOMELESS
-0.09
-0.17
0.17
-0.08
0.52
HOSTEL
0.54
0.24
0.09
0.45
0.20
0.01
SERVE_I
0.05
-0.28
0.16
-0.11
-0.09
-0.04
SE_INDEX
0.04
-0.21
0.01
0.18
0.00
-0.05
SAT_ENV
0.27
0.23
0.05
-0.04
-0.04
0.81
SAT_HOUSE
0.36
0.35
0.28
-0.04
-0.12
0.66
SAT_ECON
0.38
0.56
0.20
0.04
0.03
-0.20
SAT_SERVICE
0.36
0.21
0.03
-0.10
0.04
0.81
BASICS
0.35
0.06
0.18
0.04
0.14
0.85
0.03
DEVELOP
-0.05
0.22
0.07
0.02
0.94
0.14
-0.08
BNEEDS
0.95
0.04
0.02
0.04
UPGRADE
-0.15
0.06
0.03
0.95
0.04
-0.10
SAT_LIFE
0.36
0.53
0.28
0.00
0.05
-0.28
0.07
DEPEND
-0.02
0.15
0.03
0.01
0.93
PROXH20
0.34
0.81
0.10
0.1I
0.1I
0.22
t Factor loadings> 0.50 are indicated in bold, factors> 0.70 are considered
i Variable definitions are found in Appendix A.
166
7
-0.77
-0.25
0.14
0.15
0.00
0.06
0.00
0.00
-0.26
-0.02
-0.03
-0.03
0.00
0.01
0.02
-0.13
0.03
-0.02
0.02
-0.03
0.03
0.00
0.22
-0.02
0.03
0.19
-0.03
-0.23
0.09
0.00
0.02
0.03
0.02
0.01
0.00
0.02
-0.03
0.01
0.01
-0.01
-0.09
0.03
-0.03
8
-0.15
-0.13
0.19
-0.01
0.04
0.02
0.03
0.04
-0.30
0.04
0.07
0.07
0.02
0.02
0.06
0.19
0.00
-0.01
0.00
0.06
0.01
0.00
-0.02
0.06
0.01
0.05
0.04
0.06
0.04
-0.01
0.01
0.00
0.00
0.00
-0.01
-0.02
-0.08
0.06
0.03
0.05
-0.09
-0.01
0.13
-om -0.D2
0.11
0.09
0.01
0.05
0.07
0.08
0.09
-0.01
-0.1 I
0.07
0.09
0.05
-0.1 I
0.03
0.04
0.01
-0.41
-0.12
-0.16
0.06
-0.02
-0.03
0.03
-0.05
-0.06
-0.13
0.12
0.35
0.04
-0.03
0.13
-0.04
-0.03
-0.02
-0.08
-0.04
-0.D2
-0.1 1
0.01
0.08
-0.07
-0.09
0.07
-0.13
significant.
9
-0.05
-0.D2
-0.16
0.14
-0.06
0.06
-0.04
-0.08
0.39
-0.17
-0.20
-0.22
0.00
0.00
-0.01
-0.32
-0.06
0.18
0.14
-0.20
0.1I
-0.07
-0.29
-0.14
0.14
-0.15
-0.05
0.03
0.05
0.14
0.13
0.1I
0.10
0.19
0.12
0.18
0.03
0.00
0.1I
-0.15
0.21
0.09
-0.43
0.15
0.07
0.25
0.17
0.09
0.07
0.23
0.06
0.05
0.06
-0.84
-0.90
-0.05
-0.01
0.03
0.07
0.14
-0.D2
-0.1 I
-0.09
0.18
-0.01
0.15
10
0.14
0.08
0.18
-0.22
-0.02
0.07
-0.01
-0.03
0.14
-0.02
-0.03
-0.06
-om
-0.01
-0.07
-0.08
-0.03
0.01
0.03
-0.04
0.05
-0.06
0.16
-0.02
0.07
0.11
-0.10
-0.1 I
0.13
-0.02
0.12
0.09
0.09
0.05
-0.01
0.01
0.19
0.10
0.00
-0.05
-0.21
-0.01
-0.32
0.09
0.11
0.04
0.00
0.21
0.17
-0.06
0.04
0.34
-0.09
0.04
0.00
0.03
0.07
0.02
0.07
0.03
-0.1 I
-0.04
-0.06
-0.17
-0.12
-0.05
11
0.03
0.01
-0.04
0.02
0.04
-0.06
0.03
0.05
-0.08
-0.02
0.02
0.02
0.02
0.03
0.22
0.04
0.01
0.19
0.08
0.01
0.05
0.01
0.01
0.00
0.05
-0.07
0.1I
-0.37
-0.09
0.18
0.07
0.01
-0.05
0.12
0.09
0.19
-0.16
-0.20
0.09
0.02
0.39
0.13
0.25
0.00
-0.12
0.56
0.31
0.06
-0.20
0.45
0.49
0.06
-0.09
-0.08
-0.06
0.09
0.14
0.11
0.10
0.00
0.02
0.00
-0.02
0.06
0.04
-0.03
12
0.18
0.33
-0.72
0.20
-0.09
0.07
-0.08
-0.1I
0.18
-0.18
-0.22
-0.23
-0.03
-0.03
-0.16
-0.35
-0.06
0.00
0.04
-0.22
0.09
-0.1 I
-0.42
-0.16
0.08
-0.21
0.13
-0.33
0.16
0.07
0.07
0.10
0.10
0.06
0.03
0.05
0.12
0.17
0.08
-0.17
-0.12
0.04
-0.50
0.02
0.17
0.01
0.04
0.13
0.16
0.05
-0.01
-0.32
-0.07
-0.10
-0.07
0.12
-0.04
-0.33
0.06
0.07
0.12
0.07
0.09
-0.29
0.12
0.07
Table 6.2: Continued.
0.10
ELECTRIC
0.36
0.13
0.78
0.Q3
REFUSE
0.36
0.84
0.14
FTOILET
0.27
0.01
0.17
0.87
A_LITERACY
0.34
0.57
0.50
0.08
F_L1TERACY
0.20
0.60
0.60
-om
-0.32
FOR]ER
-0.03
-0.13
-0.08
GRS]ER
-0.49
0.16
0.10
-0.11
WET]ER
0.01
-0.08
0.14
0.00
-0.36
LOWI]ER
0.14
-0.31
0.02
PLNT]ER
-0.08
-0.09
0.29
-0.02
0.31
DRY]ER
0.14
0.06
-0.08
-0.1 9
IRR]ER
0.04
0.14
0.15
0.33
URB]ER
0.43
0.03
0.25
0.42
0.38
PERCAPlNC
0.52
0.13
TOTPOLSTA
0.38
0.27
0.03
0.77
TOTPOSTOF
0.31
0.21
0.11
0.85
BEDSTOT
0.36
0.09
0.04
0.86
0.30
TELSHAREPR
0.36
0.08
0.55
f Factor loadings> 0.50 are indicated in bold, factors>
I Variable definitions are found in Appendix A.
0.13
-0.01
0.09
0.11
-0.Q7
0.13
-0.06
0.15
-0.09
0.05
0.14
0.13
-0.05
0.14
-0.01
0.12
-0.Q2
0.06
-0.09
0.11
-0.Q7
-0.05
-0.27
0.81
-0.34
-0.24
0.03
-0.62
-0.87
-0.04
0.24
0.00
-0.15
-0.06
0.13
0.18
-0.Q3
0.15
0.10
-0.09
0.87
-0.10
0.08
0.02
-0.62
-0.09
-0.05
-0.19
0.Q7
0.Q7
-0.17
0.72
-0.35
0.10
-0.06
-0.01
-0.03
0.00
-0.03
0.08
0.Q7
0.08
0.08
0.01
0.00
0.14
-0.04
-0.10
-0.16
0.19
0.00
0.06
0.70 are considered significant.
The housing variable refers to low cost high-density
modem urban living conditions.
0.11
0.13
0.17
-0.17
-0.12
-0.10
0.06
-0.11
-0.16
0.02
0.09
-0.06
0.11
0.00
-0.Q7
0.02
0.Q3
-0.05
-0.03
0.05
0.02
0.24
0.22
0.04
0.04
0.00
0.18
-0.81
-0.02
-0.01
0.14
0.03
-0.03
0.01
-0.02
0.09
-0.05
0.00
-0.Q3
0.09
0.Q7
0.11
-0.06
-0.04
-0.04
-0.01
-0.06
0.17
0.06
0.13
0.12
0.08
-0.Q7
0.06
0.01
0.10
0.10
-0.05
-0.06
0.02
0.28
-0.02
-0.72
0.14
0.05
0.35
0.10
0.23
0.08
0.Q3
-0.08
0.33
living conditions that tend to represent
Urban and semi-urban areas tend to have more health, postal,
and crime prevention facilities proportional to population density.
This pattern is illustrated by
the high factors found in the Durban and Pietermaritzburg magisterial districts versus many of the
former KwaZulu homeland and rural districts with low populations (Figure 6.2d).
Number of employed people in renewable resource industries (RR_INDST, 0.94) and the
percentage of commercial dryland agriculture (DRY_PER, 0.87) possess the highest loadings for
this factor, these two variables being clearly interrelated.
The only other moderately significant
loading is the total number of people considered homeless (HOMELESS,
0.52), suggesting a
relationship between commercial agriculture and the migrant African labor it employs.
Figure
6.2e highlights the high scoring magisterial districts along the coasts consisting of sugarcane
farming.
This factor has a clear association with human population growth and urbanization.
loadings are with the 1996 population
density (POPDEN, 0.86) and the percentage
High
of land
classified as urban CURB_PER, 0.72). The result for this factor should be compared to the size of
the magisterial
district that the indicators are calculated by, as the results are clearly scale
dependent. The high loading magisterial districts of Chatsworth and Umlazi (see also Figure 1.10
and Table 1.8) shown in Figure 6.2f illustrate the size scale problem, but the variable pattern is
reliable.
D
< -1.0
D
-1.0 - 0.0
III
0.0 -1.0
•
1.0 >
Figure 6.2: Factor patterns of variation derived from principal component analysis (peA) of the
socio-economic-environmental
indicator data set, where shading indicates factor scores.
The highest loadings on this axis represent the percentage of land covered by waterbodies
and wetlands (WET_PER, -0.87) and the percentage of land conserved in protected areas
(PARK_PER, -0.77). This pairing of indicators reflects the policy of protected areas being
developed around dam constructions (DWA, 1986), as well as the protection of the major
wetlands in KwaZulu-Natal's north coast regions. This was done historically to combat malaria
(Pringle, 1982) in the region. The pattern of factor scores (Figure 6.2g) illustrates the same results
as outlined in Chapter 3, however the PCA extracts the detail of the historical nature of
conversation development in the province better. The interior areas along the Tugela River,
central Zululand, Midlands, and south coast regions are largely under protected.
Factor 8, the vegetation structure/land-use
dimension
This axis relates to the duality in major vegetation structural types covering the province
(Figure 6.2h; also see Figure 1.3 and 1.7 for detail). The high loadings are with percentage land
covered in forest and woodland (FOR_PER, 0.81) and the moderate inverse represented by
percentage of land covered in grassland (GRS_PER, -0.62) and the percentage of land under
commercial irrigated agriculture (IRR_PER, -0.62). This relationship contradicts the perception
in the province that grassland and plantation forestry cover are related to one another (e.g.,
Armstrong et aI., 1997). The results provide evidence for agriculture as the primary related
transformation agent within grasslands in the province and supports earlier conjectures by
Fairbanks et ai. (2000) for South Africa as a whole.
The socio-economic index (SE_INDEX, -0.90) and servIce index (SERVE_I, -0.84)
loaded the highest on axis of factor nine. These two indices provide a global view to encapsulate
the development profile and needs of the magisterial districts. The axis displays a socio-economic
and service gradient, displaying districts requiring development intervention. The pattern on this
axis generally makes sense with larger areas of the former homeland areas requiring development
assistance and the former White controlled areas and Durban economic core requiring less
assistance (Figure 6.2i). Two districts, Babanango and Kranskop, do not quite fit the gradient
pattern, but were both noted in the original data set (Kok et aI., 1997) as having tentative survey
results due to small sample sizes for these indicators. In addition, the eigenvalues and importance
of the variance explained on this axis is very low in comparison to the previous factors.
The percentage of land covered in exotic woodlots and commercial plantations
(PLNT_PER, -0.81) is the only significant loading on this axis. Figure 6.2j clearly shows the
pattern of the commercial forestry plantation sector and the importance of woodlots in the
homelands not endowed with sufficient natural woodlands (particularly in the southern midlands
and south coast). As explained for factor nine, this axis does not contribute significantly to the
variation within the province.
The loading on this factor is moderately significant with a mmor contribution to
explaining the variation. It relates to the total number of people living in retirement villages and
retirement holiday homes (RETIRE, -0.56). The eigenvalue is very low and the contribution to
overall explanation to the variation in the province is negligible. The pattern of the axis is,
however, interesting as it highlights areas with high numbers of pensioners residing along the
south coast and Midlands regions (Figure 6.2k). Each of these areas is prized for their scenic
natural beauty and tranquil living conditions (see www.kzn-deat.gov.za/tourism; www.tourismkzn.org for details).
The last factor had high loadings represented by two variants of percentage land covered
by low intensity transformation (M_PER, -0.72; LOWCPER, -0.72). This result is surprising
since there is a strong difference in levels of land transformation between the former KwaZulu
homeland areas and the original white managed regions. The duality was obviously not strong
enough to be extracted earlier in the principal component analysis, or has been distorted with the
zoning created for the 1996 census. The other loading barely considered as related, regards the
total number of people living in traditional built homes (TRADHOME, -0.50), which is associated
with subsistence agriculture and degraded rangelands. The eigenvalue and contribution to the
overall variance is, however, the lowest and could be dropped from the overall results without any
loss of provincial description. It is shown because it does describe and highlight spatially (Figure
6.21)the levels of low intensity transformation within the province. The Msinga and Ndwendwe
magisterial districts have the largest factor scores (see Figure 1.10 and Table 1.8).
Significant eigenvalues were calculated for five axis factors from the landscape mosaic
pattern indicators explaining 88% of the variation among indicators in the dataset (Table 6.1).
The factor loadings are illustrated in Table 6.3. In KwaZulu-Natal magisterial districts, indicators
depicting contagion and texture measures of diversity and dominance dominated the first axis
(Figure 6.3a). The latter strong inverse association with the diversity and dominance measures
indicated that magisterial districts in the Maputaland, coastal, Midlands and parts of the Zululand
region are characterized by a high population diversity of classes in relative equity. Whereas one
coastal district (dominated by sugarcane) and much of the districts in the interior grassland areas
are characterized by large contiguous patches. The second axis gradient contrasted districts with
large mean core areas per patch of LCLU versus those districts with large densities of core area
patches (Figure 6.3b). This axis largely reflects class-fragmented
lower class fragmentation
districts from districts with
and large patches. Axis three characterizes the higher complexity of
patch shapes in mainly the Midlands and Zululand regions versus districts with richer and higher
densities of classes found in the economic core area and Maputaland region (Figure 6.3c). Class
richness is a product both of exposure to human interference and intensity of human activity, and
components of natural landscape diversity found along the north coast. Indicators of distances
between patches (Figures 6.3d and e) dominate the fourth and fifth factors.
As with other
research experiences (e.g., Cain et aI., 1997; Riitters et aI., 1995) the results here confirm from a
statistical point of view that, most landscape mosaic pattern indicators actually measure one of
just a few independent dimensions of pattern.
Many indicators are redundant, which appears to
dampen the need to calculate many pattern metrics in this region of South Africa.
Figure 6.4a, b, and c presents the results obtained by hierarchical and k-means clustering.
Geographically
contiguity.
compact groups were obtained without applying any constraint of geographic
This indicates strong regional trends in all the datasets subject to the analysis.
The
general structure consists of clumped groups of magisterial districts following a pattern from the
coast to the Drakensberg
Escarpment
or from the northern Zululand woodlands to the high
grassland regions. The use of magisterial
districts suggests that the analysis of LCLU and
landscape pattern indicators support claims by Cain et aI. (1997) that land patterns are more
Table 6.3: Factor loadings from principal component analysis with a varimax rotation for the
landscape pattern indicators derived from the 1996 magisterial districts. t
Indicator*
LPI
NP
PD
MPS
PSSD
PSCV
MSI
AWMSI
FD
MPFD
AWMPFD
TCA
NCA
CAD
MCAPP
PCASD
PCACV
MAPDC
DCASD
DCACV
TCA]
MCA]
MNND
NNSD
NNCV
MPI
SHDI
SOl
MSDI
CR
CRD
SHEI
SEI
MSEI
II
CI
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
0.95
0.38
0.36
-0.30
0.86
0.83
-0.79
0.66
-0.63
-0.65
0.40
0.71
-0.14
0.02
-0.22
0.86
0.83
0.19
0.72
0.86
0.61
-0.65
0.01
-0.05
-0.16
0.53
-0.94
-0.97
-0.97
-0.09
0.25
-0.97
-0.97
-0.97
-0.78
0.96
-0.11
-0.74
-0.76
0.90
0.21
-0.22
0.46
-0.30
0.08
0.42
-0.14
0.49
0.11
0.35
-0.48
0.07
0.37
0.47
-0.09
0.59
0.21
-0.14
0.84
-0.10
0.28
-0.27
0.06
0.34
0.46
0.03
0.56
0.37
0.01
0.08
-0.04
-0.02
0.16
0.65
0.01
0.04
-0.05
0.22
-0.76
-0.04
0.02
-0.07
-0.02
0.03
-0.05
0.17
-0.08
-0.01
0.08
0.10
-0.14
0.07
-0.19
-0.23
0.17
0.05
0.01
-0.11
0.01
0.08
0.10
0.13
0.13
0.11
0.01
0.13
0.71
0.48
-0.01
-0.10
0.25
0.11
0.12
0.79
-0.16
-0.01
0.05
-0.05
0.22
0.03
-0.10
-0.11
-0.07
0.19
0.03
-0.10
0.30
0.09
0.41
0.28
0.12
-0.07
-0.03
0.04
0.16
0.02
-0.10
0.05
-0.07
-0.13
-0.19
0.20
0.24
0.74
0.90
0.05
0.08
0.08
0.08
-0.01
-0.12
0.04
0.06
0.05
-0.14
-0.04
-0.85
-0.92
0.93
0.23
-0.24
0.92
0.34
-0.26
0.70
0.47
0.58
0.43
0.07
0.10
0.01
0.D7
0.09
-0.06
-0.25
0.02
0.06
0.09
0.11
0.04
f Factor loadings>
0.50 are indicated in bold, those factors>
• Variable definitions are found in Chapter I, Table 1.5.
0.70 are considered significant.
homogeneous within areal units that roughly correspond to physiographic and political economy
processes that determine land cover pattern over large regions. The magisterial districts are
defined not unlike catchments, as their boundaries tend to follow major rivers and physiographic
divides, with minor exceptions in the northern Zululand region.
The unimodal ordination and hierarchical clustering analyses derived five "co-evolved"
LCLU regions within the province (Figure 6.4a).
These regions correspond to the spatial
economic development of the province emanating outwards from the Durban metropolitan
economic core. Durban harbor started as the original colony by the British on the east coast of
South Africa in 1835, with Pietermaritzburg being founded just inland from Durban in 1838.
Since 1911, Durban has witnessed an economic consolidation due to its strategic location
Factor scores
D
D
II
•
Figure 6.3: Factor patterns of variation
landscape mosaic pattern indicators.
derived from principal
and proximity to the Gauteng province mining-industrial
<-1.0
-1.0 - 0.0
0.0- 1.0
component
metropolitan
analysis of the
region (Browett, 1976).
This proximity has also driven its emergence as the eighth busiest port in Africa (see Christopher,
1982). At the analysis scale of magisterial district, the pattern is similar to Von Thtinen's (1826;
see Bradford and Kent, 1986) prediction of zones of intensity in economic activity, diffusing
outwards via an economic core-periphery pattern as defined by Friedmann (1972).
for KwaZulu-Natal
The pattern
is very similar to the development pattern described for New South Wales,
Australia by Rutherford et a1. (1966; see Haggett et a1., 1977). In this case the economic core is
based at Sydney (with its industrial habor) and renewable resource zones emanate out from the
core to the interior bush. The regionalization ofKwaZulu-Natal
industrial core to sugarcane-forestry,
agriculture-forestry,
illustrates a gradient from urban-
and rural African subsistence regions.
Co-evolved regions
~
(I) Subsistence: woodland
(2) Agriculture-forestry: grassland
D
••
(3) Subsistence: grassland-thicket
(4) Sugan:ane-forestry: grassland-thicket
(5) Urban-industrial: thicket-grassland
••
D
D
D
~
Socio-economic
regions
(I) Economic core
(2) Economic-urlxm corridor
(3) Peri-urban
(4) Ex-homeland develoJX11entregions
(5) Corrmercial sugarcane-retirerrent & holiday regions
(6) Renewable resources growth regions
~...
(1) Sparsely developed-irrigated agriculture & mining
§
(8) Underdeveloped deep rural regions
Landscape mosaic
classes
~...
•
•
2
~
3
D
4
5
In the case of this province, the policy of homeland separate development created by the Exapartheid State distorts the land-use intensity gradient (see Fair and Schmidt, 1974). In particular
the subsistence:grassland-thicket zone has three districts, Umbumbulu, Ndwendwe, and
Mapumulo (Figure 1.10), situated within near proximity of two major economic zones. These
districts are not consistent with the economic intensity development pattern. These districts
represent ex-homeland regions created close to the White controlled commercial farming and
industrial core areas for African labor exchange purposes, owing to their failure to develop
economic core activities in their own right due to past government economic policy (see Fair and
Schmidt, 1974; Board, 1976).
The five zones were also analyzed for their association with dominant landscape
structures as developed in Chapter 3. Figure 6.5 details the dominant landscapes found within
each magisterial district, while the dominant landscapes identifying the five co-evolved regions
are: (1) coastal undulating/flat dry; (2) highland undulating/flat moist; (3) highlands
undulating/flat dry; (4) lowlands undulating/flat moist; and (5) coastal undulatinglflat moist. The
most striking feature of this result is the poor moisture regimes of the ex-KwaZulu homeland
areas, which would have partly hampered commercial dryland agriculture and plantation forestry
development (and therefore the first stages of economic development).
An ANOVA was
performed to determine whether there were statistically significant differences in co-evolved
regions based on the variety of landscape types (Figure 3.5) found within a magisterial district.
The hypothesis considered was that the diversity of LCLUs found in a district was dependent on
the variety of the original landscapes found within that district. Since the regions were developed
from ordination axes of the diversity and abundance of LCLU types this could be tested (a
Spearman rank correlation of landscape variety against DCA axis 1 was -0041). The analysis
postulated that the properties of the physical environment still significantly affect spatial patterns
of human activity, despite uneven development patterns determined by past government policy.
The landscape diversity explained 41% (N= 52; P < 0.0001) of the spatial structure derived from
LCLU analysis. By removing current magisterial districts that were completely ex-KwaZulu
homeland territory (see Figure 1.lOb), a moderate increase to 52% (N = 37; P < 0.0001) could be
explained by landscape diversity. It can be tentatively concluded that the co-evolved regions of
development within KwaZulu-Natal are partly due to physical environment constraints and
opportunities.
Nevertheless, former apartheid policy and other economic development
instruments have added distortions to the present pattern.
Dominant landscape type
•
•
D
D
:0
· .
[ill]
8
Coastal I.D1dulating/flat dry
Coastal I.D1dulating/flat moist
Lowlands mountainousihilly
dry
Lowlands mountainous/hilly
rrnist
Lowlands undulating/flat
dry
Lowlands undulating/flat
moist
Highlands mountainous/hilly
moist
~
Highlands undulating/flat
dry
~
~
~
Highlands undulating/flat
moist
..
Afro Alpine mOl.D1tainouslhilly wet
Figure 6.5: The landscape types identified in Chapter 3 are used to identify the dominant class for
each magisterial district based on a simple majority.
Figure 6.4b illustrates the result obtained from k-means clustering after examining for
values of k (the number of clusters) from 5 to 13. There was strong regional socio-economicenvironmental trends in the data and the clustering generally showed that above thirteen groups,
many local minima are found with very similar values for the within-group
sums of squares
criterion, so that no single clustering structure emerges; this is why the analysis was stopped at
thirteen groups.
The maps for each clustering criterion were examined visually and with the
hierarchical linkage plots. A natural level of eight homogeneous clustered regions was selected,
as these could be explained based on other independent criteria (see Legendre and Legendre,
1998). Using terminology developed by Mydral (1958) the significance of the results from the
PCA and clustering of the socio-economic-environmental
indicators is addressed.
Myrdal (1958)
coined the terms spread and backwash to describe the various economic and social effects
associated with unbalanced growth, with spread effects being beneficial to the periphery and
backwash effects as detrimental.
The relative strengths and spatial distribution of these effects
determine whether the gap between core and periphery is widening or narrowing.
Spread effects
are the mechanism whereby growth is transmitted from core to periphery (i.e., growth at the core
may be expected to generate demands for the products of the periphery).
On the flipside,
backwash effects show evidence that growth may seem to be transmitted to the periphery but
instead are actually refocusing growth at the core (i.e., capital transferred to the periphery in
return for its products 'leaks' back to the core because it is spent on goods which the periphery
cannot provide).
PCA factors one and two show evidence of the spread effect. The first factor can be
regarded as society's ability to consume and the volume of entrepreneurial activity, which would
be absent or poorly represented in areas where backwash effects are dominant. In the case of
KwaZulu-Natal, the backwash effects are most pronounced near the economic core area (Figure
6.2b), especially within ex-KwaZulu homeland areas. The overflow of wealth has ensured that
the spread effects are strongest in the magisterial districts in the immediate vicinity of the core
(Figure 6.2a).
The landscape mosaic pattern indicators clustered tightly together into five classes (Figure
6.4c) which were then defined on a continuum based on their ability to conserve resources and
maintain biodiversity as defined by Ludwig (1999), described in Chapter 2 (Figure 2.5), and by
mosaic pattern characteristics outlined by Forman (1995). Class three represents many of the
economically active areas, but contain similar landscape mosaic structure as Maputaland and
other poor diverse homeland districts, such as Msinga. Classes one and two represent magisterial
districts that would be considered to be functioning well for conservation purposes except for the
Lower Tugela district which is completely covered in sugarcane, which mimics natural grassland
pattern.
Several of the magisterial districts consisting of ex-KwaZulu homeland areas have
landscape pattern compositions that are moderate to weak in functionality. These districts had
higher proportions of high intensity and low intensity transformed land than the other exhomeland areas. Results from a contingency table analysis were used to compare the landscape
pattern against the socio-economic regions and co-evolved regions. The hypothesis being that the
landscape patterns and spatial distribution among the magisterial districts are dependent on the
past and present socio-economic development. The p value for the Pearson chi-square (p <
0.0001, d.f. 35) for the landscape ranked clusters against the co-evolved regions and against the
socio-economic regions (p < 0.0001, d.f. 42) leads one to believe that the landscape clusters are
dependent of the economic clusters (Ho: no interaction between landscape pattern and human
economic development). Since the significance tests could be suspect because of sparsely
populated cells, Cramer V and contingency coefficients (Systat 8.1, 1998) were calculated to
adjust for misleading interpretations. They were, respectively, 0.59 and 0.76 for landscape versus
co-evolved regions and, 0.76 and 0.84 for landscape versus socio-economic region, with numbers
approaching one conveying dependency among the tested variables. In this case, the socioeconomic regions appear to provide a better predictor of magisterial district landscape "health"
than the calculated co-evolved regions derived from LCLU data. The measurement of the total
mosaic of LCLUs do become confounded in the clustering operation because of pattern
similarities found in districts with complete habitat versus habitat replaced with one agriculture
cover type or diverse areas mimicking high intensity human-use areas.
The CN2 algorithm produced stable results of if-then rules for identifying all the data set
clusters against every other data set used in the analysis.
Because the pattern search algorithm
uses exact rule definitions, based on the data resolution provided to it there is a tendency for the
if-then results to appear overly precise in their definition.
For example, many of the variables are
defined to two decimal places, however this may be spurious precision, and thus the results should
be interpreted in a more generalized manner.
Tables 6.4 through 6.7 provide the if-then rule
complexes for each cluster region, many of the regions required up to four sets of rules for
definition.
This pattern recognition algorithm makes it clear that by its procedures many of the
"homogeneous" clusters have internal variation that cannot be defined by one rule set and instead
provide evidence towards local minima requiring separate definitions within many of the cluster
groups for either data set. This is more than likely an artifact of the processing of the regional
clusters based on the classification of linearly derived factor and correspondence
which will integrate correlated variables into one dimension.
analysis scores,
The pattern recognition algorithm
uses the original raw values to develop rules for another data sets regional clusters, therefore
loosing any previous indicator variable relationships between data set groups.
Table 6.4 illustrates if-then rules for the landscape pattern indicators that identify the
socio-economic regions. The pattern in the cluster definitions shows that classes one and two can
be defined by shape and nearest neighbor indicators, class three by patchiness,
classes four
through seven by patch size and feature richness, and class eight by patch shape complexity and
landscape
diversity.
Table 6.5 provides
detrimental effects on the environment.
further evidence
of Western
style development's
Class one is telling, essentially to have near-pristine
grassland landscapes the people living there should remain without basic standards of living (e.g.,
running water, electricity,
sanitation, etc.), and therefore stymied development
and stagnant
poverty. In contrast, classes six and seven are identified by human population density, especially
the amount of men that infiltrate the economic core, leaving their families behind in the rural
areas, which explains the presence of the low male to female ratios in classes three and four (see
example systems model outlined in Chapter 2). Tables 6.5 and 6.6 provide the rules derived from
the socio-economic-environmental
regions.
and landscape pattern indicators for defining the co-evolved
In Table 6.6 class three is extremely variable showing the extreme differences in
development paths taken within various areas of the former KwaZulu and Transkei homeland
territories. Class one shows the 'cut off from society' and wilderness character of the Maputaland
Table 6.4: If-then rules of landscape pattern indicators describing clusters developed by PCA
classification of the socio-economic-environmental indicators. t
Class
Rule set I
Rule set 2
Rule set 3
MPS <518
MNND> 1949
2
MPS > 251
PSSD < 1404
3
NP<71
4
PSSD < 2506
MAPDC>438
186.00 < NP < 198
5
MAPDC> 657
MPS > 391
TCA] <78%
6
LPI < 24
NCA < 1053
MAPDC<438
CR> 12
LPI < 34
MCA] >34%
CR> 15
7
MPS< 553
PSCV> 866
PSSD < 3135
PCACV> 762
8
PSSD > 2507
FD> 1.28
MAPDC> 398
SHDI < 1.83
MSI> 1.73
MNND< 1058
NNCV< 184
FD< 1.27
MCA] > 34%
t Variable definitions are found in Chapter I, Table 1.5.
Table 6.5: If-then rules of socio-economic-environment indicators describing clusters developed
by PCA classification of the landscape mosaic pattern indicators.t
Class
Rule set 1
Rule set 2
SHCK_BCK > 26%
GRS]ER> 50%
RATIO MF> 1.0
2
T]ER> 0.4%
HOSTEL<45
PERCAPINC < 1726
T]ER>21%
PERCAPINC < 2504
3
TRADHOME < 24824
ROOM>255
SERVE_I> 83%
SE_INDEX < 134
M]ER> 1.4%
RETlRE>63
UPGRADE> 80%
PARK]ER> 5%
DEVELOP < 75%
4
HOMELESS < 13
DRY]ER> 14%
UT]ER<54%
SERVE 1>99%
NR_INDST> 2961
RETIRE> 1129
5
POPDEN>27
t Variable definitions are found in Appendix A.
Table 6.6: If-then rules of socio-economic-environment
by DCA classification of the LCLU abundance data.t
Class
Rule set 1
Rule set 2
PARK]ER> 7%
TELSHAREPR < 0.33
RR INDST < 127
2
GRS]ER>
POPDEN < 0.70
PLNT]ER > 19%
3
RR_INDST> 137
A_LITERACY < 81%
LOWI]ER> 25%
UT]ER> 51%
RATIO MF < 0.83
FOR]ER> 26%
4
DRY]ER>
RR_INDST> 5823
5
PROXH20 > 92%
49%
18%
indicators describing clusters developed
POPTOTAL > 220366
WET]ER> 3%
POPTOTAL> 193529
DEPEND < 80.50
t Variable definitions are found in Appendix A.
Table 6.7: If-then rules of landscape pattern indicators describing clusters developed by DCA
classification of the LCLU abundance data. t
Class
Rule set 1
Rule set 2
LPI < 13
MPFD> 1.08
2
MPS < 553
PSCV> 840
PSCV < 669
FD> 1.29
MPFD < 1.07
3
MPFD> 1.07
NNCV> 160
SHDI < 1.73
NP < 701
PSSD < 5426
FD> 1.30
MCAPP> 393
4
LPI> 17
MPS > 263
DCACV>451
II> 58
MAPDC> 657
5
PSSD < 1404
LPI < 47
MPI> 2645
LPI < 52
NNCV< 158
t Variable definitions are found in Chapter I, Table 1.5.
region with low numbers of available telephone shares and extensive conservation areas. Class
five is interesting as it illustrates that the high access to safe running water is the most important
indicator defining the economic core.
The class rule definitions in Table 6.7 follow the same
general pattern as for class definitions in Table 6.4.
The priority conservation grid cells based on the "ideal" conservation
system selection
derived in Chapter 4 (Figure 4.7) were overlaid on the magisterial districts to identify districts that
had more than a third of at least one cell covering them. Figure 6.6 details the magisterial districts
that would be required
conservation.
to implement
landscape
management
plans
for sustainable
avian
The first thing to note from this comparison is the grouping nature of the grid cells,
which seem to be defined by major river basin boundaries.
The Drakensberg group is defined by
the Mooi and Buffalo River valleys, Maputaland is defined as areas north of the Black and White
Mfolozi Rivers, the Midlands group is defined south of the Tugela and Mooi River valleys but
north of the Urnzimkulu River valley, and the Zululand group is nestled between the Tugela, and
White Mfolozi rivers with cells primarily along the entire length of the Mhlatuze River.
This
result provides further evidence for previous studies conducted by Clancey (1994) and in Chapter
4 on the zoogeographic nature of river valleys as barriers to avian dispersal in South Africa, with
special reference to KwaZulu-Natal province.
Table 6.8 illustrates the comparison of magisterial district against the socio-economic
condition and landscape mosaic. Six of the magisterial districts requiring conservation action are
considered
severely underdeveloped
African rural areas, and they cover three of the bird
conservation regions. Nine of the magisterial districts are classed as sparsely developed-irrigated
agriculture and mining, these districts from the PCA analysis appear to be in a stagnant economic
development
cycle but with healthy landscape
mosaic structure
for biodiversity
(i.e., low
landscape diversity and high contagion of natural vegetation cover). Of concern are the third of
the magisterial districts that are classified as renewable resource growth regions.
These areas
represent
and human
the "economic
frontier" of the province,
resource potential for further economic development.
consisting
of environmental
These districts contain already poor to fair
landscape mosaic patterns, but areas like Eshowe are already showing a trend towards low
landscape diversity with high local diversity caused by commercial agricultural development and
over grazing.
economic
These districts will require immediate integration of landscape conservation and
development
conservation,
plans
to ensure
an equitable
and human developmental needs.
trade-off
between
avian
diversity
Chatsworth and Umlazi represent districts that
are fully dysfunctional for long-term avian diversity maintenance; with poor landscape structure
representing relict stands of habitat (e.g., McIntyre and Hobbs, 1999).
Many of the districts require that particular physiographic features will be required to be
protected.
These areas include the central and northern sections of the Drakensberg escarpment,
the Lebombo Mountains, and sections of eleven major rivers or their river mouths (Table 6.8). In
particular, entire river stretches that must be buffered and considered in an avian conservation
plan include the Mhlatuze,
Pongola and Mkuze Rivers, with the Mhlatuze River mouth at
Richard's bay also requiring attention.
Other river sections include the entire upper reaches of
Settlement hierarchy
0
Economic core
0
Major metropolitan center
0
Development nodes
a
Regional marlcet centers
Periphery towns
IV
IV
•
D
D
Major rivers
Avian physiograPtic
Current p-otected areas
Priority con~rvation
Magisterial districts
•
Avian conservation regions
zones
areas
D
D
D
~
Drakensberg
Maputaland
Midlands
ZuluJand
East coast
Figure 6.6: Priority avian conservation areas from the "ideal" model developed in Chapter 4,
associated magisterial districts, and a general regionalization of the bird conservation areas by
physiographic boundaries.
Table 6.8: Magisterial districts requiring landscape conservation plans for avian conservation,
with associated socio-economic factors that will need to be addressed for sustainable conservation
(also see Table 6.5 and 6.6).
Conservation
region
Magisterial
district
Socio-economic
region
Drakensberg
Drakensberg
Drakensberg
Drakensberg
Drakensberg
Drakensberg
Drakensberg
Drakensberg
East Coast
East Coast
East Coast
Maputaland
Maputaland
Maputaland
Maputaland
Midlands
Midlands
Midlands
Midlands
Midlands
Midlands
Midlands
Zululand
Zululand
Zululand
Zululand
Zululand
Bergville
Estcourt
Dannhauser
Kliprivier
Newcastle
Paulpietersburg
Underberg
Utrecht
Durban
Hlabisa
Mtunzini·
Ingwavuma
Ngotshe
Nongoma
Ubombo
Camperdown
Ixopo
Lions River
Pietermaritzburg
Polela
Umvoti
Urnzinto
Babanango
Eshowe
Mthonjaneni
Nkandla
Vryheid
Low development
Low development
Low development
Renewable resource
Low development
Low development
Low development
Low development
Economic core
Renewable resource
Renewable resource
Underdeveloped
Renewable resource
Underdeveloped
Renewable resource
Renewable resource
Renewable resource
Low development
Renewable resource
Underdeveloped
Renewable resource
Renewable resource
Underdeveloped
Renewable resource
Underdeveloped
Underdeveloped
Low development
r The primary
I The primary
§ The primary
• Durban and
trawler refuse
and should be
.
Landscape
mosaic
3
1
4
3
I
1
I
1
3
3
3
3
3
2
3
4
4
1
3
1
3
4
2
3
2
2
1
Priority
habitats
Grass; forest; thicketf. t
Grass; thicket; woodlandt
Grass; wetland; woodlandt
Grass; forest; thicket; woodlandt. t
Grass; forestt
Grassland; forest; wetlandt. t
Grass; shrublandt
Grass; wetlandt
Forest; bay mudflats; thickett
Forest; wetland; woodlandt
Forest; thicket; bay mudflatst
Woodland; forest; wetland§' t
Woodland; forest; grass; wetland§' t
Forest; grassland; woodland§
Woodland; forest; wetland§' t
Thicket; grass; woodland
Grass; forest; thickett
Grass; forest; wetlandt
Grass; forest; wetland; thickett
Grass; forest; wetlandt
Grass; forestt
Forest; thicket; grasst
Grass; thickett
Forest; thicket; grasst
Woodland; thicket; grasst
Grass; forest; thickett
Grass; woodland; forest
conservation area in this magisterial district should be along the escarpment.
conservation area in this magisterial district should include areas along major rivers or river mouths; see Figure 1.2a.
conservation area in this magisterial district should include the Lebombo Mountains .
Richards bay are outliers for conservation action as noted in Chapter 4, because of the use of their mudflats and fishing
by a large diversity of southern palearctic ocean birds. These bays, however. were important to birds in historical times
restored.
the Tugela, Umvoti, and Mgeni Rivers (but includes the Mgeni River mouth), as well as the
middle stretches of the Mkomasi and Mzimkhulu Rivers.
As discussed in earlier chapters, the long-term conservation
of biological diversity is
dependent not only on the establishment of representative protected areas, but also on maintaining
hospitable
Hansen
environments
and viable populations
et al., 1991; Shafer,
within managed landscapes
1994). Since the landscape
mosaic pattern
(Western,
1989;
metrics become
inseparable at times when distinguishing between areas of complete or diverse natural habitat and
areas of heavy human influence other representations
are required.
Table 6.9 presents results
depicting the state of fragmentation in the non-degraded major vegetation classes in the province.
These vegetation types were used earlier in a pattern analysis, which showed evidence of their
relationship to ecologically grouped bird diversity patterns (see Chapter 5). Figure 6.7 illustrates
the habitat rating results for each vegetation class and the total habitat connectivity within each
magisterial district. The habitat connectivity map (Figure 6.7e) clearly displays a similar pattern
developed earlier from a detrended correspondence
and classification of the socio-economic-enviromental
analysis of the LCLU data set and the PCA
indicators.
The loss of core and connected
habitats emanates out from the Durban economic core as far north as Richards bay and then
mostly skewed to the south along the coast and in the Midlands region.
Earlier evidence from
Chapter 5 showed that the bird assemblages along the East coast and Midlands regions were being
dominated by generalist species able to exploit the smaller patches left inhospitable for interior
core habitat specialist birds. The Nqutu magisterial district is the only outlier in this explanation.
Table 6.9: Magisterial districts requmng landscape conservation plans and the associated
vegetation habitat ratings derived from pattern indicators. Habitat connectivity rating is provided
using all habitat types to derive measure.
Magisterial
district
Bergville
Estcourt
Dannhauser
Kliprivier
Newcastle
Paulpietersburg
Underberg
Utrecht
Durban t
Hlabisa
Mtunzini t
Ingwavuma
Ngotshe
Nongoma
Ubombo
Camperdown
Ixopo
Lions River
Pietermaritzburg
Polela
Umvoti
Urnzinto
Babanango
Eshowe
Mthonjaneni
Nkandla
Vryheid
Habitat
connectivity
Poor
Fair
Poor
Poor
Poor
Poor
Poor
Poor
None
Moderate
None
Moderate
Good
Moderate
Good
Poor
None
None
None
None
Fair
None
None
None
Good
None
Good
Poor
Poor
Moderate
Poor
Moderate
Moderate
Poor
None
Moderate
Good
Good
Moderate
Good
Poor
Fair
None
Fair
Moderate
Moderate
Moderate
Poor
Moderate
None
Poor
None
Good
Poor
Moderate
Moderate
Moderate
Fair
Poor
Moderate
Poor
Poor
Poor
Moderate
Moderate
Good
Poor
Moderate
Good
Moderate
Moderate
Poor
Moderate
Poor
Fair
Good
Fair
Moderate
Moderate
Moderate
Poor
Good
Good
Poor
Fair
Good
Fair
Good
Good
Poor
Poor
Poor
Moderate
Moderate
Moderate
Poor
Poor
Moderate
Moderate
Poor
Moderate
Moderate
Poor
Fair
Moderate
Moderate
Fair
Good
Fair
Good
Moderate
Fair
Good
Moderate
Good
Good
Poor
Moderate
Moderate
Moderate
Good
Moderate
Fair
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Good
Moderate
Good
Fair
Good
t Durban and Richards
bay are outliers for conservation action as noted in Chapter 4, because of the use of their mudflats and fishing
trawler refuse by a large diversity of southern palearctic ocean birds. These bays, however, were important to birds in historical times
and should be partially restored.
Many African cultures do not distinguish between the natural and human realm, as there
is no clear-cut separation
(Western,
1989).
This contrasts the Western idea of nature and
segregating it from humanity in formal parks and reserves, when most of the world's biodiversity
is outside parks interacting (negatively and positively) with humans.
The human realm occupies
95% of the Earth's surface and will one way or another affect the future of nature far more than
the diminutive parks. Overall the formal mCN protected area categories of strict nature reserve,
Settlement hierarchy
D
Economic core
0
MYor lretrqX>li1llncenter
0
D:velopIrellt nodes
0
Regional marlcet cen1erS
AI
•
•
Qnreli JTotected areas
Habitat rating
Perip!JerytO'MlS
D
D
Mljor rivers
~
Good
Fair
MxIerate
Poor
NotPresert
Figure 6.7: The following maps present a rating of the vegetation habitats: (a) to (d) based on
patch size and fragmentation, and (e) is the habitat connectivity rating considering all available
vegetation types residing in a magisterial district. The districts in a poor to moderate state (e)
largely reside along the coast and in the Midlands region. These areas were shown in the analyses
of Chapter 5 to be undergoing significant changes in bird assemblage structure because of high
intensity transformation.
Settlement hierarchy
o
Economic core
o
Major metrqJOlitan center
a
L\lvelopment nodes
Regioml market centers
Periphery
/\/
toWlS
Major rivers
Protected area categories
II: National Parle
D
D
D
D
III: Natural Monwnent
IV: Habita1JSpecies Mar!aglment Area
V: Protected Landscape/Seascape
VII: ArthropologicallHabitat'Species
Managed Reserve
wilderness area, and national park will need to be extended to allow for more models of
conservation aligned with the co-evolutionary dynamic elucidated for an area.
Figure 6.8
proposes a scenario of at least five types of protected area categories (IDeN,
1997,
http://www.wcmc.org.uk/protected_areas/data/define.htrn)
avian
that
should
be
used
for
conservation across KwaZulu-Natal based on the analysis documented here. A new category VII
is proposed, which has as its primary objective the maintenance of cultural and traditional
attributes in combination with habitat/species management. This new category ensures that the
African cultural heritage and co-evolution within particular areas is both rewarded for its positive
effects on avian biodiversity preservation and left to be managed for sustainable biodiversity
conservation as basic human development needs are met.
The analysis reported in this chapter includes the landscape pattern and socio-economicenvironmental indicators that would be available for use in most countries of the world.
Therefore, the approach presented here can be extended to other areas, a property that is usually
not found in studies of a transdisciplinary nature. This study explores our ability to develop better
models for explaining human-conservation interactions that will be required to arrive at
sustainable regional biodiversity conservation goals in developing regions of the world.
Conservation, in this thesis, is considered a general landscape principle required across the whole
region and demanding greater importance for the future sustainability of biodiversity than simply
the use of formally protected areas. The shift to integrated landscape conservation management
relies on two equally important assumptions. First, by acknowledging that humans are an integral
part of any ecosystem (Cronon, 2000) and may be considered a "keystone species" in their own
right (sensu O'Neill and Kahn, 2000), and second, that planning and management of the total
human modified landscape matrix for biodiversity functionality will ensure biodiversity
persistence both within formal protected areas and across a region (e.g., Forman, 1995).
The separate analyses of each data set have suggested a close dependency between socioeconomic development, physical environment, and the measured landscape mosaic patterns. It is
clear that the properties of the physical environment do significantly affect spatial patterns of
human activity, but culture, personality, resource opportunities/constraints, and policy may
significantly affect the evolving character of an area. Only the first four principal components
were important in explaining the socio-economic-environmental situation of strong modernization
and development needs gradients in the study area. Core-periphery structure of economic space
provides the best model for explaining the past and present co-evolution of the landscape. Issues
of unbalanced economic spatial growth can be explained by resource availability, access to
infrastructure and former separate development policies (e.g., Fair and Schmidt, 1974).
In spite of these strong continuous gradients and distortions, the hierarchical and k-means
clustering classification methods identify several homogeneous regions, which are isolated from
one another by stable transition zones. The derived regions from each data set analyzed are
explainable and comparable to other regional geographic studies conducted in Africa and to
known issues in the province.
The overall results can be compared to broadly similar studies by
Forde (1968) in Ghana, Soja (1970) in Kenya, Lea (1972) in Swaziland, and Wienand (1973) in
Nigeria.
The results for Ghana, Kenya, and Nigeria also reached a similar conclusion by
uncovering the structure and pattern of modernization and urbanization as the major landscape
drivers.
In the case of Swaziland (at least in 1972) the rural dimension of land ownership by
Africans versus the European population was the main driver, with modernization
gradients not
dominating at that time.
The identified backwash effects are problematic because they still reflect past separate
development
policies of the former apartheid South Africa.
Attempts to disperse economic
activities might be expected to produce a uniform distribution of spread effects in the future (i.e.,
Reconstruction
and Development Plan; Spatial Development Initiatives), but the data used in this
study from 1996-97 does not yet show evidence of this. The economic development surface of
KwaZulu-Natal
(Figure 6.4b) resembles the hypothesis provided by Weinand (1973) for the
economic development
of Nigeria.
Spread and backwash effects appear to decline in parallel
throughout the province as distance from the Durban economic core increases.
The economic
frontier of the province provides evidence for a secondary core area (renewable resources growth
regions), but beyond this there lies an extensive 'sparsely developed'
mining region within the high grasslands and a very underdeveloped
thicket and woodland demarcated
areas.
irrigated agriculture and
rural African region in the
This last region not only lacks substantial modem
economic activity, but also is unable to provide the infrastructure
necessary
to attract such
activities.
The use of LCLU to derive co-evolved areas of human-cultural influence and productivity turned
out to be an important aspect of the space economy description.
In this respect, satellite remote
sensing derived LCLU patterns and the assumption of human land-use as a unimodal response
model across a region derived explainable results.
Like most spatial analysis techniques, some
degree of caution needs to be exercised over the use of the results and the limitations of this
approach
need to be noted.
accompanying
The correct use of classified
error and accuracy statistics.
satellite
imagery must have
Fairbanks et al. (2000) noted that one of the key
issues related to mapping accuracy, with the development of the South African National LandCover (NLC) database, is landscape complexity in terms of the mapping scale used. For example,
sheets containing complex patterns and gradients of natural and degraded vegetation types (e.g.,
2830 Richards Bay mapsheet, KwaZulu-Natal
Province) were significantly harder to map than
those containing significantly more (and often smaller) polygons that were based on uniform
cover types with clearly definable boundaries
Province grasslands).
(e.g., dryland maize cultivation
The use and interpretation
of ecologically
in Free State
based ordination
methods
assumes that all features of a study site have been correctly and thoroughly recorded (e.g., Gauch,
1982; Jongman et al., 1995). The NLC had only thirty-one LCLU types defined for mapping
(Thompson, 1996; Fairbanks and Thompson, 1996), and these classes were only the ones that
could be reliably interpreted from 1:250 000 scale satellite imagery. Therefore, other important
land-uses that define regions of the province (e.g., sheep and cattle grazing, horse farms, game
farms, etc.) are not recorded and would most likely alter the results.
Socio-economic and biophysical factors interact to yield important landscape changes
(Turner, 1989). The landscape mosaic pattern indicators provided important results for analyzing
intensively managed landscapes and where the human impacts on the landscape are more
noticeable. Of course, the effect of changing the scale of analysis would have an important
impact on the results because all measures of landscape pattern are scale dependant (Turner et al.,
1989; Haines-Young and Chopping, 1996). Nevertheless, relationships were evident between
recorded socio-economic variables and the measured landscape mosaic pattern indicators.
Contagion was low in the economically active areas and some ex-homeland areas suggesting that
private and communal land is characterized by landscape with much spread out class types and
that economically stagnant areas in the highland grasslands consist of landscapes with contiguous
patches. An exception to this was found in the coastal magisterial district of Lower Tugela, which
is dominated by continuous sugarcane plantations. The landscape structure and its functioning
can have serious implications for the health of a landscape and its biodiversity. More survey and
temporal analysis work will be needed to fully understand the association between patterns,
landscape health, and biodiversity, as investigated in Chapter 5. It is necessary not only to
measure diversity, dominance and patch size but also to examine other indexes to evaluate
changes in landscape structure.
Remote sensing and socio-economic information integrated
within a GIS framework can form useful surrogates for monitoring the status and condition of
landscapes and therefore have the potential to be useful indicators of environmental health (Frohn
et al., 1996; Wood and Skole, 1998; O'Neill et a1., 1999; Amissah-Arthur et a1.,2000). The case
study conducted provides a framework within which to consider issues of fragmentation with
other spatial process in the context of land transformation or landscape change. A temporal
analysis should be the next step to describe change in the landscapes or model them as a sequence
of mosaics.
The required conservation action and socio-economic tension within the province proved
to be quite telling. Fifteen of the magisterial districts required as priority bird diversity area are
underdeveloped with needy populations. A landscape management system should be developed
to provide quality core habitat for the bird diversity, while acknowledging the human impacts and
influences happening around the target habitat areas, and addressing the basic human needs
required in a region (i.e., access to water, sanitation, and communication services, support to
women). The first criterion is to adequately represent all species of the target taxa, and second to
represent the associated environmental processes as proposed in Chapter 4.
Third, their
arrangement on the landscape would have to be based on ideals of persistence, therefore
representing appropriate habitat patch characteristics identified in Chapter 5 and protection of
indispensable patterns. These top-priority patterns for protection, with no known substitute for
their ecological benefits, are wide vegetated corridors protecting the identified major rivers, the
natural environmental heterogeneity along the Drakensberg Escarpment, rivers mouths, larger
estuary mudflats, connectivity for movement of species among large patches, and small patches
and corridors providing heterogeneous bits of nature throughout the economic core and
developing areas. Finally, since planning and management will more than likely need to be
conducted on communal and private lands for off-reserve management, the socio-economic
realities of the areas will need to be addressed to reduced social tension, which may ensure
sustainable conservation initiatives. Human "quality of life" development should be allowed to
go forward to the extent that they are compatible with the goal of maintaining native species and
ecosystem diversity. The development of strategically "designed" future landscape plans should
be a requirement to share ideas with communities on pathways to balancing conservation with the
regional development needs. In particular, three primary landscape design options should be
invoked.
First, future development should increase densities in currently developed areas.
Second, landscape or regional differences within a magisterial district should be recognized, with
subsequent new developments concentrating in areas that are already developed for agriculture or
more intensive human activities, while undeveloped or sensitive habitats in the other areas of a
district should be conserved. Finally, the former homeland magisterial districts appear to be
operating well for bird habitat needs, but the overly dominant effects of degraded lands may
changes this situation in the future. These partially impacted lands should be given time to rest
and be restored. Future human needs developments that will be required in these districts should
follow the first two rules to reduce the extents of human impacts. These recommendations for
KwaZulu-Natal province generally follow the "aggregate-with-outliers" principle developed by
Forman (1995), and Forman and Collinge (1995).
Land is best arranged ecologically by
aggregating land-uses, yet maintaining small patches and corridors of nature throughout
developed areas, as well as outliers of human activity spatially arranged along major zones
between land-uses. Thus, a magisterial district containing a variance in grain size, especially
coarse and fine grain, appears to be an important spatial configuration. The first two landscape
mosaic classes (Figure 6.4c) adhere to this principle, while the third class reflects districts with
increasing fine grained pattern from the increased perforations of the habitat by human land-uses.
Magisterial districts in class five reflect homogeneous coarse grain classes of human land-use
with minor components
of remaining habitat.
These pathways of landscape change through
human input generally follow the co-evolutionary framework outlined in Chapter 2.
Results from the examination of KwaZulu-Natal's
and landscape mosaic pattern characteristics
socio-economic-environmental,
provide support for the examination
system development paths in conservation analysis.
LCLU
of cultural
Along with economic geography models,
culture should also be assigned a central role in any theory purporting to characterize the process
of land-use
intensification
among rural African communities.
Unfortunately,
the analysis
conducted here is limited by the absence of a temporal component, which could allow a predictive
element.
As an example of the interactions between these factors, regional economic shifts can
bring population redistribution,
which in turn, affects biodiversity
through attendant land-use
change. The spatial and statistical models will need to be applied again with a temporal data set
of factors to develop possible alternative
land-use "futures"
resulting
from various human-
environment interactions.
Physical location and transportation costs often determine the profitability of an economic
activity.
In turn, that economic activity is the primary determiner of landscape pattern and
change. There is tremendous opportunity for conservation science and landscape ecology to take
advantage of the well-developed
theories of human and economic geography (Thoman et aI.,
1962; Bradford and Kent, 1986; Healey and Ilbery, 1990). Other applicable areas, however not
included in this study, include central place theory (e.g., Christaller, 1933; Berry and Pred, 1961),
location theory (e.g., Isard, 1956; Hall, 1966; Smith, 1971), and market area analysis (e.g., Losch,
1954).
Location theory, for example, considers the value of various products and the cost of
transporting them to a central market. The theory then predicts which product will be grown close
to the market and which can be profitably grown at greater distances.
These theories should be
able to drive models of land-use change and assess producers and consumers ability to optimize
their use of resources on the landscape, which can then be used to develop "spatial solutions" to
protect biodiversity.
The integration of human geography with landscape ecology seems to hold
the potential for major breakthroughs
in our understanding
of landscapes (e.g., Behrens, 1996;
O'Neill, 1999) and its application to sustainable biodiversity conservation strategies.
Re-integrating
nature problematical
society with the environment and the goals of conservation is by its very
in as much as several potential solutions always appear in any aspect of
societal life (i.e., cultural, religious, political, and economical) and how the environment may be
addressed (e.g., Cronon, 2000).
History and present experience
show that controlling these
problems in the context of the modem industrial complex is through mutual discussion and
analytical discourse (see Chapter 2, Figure 2.5). An increase in the level of integration among
fragmented disciplines (e.g., geography, biology, economics, anthropology, and sociology) to
develop and arrive at multiple solutions for human and biodiversity survival within Africa could
reduce the tension around the conservation issue. The idea that anyone discipline has the correct
analytical framework for this task is severely misguided.
A realistic philosophy for conservation must be connected with human survival and the
support and participation of local communities (Tisdell, 1995). In North America and Australia
where there is still land to spare and the human population growth rate is very low, the wilderness
concept has real value (Noss, 1991). However, in Africa where land is in great demand and the
population growth rate is high, it may be very unrealistic to set aside a large area of the continent
for all time as wilderness areas (e.g., Soule and Sanjayan, 1998; Musters et al., 2000). The Peace
Parks concept, however, may eventually prove this position wrong, but Peace Parks are being
designed under the premise of multiple-use areas with conservation principles (http://www.peaceparks.org.za). Nevertheless, the aesthetic appeal of wilderness and biodiversity seems insufficient
in itself to justify perpetuating land-use at a level below the optimum, however as outlined any
land-use must be planned equitably and ecologically within the overall constraints of the sociocultural-economic and biotic landscape. When it comes down to the real point there are only two
valid arguments to advance in support of biodiversity-its ecological value and its economic value
to human "quality of life". The dictum for the developing countries of the world should become
'conservation as if people mattered and development as if nature mattered' (e.g., Adams and
McShane, 1996).
A river, with its waterfalls, wetlands and meadows, a lake, a hill, a cliff
or individual rocks, a forest, and ancient trees standing singly ... If the
inhabitants of a town were wise, they would seek to preserve these
things, though at a considerable expense; for such things educate far
more than any hired teachers or preachers, or any present recognized
system of school education. I do not think them fit to be the founder of
a state or even ofa town who does not foresee the use of these things...
The studies documented
in this thesis offer a series of conservation
develops a framework for understanding
human impact on the KwaZulu-Natal
analysis strategies.
approaches and
and assessing landscape morphologies
derived from
province, South Africa using both standard and original
The use of avian biodiversity as a bioindicator for levels of human impact
provided a rather telling description of landscape development
over the last 25 years.
The
analyses presented are intended to provide a framework derived from empirical results for
subsequent more-detailed and quantitative studies.
Evaluating environmental
change requires analysis of various relationships
between humans and biota/nature, focussing on their reciprocal impacts.
over time,
Elucidating the history
of the environment and changes which have taken place or which are likely to occur in the future
requires knowledge of not only natural processes, but also human activity as well. To date many
of the theories and techniques developed to make conservation more efficient miss the point that
there is a paradox of management in conservation.
The paradox states that the probability of
having a significant effect is greater in small areas, whereas the probability of successful longterm management is greater in large areas. For example, we can see the result of protecting a rare
butterfly or plant in a local grassland, but at the same time, over human generations the chance of
finding the butterfly of plant at that same spot is low, whereas the region is likely to continue in
similar form. Therefore, the prescriptive approaches to conservation including reserve selection
algorithms, gap analysis, and other computerized
conservation
planning (Prendergast
et aI., 1999).
approaches have only limited potential for
Both landscape-level
(Le., top down) and
species-level (i.e., bottom-up) approaches are required for practical conservation.
than less knowledge is required to make conservation
More rather
decisions, which in turn should remain
flexible. The role of adaptive environmental management to address local surprise and emergent
regional change should be required for management of the total human landscape (e.g., Holling,
1986).
This would acknowledge
the dimensions
of evolution,
instability,
and change
in
addressing the biodiversity crisis. Evolution in human systems is a continual, imperfect learning
process, spurred by the difference between expectation and experience,
but rarely providing
enough information for a complete understanding.
Consequently, adaptive management becomes
a social as well as scientific process.
Like co-evolution, adaptive environmental management is on going. Most people think
of the right policy and its proper implementation as setting a system on the right trajectory once
and forever. For example, this is the case in reserve selection analysis or gap analysis where an
overall strategy for conservation
biodiversity.
is pushed on a public as the final solution to conserve
Or management procedures are proposed that offer a final and lasting solution to an
environmental problem.
Adaptive environmental management helps get people out of this mode,
it also does not make the distinctions between scientific, expert and experiential knowledge that
are typically made.
modelers
So, like the co-evolutionary
and technologists
contributes
approach, shared learning among disciplines,
to the adaptive approach
required
for total human
landscape conservation, while the co-evolutionary approach adds in more of a social dimension
required to understand tensions and emergent change.
A better understanding
of biodiversity risk for models of conservation
assessment and
prioritization was presented. A theoretical foundation for the relationships between categories of
social, economic and environmental indicator variables in models for biodiversity risk assessment
were developed
to enhance the complex biodiversity
conservation
debate.
The analytical
framework proposed could be used to gauge the sustainability of existing and future biodiversity
conservation areas while remaining open to its own evolution when new knowledge is acquired.
An underlying assumption of the approach is that a co-evolutionary
social, cultural, economic and environmental
manner.
relationship exists between
systems, and that they cannot be addressed in a
reductionist
and deterministic
Current
conservation
deals with symptoms of environmental
ecosystem
management
and biodiversity
degradation rather than its causes.
Co-
evolutionary theory integrated into the larger analytical framework and principles of landscape
ecology was used to demonstrate
diversity.
the development
of landscapes
and their effects on avian
Landscape ecology has emerged as a discipline whose primary focus is the analysis of
the ecological effects of environmental
heterogeneity
and pattern on ecological process.
The
fusion of co-evolutionary theory and landscape ecology makes for an exciting scientific synthesis
in which to explain anthropogenic pressure on the landscape and ultimately to bring into question
the sustainability of biodiversity conservation within regions of the world's developing nations.
The on-going development of socio-economic systems contrasted the Western model of
development based on consumption with the rural African system. Avian diversity proved to be a
fairly convincing indicator of landscape health, and illustrated what might happen to bird diversity
and assemblage structure if development policy directs the former KwaZulu homeland areas to
the same socio-economic
"Western" consumption system as found in the "White" dominanted
economic core. Consumption growth and the on-going development of socio-economic systems
distance people from the environmental systems they are impacting (Norgaard, 1994). The
methods focused on both trends and patterns of variance in a multivariate data matrix to identify
co-evolved regions allowing identification of dominant trends as well as underlying tensions
within a defined area. Potential sources of human insecurity and development patterns at odds
with positive biodiversity survival can be identified within the co-evolved landscapes. These
sources can then be targeted for political action and be used to inform public debate.
Many factors contribute to the avian composition and change of a region, and it cannot be
expected that all relevant information can be anticipated or even fully represented in a GIS
database. However, by examining those regions not well explained by the current efforts, future
research can be targeted to better understand unique or localized effects on avian diversity. The
methodologies used in this thesis should be supported by finer scaled studies with higher accuracy
data. Landscape-level study provides a means to quantify and monitor broad-scale changes
related to biodiversity and ecosystem processes.
Species- or population level analysis can
contribute a more mechanistic understanding of the impact of landscape change, while broader
scale investigations provide information on broad-scale patterns that can enhance or constrain the
conservation of biological diversity.
The work presented here is planned as the beginning of an ongoing research effort, and
opens pathways to a much larger array of future research directions. The author recognizes that
this research effort represents only a limited set of conditions within the synthesis of diverse
information that will be required to develop realistic expectations for the task of sustainable
biodiversity conservation. How extensible are the various approaches? What level of accuracy is
required? How much error is allowed in databases from the socio-economic and ecological
sectors? What is the optimal mix of computational and interpretative capabilities for producing
high quality socio-economic, ecological, and conservation information? The problems call for
interdisciplinary research to produce information useful to the development of conservation
strategies, land characterization, extent of anthropogenic stress, and climate change models.
The best hope for all species is linked to a single uncompromisable human goal- the
improvement of human welfare. Our future, and that of wildlife, is not an inevitability, but rather
a matter of foresight, choice and action (Western, 1989) in directing the landscape changes to
come in a sustainable manner through ecologically responsible spatial planning. These choices
and action can only come from an approach based on co-evolutionary thought and shared learning
among disciplines, system modelers and appropriate technology. As an understanding of relevant
scales and types of information evolves and the power of synergistic relationships between
available data is harnessed, the development of a regional management strategy to support
conservation across the total human landscape may become a possibility.
Abbitt, R.lF., Scott, J.M., and Wilcove, D.S., 2000. The geography of vulnerability: incorporating
species geography and human development pattern into conservation planning.
Biological
Conservation 96:169-175.
Acocks, J.P.H., 1953.
28:1-192.
Veld types of South Africa. Memoirs of the Botanical Survey of South Africa
Adams, lS. and McShane, T.O., 1996. The myth of wild Africa: Conservation without illusion.
University of California Press, Berkeley, CA
Allan, D.G., 1994. The abundance and movements of Ludwig's bustard. Ostrich 65:95-105.
Allan, D.G., Sinclair, le., and Rossouw, l, 1999. The waterbirds (aves) of Durban bay: current status
and historical trends. Durban Museum Novitates 31: 1-21.
Allan, D.G., Harrison, lA, Navarro, R.A, van Wilgen, B.W., and Thompson, M.W., 1997. The impact
of commercial afforestation on bird populations in Mpumalanga province, South Africa- insights
from bird atlas data. Biological Conservation 79: 173-185.
Allen, P.M., 1993. Spatial evolution in complex systems. In Nonlinear evolution of spatial economic
systems, P. Nijkamp and A Reggioni (Eds.), pp. 178-202. Springer-Verlag, London, UK.
Allen, T.F.H.and Starr, T.B., 1982. Hierarchy: Perspectives for ecological complexity. University of
Chicago Press, Chicago, 11.
Amissah-Arthur, A., Mougenot, B., and Loireau, M., 2000. Assessing farmland dynamics and land
degradation on Sahelian landscapes using remotely sensed and socioeconomic data. International
Journal of Geographical Information Systems 14:583-599.
Andren, H., 1994. Effects of habitat fragmentation on birds and mammals in landscapes with different
proportions of suitable habitat: a review. Oikos 71 :355-366.
Anselin, 1., 1995. Local indicators of spatial association- LISA Geographical Analysis 27:93-115.
Anselin, 1., 1999. Spacestat: A software program for the analysis of spatial data, Version 1.96.
Regional Research Institute, West Virginia University, Morgantown, WV.
Armstrong, A, Benn, G.A, Bowland, AE., Goodman, P.S., Johnson, D.N., Maddock, AH., and ScottShaw, e.R., 1998. Plantation forestry in South Africa and its impact on biodiversity. South African
Forestry Journal 182:59-65.
Armstrong, A, Benn, G.A, Coke, M., Goodman, P.S., Johnson, D.l, Langley, D.N., Mckean, S.,
Mckenzie, C., Mwanyama, N., and Scott-Shaw, C.S., 2000. Determining the conservation value of
land in KwaZulu-Natal. Biodiversity Division, KwaZulu-Natal Nature Conservation Services,
Department of Water Affairs and Forestry Report, Pretoria, South Africa.
Askins, R.A, Philbrick, M.J., and Sugeno, D.S., 1987. Relationship between the regional abundance of
forest and the composition of forest bird communities. Biological Conservation 39:129-152.
Askins, R.A, Lynch, IF., and Greenberg, R., 1990. Population declines in migratory birds in eastern
North America. Current Ornithology 7:1-57.
Askins, R.A, 1995. Hostile landscapes and the decline of migratory songbirds. Science 267:1956-1957.
Austin,
M.P. and Margules,
e.R.,
1986. Assessing
representativeness.
In Wildlife conservation
evaluation, M.B. Usher (Eds.), pp. 45-67. Chapman and Hall, London, UK.
Bailey, R.G., 1996. Ecosystem geography. Springer-Verlag, New York, NY.
Baillie, S.R., Sutherland, W.J., Freeman, S.N., Gregory, R.D., and Paradis, E., 2000. Consequences of
large-scale processes for the conservation of bird populations. Journal of Applied Ecology
37(Suppl. 1):88-102.
Baker, W.L., 1992. The landscape ecology of large disturbances
nature reserves. Landscape Ecology 7:181-194.
in the design and management of
Baker, W.L., 1993. Spatially heterogeneous multi-scale response of landscapes to fire suppression.
Oikos 66:66-71.
Balmford, A and Gaston, K.J., 1999. Why biodiversity surveys are good value. Nature 398:204-205.
Balmford A, Mace G.M., and Ginsberg J.R., 1998. The challenges to conservation in a changing world:
putting processes on the map. In Conservation in a changing world, G.M. Mace, A Balmford, and
J.R. Ginsberg (Eds.), pp. 1-28. Cambridge University Press, Cambridge, UK.
Barbault, R., 1995. Biodiversity dynamics: from population and community ecology approaches to a
landscape ecology point of view. Landscape and Urban Planning 31:89-98.
Barnes, K.N., 1998. The important bird areas of southern Africa. Birdlife South Africa, Johannesburg,
South Africa.
Bascompte, J. and Sole, R.V., 1996. Habitat fragmentation and extinction thresholds in spatially explicit
models. Journal of Animal Ecology 65:465-473.
Baudry, J., 1993. Landscape dynamics and farming systems: Problems of relating patterns and
predicting ecological changes. In Landscape ecology and agroecosystems, R.G.H. Bunce, L.
Ryszkowski, and M.G. Paoletti (Eds.), pp. 21-40. Lewis Publishers, London, UK.
Bedward, M., Pressey, R.L., and Keith, D.A, 1992. A new approach for selecting fully representative
reserve networks: addressing efficiency, reserve design and land suitability with an iterative
analysis. Biological Conservation 62:115-125.
Behrens, C.A, 1996. A formal justification for the application of GIS to the cultural ecological analysis
of land-use intensification and deforestation in the Amazon. In Anthropology, Space, and
Geographic Information Systems, M. Aldendeffer and H.D.G. Maschner (Eds.), pp. 55-77. Oxford
University Press, Oxford, UK.
Benefield, C.B. and Bunce, R.G.H., 1982. A preliminary visual presentation of land classes in Britain.
Merlewood Research and Development Paper 91, ITS, Grange-over-Sands, UK.
Bennett, J.W., 1976. The ecological
Pergamon Press, New York, NY.
transition:
Cultural anthropology
and human adaptation.
Benson, C.W., Stuart-Irwin, M.P., and White, C.M.N., 1962. The significance of valleys as avian
zoogeographical barriers. Annals of the Cape Provincial Museums 2:155-189.
Bernert, J.A, Eilers, J.M., Sullivan, TJ., Freemark, K.E., and Ribic, c., 1997. A quantitative method
for delineating regions: an example for the western com belt plains ecoregion of the USA
Environmental Management 21:405-420.
Berry, RJ.L., 1970. Commuting patterns: Labor market participation and regional potential. Growth
and Change 4:3-10.
Berry, BJ.L. and Pred, A, 1961. Central place studies: A bibliography.
Institute, University of Pennsylvania, Philadelphia, PA.
Regional Studies Research
Bishop, Y., Fienberg, S., and Holland, P., 1975. Discrete multi-variate analysis: Theory and practice.
Massachusetts Institute of Technology Press, Cambridge, MA
Blankson, E.J. and Green, B.H., 1991. Use of landscape classification as an essential prerequisite to
landscape evaluation. Landscape and Urban Planning 21:149-162.
Board, C., Davies, R.J., and Fair, TJ.D., 1970. The structure of the South African space economy.
Regional Studies 4:367-392.
Bohning-Gaese, K., 1997. Determinants of avian species richness at different spatial scales. Journal of
Biogeography 24:49-60.
Borcard, D., Legendre, P., and Drapeau, P., 1992. Partialling out the spatial component of ecological
variation. Ecology 73:1045-1055.
Botkin, D.B., Woodby, D.A., and Nisbet, R.A, 1991. Kirtland's warbler habitats: a possible early
indicator of climatic warming. Biological Conservation 56:63-78.
Bradford, M.G. and Kent, W.A, 1986. Human geography: Theories and their applications. Oxford
University Press, Oxford, UK.
Bray, 1.R. and Curtis, J.T., 1957. An ordination
Wisconsin. Ecological Monographs 27:325-349.
of the upland forest communities
of southern
Breininger, DR, Larson, V.L., Duncan, B.W., Smith R.B., Oddy, D.M., and Goodchild, M.F., 1995.
Landscape patterns of Florida scrub jay habitat use and demographic success. Conservation
Biology 9: 1442-1453.
Breitburg, D.L., Baxter, J.W., Hatfield, e.A, Howarth, R.W., Jones, e.G., Lovett, G.M. and Wigand,
C., 1998. Understanding effects of multiple stressors: ideas and challenges.
In Successes,
limitations, and frontiers in ecosystem science, M.L. Pace and P.M. Groffinan (Eds.), pp. 416-431.
Springer-Verlag, New York, NY.
Brooker, L., Brooker, M., and Cale, P., 1999. Animal dispersal in fragmented habitat: measuring habitat
connectivity,
corridor
use,
and
dispersal
mortality.
Conservation
Ecology
www.consecol.org/vo13/issl/art4.
Browett, 1.G., 1976. The application of a spatial model to South Africa's development regions. South
African Geographical JournaI58:118-129.
Browett, J.G. and Fair, T.1.D., 1974. South Africa, 1870-1970: a view of the spatial system.
African Geographical Journal 56:111-120.
South
Brown, C.1. and Barnes, P.R., 1984. Birds of the Natal alpine belt. The Lammergeyer 33: 1-13.
Brown, J.H., 1984. On the relationship
Naturalist 124:255-279.
between abundance and distribution
of species. American
Brown, J.H., 1995. Macroecology. University of Chicago Press, Chicago, IL.
Burel, F., Baudry, 1., Buter, A., Clergeau, P., Delettre, Y, Le Coeur, D., Dubs, F., Morvan, N., Paillat,
G., Petit, S., Thenail, e., BruneI, E., and Lefeuvre, J.C., 1998. Comparative biodiversity along a
gradient of agricultural landscapes. Acta Oecologica 19:47-60.
Burkey, T.V., 1989. Extinction in nature reserves: the effect of fragmentation
migration between reserve fragments. Oikos 55:75-81.
Burnett, M.R., August, P.V., Brown, 1.H., and Killingbeck, K.T.,
geomorphological heterogeneity on biodiversity, I. a patch-scale
Biology 12:363-370.
and the importance of
1998.
The influence of
perspective.
Conservation
Burroughs, P.A, 1986. Principles of geographical information systems for land resources assessment.
Oxford University Press, Oxford, UK.
Caico, S. L., Scott, J. M., Butterfield, B., and B. Csuti., 1995. A gap analysis of the management status
of vegetation ofIdaho (U.S.A). Conservation Biology 9:498-511.
Cain, D.H., Riitters, K., and Orvis, K., 1997. A multi-scale analysis of landscape statistics. Landscape
Ecology 12:199-212.
Cairns, J., Jr., 1988. Restoration ecology: the new frontier. In Rehabilitating
Cairns, Jr. (Ed.), pp. 1-11. CRC Press, Boca Raton, FL.
damaged ecosystems, J.
Campbell, J., 1988. The power of myth. B.S. Flowers (Ed.), Doubleday, New York, NY.
Carpenter, S.R., 1990. Large-scale perturbations: opportunities for innovation. Ecology 71:2038-2043.
Chapin, F.S. III, Zavaleta, B.S., Eviners, V.T., Naylor, R.L., Vitousek, P.M., Reynolds, H.L., Hooper,
D.U., Lavorel, S., Sala, O.E., Hobbie, S.E., Mack, M.e., and Diaz, S., 2000. Consequences of
changing biodiversity. Nature 405: 234-242.
Chapman, K., 1979. People, pattern, and process: An introduction
Arnold, London, UK.
to human geography.
Edward
Christaller, W., 1933. Central places in Southern Germany. Jena, Germany.
Christian, C.S. and Stewart, G.A, 1953. General report on survey of Katherine-Darwin
Research Series 1, CSIRO, Canberra, Australia.
region. Land
Christopher, AJ., 1982. South Africa. The World's Landscapes Series, Longman Press, London, UK.
Church, R.L., Stoms, D.M., and Davis, F.W., 1996. Reserve selection as a maximal covering location
problem. Biological Conservation 76:105-112.
Clancey, P.A., 1964. The birds of Natal and Zululand. Oliver and Boyd, Edinburgh, UK.
Clancey, P.A, 1994. Combined biogeographic role of river valleys and aridity in southern African bird
distribution. Durban Museum Novitates 19: 13-29.
Clark, P. and Niblett, T., 1989. The CN2 induction algorithm. Machine Learning Jouma13(4):261-283.
Clark, P. and Boswell, R. 1991. Rule induction with CN2: some recent improvements. Proceeding of the
Fifth European Conference in Machine Leaning, Y. Kodratoff (Ed.), pp. 151-163 . Springer-Verlag,
Berlin, DE.
Clark, W.C. and Munn, R.E. (Eds.), 1986.
University Press, Cambridge, UK.
Sustainable
development
of the biosphere.
Cliff, A.D. and Ord, lK., 1981. Spatial processes: Models and applications.
Cody, M.L., 1981. Habitat selection in birds: the roles of vegetation
productivity. BioScience 31: 107-113.
Cambridge
Pion, New York, NY.
structure,
competitors,
and
Cody, M.L. (Ed.), 1985. Habitat selection in birds. Academic Press, San Diego, CA
Cody, M.L., 1993. Bird diversity components within and between habitats in Australia. In Species
diversity in ecological communities: Historical and geographical prospectives, R.E. Ricklefs and
D. ScWuter (Eds.), pp. 147-158. University of Chicago Press, Chicago, IL.
Cole, M., 1961. South Africa. Methuen, London, UK.
Congalton, R.G., 1991. A review of assessing the accuracy of classifications
Remote Sensing of Environment 37:35-46.
of remotely sensed data.
Congalton, R., Oderwald, R., and Mead, R., 1983. Assessing Landsat classification accuracy using
discrete multivariate statistical techniques. Photogrammetric Engineering and Remote Sensing,
49: 1671-1678.
Connell, J.H., 1978. Diversity in tropical rain forests and coral reefs. Science 199: 1302-1310.
Conservation International, 1998. Conservation International1ists
world. Http://www.conservation.org.
the countries of mega diversity in the
Cowan, G.I. and Marneweck, G.C., 1996. South African National Report to the RAMSAR Convention
1996. Department of Environmental Affairs and Tourism, Pretoria, South Africa.
Cowling, R.M., Richardson, D.M., and Pierce, S.M. (Eds.), 1997.
Cambridge University Press, Cambridge, UK.
Vegetation of Southern Africa.
Cowling, R.M., Pressey, R.L., Lombard, AT., Desmet, P.G., and Ellis, AG., 1999. From representation
to persistence: requirements for a sustainable system of conservation areas in the species-rich
Mediterranean-climate desert of Southern Africa. Diversity and Distributions 5:51-71.
Cressie, N.AC.,
1993. Statistics for spatial data. Wiley, New York, NY.
Cronon, W., 2000. Resisting monliths and tabulae rasae. Ecolgoica1 Applications 10:673-675.
Cyrus, D. and Robson, N., 1980. Bird atlas of Natal. University
KwaZulu-Natal, South Africa.
of Natal Press, Pieterrnaritzburg,
Csuti, B., Polasky, S., Williams, P.H., Pressey, R.L., Camm, J.D., Kershaw, M., Kiester, AR., Downs,
B., Hamilton, R., Huso, M., and Sahr, K., 1997. A comparison of reserve selection algorithms
using data on terrestrial vertebrates in Oregon. Biological Conservation 80:83-97.
Dacey, M.F., 1964. Modified poisson probability law for point patterns more regular than random.
Annals of the Association of American Geographers 54:559-565.
Dahlberg, A.c., 1996. Interpretations of environmental change and diversity: A study from North East
District, Botswana.
Unpublished Ph.D., Department of Geography, Stockholm University,
Sweden.
Daily, G.c. (Ed.), 1997. Nature's services: societal dependence on natural ecosystems. Island Press,
Washington, D.C., US.
Dale, V.H., O'Neill, R.V., Pedlowski, M., and Southworth, F., 1993. Causes and effects of land-use
change in central Rondonia, Brazil. Photogrammetric Engineering and Remote Sensing 59:9971005.
Dale, V.H., O'Neill, R.V., Southworth, F., and Pedlowski, M., 1994.
Amazonian settlement of Rondonia. Conservation Biology 8: 196-206.
Modeling
in the Brazilian
Dale, V.H., Pearson, S.M., Offerman, H.L., and O'Neill, R.V., 1994. Relating patterns of land-use
change to faunal biodiversity in central Amazon. Conservation Biology 8: 1027-1036.
Dasgupta, P., 1993. An inquiry into well-being and destitution. Oxford University Press, Oxford, UK.
Davis, F.W. and Stoms, D.M., 1996. Sierran vegetation: A gap analysis. In: Sierra Nevada Ecosystem
Project: Final Report to Congress. Volume 2, Assessments and Scientific Basis for Management
Options, pp. 671-689. Davis: University of California, Centers for Water and Wildland Resources.
Davis, F.W., Stoms, D.M., Church, R.L., Okin, W.J., and Norman Johnson, K., 1996. Selecting
biodiversity management areas. In Sierra Nevada Ecosystem Project: Final report to Congress,
Volume II, Assessments and scientific basis for management options. Davis: University of
California, Centers for Water and Wildland Resources.
De Agar, P.M., De Pablo, c.L., and Pineda, F.D., 1995. Mapping the ecological structure ofa territory:
a case study in Madrid (central Spain). Environmental Management 19:345-357.
Dent, M.A., Lynch, S.D., and Schulze, R.E., 1989. Mapping mean annual and other rainfall statistics
over Southern Africa. WRC Report 109/1/89, Water Research Commission, Pretoria, South Africa.
Denton, S.R. and Barnes, B.V., 1987. Tree species distributions related to climatic patterns in Michigan.
Canadian Journal of Forest Research 17:613-629.
Derleth, E.L., McAuley, D.G., and Dwyer, T.J., 1989. Avian community response to small-scale habitat
disturbance in Maine. Canadian Journal of Zoology 67:385-390.
De Velice, R.I., De Velice, J.W.R., and Park, G.N., 1988. Gradient analysis in nature reserve design: a
New Zealand example. Conservation Biology 2:206-217.
Diamond, J.M., 1988. Factors controlling species diversity: overview and synthesis. Annals of the
Missouri Botanical Gardens 75: 117-129.
Diamond, J., 1997. Guns, germs, and steel: Thefates of human societies.
Vintage, London, UK.
Doak, D.F., 1995. Source-sink models and the problem of habitat degradation:
applications to the Yellowstone Grizzly. Conservation Biology 9:1370-1379.
Donald, P.F. and Fuller, R.F., 1998 Ornithological
Study 45:129-145.
general models and
atlas data: a review of uses and limitations. Bird
Dovers, S.R. and Handmer, J.W., 1993. Contradictions
20:217-222.
in sustainability.
Environmental
Conservation
Dufrene, M. and Legendre, P., 1997. Species assemblages and indicator species: the need for a flexible
asymmetrical approach. Ecological Monographs 67:345-366.
Dutilleul, P., 1993. Modifying the t Test for assessing the correlation between two spatial processes.
Biometrics 49:305-314.
DWA, 1986. Management of the water resources of the Republic of South Africa. Department of Water
Affairs, Government Printer, Pretoria, South Africa.
DWAF, 1996. Hydrologic catchments of South Africa, Swaziland
Commission, Department of Water Affairs and Forestry, Pretoria.
and Lesotho.
Water Research
Dyer, T.G.J. and Tyson, P.D., 1977. Estimating above and below normal rainfall periods over South
Africa, 1972-2000. Journal of Applied Meteorology 16:145-147.
Ehrlich, P. R. and Wilson, E.O., 1991. Biodiversity studies: science and policy. Science 253:758-762.
Ehrlich, P.R., Ehrlich, A.H., and Daily, G., 1995. The stork and the plough. Yale University Press, New
Haven.
Ellery, W.N., Scholes, R.I., and Mentis, M.T., 1992. An initial approach to predicting the sensitivity of
the South African grassland biome to climate change. South African Journal of Science 87:499503.
Ellery, W.N., Scholes, R.I., and Scholes, M.C., 1995. The distribution of sweetveld and sourveld in
South Africa's grassland biome in relation to environmental factors. African Journal of Range and
Forage Science 12:38-45.
Elton, C.S., 1958. The ecology of invasions by animals and plants. Chapman and Hall, London, UK.
ESRI, 1998. Are, Grid and ArcView GIS command references.
Institute, Redlands, CA.
Environmental
Systems Research
Everard, D.A., Midgley, J.J., and van Wyk, G.F., 1995. Dynamics of some forests in KwaZulu-Natal,
South Africa, based on ordinations and size-class distributions. South African Journal of Botany
61:283-292
Fagan, W.F. and Kareiva, P.M., 1997. Using compiled species lists to make biodiversity comparisons
among regions: a test case using Oregon butterflies. Biological Conservation 80:249-259.
Fahrig, L., 1997. Relative effects of habitat loss and fragmentation on population extinction. Journal of
Wildlife Management 61:603-610.
Fair, T.J.D., 1965. The core-periphery concept and population
South African Geographical Journal 47:59-71.
growth in South Africa, 1911-1960.
Fair, T.J.D., 1975. The National Physical Development Plan (NPDP): a summary and a review. South
African Geographical Journal 57:126-134.
Fair, T.I.D., 1976. Polarisation, dispersion and decentralization
South African Journal of Geography 58:40-56.
in the South African space economy.
Fair, T.I.D. and Schmidt, C.F., 1974. Contained urbanization: a case study. South African Geographical
Journal 56:155-166.
Fairbanks, D.H.K., 1995. An afforestation-potential
land evaluation for the Eastern Transvaal,
KwaZulu-Natal, and Eastern Cape Provinces. Department of Water Affairs and Forestry, Report
FOR-DEA 873. Division of Water, Environment and Forestry, CSIR, Pretoria, South Africa.
Fairbanks, D.H.K. and Srnith, R.E., 1996. Report and maps on the afforestation-potential
land
evaluation for the Northern Province, Western Cape Province, Ciskei and Transkei. Department of
Water Affairs and Forestry, Report FOR-DEA 940, Division of Water, Environment, and Forestry,
CSIR, Pretoria, South Africa.
Fairbanks, D.H.K., 2000. Physio-clirnatic
Ecology 149:71-89.
classification
of South Africa's
woodland
biome. Plant
Fairbanks, D.H.K. and Benn, G.A., 2000. Identifying regional landscapes for conservation planning: a
case study from KwaZulu-Natal, South Africa. Landscape and Urban Planning 50:237-257.
Fairbanks, D.H.K. and Thompson, M.W., 1996. Assessing land-cover map accuracy for the South
African land-cover database. South African Journal of Science 92:465-470.
Fairbanks, D.H.K. and McGwire, K.C., 2000. Coarse-Scale Gradient Analysis of Environmental Factors
in Relation to Plant Species Diversity for Vegetation Communities of California. Geographic
Information Science 6(1): 1-13.
Fairbanks, D.H.K., McGwire, K.C., Cayocca, K.D., LeNay, J.P., and Estes, J.E., 1996. Sensitivity to
climate change of floristic gradients in vegetation communities. In GIS and environmental
modeling: Progress and research issues, M.F. Goodchild, L.T. Steyaert, B.O. Parks, C. Johnston,
D. Maidment, M. Crane, and S. Glendinning (Eds.), pp. 135-140. GIS World Books, Ft. Collins,
CO.
Fairbanks, D.H.K., Thompson, M.W., Vink, D.E., Newby, T.S., van den Berg, H.M., and Everard, D.A.,
2000. The South African land-cover characteristics database: a synopsis of the landscape. South
African Journal of Science 96:69-82.
Fairhead, J. and Leach, M., 1996. Misreading the African landscape: Society and ecology in a forestsavanna mosaic. Cambridge University Press, Cambridge, UK.
Faith, D.P., 1995. Biodiversity and regional sustainability analysis.
Ecology, Canberra, Australia.
CSIRO, Division of Wildlife and
Faith, D.P. and Norris, R.H., 1989. Correlation of environmental variables with patterns of distributiong
and abundance of common and rare freshwater macroinvertebrates. Biological Conservation
50:77-98.
Faith, D.P. and Walker, P.A., 1996a. Environmental diversity: on the best-possible use of surrogate
data for assessing the relative biodiversity of sets of areas. Biodiversity and Conservation 5:399415.
Faith, D.P. and Walker, P.A., 1996b. Integrating conservation and development: incorporating
vulnerability into biodiversity-assessment of areas. Biodiversity and Conservation 5:417-429.
Faith, D.P. and Walker, P.A., 1996c. Integrating conservation and development: effective trade-offs
between biodiversity and cost in the selection of protected areas. Biodiversity and Conservation
5:431-446.
Farina, A., 1997. Landscape structure and breeding bird distributions in a sub-Mediterranean agroecosystem. Landscape Ecology 12:265-378.
Farina, A., 2000. The cultural landscape as a model for the integration of ecology and economics.
BioScience 50:313-320.
Fielding, A.H. and Bell, J.F., 1997. A review of methods for the assessment of prediction errors in
conservation presence/absence models. Environmental Conservation 24:38-49.
Flather, C.H., 1996. Fitting species-accumulation functions and assessing regional land-use impacts on
avian diversity. Journal of Biogeography 23:155-168.
Flather, C.H. and Sauer, J.R., 1996. Using landscape ecology to test hypotheses about large-scale
abundance patterns in migratory birds. Ecology 77:28-35.
Folke, C., Holling, C.S., and Perrings, C., 1996. Biological diversity, ecosystems, and the human scale.
Ecological Applications 4:1018-1024.
Foody, G.M., 1992. On the compensation for chance agreement in image classification accuracy
assessment. Photogrammetric Engineering and Remote Sensing 58:1459-1460.
Forde, E.R., 1968. The population of Ghana, a study of the spatial relationships of its socio-cultural
and economic characteristics. Northwestern University Press, Evanston, IL.
Forester, DJ. and Machlis, G.E., 1996. Modeling human factors that affect the loss of biodiversity.
Conservation Biology 10:1253-1263.
Forman, R.T.T., 1989. Ecologically sustainable landscapes: the role of spatial configuration. In
Changing landscapes: An ecological perspective, I.S. Zonneveld and R.T.T. Forman (Eds.), pp.
261-278. Springer-Verlag, New York, NY.
Forman, R.T.T., 1995. Land mosaics: The ecology of landscapes and regions. Cambridge University
Press, Cambridge, UK.
Forman, R.T.T., 2000. Estimate of the area affected ecologically by the road system in the United
States. Conservation Biology 14:31-35.
Forman, R.T.T., and Alexander, L.E., 1998. Roads and their major ecological effects. Annual Reviews
of Ecology and Systematics 29:207-231.
Forman, R.T.T. and Collinge, S.K., 1995. The 'spatial solution' to conserving biodiversity in
landscapes and regions. In Conservation offaunal diversity in forested landscapes, R.M. DeGraaf
and R.I. Miller (Eds.), pp. 537-568. Chapman and Hall, London, UK.
Forman, R.T.T. and Collinge, S.K., 1997. Nature conserved in changing landscapes with and without
spatial planning. Landscape and Urban Planning 37:129-135.
Forman, R.T.T., Galli, A.E., and Leck, C.F., 1976. Forest size and avian diversity in New Jersey
woodlots with some land use implications. Oecologia 26: 1-8.
Forman, R.T.T and Godron, M., 1986. Landscape ecology. John Wiley and Sons, New York, NY.
Franklin, J. F., 1993. Preserving biodiversity: species, ecosystems, or landscapes?
Applications 3:202-205.
Ecological
Franklin, J.F. and Forman, R.T.T., 1987.Creating landscape patterns by forest cutting: ecological
consequences and principles. Landscape Ecology 1:5-18.
Freemark, K., 1995. Assessing effects of agriculture on terrestrial wildlife: developing a hierarchical
approach for the US EPA Landscape and Urban Planning 31:99-115.
Friedman, J., 1972. A general theory of polarized development. In Growth centers in regional
economic development, N.M. Hansen (Ed.). Free Press, New York, NY.
Friedmann, J. and Alonso, W. (Eds.), 1974. Regional development and planning: A reader.
Massachusetts Institute of Technology Press, Boston, MA
Freitag, S. and van Jaarsveld, AS., 1995. Towards conserving regional mammalian species diversity: a
case study and data critique. South African Journal of Zoology 30:136-144.
Freitag, S. and van Jaarsveld, AS., 1997. Relative occupancy, endemism, taxonomic distinctness and
vulnerability: prioritising regional conservation actions. Biodiversity and Conservation 6:211-232.
Freitag, S., van Jaarsveld, AS. and Biggs, H.C., 1997. Ranking biodiversity areas: an iterative
conservation value-based approach. Biological Conservation 82:263-272.
Freitag, S., Hobson, C., Biggs, H.C., and van Jaarsveld, AS., 1997. Testing for potential survey bias:
the effects of roads, urban areas and nature reserves on a southern African mammal data set.
Animal Conservation 1:119-127.
Frohn, R.C., McGwire, K.c., Dale, V.H., and Estes, J.E., 1996. Using satellite remote sensing analysis
to evaluate a socio-economic and ecological model of deforestation in Rondonia, Brazil.
International Journal of Remote Sensing 17:3233-3255.
Furness, R.W. and Greenwood, J.J.D., 1993. Birds as monitors of environmental change. Chapman and
Hall, London, UK.
Gadgil, M., Berkes, F., and Folke, C., 1993. Indigenous knowledge for biodiversity conservation.
Ambio 22:151-156.
Gallant, AL., Binnian, E.F., Omernik, J.M., and Shasby, M.B., 1995.
Professional Paper 1567, US Geological Survey, Washington, D.C.
Ecoregions of Alaska.
Gardner, R.H., Milne, B.T., Turner, M.G., and O'Neill, R.V., 1987. Neutral models for the analysis of
broad-scale landscape pattern. Landscape Ecology 1:19-28.
Gardner, R.H., O'Neill, R.V., Tuner, M.G., and Dale, V.H., 1989. Quantifying scale-dependent effects
of animal movement with simple percolation models. Landscape Ecology 3:217-227.
Gaston, K.J., 1990. Patterns in the geographical ranges of species. Biological Review 65:105-129.
Gaston, KJ. and Lawton, J.H., 1990. Effects of scale and habitat on the relationship between regional
distribution and local abundance. Oikos 58:329-335.
Gauch, Jr., H.G., 1982. Multivariate analysis in community ecology. Cambridge University Press,
Cambridge, UK.
Gertenbach, W.P.D., 1983. Landscapes of the Kruger National Park. Koedoe 26:9-121.
Giampietro, M., 1994. Using hierarchy theory to explore the concept of sustainable development.
Futures 26:197-207.
Goodland, R., 1995. The concept of environmental sustainability. Annual Review of Ecology and
Systematics 26:1-24.
Gould, P.R., 1970. Tanzania 1920-63: the spatial impress of the modernization process. World Politics
22:149-170.
Green, B.H., Simmons, E.A, and Woltjer, I., 1996. Landscape conservation: Some steps towards
developing a new conservation dimension. IUCN-CESP Landscape Conservation Working Group.
Guillet, A and Crowe, T.M., 1985. Patterns of distribution, species richness, endemism and guild
composition of waterbirds in Africa. African Journal of Ecology 23:89-120.
Guillet, A and Crowe, T.M., 1986. A preliminary investigation of patterns of distribution and species
richness of southern African waterbirds. South African Journal of Wildlife Research 16:65-81.
Gustafson, E.I. and Parker, G.R., 1992. Relationships between land-cover proportion and indices of
landscape spatial pattern. Landscape Ecology 7:101-110.
Hagerstrand, T., 1956. Innovation diffusion as a spatial process.
University of Chicago Press, Chicago, IL.
Translated by A. Pred, 1967,
Haggett, P., Cliff, A.D., and Frey, A, 1977. Locational models. Edward Arnold, London, UK.
Haila, Y. and Margules, C.R., 1996. Survey research in conservation biology. Ecography 19:323-331.
Haines-Young, R. and Chopping, M., 1996. Quantifying landscape structure: a review of landscape
indices and their application to forested landscapes. Progress in Physical Geography 20:418-445.
Hall, M., 1987. The changing past: Farmers, kings and traders in Southern Africa, 200-1860. David
Philip, Cape Town, South Africa.
Hall, P. (Ed.), 1966. Von Thiinen's isolated state. Pergamon Press, Oxford, UK.
Hansen, AJ. and di Castri, F. (Eds.), 1992. Landscape boundaries: Consequences for biotic diversity
and ecologicaljlows.
Springer-Verlag, New York, NY.
Hansen, AJ., Spies, T.A., Swanson, F.J., and Ohmann, J.L., 1991. Lessons from natural forests:
implications for conserving biodiversity in natural forests. BioScience 41 :382-392.
Hansen, A.J. and Urban, D.L., 1992. Avian response to landscape pattern: the role of species life
histories. Landscape Ecology 7:163-180
Hansen, N.M., 1967. Development pole theory in a regional context. Kyklos 20:109-127.
Hanski, I., Kouki, J., and Halkka, A, 1993. Three explanations of the positive relationship between
distribution and abundance of species. In Species diversity in ecological communities: Historical
and geographical perspectives, R.E. Ricklefs and D. Schulter (Eds.), pp. 108-116. University of
Chicago Press, Chicago, IL.
Hansson, L. and Angelstam, P., 1991. Landscape ecology as a theoretical basis for nature conservation.
Landscape Ecology 5:191-201.
Harner, R.F. and Harper, K.T., 1976. The role of area, heterogeneity, and favorability in plant species
diversity of pinyon-juniper ecosystems. Ecology 57:1254-1263.
Harrison, J.A, 1992. The Southern African Bird Atlas Project- five years of growth. South African
Journal of Science 88:410-413.
Harrison, J.A and Martinez, P., 1995. Measurement and mapping of avian diversity in southern Africa:
implications for conservation planning. IBIS 137:410-417.
Harrison, J.A, Allan, D.G., Underhill, L.G., Herremans, M., Tree, A.I., Parker, V., and Brown, C.I.,
1997. The atlas of Southern African birds, Volumes 1 and 2. BirdLife South Africa, Johannesburg,
South Africa.
Healey, M.J. and Ilbery, B.W., 1990. Location and change: Perspectives on economic geography.
Oxford University Press, Oxford, UK.
Hobbs, R.J., 1993. Effects of landscape fragmentation on ecosystem processes in the Western
Australian wheatbelt. Biological Conservation 64:193-201.
Hobbs, R.I., Saunders, D.A, and Arnold, G.W., 1993. Integrated landscape ecology: a Western
Australian perspective. Biological Conservation 64:231-280.
Holling, C.S., 1973. Resilience and stability of ecological systems. Annual Review of Ecology and
Systematics 4:1-23.
Holling, C.S., 1986. The resilience of terrestrial ecosystems: local surprise and global change. In
Sustainable development of the biosphere, W.C. Clarke and R.E. Munn (Eds.), pp. 292-317.
Cambridge University Press, Cambridge, UK.
Holling, C.S., 1996. Surprise for science, resilience for ecosystems, and incentives for people.
Ecological Applications 6:733-735.
Host, G.B., Polzer, P.L., Mladenoff, DJ., White, M.A., and Crow, T.R., 1996. A quantitative approach
to developing regional ecosystem classifications. Ecological Applications 6:608-618.
Howard, P.c., Viskanic, P., Davenport, T.R.B., Kigenyi, F.W., Baltzer, M., Dickinson, C.J., Lwanga,
J.S., Matthews, R.A., and Balmford, A., 1998. Complementarity and the use of indicator groups
for reserve selection in Uganda. Nature 394:472-475.
Hunter, M.L., Jacobson, G.L., and Webb, T., 1988. Palaeoecology and the coarse-filter
maintaining biological diversity. Conservation Biology 4:375-385.
approach to
Huntley, B.J., 1989. Biotic diversity in Southern Africa: Concepts and conservation. Oxford University
Press, Oxford, UK.
Huntley, B.J., 1998. The dynamic response of plants to environmental change and the resulting risks of
extinction. In Conservation in a changing world, G.M. Mace, A. Balmford and J.R. Ginsberg
(Eds.), pp. 69-86. Cambridge University Press, Cambridge, UK.
Isard, W., 1956. Location and space economy. John-Wiley and Sons, New York, NY.
Jenks, G.F., 1963. Generalization
Geographers 53:15-26.
in statistical mapping. Annals of the Association
Jizhong, Z., Shijva, M., and Changming,
modelling 59:151-163.
C., 1991.
of American
An index for ecosystem diversity. Ecological
Johnston, R.J., 1980. Multivariate statistical analysis in geography. Longman Press, London, UK.
Jongman, R.H.G., ter Braak, CJ.F., and van Tongeren, O.F.R., 1995. Data analysis in community and
landscape ecology. Cambridge University Press, Cambridge, UK.
Joubert, A.M. and Hewitson, B.c., 1997. Simulating present and future climates of Southern Africa
using general circulation models. Progress in Physical Geography 21 :51-78.
Kirkpatrick, J.B., 1983. An iterative method for establishing priorities for the selection of nature
reserves: an example from Tasmania. Biological Conservation 25:127-134.
Kodric-Brown, A. and Brown, J.H., 1993. Incomplete data sets in community
biogeography: a cautionary tale. Ecological Applications 3 :736-742.
ecology
and
Kok, P., Ndlovu, B., and O'Donovan, M. 1997. Development indicators for promoting good governance
in KwaZulu-Natal, Volume 1: Research findings.
Human Sciences Research Council, HSRC
Printers, Pretoria, South Africa.
Kolasa, J. and Rollo, C.D., 1991. Introduction: the heterogeneity of heterogeneity: a glossary. In
Ecological heterogeneity, J. Kolasa and S.T.A. Pickett, pp. 1-23. Springer-Verlag, New York, NY.
Kotliar, N.B. and Wiens, J.A., 1990. Multiple scales of patchiness and patch structure: a hierarchical
framework for the study of heterogeneity. Oikos 59:253-260.
Kratz, T.K., Benson, B.J., Blood, E.R., Cunningham, G.L., and Dahlgren, R.A., 1991. The influence of
landscape position on temporal variability in four North American ecosystems. American
Naturalist 138:355-378.
Krebs, C.J., 1989. Ecological methodology. Harper Collins Publishers, New York, NY.
Krummel, J.R., Gardner, R.H., Sugihara, G., O'Neill, R.V., and Coleman, P.R., 1987. Landscape
patterns in a disturbed environment. Oikos 48:321-340.
Lamberson, R.H., McKelvey, R., Noon, B.R., and Voss, c., 1992. A dynamic analysis of northern
spotted owl viability in a fragmented forest landscape. Conservation Biology 6:505-512.
Lapin, M. and Barnes, B.V., 1995. Using the landscape ecosystem approach to assess species and
ecosystem diversity. Conservation Biology 9: 1148-1158.
Laszlo, E., 1987. Evolution: the grand synthesis. Shambha1a, New Science Library, Boston, MA.
Latham, R.E. and Ricklefs, R.E., 1993. Global patterns of tree species richness in moist forests: energydiversity theory does not account for variation in species richness. Oikos 67:325-333.
Laurance, W.F., 1994. Rainforest fragmentation
and the structure of small mammal communities in
tropical Queensland. Biological Conservation 69:22-32.
Laurance, W.F. and Yensen, E., 1991. Predicting the impacts of edge effects in fragmented habitats.
Biological Conservation 51 :77-92.
Lea, J.P., 1972. The differentiation of the rural periphery in Swaziland: a multivariate analysis.
African Geographical Journal 54:105-123.
Legendre, P. and Legendre, L., 1998. Numerical ecology.
Environmental Modelling 20, Elsevier, New York, NY.
Second
Li, H., Franklin, J.F., Swanson, F.J., and Spies, T.A., 1993. Developing
patterns: a simulation approach. Landscape Ecology 8:63-75.
Liversidge, R., 1962. Distribution
Museums 2:143-151.
edition,
South
Developments
alternative
in
forest cutting
of birds in relation to vegetation. Annals of the Cape Provincial
Lombard, A.T., 1995. The problems with multi-species conservation: do hotspots, ideal reserves and
existing reserves coincide? South African Journal of Zoology 30: 145-163.
Lombard, A.T., Cowling, RM., Pressey, R.L., and Mustart, PJ., 1997. Reserve selection in a speciesrich and fragmented landscape on the Agulhas Plain, South Africa. Conservation Biology 11: 11011116.
Losch, A., 1954. The economics of location. Yale University Press, New Haven, CT.
Low, A.B. and Rebelo, A. (Eds.), 1996. Vegetation of South Africa, Lesotho,
Department of Environmental Affairs and Tourism, Pretoria, South Africa.
and Swaziland.
Lubchenco, J.A., Olson, A.M., Brubaker, L.B., Carpenter, S.R., Holland, M.M., Hubbell, S.P., Levin,
S.A., MacMahon, J.A., Matson, P.A., Melillo, J.M., Mooney, H.A., Peterson, C.H., Pulliam, H.R,
Real, L.A., Regal, P.J., and Risser, P.G., 1991. The sustainable biosphere initiative: an ecological
research agenda. Ecology 72:371-412.
Ludwig, J.A., 1999. Disturbance and landscapes: the little things count. In Issues in Landscape Ecology,
J.A. Wiens and M.R. Moss (Eds.), pp. 59-63. International Association for Landscape Ecology,
Pioneer Press, Greeley, CO.
Ludwig, J.A. and Reynolds, J.F., 1988. Statistical ecology. Chapman and Hall, New York, NY.
Lynch, J.F. and Whigham, D.F., 1984. Effects of forest fragmentation on breeding bird communities in
Maryland, USA. Biological Conservation 28:287-324.
McArdle, B.H., 1990. When are rare species not there? Oikos 57:276-277.
MacArthur, RH.,
68:387-397.
1964.
Envionmental
factors affecting bird species diversity.
American Naturalist
McDonnell, M.J. and Pickett, S.T.A., 1990. Ecosystem structure and function along urban-rural
gradients: an unexploited opportunity for ecology. Ecology 71:1232-1237.
McDonnell, M.J. and Pickett, S.T.A., 1993. Humans as components of ecosystems:
subtle human effects and populated areas. Springer-Verlag, New York, NY.
The ecology of
McDonnell, M.J., Pickett, S.T.A., and Pouyat, R.V., 1993. The application of the ecological gradient
paradigm to the study of urban effects. In Humans as components of ecosystems, M.J. McDonnell
and S.T.A. Pickett (Eds.), pp. 175-189. Springer-Verlag, New York, NY.
McDonnell, M.J., Pickett, S.T.A., Pouyat, R.V., Ziperer, W.C., Parmelee, RW., Carreiro, M.M., and
Medley, K., 1995. Ecosystem processes along an urban-to-rural gradient. Urban Ecosystems 1:2136.
Mace, G.M., Balmford, A., Boitani, L., Cowlishaw, G., Dobson, A.P., Faith, D.P., Gaston, KJ.,
Humphries, C.J., Lawton, J.H., Margules, C.R., May, R.M., Nicholls, A.D., Possingham, H.P.,
Rahbek, c., van Jaarsveld, A.S., Vane-Wright, R.I., and Williams, P.H., 2000. From hotspots
towards conservation consensus. Nature 405:393.
McGarigal, K. and McComb, W.C., 1995. Relationship between landscape structure and breeding birds
in the Oregon Coast Range. Ecological Monographs 65:235-260.
McGarigal, K., and Marks, BJ., 1995. FRAGSTATS: Spatial pattern analysis program for quantifying
landscape structure. U.S. Forest Service General Teclmical Report PNW 351.
McKenzie, N.L., Belling, L., Margules, C.R., and Keighery, GJ., 1989. Selecting representative
reserve systems in remote areas: a case study in the Nullarbor region, Australia.
Biological
Conservation 50:239-261.
MacLean, G., 1984. Robert's book of South African Birds. The Trustees of the John Voelcker Bird Book
Fund, Cape Town, South Africa.
Maddock, A and du Plessis, M.A, 1999. Real world approaches to the selection of conservation areas.
Biodiversity and Conservation 8:603-615.
Maddock, A. and Benn, G.A, 2000. Identification of conservation-worthy
South Africa. Conservation Biology 14: 155-166.
areas in northern Zululand,
Magurran, AE., 1988. Ecological diversity and its measurement. Princeton University Press, Princeton,
NJ.
Margules, C.R., 1986. Conservation evaluation in practice. In Wildlife conservation
Usher (Ed.), pp. 289-314. Chapman and Hall, London, UK.
Margules, C.R., 1989. Introduction to some Australian
Biological Conservation 50: 1-11.
developments
evaluation M.B.
in conservation
evaluation.
Margules C.R. and Pressey RL., 2000. Systematic conservation planning. Nature 405:243-253.
Margules, C.R. and Redhead, T.D., 1995. BioRap: Rapid assessment of biodiversity of priority areas.
CSIRO, Division of Wildlife and Ecology, Canberra, Australia.
Margules, C.R. and Usher, M.B., 1981. Criteria used in assessing wildlife conservation potential: a
review. Biological Conservation 21:79-109.
Margules, C.R., Creswell, LD., and Nicholls, AO., 1994. A scientific basis for establishing networks of
protected areas. In Systematics and conservation evaluation, P.L. Forey, c.J. Humphries, and R.L
Vane-Wright (Eds.), pp. 327-350. Clarendon Press, Oxford, UK.
Margules, C.R., Nicholls, AO., and Pressey, R.L., 1988. Selecting networks of reserves to maximise
biological diversity. Biological Conservation 25: 127-134.
Matheron, G., 1963. Principles ofgeostatistics.
Economic Geology 58:1246-1266.
Mathsoft, 1999. S-Plus Guide to Statistics, Version 4.5. Matbsoft, Seattle, WA
Maurer, B.A and Heywood, S.G., 1993. Geographic range fragmentation and abundance in Neotropical
migratory birds. Conservation Biology 7:501-509.
Marsh, G.P., 1864. The Earth as modified by human action. Scribner, Armstrong and Co., New York,
NY.
Meffe, G.K. and Carroll,
Sunderland, MA.
C.R.,
1997. Principles
of conservation
biology.
Menge, B.A and Olson, AM., 1990. Role of scale and environmental
community structure. Trends in Ecology and Evolution 5:52-57.
Sinauer
Associates,
factors in regulation
of
Meyers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.AB., and Kent, J., 2000. Biodiversity
hotspots for conservation priorities. Nature 403:853-858.
Midgley, J.J., Cowling, R.M., Seydack, A.H.W., and Van Wyk, G.F., 1997. Forest. In Vegetation of
Southern Africa, R.M. Cowling, D.M. Richardson, and S.M. Pierce, pp. 278-299. Cambridge
University Press, Cambridge, UK.
Milne, B.T., 1992. Spatial aggregation and neutral models in fractal landscapes. American Naturalist
139:32-57.
Milton, S.J., and MacDonald, LAW., 1988. Tree deaths near tar roads in the Northern Transvaal. South
African Journal of Science 84:164-165.
Mittermeier, R.A, Myers, N., Thomsen, J.B., da Fonseca, G.AB., and Olivieri, S., 1998. Biodiversity
hotspots and major tropical wilderness areas: approaches to setting conservation priorities.
Conservation Biology 12:516-520.
1990. Paradise under pressure: St. Lucia, Kosi Bay, Sodwana, Lake Sibaya, and
Maputaland. Southern Book Publishers, Johannesburg, South Africa.
Mountain, AG.,
Munasinghe, M., 1993.
Environmental economics and sustainable development.
Environment Paper No.3, World Bank, Washington, D.C.
World Bank
M. and Cruz, W., 1995. Economy-wide policies and the environment: Lessons from
experience. World Bank Environment Paper No. 10., World Bank, Washington, D.C.
Munasinghe,
Musters, C.J.M., de Graaf, H.J., and ter Keurs, W.I., 2000. Can protected areas be expanded in Africa?
Science 287: 1759-1760.
Myrdal, G., 1958. Economic theory and underdeveloped regions. Vora, Bombay, India.
Nassauer, J.I., 1992. The appearance of ecological systems as a matter of policy. Landscape Ecology
6:239-250.
Naveh, Z., 1995. Interactions oflandscapes
and cultures. Landscape and Urban Planning 32:43-54.
Naveh, Z. and Lieberman, AS., 1993. Landscape ecology: Theory and applications. Springer-Verlag,
New York, NY.
National Research Council, 1994. Setting priorities for the human dimensions of global change.
National Academy Press, Washington, D.C.
Nichols, W.F., Killingbeck, K.T., and August, P.V., 1998. The influence of geomorphological
heterogeneity on biodiversity, II. a landscape perspective. Conservation Biology 12:371-379.
Nicholls, AO., 1998. Integrating population abundance, dynamics and distribution into broad-scale
priority setting. In Conservation in a changing world, G.M. Mace, A Balmford, and J.R. Ginsberg
(Eds.), pp. 251-272. Cambridge University Press, Cambridge, UK.
Nicholls, AO. and Margules,
Conservation 64: 164-169.
c.R.,
1993. An upgraded
reserve
Norgaard, R.B., 1981.
Sociosystem and ecosystem coevolution
Environmental Economics and Management 8:238-254.
selection
algorithm.
in the Amazon.
Biological
Journal
of
Norgaard, R.B., 1984. Coevolutionary development potential. Land Economics 60:2-12.
Norgaard, R.B., 1988. Sustainable development: a co-evolutionary view. Futures 12:606-620.
Norgaard, R.B., 1994a. The process ofloss: exploring the interactions between economic and ecological
systems. American Zoologist 34:145-158.
Norgaard, R.B., 1994b. Development betrayed: The end of progress and a co-evolutionary revisioning
ofthefuture. Routledge, New York, NY.
Noss, R.F., 1983. A regional landscape approach to maintain diversity. BioScience 33:700-706.
Noss, R.F., 1990. Indicators for monitoring biodiversity: a hierarchical approach. Conservation Biology
4:355-364.
Noss, R.F., 1991. Sustainability and wilderness. Conservation Biology 5:120-122.
Noss, R.F., 1996. Ecosystems as conservation targets. Trends in Ecology and Evolution 11:351.
Noss, R.F. and Harris, L.D., 1986. Nodes, networks, and MUMS: preserving
Environmental Management 10:299-309.
diversity at all scales.
Noss, R.F., O'Connell, M.A, and Murphy, D.D. 1997. The science of conservation planning: Habitat
conservation under the endangered species act. Island Press, Washington, D.C., US.
O'Connor, R.I., 1992. Population variation in relation to migrancy status in some North American birds.
In Ecology and conservation of Neotropical migrant landbirds, J.M. Hagan III and D.W. Johnston,
pp. 64-74. Smithsonian Institution Press, Washington, D.C.
O'Connor, R.J., Jones, M.T., White, D., Hunsaker, c., Loveland, T., Jones, B., and Preston, E., 1996.
Spatial partitioning of environmental correlates of avian biodiversity in the conterminous United
States. Biodiversity Letters 3:97-110.
Ohmann, J.L. and Spies, T.A., 1998. Regeional gradient analysis and spatial pattern of woody plant
communities of Oregon forests. Ecological Monographs 68: 151-182.
0ldand, R.H. and Eilersten, 0., 1994. Canonical correspondence analysis with variation partitioning
some comments and an application. Journal of Vegetation Science 5:117-126.
Omernik, J.M., 1987. Ecoregions of the conterminous United States.
American Geographers 77:118-125.
Annals of the Association of
Omernik, J.M., 1995. Ecoregions: a spatial framework for environmental management. In Biological
assessment and criteria: Tools for water resource planning and decision making, T. Simon and W.
Davis (Eds.), pp. 107-114. Lewis Publishers, Chelsea, UK.
O'Neill, R.V., 1999. Theory in landscape ecology. In Issues in Landscape Ecology, J.A. Wiens and
M.R. Moss (Eds.), pp. 1-5. International Association for Landscape Ecology, Pioneer Press,
Greeley, CO.
O'Neill, R.V. and Kahn, J.R., 2000. Homo economus as a keystone species. BioScience 50:333-337.
O'Neill, R.V., DeAngelis, D.L., Waide, J.B., and Allen, T.F.H.,
ecosystems. Princeton University Press, Princeton, NJ.
1986.
A hierarchical
concept of
O'Neill, RV., Hunsaker, C.T., Timmins, S.P., Jackson, B.L., Jones, K.B., Riitters, K.H., and Wickham,
J.D., 1996. Scale problems in reporting landscape pattern at the regional scale. Landscape
Ecology 11:169-180.
O'Neill, RV., Krummel, J.R, Gardner, R.H., Sugihara, G., Jackson, B., DeAngelis, D.L., Milne, B.T.,
Turner, M.G., Zygmut, B., Christensen, S.W., Dale, V.H., and Graham, R.L., 1988. Indices of
landscape pattern. Landscape Ecology 1:153-162.
O'Neill, R.V., Riitters, K.H., Wickham, J.D., and Jones, K.B., 1999. Landscape pattern metrics and
regional assessment. Ecosystem Health 5:225-233.
Opdam, P., 1991. Metapopulation theory and habitat fragmentation: a review ofholarctic
studies. Landscape Ecology 5:93-106.
breeding bird
Opdam, P., Rijsdijk, G., and Hustings, F., 1985. Bird communities in small woods in an agricultural
landscape: effects of area and isolation. Biological Conservation 34:333-352.
Opdam, P., Van Dorp, D., and ter Braak. C.J.F., 1984. The effect of isolation on the number of
woodland birds in small woods in the Netherlands. Journal of Biogeography 11:473-478.
Palmer, M., 1993. Putting things in better order: the advantages of canonical correspondence
Ecology 74:2215-2230.
Pearce, D., 1993. Blueprint 3: Measuring sustainable development.
UK.
analysis.
Earthscan Publications, London,
Pearson, S.M., 1993. The spatial extent and relative influence of landscape-level
bird populations. Landscape Ecology 8:323-334.
factors on wintering
Pearson, D.L. and Carroll, S.S., 1999. The influence of spatial scale on cross-taxon congruence patterns
and prediction accuracy of species richness. Journal of Biogeography 26:1079-1090.
Perrings, F.H. and Walters, S.M. (Eds.), 1962. Atlas of the British flora. Nelson, London, UK.
Peters, R.L., 1992. Conservation of biological diversity in the face of climate change. In Global
warming and biological diversity, RL. Peters and T.E. Lovejoy (Eds.), Yale University Press,
New Haven, CT.
Peters, R.L. and Darling, J.D., 1985. The greenhouse effect and nature reserves. BioScience 35:707-717.
Pimm, S.L., 1984. The complexity and stability of ecosystems. Nature 307:321-326.
Pimm, S.L., 1991. The balance of nature? University of Chicago Press, Chicago, IL.
Pimm, S.L., Russell, G.J., Gittleman, J.L., and Brooks, T.M., 1995. The future of biodiversity. Science
269:347-350.
Poynton, J.C., 1961. Biogeography of south-east Africa. Nature 189:801-803.
Prendergast, J.R., Quinn, R.M., and Lawton, J.H., 1999. The gaps between theory and practice in
selecting nature reserves. Conservation Biology 13:484-492.
Prendergast, J.R., Wood, S.N., Lawton, J.H., and Eversham, B.C., 1993. Correcting for variation in
recording effort in analyses of diversity hotspots. Biodiversity Letters 1:39-53.
Prendergast, J.R, Quinn, R.M., Lawton, J.H., Eversham, B.C, and Gibbons, D.W., 1993. Rare species,
the coincidence of diversity hotspots and conservation strategies. Nature 365:335-337.
Prendergast, J.R. and Eversham, B.C., 1997. Species richness covariance in higher taxa: empirical tests
of the biodiversity indicator concept. Ecography 20:210-216.
Pressey, R.L., 1994. Ad hoc reservation: forward or backward steps in developing representative
reserve systems. Conservation Biology 8:662-668.
Pressey, R.L., 1997. Priority conservation areas: towards an operational defInition for regional
assessments. In National parks and protected areas: Selection, delimitation and management, J,J.
Pigram and RC Sundell (Eds.), pp. 337-357. Centre for Water Policy Research, University of
New England, Arrnidale, Australia.
Pressey, R.L., Bedward, M., and Keith, D.A, 1994. New procedures for reserve selection in New South
Wales: maxirnising the chances of achieving a representative network.
In Systematics and
conservation evaluation, P.L. Forey, CJ. Humphries and RL. Vane-Wright (Eds.), pp. 351-374.
Oxford University Press, Oxford, UK.
Pressey, R.L., Ferrier, S., Hager, T.C, Woods, C.A, Tully, S.L., and Weiman, K.M., 1996. How well
protected are the forests of north-eastern New South Wales?- Analyses of forest environments in
relation to formal protection measures, land tenure, and vulnerability to clearing. Forest Ecology
and Management 85:311-333.
Pressey, RL., Humphries, C.J., Margules, CR., Vane-Wright, R.I., and Williams, P.H., 1993. Beyond
opportunism: key principles for systematic reserve selection. Trends in Ecology and Evolution
8:124-128.
Pressey, R.L., Johnson, I.R., and Wilson, P.D., 1994. Shades of irreplaceability: towards a measure of
the contribution of sites to reservation goal. Biodiversity and Conservation 3:242-262.
Pressey, R.L. and Nicholls, A.a., 1989. EffIciency in conservation evaluation: scoring versus iterative
approaches. Biological Conservation 50: 199-218.
Pressey, R.L., Possingham, H.P., and Margules, CR., 1996. Optirnality in reserve selection algorithms:
When does it matter and how much? Biological Conservation 76:259-267.
Pressey, R. L., Possingham, H.P., and Day, J.R, 1997. Effectiveness of alternative heuristic algorithms
for identifying
indicative minimum requirements
for conservation
reserves. Biological
Conservation 80:207-219.
Preston, F.W., 1948. The commonness, and rarity, of species. Ecology 29:611-627.
Preston, F.W., 1960. Time and space and the variation of species. Ecology 41 :785-790.
Pringle, J., 1982. The conservationists
South Africa.
and the killers. T.V. Bulpin and Books of Africa, Cape Town,
Puzo, B. 1978. Patterns of man- land relations. In Biogeography and Ecology of Southern Africa,
Volume 2, M.J.A Werger and AC. van Bruggen, pp. 1051-1112. Junk Publishers, The Hague,
Netherlands.
Quinn, J.F. and Harrison, S.P. 1988. Effects of fragmentation
evidence from biogeographic patterns. Oecologia 75:132-140.
and isolation
on species richness:
Rapport, D.J., Gaudet, C, Karr, J.R, Baron, J.S., Bohlen, C., Jackson, W., Jones, B., Naiman, R.J.,
Norton, B., and Pollock, M.M., 1998. Evaluating landscape health: integrating societal goals and
biophysical process. Journal of Environmental Management 53: 1-15.
Rebelo, A.G. and Siegfried, W.R., 1990. Protection of Fynbos vegetation: ideal and real-world options.
Biological Conservation 54: 15-32.
Rebelo, AG. and Siegfried, W.R., 1992. Where should nature reserves be located in the Cape floristic
region, South Africa?
Models for the spatial configuration of a reserve network aimed at
maxirnising the protection of floral diversity. Conservation Biology 6:243-252.
Redclift, M.R., 1987. Sustainable development: Exploring the contradictions. Metheun, New York, NY.
Reid, W.V., 1994. Setting objectives for conservation evaluation. In Systematics and conservation
evaluation, P.I. Forey, c.J. Humphries, and R.I. Vane-Wright (Eds.), pp. 1-13. Claredon Press,
Oxford, UK.
Reid, W.V., 1998. Biodiversity hotspots. Trends in Ecology and Evolution 7:275-280.
Reid, W.V., McNeely, J.A, Tunstall, D.R, Bryant, D.A, and Winograd, M., 1993. Biodiversity
indicators/or policy-makers. World Resources Institute, Washington D.C., US.
Reijen, R., Foppen, R., Ter Braak, C., and Thissen, J., 1995. The effects of car traffic on breeding bird
populations in woodland. III. Reduction of density in relation to the proximity of main roads.
Journal of Applied Ecology 32:187-202.
Reyers, B., Fairbanks, D.H.K., van Jaarsveld, AS., and Thompson, M.W., 2000. South African priority
conservation areas: a coarse filter approach. Biodiversity and Conservation in review.
Reyers, B., van Jaarsveld, AS., and Kruger, M., 2000. Complementarity
strategy. Proceedings of the Royal Society London Series B 267: 1-9.
as a biodiversity
indicator
Richardson, H.W., 1973. Regional growth theory. MacMillan, London, UK.
Ricklefs, R.E. and Schluter, D., 1993. Species diversity: regional and historical influences. In Species
diversity in ecological communities: Historical and geographical prospectives, R.E. Ricklefs and
D. Schluter (Eds.), pp. 350-363. University of Chicago Press, Chicago, IL.
Risser, P.G., 1995. Biodiversity and ecosystem function. Conservation Biology 9:742-746.
Ritters, K.H., O'Neill, R.V., Hunsaker, C.T., Wickham, J.D., Yankee, D.H., Timmins, S.P., Jones, K.R,
and Jackson, B.L., 1995. A factor analysis of landscape pattern and structure metrics. Landscape
Ecology 10:23-39.
Robbins, C.S., Dawson, D.K., and Dowell, B.A, 1989a. Habitat area requirements of breeding forest
birds of the Middle Atlantic states. Wildlife Monograph 103.
Robbins, C.S., Sauer, J.R., Greenberg, R.S., and Droege, S., 1989b. Population declines in North
American birds that migrate to the neotropics. Proceedings of the National Academy of Sciences
(USA) 86:7658-7662.
Robertson, A, Simmons, R.E., Jarvis, AM., and Brown, c.J., 1995. Can bird atlas data be used to
estimate population size? A case study using Namibian endemics. Biological Conservation 71:8795.
Rodrigues AS.L., Gregory R.D., and Gaston K.J., 2000. Robustness of reserve selection procedures
under temporal species turnover. Proceedings of the Royal Society London Series B 267:49-55.
Rodrigues AS.L., Tratt R., Wheeler RD., and Gaston K.J., 1999. The performance of existing networks
of conservation areas in representing biodiversity. Proceedings of the Royal Society London Series
B 266:1453-1460.
Romme, W.H., 1982. Fire and landscape diversity in subalpine forests of Yellowstone National Park.
Ecological Monographs 52:199-221.
Rotenberry, J.T., 1978. Components of avian diversity along a multifactorial climatic gradient. Ecology
59:693-699.
Rotenberry, J.T. and Wiens, J.A, 1980. Habitat structure, patchiness, and avian communities in North
American steppe vegetation: a multivariate analysis. Ecology 61:1228-1250.
Rotenberry, J.T., 1986. Habitat relationships of shrub steppe birds: even 'good' models cannot predict the
future. In Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates, J. Verner, M.L.
Morrison, and c.J. Ralph, pp. 217-221. University of Wisconsin Press, Madison, WI.
Rutherford, J., Logan, M.I., and Missen, GJ., 1966. New viewpoints in economic geography. Sydney.
Rutherford, M.C. and Westfall, R.H., 1986. Biomes of southern Africa: an objective characterization.
Memoirs of the Botanical Survey of South Africa 54: 1-98.
Saetersdal, M. and Birks, H.J.R, 1993. Assessing the representativeness of nature reserves using
multivariate analysis: vascular plants and breeding birds in deciduous forests, Western Norway.
Biological Conservation 65:121-132.
Samways, M.J., 1990. Land forms and habitat refugia in the conservation of montane grasshoppers in
Southern Africa. Biological Conservation 4:375-382.
Sauer, C.O., 1925. The morphology of landscape. University of California Publications
2:19-53.
in Geography
Sauer, e.O., 1956. The agency of man on the earth. In Man's role in changing the face of the Earth,
Volumes 1 and 2, W.L. Thomas, Jr., e.O. Sauer, M. Bates, and L. Mumford (Eds.), pp. 49-69.
University of Chicago Press, Chicago, IL.
Scarpace, F., Quirk, B., Kiefer, R., and Wynn, S., 1981. Wetland mapping from digitized aerial
photography. Photogrammetric Engineering and Remote Sensing 52:829-838.
Scholes, R.I. and Walker, RH., 1993. An African savanna: Synthesis of the Nyslvley study. Cambridge
University Press, Cambridge, UK.
Schoeman, J.L. and Scotney, D.M., 1987. Agricultural potential as determined by soil, terrain, and
climate. South African Journal of Science 83:260-268.
Schulze, RE., 1998. South African atlas of agrohydrology
Water Research Commission, Pretoria, South Africa.
and climatology.
WRC Report TT82/96,
Schwabe, C.A, Illing, B., and Wilson, D.e., 1996. Service need and provision within KwaZulu-Natal.
Human Sciences Research Council, HSRC Printers, Pretoria, South Africa.
Schwartz, M.W., 1999. Choosing the appropriate scale of reserves for conservation. Annual Review of
Ecology and Systematics 30:83-108.
Scott, J.M., Davis, F., Csuti, B., Noss, R, Butterfield, B., Groves, C., Anderson, H., Caicco, S.,
D'Erchia, F., Edwards Jr. T.e., Ulliman, J., and Wright, RG., 1993. Gap Analysis: a geographic
approach to protection of biological diversity. Wildlife Monographs 123:1-41.
Shackleton, C.M., 2000. Comparison of plant diversity in protected and communal
Bushbuckridge lowveld savanna, South Africa. Biological Conservation 94:273-285.
Shafer, C.L., 1990. Nature reserves: Island theory and conservation practice.
Press, Washington, D.C.
lands in the
Smithsonian Institution
Shafer, C.L., 1994. Beyond park boundaries. In Landscape planning and ecological networks, E.A
Cook and H.N. van Lier (Eds.), pp. 201-223. Elsevier, Amsterdam.
Sharrock, J.T.R., 1976. The atlas of breeding
Berkhamsted, UK.
birds in Britain and Ireland. T.and AD.
Poyser,
Smith, D., 1971. Industrial location. London, UK.
Soberon, J. and Llorente, J., 1993. The use of species accumulation
species richness. Conservation Biology 7:480-488.
functions for the prediction of
Soja, E.W., 1968. The geography of modernization in Kenya. Syracuse University Press, Syracuse, NY.
Soule, M.E., 1991. Conservation: tactics for a constant crisis. Science 253:744-750.
Soule, M.E. and Sanjayan, M.A, 1998. Conservation targets: do they help? Science 279:2060-2061
Soule, M.E. and Simberloff, D., 1986. What do genetics and ecology tell us about the design of nature
reserves? Biological Conservation 35:19-40.
South African Government, 1975. National physical development plan. Department of Planning and the
Environment, Government Printer, Pretoria, South Africa.
South African Government, 1997. White paper on the conservation and sustainable use of South
Africa's biological diversity. Department of Environmental Affairs and Tourism, Government
Printers, Pretoria, South Africa.
South African Government, 1997. White paper on South African land policy.
Affairs, Government Printers, Pretoria, South Africa.
Department of Land
South African Surveyor General, 1993. Digital elevation model for South Africa.
Affairs, Division of Surveying and Mapping, Cape Town, South Africa.
Department of Land
South African Surveyor General, 1993. 1:500 000 scale road network.
Division of Surveying and Mapping, Cape Town, South Africa.
Department of Land Affairs,
Southworth, F., Dale, V.H., and O'Neill, R.V., 1991. Contrasting patterns of land-use in Rondonia,
Brazil: simulating the effects of carbon release. International Social Science Journal 43 :681-698.
Stats SA, 1998. The 1996 South African population census. Stats SA, Pretoria, South Africa.
Stoms, D.M., 1994. Scale dependence of species richness maps. Professional Geographer 46:346-358.
Stoms, D.M., 2000. GAP management
Landscape Ecology 15:21-33.
status and regional
indicators
of threats to biodiversity.
Strittholt, J. R. and Boerner, R.E.J., 1995. Applying biodiversity gap analysis in a regional nature
reserve design for the edge of Appalachia, Ohio (U.S.A.). Conservation Biology 9:1492-1505.
Systat, 1998. SYSTAT' the system for statistics. Version 8.1. SPSS Inc., Chicago, IL.
Taggart, J.B., 1994. Ordination as an aid in determining priorities for plant community protection.
Biological. Conservation 68:135-141.
Temple, S.A., 1986. Predicting impacts of habitat framgentation on forest birds: a comparison of two
models. In Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates, J. Verner, M.L.
Morrison, and c.J. Ralph, pp. 301-304. University of Wisconsin Press, Madison, WI.
Ter Braak, C.J.F., 1988. Partial canonical correspondence analysis. In Classification and related
methods of data analysis, H.H. Bock (Ed.), pp. 551-558. Pudoc, Wageningen, Netherlands.
ter Braak, C.J.F. and Prentice, I.C., 1988. A theory of gradient analysis. Advances
Research 18:272-327.
in Ecological
ter Braak, C.J.F. and Smilauer, P., 1998. CANOCO reference manual and user's guide for Windows:
Software for canonical community ordination (version 4). Microcomputer Power, Ithaca, NY.
Thoman, R.S., Conkling, E.C, and Yeates, M.H., 1962. The geography of economic activity. McGrawHill, New York, NY.
Thompson, M.W., 1996. A standard land-cover classification scheme for remote-sensing
South Africa. South African Journal of Science 92:34-42.
application in
Thorrington-Smith, Rosenberg, and McCrystal, 1978. Towards a plan for KwaZulu: A preliminary
development plan. KwaZulu Government, Ulundi.
Tilman, D., 1996. Biodiversity: Population versus ecosystem stability. Ecology 77:350-363.
Tilman, D., Wedin, D., and Knops, J., 1996. Productivity and sustainability influenced by biodiversity
in grassland ecosystems. Nature 379:718-720.
Tisdell, c.A., 1995. Issues in biodiversity conservation
Environmental Conservation 22:216-222.
including the role of local communities.
Trombulak, S.C., and Frissell, c.A., 2000. Review of ecological effects of roads on terrestrial and
aquatic communities. Conservation Biology 14: 18-30.
Turner, M.G., 1989. Landscape ecology: the effect of pattern on process. Annual Review of Ecology
and Systematics 20:171-197.
Turner, M.G., 1990. Spatial and temporal analysis oflandscape
patterns. Landscape Ecology 4:21-30.
Turner, M.G. and Gardner, R.H. (Eds.), 1991. Quantitative methods in landscape ecology: The analysis
and interpretation of landscape heterogeneity. Springer-Verlag" New York, New York, US.
Turner, M.G. and Ruscher, C.L., 1988. Changes in the landscape patterns in Georgia, USA. Landscape
Ecology 1:241-251.
Turner, M.G., Wear, D.N., and Flamm, R.O., 1996. Land ownership and land-cover change in the
southern Appalachian higWandsand the Olympic Peninsula. Ecological Applications 6:1150-1172.
Turner, M.G., O'Neill, R.V., Gardner, R.H., and Milne, RT., 1989. Effects of changing spatial scale on
the analysis oflandscape pattern. Landscape Ecology 3:153-162.
Underhill, L.G., Oatley, T.R, and Harrison, J.A, 1991. The role of large-scale data collection projects
in the study of southern African birds. Ostrich 62:124-148.
UNCED, 1992. United Nations convention on biological diversity. United Nations Commission on
Environment and Development, New York, New York, US.
Urban, D.L., O'Neill, R.V., and Shugart, H.H. Jr., 1987. Landscape ecology. BioScience 37:119-127.
Van Jaarsveld, AS., 1996. Towards a co-evolutionary framework for implementing sustainable
development. International Journal of Sustainable Development and World Ecology 3: 15-37.
Van Jaarsveld, AS., Mutembwa, AM., Rangasamy, L., and Hay, D.G., 1996. A global security
approach towards implementing sustainable development in southern Africa. International Journal
of Sustainable Development and World Ecology 3:35-48.
Van Jaarsveld, AS. and Lombard, AT., 1995. Towards the establishment of a national environmental
information network. South African Journal of Science 91:9-10.
Van Jaarsveld, AS., Chown, S.L., Erasmus, B.F.N., Kshatriya, M., and Wessels, K.J., 2000.
Vulnerability
and adaptation
assessment
of South African
animal
taxa to climate
change.
Conservation Planning Unit, Department of Zoology and Entomology, University of Pretoria,
Department of Environmental Affairs and Tourism Report, Pretoria, South Africa.
Van Jaarsveld, AS., Freitag, S., Chown, S.L., Muller, c., Koch, S., Hull, H., Bellamy, C., Kruger, M.,
Endrody-Younga, S., Mansell, M.W., and Scholtz, C.H., 1998. Biodiversity Assessment and
Conservation Strategies. Science 279:2106-2108.
Verlinden, A, 1997. Human settlements and wildlife distribution in the southern Kalahari of Botswana.
Biological Conservation 82:129-136.
Villard, M-A and Maurer, RA, 1996. Geostatistics as a tool for examining hypothesized declines in
migratory songbirds. Ecology 77:59-68.
Vincent, J., 1959. Bird protection in South Africa, with particular reference to Natal. Ostrich, Sup. No.
3.
Virolainen, K.M., Virola T., Suhonen J., Kuitunen M., Lammi A., and Siikamaki P., 1999. Selecting
networks of nature reserves: methods do affect the long-term outcome. Proceedings of the Royal
Society London Series B 266: 1141-1146.
Von Bertalanffy, L., 1968. General systems theory: Foundations,
George Braziller, New York, NY.
development,
and applications.
Von Thiinen, J., 1826. The isolated state in relationship to agriculture and political economy. G.
Fischer, Jena, Germany.
Vos, C.C. and Opdam, P. (Eds.), 1992. Landscape ecology of a stressed environment. Chapman and
Hall, London, UK.
Waldrop, 1994. The science of complexity. Penguin Books, New York, NY.
Walker, RH., 1995. Conserving biological diversity through ecosystem resilience. Conservation
Biology 9:747-752.
Wear, D.N., Turner, M.G., and Flamm, R.O., 1996. Ecosystem management with multiple owners:
landscape dynamics in a southern Appalachian watershed. Ecological Applications 6:1173-1188.
Weber, A, 1909. Theory of the location of industries. Translated by c.J. Friedrich, 1929, University of
Chicago Press, Chicago, IL.
Webster, R. and Oliver, M.A, 1990. Statistical methods in soil and land resource survey. Oxford
University Press, Oxford, UK.
Weinand, H.C., 1973. Some spatial aspects of economic development in Nigeria. Journal of
Developing Areas 7:247-263.
Wessels, KJ., Reyers, B., Kruger, M., and van Jaarsveld, AS., 2000. Incorporating land-cover
infonnation into regional biodiversity assessments in South Africa. Animal Conservation 3: 67-79.
Western, D., 1989. Conservation without parks: wildlife in the rural landscape. In Conservation for the
twenty-first century, D. Western and M. Pearl (Eds.), pp. 158-165. Oxford University Press,
Oxford, UK.
White, D.S., Minotti, P.G., Barczak, M.J., Sifneos, J.C., Freemark, K.E., Santelmann, M.V., Steinitz,
C.F., Kiester, A.R., and Preston, E.M., 1997. Assessing risks to biodiversity from future landscape
change. Conservation Biology 11:349-360.
White, P.S. and Bratton, S.P., 1980. After preservation: philosophical and practical problems of change.
Biological Conservation 18:241-255.
Whittaker, R.H., 1960. Vegetation
Monographs 30:279-338.
of the Siskiyou Mountains,
Oregon and California.
Ecological
Whittaker, R.H., 1965. Dominance and diversity in land plant communities. Science 147:250-260.
Whittaker, R.H., 1967. Gradient analysis of vegetation. Biological Reviews 42:207-264.
Whittaker, RH., 1977. Evolution of species diversity in land communities.
67.
Evolutionary Biology 10:1-
Wickham, J.D., Jones, K.B., Riitters, K.H., O'Neill, R.V., Tankersley, R.D., Smith, E.R., Neale, AC.,
and Chaloud, D.J., 1999.
An integrated assessment of the U.S. Mid-Atlantic region.
Environmental Management 24: 553-560.
Wickham, J.D., Riitters, K.H., O'Neill, RV., Jones, K.B., and Wade, T.G., 1996. Landscape contagion
in raster and vector environments.
International Journal of Geographic Infonnation Systems
10:891-899.
Wickham, J.D., Wu, J., and Bradford, D.F., 1997. A conceptual framework for selecting and analyzing
stressor data to study species richness at large spatial scales. Environmental Management 21 :247257.
Wiens, J.A, 1973. Pattern and process in grassland bird communities. Ecological Monographs 43:237270.
Wiens, J.A, 1974. Habitat heterogeneity and avian community structure in North American grasslands.
American Midland Naturalist 91 :195-213.
Wiens, J.A., 1981. Scale problems in avian censusing. Studies in Avian Biology 6:513-521.
Wiens, J.A, 1984. On understanding a non-equilibrium world: myths and reality in community patterns
and processes. In Ecological communities, conceptual issues and the evidence, D.R Strong Jr., D.
Sirnberloff, L.G. Abele, and A. B. Thistle (Eds.), pp. 439-457. Chapman and Hall, New York,
NY.
Wiens, J.A, 1989a. The ecology of bird communities.
Cambridge, UK.
Volumes 1 and 2. Cambridge University Press,
Wiens, J.A, 1989b. Spatial scaling in ecology. Functional Ecology 3:385-397.
Wiens, J.A, 1992. What is landscape ecology, really? Landscape Ecology 7:149-150.
Wiens, J.A and Rotenberry, J.T., 1981. Habitat associations and community
shrub steppe environments. Ecological Monographs 51 :21-41.
structure of birds in
Wiens, J.A and Rotenberry, J.T., 1985. Response of breeding passerine birds to rangeland alteration in
a North American shrub steppe locality. Journal of Applied Ecology 22:655-668.
Wiens, J.A, Rotenberr, J.T., and Van Home, B., 1987. Habitat occupancy configurations
American shrubsteppe birds: the effects of spatial scale. Oikos 48:132-147.
Wiken, E., 1986. Terrestrial ecozones of Canada. Ecological
Environment Canada, Ottawa, Canada.
Land Classification
of North
Series No. 19,
Williams, P.H. and Gaston, KJ., 1998. Biodiversity indicators: graphical techniques, smoothing and
searching for what makes relationships work. Ecography 21 :551-560.
Williams, P.H., Gibbons, D.W., Margules, C.R., Rebelo, A., Humphries, c.J., and Pressey, R.L., 1996.
A comparison of richness hotspots, rarity hotspots, and complementary areas for conserving
diversity of British birds. Conservation Biology 10: 155-174.
Wilson, E.O. (Ed.), 1988. Biodiversity. National Academy Press, Washington, D.C.
Wilson, M.F., 1974. Avian community organization and habitat structure. Ecology 55:1017-1029.
Winterbottom, J.M., 1959. Some zoo-geographical
No.3.
remarks on the South African avifauna. Ostrich Sup.
Winterbottom, J.M., 1962. A note on zoogeographical limits in South-East Africa, as suggested by the
avifauna. Annals of the Cape Provincial Museums 2:152-154.
With, K. and Crist, T.O., 1995. Critical thresholds in species' responses to landscape structure. Ecology
76:2446-2459.
Wood, C.H. and Skole, D., 1998. Linking satellite, census, and survey data to study deforestation in the
Brazilian amazon. In People and pixels: Linking remote sensing and social science, D. Liverrnann
(Eds.), pp. 70-93. National Academy Press, Washington, D.C.
Woodcock, C. and Strahler, A., 1987. The factor of scale in remote sensing. Remote Sensing of
Environment 21 :311-332.
Woodwell, G.M. and Smith, H.H., 1969. Diversity and stability in ecological
Symposium in Biology 22. Brookhaven National Laboratory, Upton, NY.
WCED, 1987. Our common future. World Commission on Environment
University Press, Oxford, UK.
systems. Brookaven
and Development,
Oxford
Wright, D.H., 1991. Correlations between incidence and abundance are expected by chance. Journal of
Biogeography 18:463-466.
Wright, R.G., MacCracken, J.G., and Hall, J., 1994. An ecological evaluation of proposed new
conservation areas in Idaho: evaluating proposed Idaho national parks. Conservation Biology
8:207-216.
Wright, R.G., Murray, M.P., and Merrill, T., 1998. Ecoregions
Biological Conservation 86:207-213.
as a level of ecological
Yu, K., 1996. Security patterns and surface model in landscape ecological planning.
Urban Planning 36:1-17.
analysis.
Landscape and
K.S. and Young, K.R., 1998. Nature's geography: New lessons for conservation in
developing countries. University of Wisconsin Press, Madison, WI.
Zimmerer,
Zube, E., 1987. Perceived land-use patterns and landscape values. Landscape Ecology 1:37-45.
The following table describes the socio-economic and environmental variables used in
Chapter 6. References to sources are also provided.
Central Statistical Services, 1998. The 1996 South African population census. Stats SA, Pretoria,
South Africa.
Fairbanks, D.H.K., Thompson, M.W., Vink, D.E., Newby, T., van den Berg, H.M., and Everard,
D.A.,2000. The South African landcover characteristics database: a synopsis of the
landscape. South African Journal of Science 96:69-84.
Kok, P., Ndlovu, B., and O'Donovan, M., 1997. Development indicators for promoting good
governance in KwaZulu-Natal, Volume 1: Researchfindings.
Human Sciences Research
Council, HSRC Printers, Pretoria, South Africa.
Schwabe, c.A., TIling, B., and Wilson, D.C., 1996. Service need and provision within KwaZuluNatal. Human Sciences Research Council, HSRC Printers, Pretoria, South Africa.
POPTOTAL
POPDEN
MALE96
FEMALE96
RATIO MF
AGE_0_4
AGE 0 5
AGE_5_14
AGE_I 5_44
AG_15_64
AG_65_99
NO_SCHL
IN_SCHL
NO_DEGRE
YES_DEGR
CHLDNWRK
EMPLOYED
UNMPLOY
DEP_RAT
POVERTY
ABV]OVR
PO V_RAT
RR_INDST
NR_INDST
MANUFAC
ENERGY
CONSTRUC
TRADE
TRAN COM
BUS_SERV
SOC_SERV
PRIVATE
EXT_ORG
REP]ORG
IND_NEC
NO_APP
NA_INST
HOUSE
TRADHOME
FLAT
TOWN
RETIRE
ROOM
SHCK_BCK
SHCK EW
Population census
Population density
Male population
Female population
Ratio of males to females
Total number of children in the 0 - 4 years of age class
Total number of children in the 0 - 5 years of age class
Total number of school age children in the 5 - 14 years of age class
Total number of people in the IS - 44 years prime working age class
Total number of people in the IS - 64 years complete working age class
Total number of people in the 65 - 99 years retirement age class
Total number of children (5 - 14 years) not in school
Total number of children (5 - 14 years) in school
Total number of people with no highschool diploma
Total number of people with a highschool diploma
Total number of children (5 - 14 years) working
Total number of people (IS - 64 years) formally employed
Total number of people (15 - 64 years) not employed
Dependency ratio (children 0 - 14 years / total number employed)
Total number of people living in poverty (less than RI8 000 per annum)
Total number of people living above poverty line (greater than R 18 000 per annum)
Ratio of those in poverty to those living above poverty line
Total number of people employed in renewable resource industries (e.g., agriculture, forestry, etc.)
Total number of people employed in non-renewable resource industries (e.g., mining and quarrying)
Total number of people employed in the commercial manufacturing sector
Total number of people employed in the energy production sector
Total number of people employed in the building construction sector
Total number of people employed in the wholesale and retail trade sector
Total number of people employed in the transport and communications sector
Total number of people employed in the business services sector (e.g., insurance, banks, real estate)
Total number of people employed in the social services sector
Total number of people employed in private households
Total number of people employed in exterritorial organizations
Total number of people employed in representative foreign governments (e.g., diplomatic, NGO)
Total number of people employed in industry NEC or unspecified
Total number of people employed in non-applicable industries by definition in census
Total number of people employed in non-applicable institutions by definition in census
Total number of people living in a formal Western style modern house
Total number of people living in a African traditional house (e.g., made from natural materials)
Total number of people living in flats in blocks (e.g., apartments)
Total number of people living in townhouses or duplexes (e.g., condominiums)
Total number of people living in retirement villages or holiday retirement homes
Total number of people living in a room of a shared house or flat
Total number of people living in a informal dwelling/shack on private property (e.g., backyard or farm)
Total number of people living in a informal dwelling/shack illegally or on town council land
FLATLET
CARAVAN
HOMELESS
HOSTEL
Total
Total
Total
Total
number
number
number
number
of
of
of
of
people
people
people
people
living
living
living
living
in a room or flatlet on shared property
in a caravan or tent
homeless
in a workers hostel or institution (e.g., mining, mental hospital, prison)
1996-97 HSRC
data
SE_lNDEX
SAT_ENV
SAT_HOUSE
SAT_ECON
SAT_SERVICE
BASICS
DEVELOP
BNEEDS
UPGRADE
SAT_LIFE
DEPEND
PROXH20
ELECTRIC
REFUSE
FTOILET
A_LITERACY
F_LITERACY
PERCAPINC
TOTPOLSTA
TOTPOSTOF
BEDS TOT
TELSHAREPR
Service index is a composite index based on the following variables: ratio of population to police stations, post
offices and hospital beds; ratio of road length to district area; ratio of 6 - 17 year olds to school; the percentage of
dwellings that are fully serviced, informal, electrified formal, electrified informal, and number of telephone
shares
Socio-economic index is a composite index based on the following variables: poverty gap, pupi1:teacher ratio,
dependency ratio, total number of households, and the population density
Satisfaction with the general environment and attractiveness of the area ('Yo)
Satisfaction with the household's position (especially in a community context) ('Yo)
Satisfaction with the economic situation ('Yo)
Satisfaction with local facilities and services ('Yo)
Access to or possession of basic items (running water, electricity, flush toilet, and fridge in home) ('Yo)
Need for improving the general development situation ('Yo)
Need for addressing people's basic needs (provide clean water, healthy environment, health care, nutrition and job
creation) ('Yo)
Need for upgrading of infrastructure ('Yo)
Satisfaction with life on the whole ('Yo)
Need for improving administrative dependability and equity ('Yo)
Proximity to safe water ('Yo)
Availability of electricity in homes ('Yo)
Access to refuse removal and waste disposal services ('Yo)
Proportion of households with access to a flush toilet (measure of sanitation) ('Yo)
Adult literacy rate (total number of people with a minimum of five years schooling) ('Yo)
Functional literacy rate (ability to read or write, but ability may not have been attained formally) ('Yo)
Total per capita income
Total number of police stations
Total number of post offices
Total number of hospital beds
Total telephone shares
Fairbanks et a/.
2000
FOR]ER
GRS]ER
WET]ER
LOWI]ER
PLNT]ER
DRY]ER
lRR]ER
URB]ER
UT]ER
M_PER
T_PER
1998 KZNNCSt
Percentage
Percentage
Percentage
Percentage
Percentage
Percentage
Percentage
Percentage
Percentage
Percentage
Percentage
ofland covered by forest and woodland
of land covered by grassland
ofland covered by waterbodies and wetlands
ofland covered by subsistence agriculture
ofland covered by exotic woodlots and commercial plantations
ofland covered by commercial dryland agriculture
ofland covered by commercial irrigated agriculture
of land covered by urban-residential or industrial land-use
land untransformed (e.g., "natural" state)
land under low intensity transformation
land under high intensity transformation
The following manuscript is in revision for Diversity and Distributions.
outlines the method used to rank vegetation priority areas used in Chapter 3.
This small study
1
Conservation Planning Unit
Department of Zoology and Entomology
University of Pretoria
Pretoria, 0002
South Africa
2
Centre for Environmental Studies
University of Pretoria
Pretoria, 0002
South Africa has an important responsibility
inadequate conservation
coarse-filter
to global biodiversity
conservation,
area network for addressing this responsibility.
but a largely
This study employs a
approach based on 68 potential vegetation units to identify areas that are largely
transformed, degraded or underrepresented
in formal national and provincial protected areas. The
assessment highlights broad vegetation types that are currently threatened by human impacts or a
lack of protection. Most vegetation types contain large tracts of natural vegetation, with little
degradation and transformation.
Regions in the grasslands, fynbos and forest biomes are worst
affected. Very few of the vegetation types are adequately protected according to the IUCN's 10%
protected
area conservation
target, with the fynbos and savanna biomes containing
a few
vegetation types that do achieve this goal. In addition to this current vulnerability assessment the
ecological effect of the national road network is also evaluated. This provides an indication of the
remaining untransformed area threatened by the road-effect zones in each vegetation type and can
be used as a measure of potential threat facing that vegetation type due to future impacts and land
use changes. An average of 5.5% of the area of each vegetation type is exposed to road-effects.
Many of the grasslands, fynbos and thicket vegetation types face not only current land use threats,
but may also be exposed to future threats due to a large road-effect zone. This investigation
identifies
areas where limited conservation
resources
should be concentrated
by identifying
vegetation types with high levels of current and potential anthropogenic land use and inadequate
conservation efforts.
South Africa contains a wealth of biodiversity within its borders, unequalled by other temperate
regions. With almost 50% of the world's biodiversity
falling within tropical forests (Myers,
1997), of which South Africa has none (Midgley et al., 1997), the country's contribution to global
biodiversity is unexpectedly large. Inventoried species total over 250 mammals, 790 birds, 303
reptiles,
95 amphibians,
Groombridge,
94 freshwater
fish and 23420 flowering
plants
(Cowling,
1994; WRI, 1994; van Jaarsveld, 2000). Thus one begins to understand how South
Africa earned its place in the top 25 most biodiverse nations in the world (WCMC,
Conservation
1989;
International,
1992;
1998). In addition to this ranking South Africa harbours the fifth
highest number of plant species in the world, with the Cape Floristic Region being recognised as
one of the six floral kingdoms of the world, and one of the 25 hotspots of global biodiversity
(Myers et al., 2000). These hotspots are areas of importance to conservation because of high
levels of species richness, endemism and threat (Myers et al., 1990; 2000). The Cape Floristic
Region is one of the few hotspots to fall entirely within one country. It contains 8200 plant
species of which 5682 are endemic and has already lost approximately
30.3% of its primary
vegetation (Fairbanks et al., 2000; Myers et al., 2000). The Succulent Karoo is another hotspot
falling partly within the boundaries of South Africa (Lombard et al., 1999; Myers et at., 2000).
Although its responsibility towards global biodiversity conservation is large, South Africa
with only 4.8% (DEAT, 1996) (Figure la) of its land surface under formal protection falls far
short of the mCN's nominal recommendation of 10% protected area coverage. This coverage also
lags behind the 10% average attained by the rest of sub-Saharan Africa, with Botswana reaching
18.5%, Mozambique
12.7% and Namibia 12.4% (WRI, 1994; McNeely 1994; Siegfried et aI.,
1998). A moderately
expanding
human population
(Central
Statistical
Survey,
1998) and
associated land transformation in South Africa (mainly urbanisation, cultivation and afforestation
(Hoffmann,
1997» leaves 79% of the country covered with natural woody and grassland
vegetation communities (Figure Ib) (Fairbanks et al., 2000). Waterbodies and wetlands cover less
than one percent of the land surface area, with human land uses making up the remaining 20%
(Fairbanks et al. 2000). Fairbanks et al. (2000) demonstrate that along with the approximately
30% transformation
in the fynbos biome, the savanna and grassland biomes are about 10% and
26% transformed and degraded by human land uses respectively (Figure Ie) (see also Thompson
et al., In Review). In addition to this there are a total of 1176 species presently recognised as
threatened
(WRI, 1994; van Jaarsveld, 2000). Thus with these valuable and often endemic
biodiversity resources, facing ever-increasing
threats from human-induced
land transformation,
and mostly inadequate conservation efforts to stem these threats, South Africa has an obvious
responsibility to do more towards the conservation of biodiversity (van Jaarsveld, 2000).
Most of South Africa's existing protected areas were proclaimed in an ad hoc fashion,
usually because they contained areas with high scenic or tourism potential, contained endemic
diseases and did not conflict with other forms of land use (Pringle, 1982; Freitag et al., 1996;
Pressey et aI., 1993). Because this form ofland allocation to conservation is highly inefficient and
fails to effectively
conserve biodiversity,
several techniques
have been developed
for the
systematic selection of land with a high conservation value, i.e. with high levels of biodiversity
and large anthropogenic
threats facing that biodiversity
& Pressey, 2000). However,
Margules
these techniques
(for reviews
see Williams,
1998;
require data on the distribution
of
biodiversity and threats facing biodiversity in order to identify areas important to conservation.
Because the biodiversity
of a region can never be fully observed and inventoried,
species
distribution data are often used as a surrogate or substitute measure of biodiversity. This form of
data however, has a large number of shortcomings associated with it. These include inadequate
taxonomical knowledge of the groups employed, biased sampling efforts and lack of congruency
between taxa (van Jaarsveld et ai, 1998; Maddock & du Plessis, 1999, Fairbanks & Benn, 2000;
Reyers et al., 2000).
In recent years, the focus for conservation has shifted, with recommendations
towards a more
holistic approach of protecting biodiversity in the aggregate, the so-called 'coarse-filter' approach
(Noss, 1987; Noss, 1990). The goal of coarse-filter conservation is to preserve all or most species
in a region by protecting sufficient (>20000 ha) samples of every plant community type (see Scott
et al., 1993). Other hierarchical methods have included species assemblages,
land facets, or
landscapes (Pressey; 1994; Pressey & Logan, 1994; Wessels et al., 1999; Fairbanks & Benn,
2000). At a national scale South Africa has a few databases of broader surrogates for biodiversity,
including Acocks' Veld Types (Acocks, 1988) and the more recent Vegetation of South Africa,
Lesotho
and Swaziland
biological resources
(Low & Rebelo,
1996; McDonald,
1997). Acocks (1988) defined
from a purely agricultural potential perspective,
while Low and Rebelo
(1996) looked at the definition of these resources from a management and potential use angle.
These vegetation units were defined as having, "... similar vegetation structure, sharing important
plant species, and having similar ecological processes." Thus, these are units that would have
potentially
occurred today, were it not for all the major human-made
agriculture and urbanisation.
transformations
e.g.
Therefore the Low and Rebelo (1996) vegetation map contains
significant potential for acting as a broad scale surrogate of South African biodiversity and for
identifying land important to biodiversity conservation.
Before the Low and Rebelo (1996) map can be used one has to differentiate between the potential
vegetation cover of regions (as defined by Low & Rebelo, 1996) and that which is in reality found
in the region, In other words one needs an indication of current natural vegetation pattern, degree
of transformation,
and amount of protection afforded each vegetation type before one can decide
ifit constitutes a conservation priority (Rebelo, 1997). As Low and Rebelo (1996) point out "there
is little point in setting aside more of a vegetation type with vast expanses in pristine condition,
while ignoring the last patches of a type which is not yet conserved." Low and Rebelo (1996)
provide some estimates of protection and transformation data, however as they admit, "these are
woefully incomplete".
Thus, some indication of current land-cover
(the suite of natural and
human-made features that cover the earth's immediate surface) at a national scale is required for
effective land-use planning, sustainable resource management, environmental research and in this
instance conservation planning (Rebelo, 1997; Fairbanks et ai" 2000).
To this end the advent of the National Land-cover
(NLC) database is of extreme
relevance. This national database was derived using manual photo-interpretation
techniques from
a series of 1:250,000 scale geo-rectified hardcopy satellite imagery maps, based on seasonally
standardised, single date Landsat Thematic Mapper (TM) satellite imagery captured principally
during the period 1994-95 (Fairbanks & Thompson, 1996). It provides the first single standardised
database of current land-cover information for the whole of South Africa, Lesotho and Swaziland
(Fairbanks et al., 2000). For the purpose of the present study the 31 land-cover classes were
reclassified into three categories: natural, degraded and transformed land-cover (Table 1). Natural
land-cover included all untransformed
vegetation, e.g. forest, woodland, thicket and grassland.
The degraded land-cover category was dominated by degraded classes of land-cover. These areas
have a very low vegetation cover in comparison with the surrounding natural vegetation cover and
were typically associated with rural population centres and subsistence level farming, where fuelwood removal, over-grazing and subsequent soil erosion were excessive (Thompson 1996). The
transformed
category consisted
of areas where the structure and species composition
were
completely or almost completely altered which includes all areas under crop cultivation, forestry
plantations, urbanised areas, and mines/quarries.
The databases of potential vegetation cover and current land-cover, along with a map of
protected areas in South Africa, were overlaid in a geographic information
system (GIS) to
determine the extent of natural, degraded, transformed and protected area within each of the 68
vegetation types identified in Low and Rebelo (1996). In addition to this the NLC database could
be used to identify the major broad categories of current threat (e.g. cultivation, forestry) facing
these respective vegetation types.
In addition to these current land use threats, one of the most widespread forms of alteration of
natural habitats and landscapes over the last century has been the construction and maintenance of
roads (Trombulak & Frissell, 2000). Road networks affect landscapes and biodiversity in seven
general ways: (1) increased mortality from road construction; (2) increased mortality from vehicle
collisions; (3) animal behaviour modification;
(4) alteration of the physical environment;
(6)
alteration of the chemical environment; and (7) increased alteration and use of habitats by humans
(from Trombulak & Frissell, 2000). These networks cover 0.9% of Britain and 1.0% of the USA
(Forman & Alexander,
ecological
1998), however the road-effect zone, the area over which significant
effects extend outward from the road, is usually much wider than the road and
roadside. Thus while the National Land-cover database provides a reasonable estimate of areas
with high
current
transformation,
vulnerability
to biodiversity
loss due to existing
anthropogenic
land
road-effect zones can be used to provide an estimate of the potential threat to
regional biodiversity through changing land uses and increased future human impacts.
Some evidence on the size of the road-effect zone is available from studies in Europe and
North America. Reijnen et ai. (1995) estimated that road-effect zones cover between 12-20% of
The Netherlands, while Forman (2000) illustrated that 19% of the USA is affected ecologically by
roads and associated traffic. The road-effect zone for South Africa was determined using a similar
method to that used by Stoms (2000) in which the spatial extent of road effects can be used as an
ecological indicator that directly represents impacts on biodiversity. For this, the road-effect zone
was used as a measure of the area potentially affected by roads. The affected distances were
estimated from the reviews mentioned above, as well as from local studies (Milton & MacDonald,
1988). Unsolicited
KwaZulu-Natal
untransformed
Fairbanks
data, which demonstrated that more than 80% of the transformed
Province
occurs within 2 km of a road, with approximately
61% of the
areas occurring within the same distance, was also used (unpublished
& G. Benn).
Therefore
national
routes
and freeways
area of
were assumed
data D.
to affect
biodiversity for a greater distance from the roadway (1 km on each side) than farm roads (100 m,
Table 2).
Road segments from the South African Surveyor General 1993 1:500,000 scale map
series files (SA Surveyor General, 1993) were buffered in a standard geographic information
system operation to the distance related to its class (Figure Id). The roads in protected areas were
excluded from this analysis as the road-effect in national parks is of little concern. A road-effect
zone was calculated
for the remaining untransformed
areas within each vegetation
type by
summing the total area within the road effect zone surrounding roads in each vegetation type and
converting to a percentage of the total remaining untransformed area in that vegetation type.
The majority of vegetation types of South Africa are not largely degraded or transformed (Table
3). Of the 68 vegetation types 61 contain more than 50% natural vegetation cover with an average
of 76.7% natural vegetation cover across all vegetation types. The vegetation types contain an
average of 5.6% degraded surface area, with all but one (Afro Mountain Grassland) being less
than 20% degraded (Table 3). Only five of the vegetation types are more than 50% transformed
by anthropogenic land uses, with an average of 17.3% being transformed within vegetation types.
Figure 2 provides a diagrammatic representation
of the current levels of transformation,
degradation and protection across all vegetation types. Similar to the findings of the coarse-scale
species-based approach used by Rebelo (1997), the grasslands and fynbos have experienced the
most transformation
(see Fairbanks et al., 2000), with the coastal indigenous forests having been
subjected to extensive transformation for its size (Figures 2a, b). The grasslands biome as well as
a few areas in the savanna biome are moderately degraded (Figure 2c).
The average amount of vegetation type currently under protection is 9.6% with only 18
vegetation types conforming to the meN's
nominal recommendation
of 10% protected area
coverage (Table 3). Only a few regions in the savanna and fynbos biomes receive adequate levels
of protection (Figure 2d).
Table 4 provides a list of vegetation types ordered according to their current vulnerability
status. This was calculated by ranking each vegetation type from one to 68 according to the
amount of area that was degraded, transformed and protected. Vegetation types were ranked from
one (lowest) to 68 (highest) according to the amount of land degraded or transformed, and from
68 (lowest) to one (highest) according to the amount of protected
area coverage. Thus a
vegetation type with large amounts of land degraded or transformed and a low level of protection
would be ranked high (close to 68) for all three columns in Table 4. The average of these three
columns could then be used as an indication of the current vulnerability status of that vegetation
type. Types with high average ranks face a high risk of biodiversity loss due to a combination of
extensively degraded and transformed areas with a low protection status.
conservation vegetation types drawn from Table 4. The Afro Mountain Grassland, Moist Cold
Highveld Grassland, Eastern Thorn Bushveld, Subarid Thorn Bushveld, Moist Upland Grassland
and Kalahari Plains Thorn Bushveld all contain large areas of degraded vegetation. These same
vegetation types (except the Kalahari Plains Thorn Bushveld) along with the Sand Plain Fynbos,
Short Mistbelt Grassland, Laterite Fynbos and Coastal Bushveld-Grassland
areas of commercial, semi-commercial
Mistbelt
Grassland
contain extensive
and subsistence dryland cultivation (Table 5). The Short
and Coastal Bushveld-Grassland
contain large areas of exotic forestry
plantations and commercial sugarcane cultivation (Table 5). Of all these priority vegetation types
only the Coastal Bushveld-Grassland
has more than 10% protected area coverage at 13.5%, but
high levels of degradation as well as high levels of transformation still make it an area of concern
along its entire latitudinal distribution. The rest of these top 10 priority vegetation types all fall
below five percent protected area coverage (Table 3).
The Shrubby
Kalahari
Dune Bushveld,
Upland
Succulent
Karoo,
Lebombo
Arid
Mountain Bushveld, Thorny Kalahari Dune Bushveld and Mopane Shrubveld are all areas of less
concern to biodiversity conservation due to a combination of low levels of land transformation
and degradation
within these vegetation types and high levels of protection
majority of these vegetation types fall above the IUCN's recommended
(Table 3). The
10% protected area
coverage, with the exception of the Upland Succulent Karoo at 4.2% (Table 3). The Mopane
Shrubveld
and Thorny
Kalahari
Dune Bushveld
include
100 and 99.6% protected
area,
respectively. These areas also contain extensive tracts of natural vegetation ranging from 83.5%
for the Thorny Kalahari Dune Bushveld to 100% for the Mopane Shrubveld (Table 3).
Low and Rebelo (1996) also provided an estimate of threat status of the vegetation types. This
included a measure of land transformed by agriculture and other uses, based on "scant information
for some of the Acocks Veld Types and should be cautiously interpreted as a rough index of
habitat loss" (Low & Rebelo, 1996). They also include an estimate of the proportion of each
vegetation type falling within conserved areas, based on an approximation
boundaries
which still require
confirmation
(Low & Rebelo,
of conservation area
1996). Following
a similar
methodology to Thompson et al. (in review), we evaluate these estimates from Low and Rebelo
(1996) as well as the calculations of protected and transformed land obtained from this study
using the National Land-cover database (Table 3). Top conservation
identified based on Low and Rebelo's
priority vegetation types
(1996) estimates in Table 3 highlight the Moist Clay
Highveld Grassland, Dry Clay Highveld Grassland, Moist Cool Highveld Grassland, Kalahari
Plateau Bushveld,
Dry Sandy Highveld Grassland,
Karroid Kalahari Bushveld,
Moist Cold
Highveld Grassland, West Coast Renosterveld, Natal Central Bushveld and Clay Thorn Bushveld
as areas of conservation concern due to high land transformation and low levels of protection. The
Mountain Fynbos, Mopane Bushveld, Lebombo Arid Mountain Bushveld, Mopane Shrubveld and
Thorny Kalahari Dune Bushveld are estimated to be areas of low priority for conservation as they
are well protected and little transformed (Table 3).
As found in Thompson et al. (in review), there is some degree of similarity in the rank
orders of vegetation types according to threat status found in this study and in Low and Rebelo's
(1996) estimates, Table 3 illustrates the differences found between them. The Low and Rebelo
(1996) estimates for land transformation and protection being consistently and significantly higher
(paired t-test for levels of transformation, t
test for levels of protection, t
= 9.00, degrees of freedom = 49, p < 0.0001; paired t-
= 3.8, degrees of freedom = 67, p < 0.01). It must however be noted
that the estimates of transformation
in Low and Rebelo (1996) included grazed areas, while the
NLC transformation category does not (Thompson et al. in review).
The road-effect zone impacts on an average of 5.5% of the remaining natural land-cover in all
vegetation
Highveld
types (Table 3), with 5 vegetation types (Mesic Succulent
Grassland,
Dune Thicket,
Eastern
Thorn Bushveld,
Thicket, Moist Clay
Rocky Highveld
Grassland)
containing between 10 and 14.2% road-effect zones (Table 3). The rest of the vegetation types lie
under this 10% level, with the Mopane Shrubveld containing no road-effect due to the fact that it
all falls entirely within the boundaries of the Kruger National Park (Table 3).
Figure 3 is a graphic representation
of the current vulnerability
status of the vegetation types
(Table 4), as well as their potential vulnerability status, measured as the ranked potential threat
facing the vegetation types due to the sizes of their road-effect zones (Table 3). This figure
demonstrates the fact that many of the grasslands, fynbos and thicket vegetation types face not
only current land use threats, but also may be exposed to future threats due to a large road-effect
zones. However, the road-effect zone used here does not consider the spatial pattern of roads. So,
although roads clearly have a significant impact on many species, meaningful indicators of roadeffects on landscapes await the attention of landscape ecologists and other scientists (Forman,
1998). As articulated by Stoms (2000), many aspects of roads affect biodiversity: road width,
traffic volume, traffic speed, vehicle miles travelled, road network
configuration,
management
of the right-of-way,
structure
or its spatial
noise levels, light disturbance,
and chemical
pollution. Most of these factors also vary over daily, weekly, and annual cycles, which may
interfere with critical behavioural periods such as breeding or migration. As such, the road-effect
zone can represent only a first order approximation attempt to capture more of the multidimensional nature of road network effects.
South Africa, with its large biodiversity conservation responsibility, faces the additional problems
of limited resources for conservation as well as pressing land reform initiatives. The land tenure
system is a problem for conservation throughout Africa and is now becoming an increasingly
demanding problem in South Africa. The almost total transfer of land in most regions of South
Africa, from government to private ownership, is possibly unique in the annals of European
co10nisation. The state by the mid 1930's had lost control over resources which in countries such
as Australia or the USA were retained by the authorities because of their unsuitability for
agriculture (Christopher, 1982). In effect the absence of state interest in land through a leasehold
system has lead to a strong demand for land and an attempt to make a living in areas highly
unsuitable for the purposes of farming. Demand for land has further driven land prices to levels
far in excess of its value as an agricultural commodity.
Therefore the limited resources of available government land and funding need to be
efficiently applied in order to ensure effective conservation as well as development opportunities.
This investigation provides an important first approximation towards identifying areas where
these limited resources should be concentrated by identifying vegetation types with high levels of
current and potential anthropogenic land use and inadequate conservation efforts in order to
constrain future spreading of transformation. As Rebelo (1997) points out, few vegetation units
are spatially uniform in terms of species composition and ecosystem processes, thus further study
within these priority areas is required to identify representative conservation sites within these
types. Although Low and Rebelo (1996) provided rough estimates of areas considered to be
facing high threats, the value of timely land-cover information on the decision making ability for
planning is evident from the present study. The advent of the National Land-cover database has
provided a much-needed standardised dataset of current land-cover to significantly improve South
African land use and conservation planning.
Further issues relevant to the identification of priority conservation areas are the scale of
conservation priority setting, and the effects of global climate change on southern African
vegetation. Rebelo (1997) points out that generally vegetation types shared with other
neighbouring nations are more adequately conserved than vegetation endemic to South Africa.
Thus a classification of vegetation types across political boundaries, as well as international cooperation are urgent requirements for future priority setting. In addition to this, future
conservation strategies will have to consider the effects of climate change on biodiversity
(Rutherford et ai., 2000). Not much is known on what these climate changes or their biological
impacts will be, but recent work has highlighted a general eastward shift in South African species
distributions as areas in South Africa dry out and warm up (Rutherford et ai., 2000; van Jaarsveld
et ai., 2000). It has also been shown that premier flagship conservation areas in South Africa are
not likely to meet their conservation
importance in any conservation-planning
goals (van Jaarsveld et ai., 2000). This is of obvious
scenario.
In many respects "lines conquer", and the South African landscape is a testament to their
power. Compasses and plumb lines, more than a force of arms, subdue landscapes, and henceforth
demarcate
control
and change. If current
development
policies
(i.e. Spatial Development
Initiatives, unstructured land reform) continue without proper equity towards conserving the most
threatened vegetation communities, in a few decades not only will the remaining "natural" areas
be gone, but the people will be even poorer for it.
We thank the Mellon Foundation, University of Pretoria, the National Research Foundation and
the South African Biodiversity Monitoring and Assessment Programme for financial assistance,
as well as GIMS® for GIS software and support.
Acocks, P.H. (1988). Veldtypes of South Africa, 3rd Edition. Memoirs of the Botanical Survey of South
Africa. 57, 1-146.
Central Statistical Survey. (1998) The People of South Africa Population Census, 1996. Report No.1 :0301-11, Pretoria.
Conservation International. (1998) Conservation
world. Http://www.conservation.org.
International
lists the countries of megadiversity
in the
Cowling, R.M., Gibbs, R.G.E., Hoffman, M.T. & Hilton-Taylor, C. (1989) Patterns of plant species
diversity in southern Africa. Biotic diversity in southern Africa: concepts and conservation (ed.
B.J. Huntley). Oxford University Press, Cape Town, South Africa.
DEAT. (1996) National
Pretoria.
Protected
Areas Database.
Department
Environmental
Affairs and Tourism,
Fairbanks, D.H.K. & Benn, G.A (2000) Identifying regional landscapes for conservation planning: A case
study from KwaZulu-Natal, South Africa. Landscape and Urban Planning. 50, 1-19.
Fairbanks, D.H.K. & Thompson, M.W. (1996) Assessing land-cover map accuracy for the South African
land-cover database. South African Journal of Science. 92,465-470.
Fairbanks, D.H.K., Thompson, M.W.M., Vink, D.E., Newby, T. S., van den Berg, H.M. & Everard, DA
(2000) The South African land-cover characteristics database: a synopsis of the landscape. South
African Journal of Science. 96, 69-82.
Forman, R.T.T. (2000) Estimate of the area affected ecologically by the road system in the United States.
Conservation Biology. 14,31-35.
Forman, R.T.T. & Alexandra, L.E. (1998) Roads and their major ecological effects. Annual Reviews of
Ecology and Systematics. 29,207-231.
Freitag, S., Nicholls, AO. & Van Jaarsveld, AS. (1996) Nature reserve selection in the Transvaal, South
Africa: what data should we be using? Biodiversity Conservation. 5, 685-698.
Groombridge, B. (ed.) (1994) Biodiversity
Cambridge, UK.
data sourcebook.
World Conservation
Monitoring
Centre,
Hoffmann, M.T. (1997) Human impacts on vegetation. Vegetation of southern Africa. (eds. R.M. Cowling,
D.M. Richardson and S.M. Pierce), pp 507-534. Cambridge University Press, UK.
Lombard, AT., Hilton-Taylor, c., Rebelo, AG., Pressey, R.L., & Cowling, R.M. (1999) Reserve selection
in the Succulent Karoo, South Africa: coping with high compositional turnover. Plant Ecology.
142,35-55.
Low, AB. & Rebelo, AG. (1996) Vegetation Map of South Africa, Lesotho and Swaziland. Department of
Environmental Affairs, Pretoria, South Africa.
Maddock, A & Du Plessis, M.A (1999) Can species data only be appropriately
biodiversity. Bioiversity & Conservation. 8,603-615.
used to conserve
Margules, C.R. & Pressey, R.L. (2000) Review article: Systematic conservation planning. Nature. 405, 243253.
McDonald, D.J. (1997) VEGMAP: a collaborative project for a new vegetation map of southern Africa.
South African Journal of Science. 93, 424-426.
McNeely,
I.A (1994) Protected areas for the 21st century: working to provide benefits to society.
Biodiversity and Conservation. 3, 390-405.
Midgley, J.J., Cowling, R.M., Seydack, AH.W. & van Wyk, G.F. (1997) Forest. Vegetation of Southern
Africa. (eds. R.M. Cowling, D.M. Richardson and S.M. Pierce), pp. 278-299. Cambridge
University Press, Cambridge, UK.
Milton, SJ. & MacDonald, LAW. (1988) Tree deaths near tar roads in the Northern Transvaal. South
African Journal of Science. 84, 164-165.
Myers, N. (1990) The biodiversity challenge: expanded hot-spots analysis. The Environmentalist.
56.
10, 243-
Myers, N. (1997) Global biodiversity II: Losses and threats. Principles of Conservation Biology, 2nd edition.
(eds. G.K. Meffe and C.R. Carroll), pp 123-158. Sinauer Assosciates, Inc, Sunderland,
Massachusetts, USA
Myers, N., Mittermeier, R.A, Mittermeier, C.G., de Fonesca, G.A.B. & Kent, I. (2000) Biodiversity
hotspots for conservation priorities. Nature. 403, 853-58.
Noss, R.F. (1987) From plant communities to landscapes in conservation inventories: A look at The Nature
Conservancy (USA). Biological Conservation. 84, 223-234.
Noss, RF. (1990) Indicators for monitoring biodiversity: a hierarchial approach. Conservation Biology. 4,
355-64.
Pressey, RL. (1994) Land classifications are necessary for conservation planning but what do they tell us
about fauna. Future of the Fauna of Western New South Wales (Eds. D. Lunney, S. Hand, P. Reed
& D. Butcher) The Royal Zoological Society of New South Wales: Chapter 3.
Pressey, R.L., Humphries, C.I., Margules C.R., Vane-Wright R.L & Williams P.H. (1993) Beyond
opportunism: key principles for systematic reserve selection. TREE. 8, 124-28.
Pressey, R.L. & Logan, V. S. (1994) Level of geographic subdivision and its effects on assessments of
reserve coverage: a review of regional studies. Conservation Biology. 8, 1037-46.
Rebelo, AG. (1997) Vegetation of southern Africa. (eds. RM. Cowling, D.M. Richardson and S.M. Pierce),
pp 571-590. Cambridge University Press, UK.
Reijen, R., Foppen, R., Ter Braak, C. & Thissen, J. (1995) The effects of car traffic on breeding bird
populations in woodland. III. Reduction of density in relation to the proximity of main roads.
Journal of Applied Ecology. 32,187-202
Reyers, B., van Iaarsveld, AS., & Kruger, M. (2000) Complementarity
Proceedings of the Royal Society of London: B. 267,505-13.
as a biodiversity indicator strategy.
Rutherford, M.C., Powrie, L.W. & Schulze, R.E. (2000) Climate change in conservation areas of South
Africa and its potential impact on floristic composition: a first assessment. Diversity and
Distributions. 5,253-262.
Edwards, T.C., Ulliman, 1. & Wright, R.G. (1993) Gap analysis: A geographic approach to
protection of biological diversity. The Journal of Wildlife Management: Wildlife Monographs No.
123.57,1-41.
Siegfried,
W.R., Benn, G.A, & Gelderblom,
implications of landscape characteristics
131-140.
C.M. (1998) Regional assessment
of African national parks. Biological
and conservation
Conservation. 84,
Stoms, D.M. (2000) GAP management status and regional indicators of threats to biodiversity. Landscape
Ecology. 15,21-33.
Thompson, M. (1996) The standard land-cover classification scheme for remote-sensing
South Africa. South African Journal of Science. 92, 34-42.
application in
Thompson, M., Vink, E.R., Fairbanks, D.H.K., Balance, A. & Shackelton, C.M. In Review. Comparison of
Extent and Transformation of South Africa's Woodland Biome from Two National Databases.
South African Journal of Science.
Trombulak, S.c. & Frissell, C.A (2000) Review of ecological effects of roads on terrestrial and aquatic
communities. Conservation Biology. 14, 18-30.
Van Jaarsveld, AS. (2000) Biodiversity: conserving the building blocks of environmental health. The
sustainability challenge for southern Africa. (ed. J. Whitman), pp 265-290. Macmillan Press Ltd,
UK.
Van Jaarsveld, AS., Chown, S.L., Erasmus, B.F.N., Kshatriya, M. & Wessels, K.J. (2000) Vulnerability
and adaptation assessment of South African animal taxa to climate change. Report to the
Department of Environmental Affairs and Tourism, Pretoria, South Africa
Van Jaarsveld, AS., Freitag, S., Chown, S.L., Muller, c., Koch, S., Hull, H., Bellamy, c., KrUger, M.,
Endrody-Younga,
S., Mansell, M.W. & Scholtz, C.H. (1998) Biodiversity assessments and
conservation strategies. Science. 279, 2106-2108.
Wessels, K.J., Freitag, S. & van Jaarsveld, AS. (1999) The use of land facets as biodiversity surrogates
during reserve selection at a local scale. Biological Conservation. 89,21-38.
WCMC. (1992) Development of a National Biodiversity Index: a discussion paper prepared
Conservation Monitoring Centre. Report on the WCMC, 15 September, 1992.
by World
Williams, P.H. (1998) Key sites for conservation: area selection methods for biodiversity. Conservation in a
Changing World: Integrating Processes into Priorities for Action. (Eds. G.M. Mace, A Balmford
& J.R. Ginsberg) pp 211-249. Cambridge University Press, UK.
National and provincial parks
DProtected
areas
~lified
National Land-cover database
l-J Transformed land-<:over
_
Degraded land-<:over
[=:J Natural land-<:over
South African biomes
_
Forest
PI
Fynbos
o Grassland
'+:1 Nama
Karoo
o Savanna
o
Succulent Karoo
_
Thicket
Buffered roads
_Roads
Figure 1: Maps of: (a) South African national and provincial protected areas (DEAT, 1996); (b)
transformed, degraded and natural land-cover; (c) biomes (Low & Rebelo, 1996); and (d) road
network buffered according to Stoms (2000).
% transformed area
00-6.56
06.56
- 11.41
11.41 - 21.61
21.61 - 89.8
% degraded area
00-1.97
01.97
- 5.07
5.07 - 11.34
_
11.34 - 36.67
% natural area
08.97
- 51.92
051.92
-76.13
076.13
- 90.24
_
90.24-100
% protected area
0-0.96
00.96
- 6.19
06.19
-14.13
_
14.13 -100
o
0
Figure 2: Diagrammatic representation oflevels of percentage (a) transformed, (b) degraded, (c)
natural and (d) protected vegetation cover within each of Low and Rebelo's (1996) vegetation
types.
16
39
0
62
+
58
'"'~c
•
52
it.
48.
0
..c
•
65
+
~
~
I-
C
:;
...
2
lIE
lIE
50
it.
44
610
53
it.
+
068
+ 40
0
54
it.
33
•
8
~
~
c
.•..~
•
5~11
<1>.
56
•
0
Q.,
49
it.
12
1~4·
2!f'+
13
Low
I
6
•
6615
+4?
25
l-
E'
36
0
if·
••
Low
•
41
670
+
29
•
60
+
20
•
II
57
•
•
59
+
4~2
O.
lIE
Forest
+
Fynbos
°
Grassland
it.
3
lIE
•
•
45
°
•
High
Figure 3: Graph of current and potential vulnerabilities
of Low and Rebelo's (1996) vegetation
types per biome. Current vulnerability measured as vulnerability rank in Table 4, potential
vulnerability
measured as ranked road-effect zone per vegetation type (Table 3). The vegetation
codes are available in Table 3.
Nama Karoo
Savanna
Succulent KarOl
Thicket
Wetlands, grassland, shrub land, bushland, thicket,
woodland, forest
Degraded land-cover
10.1%
Degraded land, erosion scars, waterbodies
Transformed land-cover
16.5%
Cultivated lands, urbanlbuilt-up areas, mines and
quarries, forestry plantations
National route
1000
Freeway
1000
Arterial
500
Main
250
Secondary (connecting and magisterial roads)
100
Other (rural road)
50
Vehicular trail (4 wheel drive route)
25
Table
3: Percentage
natural, degraded, transformed
and protected area of each of the vegetation
percentage of each vegetation type exposed to road-effect zones.
(Values in brackets indicate estimates from Low and Rebelo (J 996))
types,
as well as thl
(Vegetation types with more than 10% protected area coverage are indicated in bold)
Code
Vegetation
% natural
type
% degraded
% transformed
% protected
% roadeffect
1
Coastal Forest
89.3
1.2
9.3 (43)
1.3 (9.5)
2
Afromontane Forest
2.9
29.2 (44)
16.1 (17.6)
3
4
Sand Forest
Dune Thicket
67.9
72.3
15.6
8.5
5.8 (45)
27.6 (25)
13.0
14.8 (51)
46.7 (44.6)
10.6 (14.5)
1.5(2.1)
2.0
3.0(51)
4.6 (8.0)
7.0
4.0 (5.3)
1.2 (1.8)
14.2
62.2
6.5
6.4
1.7
11.2
6.1
6.4
5
VaHey Thicket
6
Xeric Succulent Thicket
7
Mesic Succulent Thicket
72.1
95.0
78.5
8
Spekboom Succulent Thicket
93.1
4.2
14.5(51)
2.6 (unknown)
9
Mopane Shrubveld
0.0
0.0 (0)
100.0 (100.0)
0.0
10
Mopane Bushveld
100.0
92.4
0.9
6.6 (8)
34.0 (38.3)
3.0
4.9
11
Soutpansberg Arid Mountain Bushveld
83.8
10.2
6.0 (65)
10.1 (12.6)
4.3
12
Waterberg Moist Mountain Bushveld
90.2
0.8
9.0 (28)
6.2 (8.6)
13
Lebombo Arid Mountain Bushveld
90.2
0.1
9.1 (unknown)
37.1 (38.0)
3.2
1.0
34.1 (60)
8.7 (unknown)
1.0 (0.9)
5.1
0.0 (0.2)
8.2
11.1
14
Clay Thorn Bushveld
58.7
7.1
15
Subarid Thorn Bushveld
78.7
12.6
16
17
Eastern Thorn Bushveld
Sweet Bushveld
69.7
78.3
13.8
12.0
16.5 (unknown)
9.5 (27)
0.2 (0.5)
1.8 (2.3)
18
Mixed Bushveld
14.1
16.6 (60)
3.6(3.1)
5.3
19
Mixed Lowveld Bushveld
69.3
70.4
19.8 (30)
22.5 (28.3)
20
Sweet Lowveld Bushveld
9.9
1.4
13.5 (30)
62.2 (67.3)
3.1
1.1
9.6
36.0 (76)
12.3
15.9
3.6 (36)
7.0 (9.7)
20.9 (21.5)
4.7
1.1
39.8 (unknown)
13.5 (14.0)
35.0 (87)
18.0 (80)
2.1 (3.6)
1.3 (1.6)
5.9
4.4
15.6 (35)
0.0 (unknown)
99.6 (99.8)
85.1
54.4
4.5
21
Sour Lowveld Bushveld
22
Subhumid Lowveld Bushveld
84.1
23
24
Coastal Bushveld-Grassland
43.5
Coast-Hinterland
25
Natal Central Bushveld
56.7
72.2
26
Natal Lowveld Bushveld
72.5
8.2
9.9
11.9
27
Thorny Kalahari Dune Bushveld
83.5
0.0
28
Shrubby Kalahari Dune Bushveld
Karroid Kalahari Bushveld
96.0
0.0 (55)
19.4 (19.5)
98.8
3.1
1.2
5.3
0.0
'2.2
0.0 (55)
0.1 (0.1)
3.3
Kalahari Plains Thorn Bushveld
Kalahari Mountain Bushveld
73.6
18.9
7.1 (55)
0.5 (0.5)
3.9
99.5
0.2
4.4
0.3 (25)
0.0 (0.0)
1.8(3.1)
4.6
29
30
31
Bushveld
32
Kimberley Thorn Bushveld
76.1
33
34
Kalahari Plateau Bushveid
92.7
66.3
35
Moist Clay Highveld Grassland
36
37
Dry Clay Highveld Grassland
68.2
34.9
Dry Sandy Highveld Grassland
63.5
38
Moist Sandy Highveid Grassland
39
Moist Cool Highveld Grassland
40
Rocky Highveld Grassland
41
Moist Cold Highveld Grassland
Wet Cold Highveid Grassland
42
Moist Upland Grassland
19.5 (55)
4.2 (55)
3.0
0.1
0.4
33.6 (65)
31.4 (79)
0.1
0.8
67.6
60.4
0.7
1.6
46.8
11.3
2.4
88.0
61.4
17.0
65.1 (67)
0.0 (0.0)
9.0
35.8 (65)
OJ (0.3)
31.6 (55)
0.0 (0.7)
9.1
9.4
38.0 (72)
41.8 (70)
0.7 (0.3)
9.6
0.8 (0.6)
9.4 (6.7)
6.7
9.7 (60)
21.6 (60)
Afro Mountain Grassland
36.7
Alti Mountain Grassland
87.5
8.8
3.6 (32)
45
46
67.6
7.1
4.0
6.8
5.5
10.2
11.3
25.3 (45)
1.5 (32)
11.4 (32)
North-eastern Mountain Grassland
South-eastern Mountain Grassland
0.0 (0.0)
0.8 (1.4)
0.0 (0.0)
94.5
51.9
43
44
14.1 (17.8)
7.2
2.3 (2.5)
3.3 (7.4)
0.6 (0.3)
0.0 (0.0)
11.7 (12.5)
4.1
5.5
4.8
5.7
0.8
1.2
4.6
56.9 (89)
0.9 (2.4)
7.6
5.1
12.9 (unknown)
0.1 (1.1)
7.0
3.4
47
Short Mistbelt Grassland
48
Coastal Grassland
38.5
81.7
49
Bushmanland Nama Karoo
Upper Nama Karoo
99.7
0.2
0.1 (unknown)
0.0 (0.0)
50
99.0
0.9
0.1 (unknown)
51
Orange River Nama Karoo
98.1
1.6 (unknown)
52
Eastern Mixed Nama Karoo
94.9
0.1
1.8
3.3 (unknown)
0.0 (0.0)
0.1 (1.5)
1.6 (1.1)
0.8
0.2 (unknown)
0.7 (0.2)
9.0
0.8 (unknown)
9.5 (24)
0.1 (0.0)
0.4 (0.4)
4.2 (4.4)
0.9 (1.3)
53
54
Great Nama Karoo
Central Lower Nama Karoo
99.1
90.2
5.8
4.6
7.4
5.4
6.0
4.0
4.4
55
Strandveld Succulent Karoo
Upland Succulent Karoo
86.3
97.1
2.0
56
0.7
1.7 (unknown)
57
Lowland Succulent Karoo
94.2
2.6
58
Little Succulent Karoo
89.0
2.6
3.2 (unknown)
8.4 (unknown)
59
North-western
94.0
0.0
6.0 (unknown)
0.0 (0.0)
3.0
60
Escarpment Mountain Renosterveld
98.9
80.4
0.3
1.8
0.8 (unknown)
0.0 (0.1)
17.8(11)
9.0
39.4
1.1
89.8 (97)
1.9
58.7 (32)
5.1 (3.6)
0.7 (1.8)
1.5(1.4)
2.4
5.4
88.5
0.7
10.8(11)
26.4 (26.1)
2.9
88.7
64.8
0.8
1.1
10.3 (3)
15.5 (16.1)
0.0 (0.5)
6.0
13.6 (13.8)
1.2 (1.1)
4.0
Mountain Renosterveld
61
Central Mountain Renosterveld
62
West Coast Renosterveld
63
64
South & South-west Coast
Renosterveld
Mountain Fynbos
65
Grassy Fynbos
66
Laterite Fynbos
67
Limestone Fynbos
68
Sand Plain Fynbos
87.2
34.4
7.6
34.1 (50)
5.2 (40)
8.5
57.1 (50)
3.2 (2.3)
3.9
7.7
8.1
8.8
8.6
7.1
Table 4: Vulnerability ranks of vegetation types according to percentage degraded,
transformed and protected area.
Rank Vegetation type
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Afro Mountain Grassland
Moist Cold Highveld Grassland
Eastern Thorn Bushveld
Sand Plain Fynbos
Subarid Thorn Bushveld
Short Mistbelt Grassland
Laterite Fynbos
Moist Upland Grassland
Coastal Bushveld-Grassland
Kalahari Plains Thorn Bushveld
Clay Thorn Bushveld
Dry Clay Highveld Grassland
Valley Thicket
Natal Central Bushveld
Coast-Hinterland
Bushveld
Coastal Grassland
West Coast Renosterveld
Moist Cool Highveld Grassland
Mixed Bushveld
Sour Lowveld Bushveld
South & South-west Coast Renosterveld
Moist Clay Highveld Grassland
Dry Sandy Highveld Grassland
Moist Sandy Highveld Grassland
Sweet Bushveld
North-eastern Mountain Grassland
Kimberley Thorn Bushveld
Kalahari Plateau Bushveld
Dune Thicket
Central Lower Nama Karoo
Natal Lowveld Bushveld
Strandveld Succulent Karoo
Mixed Lowveld Bushveld
Mesic Succulent Thicket
Rocky Highveld Grassland
Afromontane Forest
South-eastern Mountain Grassland
Soutpansberg
Arid Mountain Bushveld
Central Mountain Renosterveld
Coastal Forest
Spekboom Succulent Thicket
Sand Forest
Lowland Succulent Karoo
North-western Mountain Renosterveld
Little Succulent Karoo
Subhumid Lowveld Bushveld
Alti Mountain Grassland
Wet Cold Highveld Grassland
Upper Nama Karoo
Degraded Transformed
rank
rank
68
56
62
49
60
41
22
66
65
67
44
4
61
54
47
42
21
26
63
52
29
11
17
14
58
45
40
36
48
51
57
30
53
43
6
35
38
55
28
24
39
64
33
3
34
59
50
32
20
37
63
43
65
28
64
56
49
62
26
57
67
41
46
58
38
68
61
44
60
66
53
59
54
33
50
47
20
51
10
42
32
48
40
55
52
11
23
45
31
14
22
16
24
27
18
19
34
6
Protected
rank
Average
rank
65
43
52
38
64
42
67
29
15
49
40
68
34
36
30
55
45
47
26
20
35
66
51
58
32
27
31
62
17
53
13
50
8
25
44
11
48
18
22
37
39
4
41
63
28
9
16
19
59
56.67
54.00
52.33
50.67
50.67
49.00
48.33
48.00
47.33
47.33
47.00
46.33
45.33
45.33
45.00
45.00
44.67
44.67
44.33
44.00
43.33
43.33
42.33
42.00
41.00
40.67
39.33
39.33
38.67
38.00
37.33
37.33
36.33
36.00
35.00
32.67
32.33
32.00
31.67
30.67
30.67
30.00
30.00
30.00
29.67
28.67
28.33
28.33
28.33
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
Limestone Fynbos
Karroid Kalahari Bushveld
Escarpment Mountain Renosterveld
Eastern Mixed Nama Karoo
Kalahari Mountain Bushveld
Orange River Nama Karoo
Bushmanland Nama Karoo
Xeric Succulent Thicket
Great Nama Karoo
Sweet Lowveld Bushveld
Grassy Fynbos
Waterberg Moist Mountain Bushveld
Mountain Fynbos
Mopane Bushveld
Shrubby Kalahari Dune Bushveld
Upland Succulent Karoo
Lebombo Arid Mountain Bushveld
Thorny Kalahari Dune Bushveld
Mopane Shrubveld
46
23
10
27
8
5
9
31
16
25
18
15
12
19
37
13
7
2
I
21
4
9
17
8
12
5
15
7
39
35
29
36
25
3
13
30
2
1
14
54
61
33
60
56
57
23
46
3
12
21
7
6
10
24
5
2
I
27.00
27.00
26.67
25.67
25.33
24.33
23.67
23.00
23.00
22.33
21.67
21.67
18.33
16.67
16.67
16.67
14.00
2.00
1.00
Vegetation types were ranked from 1 = lowest to 68 = highest for area degraded and transformed and from 1 =
highest and 68 = lowest for protected area coverage
Table 5: Description and percentage area coverage ofland-cover
Description
Afro Mountain
Moist Cold
Eastern Thorn
Sand Plain
Subarid Thorn
Short Mistbelt
Grassland
Highveld Grassland
Bushveld
Fynbos
Bushveld
Grassland
Rank(/)
Rank (2)
Rank (3)
Rank (4)
Rank (5)
Rank (6)
46.5
69.8
34.5
78.6
39.3
0.1
0.5
0.4
0.5
30.9
5/.9
Natural land-cover
threats facing conservation priority vegetation types.
Forest plantations
Waterbodies
0.0
0.2
0.0
0.1
0.3
0.2
Dongas and sheet erosion scars
0.0
0.1
0.0
0.1
0.7
0.0
0.0
0.0
Degraded: forest and woodland
0.0
0.0
Degraded: thicket and bushland (etc)
0.0
0.0
0.0
2.3
0.6
0.0
1.3
Degraded: unimproved grassland
36.7
11.0
11.5
0.0
10.1
3.7
Degraded: shrub land and low fynbos
0.0
0.0
0.0
7.7
0.3
0.0
Cultivated:
permanent - commercial irrigated
0.6
0.0
0.0
0.1
5.2
0.1
0.0
Cultivated: permanent - commercial dryland
0.0
0.0
0.7
0.1
0.0
0.0
Cultivated: permanent - commercial sugarcane
0.0
0.0
0.1
0.0
0.0
Cultivated:
0.0
0.1
0.3
2.8
0.3
10.8
1.7
19.6
21.3
2.2
39.5
0.1
4.7
9.2
0.0
7.0
0.8
temporary - commercial irrigated
Cultivated:
temporary - commercial dryland
0.0
Cultivated:
temporary - semi-commercial
11.4
/ subsistence dryland
Urban / built-up land: residential
0.0
0.8
3.2
7.1
6.3
1.5
Urban / built-up land: residential (small holdings: woodland)
0.0
0.0
0.0
0.0
0.0
0.0
Urban / built-up land: residential (small holdings: bushland)
0.0
0.0
0.0
0.1
0.0
0.0
0.0
0.0
1.0
0.0
Urban / built-up land: residential (small holdings: shrub land)
0.0
0.0
Urban / built-up land: residential (small holdings: grassland)
0.0
0.0
0.0
0.0
0.0
0.0
Urban / built-up land: commercial
0.0
0.0
0.1
0.2
0.0
0.0
Urban / built-up land: industrial/transport
0.0
0.0
0.1
0.6
0.0
0.0
Mines & quarries
0.0
0.0
0.0
0.0
0.0
0.0
Table 5: Continued.
Laterite
Moist Upland
Coastal Bushve1d
Kalahari Plains
Fynbos
Grassland
-Grassland
Thorn Bushveld
Rank (7)
Rank (8)
Rank (9)
Rank(lO)
Natural land-cover
67.7
61.4
43.5
73.7
Forest plantations
0.1
3.9
9.3
0.0
Waterbodies
0.0
0.1
4.7
0.0
Dongas and sheet erosion scars
0.0
0.0
0.0
0.0
Degraded: forest and woodland
0.0
0.0
0.9
0.0
Degraded: thicket and bushland (etc)
0.0
0.2
7.5
18.8
Degraded: unimproved grassland
0.0
I.I
16.7
2.8
0.0
Degraded: shrubland and low fynbos
0.0
0.0
0.0
Cultivated: permanent - commercial irrigated
0.0
0.0
0.0
Cultivated: permanent - commercial dryland
0.0
0.0
0.0
0.4
Cultivated: permanent - commercial sugarcane
0.0
15.4
0.0
Description
0.0
Cultivated: temporary - commercial irrigated
0.2
0.2
1.3
0.0
0.0
Cultivated: temporary - commercial dryland
29.9
1.4
0.0
3.5
Cultivated: temporary - semi-commercial
0.0
12.7
10.2
2.7
Urban / built-up land: residential
0.3
2.0
3.1
0.7
Urban / built-up land: residential (small holdings: woodland)
0.0
0.0
0.0
0.0
Urban / built-up land: residential (small holdings: bushland)
0.0
0.0
0.9
0.0
Urban / built-up land: residential (small holdings: shrubland)
0.0
0.0
0.0
0.0
Urban / built-up land: residential (small holdings: grassland)
0.0
0.0
0.0
0.0
Urban / built-up land: commercial
0.0
0.0
0.0
Urban / built-up land: industrial/transport
0.0
0.0
0.1
0.3
Mines & quarries
0.0
0.0
0.1
/ subsistence dryland
0.0
0.1
The following tables describe the CR and ADD species that scored highest and lowest in
stepwise canonical correspondence analyses on axes I and 2 each for bird assemblage.
Associated ecological information for each species is also provided.
Table C-I: CR species that scored highest and lowest in stepwise canonical correspondence analyses on axis 1 for bird assemblages.
information is also provided.
VEGI
VEG2
Woodland
Degraded woodland
Vulnerable
Grassland
Waterbodies
Indeterminate
Coastal forest
Forest(NC)
Thicket
Uncommon
Monitoring
Waterbodies
Grassland
Prionopsscopi,(rons
Uncommon
Vulnerable
Woodland
Thicket
Forest
Urban-residential
ABUNDANCE
DISTRIBUTION
Olivetree Warbler
Hippolais olivetorum
Uncommon
Non-breeding
Pinkthroated
Macronyx ameliae
Uncommon
Batis fratrum
Locally common
Blackheaded Apalis
Apalis melanocephala
Locally common
Black Coucal
Centropus bengalensis
COMMON
low
Woodards'
Longclaw
Batis
Chestnutfronted
high
RED DATA
SPECIES
SCORE
Axis 1
All
Helmetshrike
Endemic
endemic
Bennett's Woodpecker
Campethera bennettii
Locally common
Yellowbellied Sunbird
Nectarinia
Locally common
Forest (NC)
Woodland
Yellow White-eye
Zosterops senegalensis
Rare
Coastal forest
Woodland
Greyhooded Kingfisher
Halcyon leucocephala
Uncommon
Drakensberg
Siskin
Pseudochloroptila
Common
Endemic
Sicklewinged
Chat
Cercomela sinuata
Common
Endemic
Pearlbreasted Swallow
Hirundo dimidiata
Common
Orangebreasted
Chaetops aurantius
Common
Endemic
Rockjumper
venusta
symonsi
VEG3
Indeterminate
Woodland
near-threatened
Shrubland
Grassland
Shrubland
Grassland
Degraded grassland
Degraded shrubland
Woodland
Thicket
Shrubland
Dryland agriculture
Waterbodies
Near-threatened
Grassland
Larklike Bunting
Emberiza impetuani
Common
Near-endemic
Shrubland
Grassland
Thicket
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Dryland agriculture
Uncommon
Endemic
Grassland
Common
Endemic
Grassland
Shrubland
Grassland
Dryland agriculture
Lark
Yellowbreasted
Pipit
Hemimacronyx
Francolinus
chloris
africanus
Glareola nordmanni
Locally common
Bearded Vulture
Gypaetus barbatus
Rare
Broadbilled
B1ackwinged Pratincole
Monitoring
Grassland
Summer
Pachyptila vittata
Uncommon
Greatwinged Petrel
Pterodroma macroptera
Common
Vulnerable
Ocean
Pinkthroated
Macronyx ameliae
Uncommon
Vulnerable
Grassland
Waterbodies
Lesser Gallinule
Porphyrula alieni
Locally common
Waterbodies
Woodland
Yellow White-eye
Zosterops senegalensis
Rare
Coastal forest
Woodland
Greater Frigatebird
Fregata minor
Rare
Coastal
Woodards'
Batis fratrum
Prion
Longclaw
Batis
Stierling's Barred Warbler
Calamonastes
Locally common
stierlingi
Common
VEG4
Woodland
Thickbilled
Greywing Francolin
low
Associated ecological
Ocean
Endemic
Indeterminate
Coastal forest
Woodland
Waterbodies
SCORE
Axis 1
high
COMMON
SPECIES
Bluecheeked Bee-eater
Merops persicus
Locally common
Natal Night jar
Caprimulgus
Rare
Drakensberg Siskin
Pseudochloroptila
Sicklewinged Chat
Orangebreasted
Rockjumper
ABUNDANCE
natalensis
symonsi
DISTRIBUTION
Common
Endemic
Cercomela sinuata
Common
Endemic
Chaetops aurantius
Common
Endemic
chloris
RED DATA
VEGI
VEG2
VEG3
Grassland
Woodland
Waterbodies
Vulnerable
Grassland
Waterbodies
Plantations
Near-threatened
Shrubland
Grassland
Shrubland
Grassland
Near-threatened
Yellowbreasted Pipit
Hemimacronyx
Uncommon
Endemic
Serinus albogularis
Common
Near endemic
Shrubland
Grassland
Greywing Francolin
Francolinus
Common
Endemic
Grassland
Shrubland
Bearded Vulture
Gypaetus barbatus
Rare
Cape Eagle Owl
Bubo capensis
Locally common
Black Harrier
Circus maurus
Locally common
Montagu's Harrier
Circus pygargus
Rare
Grassland
Monitoring
Endemic
Degraded shrubland
Grassland
Whitethroated Canary
africanus
Degraded grassland
VEG4
Grassland
Monitoring
Grassland
Shrubland
Near-threatened
Grassland
Shrubland
Dryland agriculture
Grassland
Dryland agriculture
Woodland
Plantations
Urban-residential
Winter
low
high
Rufousbellied Heron
Butorides rufiventris
Rare
Natal Night jar
Caprimulgus natalensis
Rare
Endemic
Grassland
Waterbodies
Blackheaded Apalis
Apalis melanocephala
Locally common
Forest (NC)
Thicket
Woodland
Waterbodies
Vulnerable
Swallowtailed Bee-eater
Merops hirundineus
Rare
Lesser Gallinule
Porphyrula alieni
Locally common
Waterbodies
Arctic Tern
Sterna paradisaea
Uncommon
Coastal
Greyrumped Swallow
Pseudhirundo
Common
Woodards' Batis
Batis fratrum
griseopyga
Locally common
Pinkthroated Longclaw
Macronyx ameliae
Uncommon
Ayres' Eagle
Hieraaetus ayresii
Rare
Drakensberg Siskin
Pseudochloroptila
Sicklewinged
Common
Endemic
Endemic
Indeterminate
Coastal forest
Vulnerable
Grassland
Near-threatened
Waterbodies
Dryland agriculture
Waterbodies
Woodland
Plantations
Shrubland
Grassland
Shrubland
Grassland
Degraded grassland
Waterbodies
Cercomela sinuata
Common
B1ackwinged Pratincole
Glareola nordmanni
Locally common
Grassland
Dryland agriculture
Rock Pipit
Anthus crenatus
Common
Endemic
Grassland
Shrubland
Greywing Francolin
Francolinus
Common
Endemic
Grassland
Shrubland
Pearlbreasted Swallow
Hirundo dimidiata
Common
Woodland
Thicket
Shrubland
Thickbilled Lark
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Dryland agriculture
Larklike Bunting
Emberiza impetuani
Common
Near-endemic
Shrubland
Grassland
Thicket
Bearded Vulture
Gypaetus barbatus
Rare
Chaetops aurantius
Common
Orangebreasted
Chat
symonsi
Woodland
Endemic
Woodland
Rockjumper
africanus
Endemic
Monitoring
Grassland
Near-threatened
Grassland
Degraded shrub land
Dryland agriculture
Waterbodies
SCORE
Axis I
Passerine
low
COMMON
SPECIES
Blackheaded Apalis
Apalis melanocephala
Stierling's Barred Warbler
Calamonastes
VEGI
VEG2
Locally common
Forest(NC)
Thicket
Common
Woodland
Zosterops senegalensis
Rare
Coastal forest
Woodland
Hypargos margaritatus
Rare
Endemic
Monitoring
Thicket
Woodland
Woodards' Batis
Batis fratrum
Locally common
Endemic
Indeterminate
Coastal forest
Greyrumped Swallow
Pseudhirundo
Pinkthroated Longclaw
Macronyx ameliae
Uncommon
Yellowbellied Sunbird
Nectarinia venusta
Locally common
Chestnutfronted
Prionops scopifrons
Uncommon
Hippolais olivetorum
Uncommon
Cercomela sinuata
Common
Endemic
Common
Endemic
Helmetshrike
Sicklewinged
Chat
griseopyga
Vulnerable
Vulnerable
Non-breeding
endemic
Waterbodies
Grassland
Waterbodies
Forest(NC)
Woodland
Urban-residential
Woodland
Thicket
Forest
Woodland
Degraded woodland
Shrubland
Grassland
Shrubland
Grassland
Pearlbreasted Swallow
Hirundo dimidiata
Common
Woodland
Larklike Bunting
Emberiza impetuani
Common
Near-endemic
Shrubland
ThickbilIed Lark
Galerida magnirostris
Common
Endemic
Shrubland
Chaetops aurantius
Common
Endemic
Hemimacronyx
Uncommon
Endemic
Rockjumper
chloris
Near-threatened
Near-threatened
Dryland agriculture
Degraded shrubland
Thicket
Shrubland
Dryland agriculture
Grassland
Thicket
Waterbodies
Grassland
Dryland agriculture
Grassland
Grassland
Serinus albogularis
Common
Near endemic
Shrubland
Redeyed Bulbul
Pycnonotus nigricans
Common
Near endemic
Grassland
Shrubland
Yellow Canary
Serinus jlaviventris
Common
Near endemic
Shrubland
Grassland
Black Coucal
Centropus bengalensis
Uncommon
Monitoring
Waterbodies
Grassland
Roseate Tern
Sterna dougallii
Locally common
Endangered
Coastal
Grassland
Sabine's Gull
Larus sabini
Rare
Coastal
Bluecheeked Bee-eater
Merops persicus
Locally common
Grassland
Woodland
Sand Plover
Charadrius leschenaultii
Locally common
Coastal
Waterbodies
Greater Frigatebird
Fregata minor
Rare
Coastal
Slenderbilled Prion
Pachyptila belcheri
Uncommon
Natal Night jar
Caprimulgus
Rare
Vulnerable
Grassland
Greatwinged Petrel
Pterodroma macroptera
Common
Vulnerable
Ocean
Curlew
Numenius arquata
Common
Greywing Francolin
Francolinus
Common
africanus
VEG4
Degraded grassland
Whitethroated Canary
natalensis
VEG3
Woodland
Pseudochloroptila
Pipit
symonsi
Common
Drakensberg Siskin
Yellowbreasted
high
RED DATA
Pinkthroated Twinspot
Orangebreasted
Nonpasserine
low
stierlingi
DISTRIBUTION
Yellow White-eye
Olivetree Warbler
high
ABUNDANCE
Woodland
Waterbodies
Ocean
Waterbodies
Coastal
Endemic
Grassland
248
Shrubland
Plantations
Urban-residential
SCORE
Axis 1
SPECIES
COMMON
ABUNDANCE
Blackwinged Pratincole
Glareola nordmanni
Locally common
Bearded Vulture
Gypaetus barbatus
Rare
DISTRIBUTION
RED DATA
VEGI
VEG2
VEG3
Grassland
Dryland agriculture
Waterbodies
Monitoring
Grassland
Grassland
Dryland agriculture
Woodland
Shrubland
Woodland
Montagu's Harrier
Circus pygargus
Rare
Pale Chanting Goshawk
Melierax canorus
Common
Ground Woodpecker
Geocolaptes olivaceus
Common
Endemic
Near-threatened
Grassland
Shrubland
Black Harrier
Circus maurus
Locally common
Endemic
Near-threatened
Grassland
Shrubland
Cape Eagle Owl
Bubo capensis
Locally common
Monitoring
Grassland
Shrubland
Urban-residential
Redbreasted Sparrowhawk
Accipiter rufiventris
Locally common
Forest
Plantations
Grassland
Pennantwinged
Macrodipteryx
Locally common
Night jar
vexillaria
Near endemic
Indeterminate
Woodland
Forest (NC)
Thicket
Vulnerable
Grassland
Waterbodies
VEG4
Dryland agriculture
Breeding
low
high
Blackheaded Apalis
Apalis melanocephala
Locally common
Pinkthroated Longclaw
Macronyx ameliae
Uncommon
Woodards' Batis
Batis fratrum
Locally common
Yellow White-eye
Zosterops senegalensis
Rare
Coastal forest
Stierling's Barred Warbler
Calamonastes
Common
Woodland
Indeterminate
Pinkthroated Twinspot
Hypargos margaritatus
Rare
Greyrumped Swallow
Pseudhirundo
Common
Southern Banded Snake Eagle
Circeatus fasciolatus
Rare
Near-threatened
Monitoring
griseopyga
Endemic
Monitoring
Coastal forest
Thicket
Waterbodies
Dryland agriculture
Woodland
Thicket
Waterbodies
Grassland
Centropus bengalensis
Uncommon
Campethera bennettii
Locally common
Woodland
Pearlbreasted Swallow
Hirundo dimidiata
Common
Woodland
Chat
Cercomela sinuata
Common
Endemic
Common
Endemic
Melierax canorus
Common
Near endemic
Chaetops aurantius
Common
Endemic
Redeyed Bulbul
Pycnonotus nigricans
Common
Near endemic
Grassland
Yellowbreasted
Hemimacronyx
Uncommon
Endemic
Grassland
Pseudochloroptila
Pale Chanting Goshawk
Orangebreasted
Rockjumper
Pipit
symonsi
chloris
Woodland
Coastal forest
Black Coucal
Drakensberg Siskin
Woodland
Woodland
Bennett's Woodpecker
Sicklewinged
Nonbreeding
low
stierlingi
Endemic
Thicket
Shrubland
Dryland agriculture
Degraded grassland
Degraded shrub land
Shrubland
Woodland
Urban-residential
Waterbodies
Shrubland
Grassland
Near-threatened
Shrubland
Grassland
Shrubland
Woodland
Near-threatened
Grassland
Larklike Bunting
Emberiza impetuani
Common
Near-endemic
Shrubland
Grassland
Thicket
Thickbilled Lark
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Dryland agriculture
Greywing Francolin
Francolinus
africanus
Common
Endemic
Grassland
Shrubland
Cory's Shearwater
Calonectris diomedea
Common
Ocean
Fleshfooted Shearwater
Puffinus carneipes
Common
Ocean
249
SCORE
Axis I
high
ABUNDANCE
DISTRIBUTION
RED DATA
VEG2
VEG3
VEG4
Grassland
Dryland agriculture
Thicket
Urban-residential
Grassland
Dryland agriculture
Waterbodies
Woodland
VEGI
COMMON
SPECIES
Sooty Shearwater
Puffinus griseus
Common
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
Whitebellied Storm Petrel
Fregetta grallaria
Rare
Ocean
Blackbellied Storm Petrel
Fregetta tropica
Common
Ocean
Collared Flycatcher
Ficedula albicollis
Rare
Woodland
Bridled Tern
Sterna anaethetus
Rare
Coastal
European Storm Petrel
Hydrobates pelagicus
Common
Coastal
Slenderbilled Prion
Pachyptila belcheri
Uncommon
Ocean
African Hobby Falcon
Falco cuvierii
Uncommon
Pennantwinged
Macrodipteryx
Night jar
Black Tern
Lesser Kestrel
Blackwinged
Pratincole
vexil/aria
Ocean
Woodland
Locally common
Chlidonias niger
Rare
Falco naumanni
Common
Glareola nordmanni
Locally common
Indeterminate
Woodland
Coastal
Monitoring
Pectoral Sandpiper
Calidris melanotos
Rare
Coastal
Montagu's Harrier
Circus pygargus
Rare
Grassland
Dryland agriculture
Falco vespertinus
Uncommon
Grassland
Woodland
Western
Redfooted
Abdim's
Stork
Ciconia abdimii
Common
Grassland
Dryland agriculture
Pasture
Redfooted
Falco amurensis
Common
Monitoring
Grassland
Dryland agriculture
Urban-residential
Greatwinged Petrel
Pterodroma macroptera
Common
Vulnerable
Ocean
Greyrumped
Pseudhirundo
Common
Woodland
Waterbodies
Dryland agriculture
Ocean
Eastern
Kestrel
Kestrel
Human
low
high
Swallow
griseopyga
Pintado Petrel
Daption capense
Common
Ayres' Eagle
Hieraaetus ayresii
Rare
Caspian Tern
Hydroprogne
Common
Yellowbellied Sunbird
Nectarinia
Olivetree Warbler
Hippolais olivetorum
caspia
venusta
Woodland
Rare
Locally common
Uncommon
Non-breeding
endemic
Waterbodies
Forest (AlC)
Woodland
Woodland
Degraded woodland
Whitewinged Tern
Childonias leucopterus
Common
Waterbodies
Southern Giant Petrel
Macronectes
Common
Ocean
Lesser Flamingo
Phoeniconaias
Sicklewinged
Cercomela sinuata
Common
Pearlbreasted Swallow
Hirundo dimidiata
Common
Larklike
Emberiza impetuani
Common
Galerida magnirostris
Common
Chat
Bunting
Thickbilled
Lark
giganteus
minor
Locally common
Near-threatened
Plantations
Coastal
Urban-residential
Waterbodies
Shrubland
Grassland
Degraded grassland
Woodland
Thicket
Shrubland
Dryland agriculture
Near-endemic
Shrubland
Grassland
Thicket
Waterbodies
Endemic
Shrubland
Grassland
Dryland agriculture
Endemic
Degraded shrubland
SCORE
Axis 1
Nonhuman
low
high
COMMON
SPECIES
ABUNDANCE
DISTRIBUTION
Redeyed Bulbul
Pycnonotus nigricans
Common
Near endemic
Montagu's Harrier
Circus pygargus
Rare
Black Harrier
Circus maurus
Locally common
Cape Eagle Owl
Bubo capensis
Locally common
Yellowrurnped Widow
Euplectes capensis
Common
South African She1duck
Tadorna cana
Locally common
Black Coucal
Centropus bengalensis
Uncommon
Blackheaded Apalis
Apalis melanocephala
Locally common
Pinkthroated Longclaw
Macronyx ameliae
Uncommon
Woodards' Batis
Batis fratrum
Locally common
Endemic
RED DATA
VEGI
VEG2
VEG3
VEG4
Grassland
Shrubland
Woodland
Urban-residential
Grassland
Dryland agriculture
Woodland
Near-threatened
Grassland
Shrubland
Dryland agriculture
Monitoring
Grassland
Shrubland
Urban-residential
Grassland
Shrubland
Dryland agriculture
Waterbodies
Shrub land
Grassland
Endemic
Monitoring
Endemic
Waterbodies
Grassland
Forest (AlC)
Thicket
Vulnerable
Grassland
Waterbodies
Indeterminate
Coastal forest
Bluecheeked Bee-eater
Merops persicus
Locally common
Grassland
Woodland
Broadbilled Roller
Eurystomus glaucurus
Locally common
Woodland
Thicket
Yellow White-eye
Zosterops senegalensis
Rare
Coastal forest
Woodland
Sand Plover
Charadrius leschenaultii
Locally common
Coastal
Waterbodies
Greater Frigatebird
Fregata minor
Rare
Coastal
Southern Banded Snake Eagle
Circeatus fasciolatus
Rare
Drakensberg Siskin
Pseudochloroptila
Common
Chaetops aurantius
Hemimacronyx
Orangebreasted
Yellowbreasted
Rockjumper
Pipit
symonsi
chloris
Near-threatened
Coastal forest
Woodland
Endemic
Near-threatened
Shrubland
Grassland
Common
Endemic
Near-threatened
Grassland
Uncommon
Endemic
Blackwinged Pratincole
Glareola nordmanni
Locally common
Greywing Francolin
Francolinus
Common
africanus
Endemic
Grassland
Dryland agriculture
Grassland
Shrubland
Gypaetus barbatus
Rare
Pale Chanting Goshawk
Melierax canorus
Common
Near endemic
Shrubland
Woodland
Whitethroated
Serinus albogularis
Common
Near endemic
Shrubland
Grassland
Serinus flaviventris
Common
Near endemic
Shrubland
Grassland
Endemic
Grassland
Shrubland
Yellow Canary
Ground Woodpecker
Geocolaptes olivaceus
Common
Thicket
Grassland
Bearded Vulture
Canary
Waterbodies
Monitoring
Near-threatened
Grassland
Waterbodies
Table B-2: CR species that scored highest and lowest in stepwise canonical correspondence analyses on axis 2 for bird assemblages.
information is also provided.
SCORE
Axis 2
All
COMMON
low
Redbilled
high
Buffalo Weaver
VEGI
VEG2
Nomadic
Woodland
Thicket
Woodland
SPECIES
ABUNDANCE
Bubalornis niger
DISTRIBUTION
RED DATA
Dickinson's Kestrel
Falco dickinsoni
Rare
Carmine Bee-eater
Merops nubicoides
Common
Thickbilled
Cuckoo
Pachycoccyx
Rare
Monitoring
Thicket
Rackettailed
Roller
Coracias spatulata
Uncommon
Vulnerable
Woodland
Monitoring
Woodland
Thicket
Woodland
Thicket
Uncommon
Woodland
Urban-residential
Coastal
audeberti
Purple Widowfinch
Vidua purpurascens
Locally common
Whitecrowned
Eurocephalus
Common
Shrike
anguitimens
Woodland
Endemic
Dusky Lark
Pinarocorys
Black Tern
Chlidonias niger
Rare
Greyhooded Kingfisher
Halcyon leucocephala
Uncommon
Fleshfooted Shealwaler
Puffinus carneipes
Common
Ocean
Cory's Shearwater
Calonectris diomedea
Common
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Sterna dougallii
Locally common
Coastal
Coastal
Roseate Tern
nigricans
Indeterminate
Woodland
Woodland
Ocean
Endangered
Coastal
Sabine's Gull
Larus sabini
Rare
European Storm Petrel
Hydrobates pelagicus
Common
Whitebellied Storm Petrel
FregeUa grallaria
Rare
European Starling
Sturn us vulgaris
Common
Bridled Tern
Sterna anaethetus
Rare
Coastal
Blackbellied Storm Petrel
FregeUa tropica
Common
Ocean
Rock Pipit
Anthus crenatus
Common
Purple Widowfinch
Vidua purpurascens
Locally common
Pearls potted Owl
Glaucidium perlatum
Common
Pachycoccyx
Rare
Uncommon
Ocean
Alien
Urban-residential
Dryland agriculture
Summer
low
Thickbilled
Cuckoo
audeberti
Rackettailed Roller
Coracias spatulata
Dickinson's Kestrel
Falco dickinsoni
Rare
Dark Chanting Goshawk
Melierax metabates
Locally common
Greyhooded Kingfisher
Halcyon leucocephala
Uncommon
Endernic
Grassland
Shrubland
Woodland
Thicket
Woodland
Thicket
Monitoring
Thicket
Woodland
Vulnerable
Woodland
Monitoring
Woodland
Woodland
Indeterminate
Woodland
Thicket
Associated ecological
VEG3
VEG4
SCORE
Axis 2
high
SPECIES
COMMON
ABUNDANCE
DISTRIBUTION
RED DATA
VEGI
Monitoring
VEG2
VEG3
Woodland
Dryland agriculture
Urban-residential
Woodland
Thicket
Waterbodies
Steelblue Widowfinch
Vidua chalybeata
Common
Grey Hombill
Tockus nasutus
Common
Grey Wagtail
Motacilla cinerea
Rare
Forest
Fleshfooted Shearwater
Puffinus carneipes
Common
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
Broadbilled Prion
Pachyptila vittata
Uncommon
Ocean
European Storm Petrel
Hydrobates pelagicus
Common
Coastal
Ocean
Pterodroma mollis
Common
Sooty Shearwater
Puffinus griseus
Common
Ocean
Wandering Albatross
Diomedea exulans
Uncommon
Ocean
Softplumaged
Petrel
Roseate Tern
Sterna dougallii
Locally common
Cory's Shearwater
Calonectris diomedea
Common
Redbilled Buffalo Weaver
Bubalornis niger
Nomadic
Purple Widowfinch
Vidua purpurascens
Locally common
Whitecrowned
Eurocephalus
Common
Endangered
Coastal
Ocean
Winter
low
high
Shrike
anguitimens
Woodland
Monitoring
Endemic
Thicket
Woodland
Thicket
Urban-residential
Dusky Lark
Pinarocorys nigricans
Uncommon
Woodland
Carmine Bee-eater
Merops nubicoides
Common
Woodland
Greyhooded Kingfisher
Halcyon leucocephala
Uncommon
Dickinson's Kestrel
Falco dickinsoni
Rare
Redbilled Helmetshrike
Prionops retzii
Common
Purple Roller
Coracias naevia
Common
Redcrested Korhaan
Eupodotis ruficrista
Common
Cory's Shearwater
Calonectris diomedea
Common
Indeterminate
Thicket
Woodland
Woodland
Woodland
Threatened
Near endemic
Monitoring
Woodland
Woodland
Thicket
Grassland
Woodland
Ocean
Roseate Tern
Sterna dougallii
Locally common
Sabine's Gull
Larus sabini
Rare
Southern Giant Petrel
Macronectes giganteus
Common
Jackass Penguin
Spheniscus demersus
Locally common
Broadbilled Sandpiper
Limicola falcinellus
Rare
Coastal
Curlew
Numenius arquata
Common
Coastal
Sand Plover
Charadrius leschenaultii
Locally common
Coastal
Subantartic Skua
Catharacta antartica
Common
Coastal
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
Endangered
Coastal
Coastal
Ocean
Endemic
Near-threatened
Coastal
Waterbodies
Shrubland
VEG4
SCORE
Axis 2
Passerine
low
high
SPECIES
COMMON
RED DATA
VEGI
VEG2
VEG3
Redbilled Buffalo Weaver
Bubalornis
niger
Nomadic
Woodland
Thicket
Nectarinia
venusta
Locally common
Forest (NC)
Woodland
Urban-residential
Chestnutfronted
Prionops scopifrons
Uncommon
Vulnerable
Woodland
Thicket
Forest
Dusky Lark
Pinarocorys
Uncommon
Woodland
Urban-residential
Purple Widowfinch
Vidua purpurascens
Locally common
Monitoring
Woodland
Thicket
Whitecrowned
Eurocephalus
Common
Woodland
Thicket
Helmetshrike
Shrike
nigricans
anguitimens
Endemic
Redbilled Helmetshrike
Prionops retzii
Common
Threatened
Steelblue Widowfinch
Vidua chalybeata
Common
Monitoring
Yellowbellied Eremomela
Eremomela icteropygialis
Common
Marico Sunbird
Nectarinia mariquensis
Common
Woodland
Collared Flycatcher
Ficedula albicollis
Rare
Woodland
European Starling
Sturn us vulgaris
Common
Urban-residential
Dryland agriculture
Thrush Nightingale
Luscinia luscinia
Uncommon
Thicket
Urban-residential
Alien
Woodland
Woodland
Dryland agriculture
Shrubland
Woodland
Urban-residential
Whinchat
Saxicola rubetra
Rare
Grassland
Woodland
Thicket
Tree Pipit
Anthus trivialis
Rare
Woodland
Forest
Plantations
Pied Mannikin
Spermestes jringi/loides
Rare
Indeterminate
Coastal forest
Urban-residential
Dryland agriculture
Spotted Thrush
Zoothero gut/ata
Uncommon
Endangered
Coastal forest
House Crow
Corvus splendens
Locally common
Alien
Urban-residential
Serinus albogularis
Common
Near endemic
Shrubland
Grassland
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Black Tern
Chlidonias niger
Rare
Thickbilled Cuckoo
Pachycoccyx
Rare
Canary
Thickbilled Lark
high
DISTRIBUTION
Yellowbellied Sunbird
Whitethroated
Nonpasserine
low
ABUNDANCE
audeberti
Coastal
Monitoring
Thicket
Woodland
Dickinson's Kestrel
Falco dickinsoni
Rare
Carmine Bee-eater
Merops nubicoides
Common
Rackettailed Roller
Coracias spatulata
Uncommon
Vulnerable
Woodland
Greyhooded Kingfisher
Halcyon leucocephala
Uncommon
Indeterminate
Woodland
Pearlspotted
Glaucidium perlatum
Common
Woodland
Thicket
Purple Roller
Coracias naevia
Common
Woodland
Thicket
Redbilled Hornbill
Tockus erythrorhynchus
Common
Woodland
Thicket
Grey Hornbill
Tockus nasutus
Common
Woodland
Thicket
Fleshfooted Shearwater
Puffinus carneipes
Common
Ocean
Owl
Woodland
Woodland
254
Dryland agriculture
VEG4
SCORE
Axis 2
COMMON
SPECIES
ABUNDANCE
Cory's Shearwater
Calonectris diomedea
Common
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
European Storm Petrel
Hydrobates pelagicus
Common
Coastal
Sabine's GulI
Larus sabini
Rare
Coastal
Roseate Tem
Sterna dougallii
LocalIy common
Blackbellied Storm Petrel
Fregetta tropica
Common
Whitebellied
Fregetta gral/aria
Rare
Ocean
Bridled Tern
Sterna anaethetus
Rare
Coastal
Sooty Shearwater
Puffinus griseus
Common
Ocean
Redbilled Buffalo Weaver
Bubalornis niger
Nomadic
Purple Widowfinch
Vidua purpurascens
LocalIy common
Eurocephalus
Common
Storm Petrel
DISTRIBUTION
RED DATA
Endangered
VEGI
VEG2
VEG3
Coastal
Ocean
Breeding
low
Whitecrowned
high
Shrike
Endemic
Rare
Dickinson's Kestrel
Falco dickinsoni
Thickbilled Cuckoo
Pachycoccyx
Carmine Bee-eater
Merops nubicoides
Common
audeberti
Woodland
Thicket
Woodland
Thicket
Woodland
Thicket
Woodland
Rare
Monitoring
Thicket
Woodland
Woodland
Rackettailed Roller
Coracias spatulata
Uncommon
Vulnerable
Greyhooded Kingfisher
Halcyon leucocephala
Uncommon
Indeterminate
Woodland
Steelblue Widowfinch
Vidua chalybeata
Common
Monitoring
Woodland
Dryland agriculture
Purple Roller
Coracias naevia
Common
Woodland
Thicket
RoseateTem
Sterna dougallii
Locally common
African Black Oystercatcher
Haematopus
Common
European Starling
Sturn us vulgaris
moquini
Common
Endangered
Alien
Spheniscus demersus
Locally common
Endemic
Morus capensis
Common
Breeding endemic
Plover
Near-threatened
Charadrius pal/idus
Uncommon
Phalacrocorax
Common
Breeding endemic
Endemic
capensis
Coastal
Urban-residential
Cape Gannet
Cape Cormorant
Woodland
Coastal
Jackass Penguin
Chestnutbanded
Nonbreeding
low
anguitimens
Monitoring
Coastal
Coastal
Monitoring
Coastal
Ocean
Green Barbet
Stactolaema olivacea
Locally common
KelpGulI
Larus dominicanus
Common
Coastal
Swift Tern
Sterna bergi
Common
Coastal
Vulnerable
Forest (AlC)
BlackTem
Chlidonias niger
Rare
Coastal
Dusky Lark
Pinarocorys nigricans
Uncommon
Woodland
255
Dryland agriculture
Urban-residential
Urban-residential
VEG4
SCORE
Axis 2
COMMON
Steppe Eagle
Aquila nipalensis
Uncommon
Lesser Spotted Eagle
Aquila pomarina
Uncommon
DISTRIBUTION
RED DATA
VEGI
Monitoring
Woodland
VEG2
VEG3
Tringa ochropus
Rare
Waterbodies
European
Merops apiaster
Common
Thicket
Woodland
Shrubland
Tree Pipit
Anthus trivialis
Rare
Woodland
Forest
Plantations
Grey Wagtail
Motacilla cinerea
Rare
Forest
Waterbodies
Olivetree
Hippolais olivetorum
Uncommon
Woodland
Degraded woodland
Oriolus oriolus
Uncommon
Woodland
Urban-residential
Plantations
Blackwinged Pratincole
Glareola nordmanni
Locally common
Grassland
Dryland agriculture
Waterbodies
Montagu's Harrier
Circus pygargus
Rare
Grassland
Dryland agriculture
Woodland
Lesser Kestrel
Falco naumanni
Common
Monitoring
Grassland
Dryland agriculture
Thicket
Pallid Harrier
Circus macrourus
Rare
Near-threatened
Grassland
Dryland agriculture
Woodland
Western Redfooted Kestrel
Falco vespertinus
Uncommon
Grassland
Woodland
Pennantwinged
Macrodipteryx
Locally common
Indeterminate
Woodland
Threatened
Grassland
Bee-eater
Warbler
Night jar
vexillaria
Comcrake
Crexcrex
Uncommon
African Hobby Falcon
F aleo cuvierii
Uncommon
Eastern Redfooted Kestrel
Falco amurensis
Common
Ruff
Philomachus pugnax
Common
Steelblue Widowfinch
Vidua chalybeata
Common
Striped Cuckoo
Clamator levaillantii
Locally common
Yellowbellied Sunbird
Nectarinia
Chestnutbacked
Eremopterix
Non-breeding endemic
Waterbodies
Dryland agriculture
Urban-residential
Woodland
Grassland
Dryland agriculture
Waterbodies
Grassland
Woodland
Dryland agriculture
Urban-residential
Woodland
Thicket
Urban-residential
Locally common
Forest(NC)
Woodland
Urban-residential
Common
Grassland
Woodland
Dryland agriculture
Common
Woodland
Thicket
Urban-residential
Thicket
Woodland
Urban-residential
Thicket
Woodland
Shrubland
Monitoring
Human
low
high
Finchlark
venusta
leucotis
Monitoring
Grey Lourie
Corythaixoides
Heuglin's
Cossypha heuglini
Common
Lesser Grey Shrike
Lanius minor
Common
Non-breeding
Sabota Lark
Mirafra sabota
Common
Near endemic
Woodland
Shrubland
Degraded woodland
Pinkbilled Lark
Spizocorys conirostris
Common
Near endemic
Grassland
Degraded grassland
Dryland agriculture
Bushveld Pipit
Anthus caffer
Common
Woodland
Degraded woodland
Fleshfooted
Puffinus carneipes
Common
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
European Storm Petrel
Hydrobates pelagicus
Common
Coastal
European Starling
Sturn us vulgaris
Common
Robin
Shearwater
concolor
VEG4
Woodland
Green Sandpiper
European Golden Oriole
high
ABUNDANCE
SPECIES
Alien
endemic
Urban-residential
Dryland agriculture
Urban-residential
Pasture
SCORE
Axis 2
COMMON
SPECIES
Subantartic
Skua
RED DATA
VEGI
VEG2
Catharacta antartica
Common
Coastal
Diomedea cauta
Common
Ocean
Thrush Nightingale
Luscinia luscinia
Uncommon
Thicket
Urban-residential
Pearlbreasted Swallow
Hirundo dimidiata
Common
Woodland
Thicket
Southern
Macronectes
giganteus
Common
Ocean
Diomedea melanophris
Common
Ocean
Giant Petrel
RedbiIled
Albatross
Bubalornis niger
Nomadic
Purple Widowflnch
Buffalo Weaver
Vidua purpurascens
Locally common
Dickinson's Kestrel
Falco dickinsoni
Rare
Thickbilled
Pachycoccyx
Rare
Cuckoo
Carmine Bee-eater
high
DISTRIBUTION
Shy Albatross
Blackbrowed
Nonhuman
low
ABUNDANCE
audeberti
Merops nubicoides
Rackettailed Roller
Coracias spatulata
Uncommon
Eurocephalus
Common
anguitimens
Woodland
Thicket
Woodland
Thicket
Woodland
Monitoring
Common
Whitecrowned
Shrike
Monitoring
Thicket
Vulnerable
Endemic
Woodland
Woodland
Black Tern
Chlidonias niger
Rare
Dusky Lark
Pinarocorys
nigricans
Uncommon
Greyhooded Kingfisher
Halcyon leucocephala
Uncommon
Cory's Shearwater
Calonectris diomedea
Common
Roseate Tern
Sterna dougallii
Locally common
Sabine's Gull
Larus sabini
Rare
Coastal
Bridled Tern
Sterna anaethetus
Rare
Coastal
Collared Flycatcher
Ficedula albicollis
Rare
Woodland
Blackbellied Storm Petrel
Fregetta tropica
Common
Ocean
Whitebellied
Fregetta grallaria
Rare
Ocean
Sooty Shearwater
Puffin us griseus
Common
Ocean
Broadbilled
Sandpiper
Limicola falcinellus
Rare
Coastal
Petrel
Pterodroma mollis
Common
Ocean
Softplumaged
Storm Petrel
Woodland
Woodland
Thicket
Coastal
Woodland
Indeterminate
Woodland
Endangered
Coastal
Ocean
Urban-residential
VEG3
VEG4
Shrubland
Dryland agriculture
Table B-3: ADD species that scored highest and lowest in stepwise canonical correspondence analyses on axis I for bird assemblages.
information is also provided.
SCORE
Axis 1
All
COMMON
SPECIES
low
Grey Wagtail
DISTRIBUTION
RED DATA
VEGI
VEG2
Forest
Waterbodies
Motacilla cinerea
Rare
Rufousbellied
Heron
Butorides rufiventris
Rare
Pinkthroated
Longclaw
Macronyx ameliae
Uncommon
Vulnerable
Grassland
Waterbodies
Centropus bengalensis
Uncommon
Rare
Monitoring
Grassland
Vulnerable
Waterbodies
Grassland
Indeterminate
Coastal forest
Black Coucal
high
ABUNDANCE
Endemic
Caprimulgus
Woodards'
Batis fratrum
Locally common
Crab Plover
Dromas ardeola
Rare
Broadbilled Roller
Eurystomus glaucurus
Locally common
Woodland
Gullbilled Tern
Gelochelidon
Rare
Coastal
Broadbilled Sandpiper
Limicola falcinellus
Rare
Pale Chanting Goshawk
Melierax canorus
Common
Near endemic
Shrubland
Woodland
Southern Grey Tit
Parusafer
Common
Endemic
Shrubland
Grassland
Mountain Pipit
Anthus hoeschi
Common
Breeding endemic
Grassland
Shrubland
Thickbilled
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Serinus flaviventris
Common
Near endemic
Shrubland
Grassland
Grassland
Lark
Yellow Canary
Sicklewinged
Chat
natalensis
nilotica
Endemic
Waterbodies
Plantations
Thicket
Coastal
Cercomela sinuata
Common
Endemic
Shrubland
Parisoma layardi
Common
Endemic
Shrubland
Drakensberg
Pseudochloroptila
symonsi
Common
Endemic
Near-threatened
Shrubland
Botha's Lark
Spizocorys fringillaris
Uncommon
Endemic
Indeterminate
Grassland
Degraded grassland
Blackchested Prinia
Prinia flavicans
Common
Near endemic
Shrubland
Thicket
Pachyptila
Uncommon
Dryland agriculture
Degraded grassland
Grassland
Urban-residential
Summer
low
high
Broadbilled
Prion
vittata
Ocean
Rufousbellied Heron
Butorides rufiventris
Rare
Pinkthroated
Macronyx ameliae
Uncommon
Vulnerable
Grassland
Waterbodies
Black Coucal
Centropus bengalensis
Uncommon
Monitoring
Waterbodies
Grassland
Natal Night jar
Caprimulgus natalensis
Rare
Vulnerable
Grassland
Waterbodies
Woodards'
Batis fratrum
Locally common
Indeterminate
Coastal forest
Longclaw
Batis
VEG4
Coastal
Layard's Titbabbler
Siskin
VEG3
Waterbodies
Natal Night jar
Oatis
Associated ecological
Endemic
Endemic
Waterbodies
Broadbilled Roller
Eurystomus glaucurus
Locally common
Woodland
Crab Plover
Dromas ardeola
Rare
Coastal
Yellow White-eye
Zosterops senegalensis
Rare
Southern Banded Snake Eagle
Circeatus fasciolatus
Rare
Thickbilled Lark
Galerida magnirostris
Common
Mountain Pipit
Anthus hoeschi
Common
Thicket
Coastal forest
Woodland
Coastal forest
Woodland
Endemic
Shrubland
Grassland
Breeding endemic
Grassland
Shrubland
Near-threatened
Plantations
Thicket
Dryland agriculture
Degraded shrub land
SCORE
Axis 1
SPECIES
ABUNDANCE
DISTRIBUTION
Serinus jlaviventris
Common
Cercomela sinuata
Common
Layard's Titbabbler
Parisoma layardi
Southern Grey Tit
Parus afer
Drakensberg
Pseudochloroptila
COMMON
Yellow Canary
Sicklewinged
Chat
Siskin
symonsi
RED DATA
VEGI
VEG2
Near endemic
Shrubland
Grassland
Endemic
Shrubland
Grassland
Common
Endemic
Shrubland
Common
Endemic
Common
Endemic
Shrubland
Grassland
Near-threatened
Shrubland
Grassland
Woodland
Indeterminate
Shrubland
Grassland
Shrubland
Thicket
VEG3
VEG4
Degraded grassland
Degraded shrubland
Pale Chanting Goshawk
Melierax canorus
Common
Near endemic
Botha's Lark
Spizocorys fringillaris
Uncommon
Endemic
Prinia jlavicans
Common
Near endemic
Motacilla cinerea
Rare
Forest
Waterbodies
Macronyx ameliae
Uncommon
Vulnerable
Grassland
Waterbodies
Rare
Endangered
Rare
Grassland
Waterbodies
Shrubland
Uncommon
Vulnerable
Waterbodies
Grassland
Waterbodies
Plantations
Degraded grassland
Degraded woodland
Dryland agriculture
Degraded shrub land
Blackchested
Prinia
Degraded grassland
Urban-residential
Winter
low
Grey Wagtail
Pinkthroated
Longclaw
Blackrumped
Buttonquail
high
Turnix hottentotta
Ephippiorhynchus
Saddlebilled Stork
senegalensis
Natal Night jar
Caprimulgus natalensis
Rare
European Storm Petrel
Hydrobates pelagicus
Common
Temminck's Courser
Cursorius temminckii
Locally common
Woodards'
Batis fratrum
Locally common
Batis
Coastal
Grassland
Endemic
Indeterminate
Coastal forest
Threatened
Woodland
Redbilled Helmetshrike
Prionops retzii
Common
Knot
Calidris canutus
Locally common
Waterbodies
Coastal
Pale Chanting Goshawk
Melierax canorus
Common
Near endemic
Shrubland
Woodland
Layard's Titbabbler
Parisoma layardi
Common
Endemic
Shrubland
Southern Grey Tit
Parusafer
Common
Endemic
Shrubland
Grassland
Sicklewinged
Cercomela sinuata
Common
Endemic
Shrubland
Grassland
Degraded grassland
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Dryland agriculture
Pseudochloroptila
Thickbilled
Chat
Lark
Common
Endemic
Shrubland
Grassland
Yellow Canary
Serinus jlaviventris
Common
Near endemic
Shrubland
Grassland
Mountain Pipit
Anthus hoeschi
Common
Breeding endemic
Grassland
Shrubland
Botha's Lark
Spizocorys fringillaris
Uncommon
Endemic
Blacknecked Grebe
Podiceps nigricollis
Locally common
Grey Wagtail
Motacilla cinerea
Rare
Redbilled Helmetshrike
Prionops retzjj
Threatened
Vulnerable
Grassland
Indeterminate
Coastal forest
Drakensberg
Siskin
symonsi
Near-threatened
Indeterminate
Grassland
Degraded grassland
Waterbodies
Grassland
Forest
Waterbodies
Passerine
low
Whitebreasted
Cuckooshrike
Coracina pectoralis
Common
Rare
Pinkthroated
Longclaw
Macronyx ameliae
Uncommon
Batis fratrum
Locally common
Woodards'
Batis
Stierling's
Barred
Yellow White-eye
Warbler
Calamonastes
stierlingi
Zosterops senegalensis
Common
Rare
Woodland
Woodland
Endemic
Waterbodies
Woodland
Coastal forest
Woodland
Shrubland
SCORE
Axis 1
COMMON
SPECIES
VEGI
VEG2
Coastal forest
Nectarinia
Rare
Endemic
Near-threatened
Thicket
Hypargos margaritatus
Rare
Endemic
Monitoring
Thicket
Woodland
Serinus citrinipectus
Rare
Monitoring
Woodland
Grassland
Southern Grey Tit
Parus afer
Common
Endemic
Shrubland
Grassland
Mountain Pipit
Anthus hoeschi
Common
Breeding endemic
Grassland
Shrubland
Cercomela sinuata
Common
Endemic
Shrubland
Grassland
Serinus flaviventris
Common
Near endemic
Shrubland
Grassland
Grassland
Twinspot
Canary
Chat
Yellow Canary
neergaardi
Galerida magnirostris
Common
Endemic
Shrubland
Layard's Titbabbler
Parisoma layardi
Common
Endemic
Shrubland
Drakensberg
Pseudochloroptila
Thickbilled
Lark
Siskin
Redeyed Bulbul
Orangebreasted
Rockjumper
Botha's Lark
Common
Endemic
Pycnonotus nigricans
symonsi
Common
Near endemic
Near-threatened
Chaetops aurantius
Common
Endemic
Near-threatened
Grassland
Spizocorys fringillaris
Uncommon
Endemic
Indeterminate
Grassland
Shrubland
Grassland
Grassland
Shrubland
Gullbilled Tern
Gelochelidon
Rare
Coastal
Limicola falcinellus
Rare
Coastal
Crab Plover
Dromas ardeola
Rare
Black Coucal
Centropus bengalensis
Uncommon
Monitoring
Waterbodies
Grassland
Natal Night jar
Caprimulgus
Rare
Vulnerable
Grassland
Waterbodies
Mongolian Plover
Charadrius mongolus
Rufousbellied
nilotica
natalensis
Butorides rufiventris
Pachyptila vittata
Uncommon
Greater
Fregata minor
Rare
Stactolaema olivacea
Locally common
Endemic
Melierax canorus
Common
Near endemic
Whitewinged Black Korhaan
Eupodotis afraoides
Common
Burchell's Courser
Cursorius rufus
Uncommon
Western Redfooted Kestrel
Falco vespertinus
Uncommon
Greywing
Francolinus
Common
Pale Chanting
Goshawk
Francolin
africanus
Endemic
Degraded grassland
Dryland agriculture
Woodland
Waterbodies
Coastal
Endemic
Vulnerable
Monitoring
Endemic
Forest (AlC)
Shrubland
Woodland
Grassland
Thicket
Shrubland
Degraded grassland
Grassland
Woodland
Grassland
Shrubland
Degraded shrub land
Gypaetus barbatus
Rare
Monitoring
Grassland
Whitewinged Flufftail
Sarothrura ayresi
Rare
Endangered
Waterbodies
Blackwinged
Glareola nordmanni
Locally common
Grassland
Dryland agriculture
Waterbodies
Greater Kestrel
Falco rupicoloides
Locally common
Grassland
Shrubland
Degraded woodland
Abdim's Stork
Ciconia abdimii
Common
Grassland
Dryland agriculture
Pasture
Waterbodies
Grassland
Waterbodies
Rufousbellied
Heron
Butorides rufiventris
Rare
Pinkthroated
Longclaw
Macronyx ame/iae
Uncommon
Urban-residential
Plantations
Bearded Vulture
Pratincole
Degraded shrubland
Ocean
Breeding
low
Dryland agriculture
Coastal
Broadbilled Prion
Frigatebird
VEG4
Coastal
Common
Rare
Heron
VEG3
Degraded grassland
Broadbilled Sandpiper
Green Barbet
high
RED DATA
Pinkthroated
SickIewinged
Nonpasserine
low
DISTRIBUTION
Neergaard's Sunbird
Lemonbreasted
high
ABUNDANCE
Endemic
Vulnerable
Dryland agriculture
SCORE
Axis I
high
SPECIES
ABUNDANCE
Redbilled Helmetshrike
Prionops retzii
Common
Woodards'
Batis fratrum
Locally common
Black Coucal
Centropus bengalensis
Uncommon
Broadbilled Roller
Eurystomus glaucurus
Locally common
Woodland
Thicket
Yellow White-eye
Zosterops senegalensis
Rare
Coastal forest
Woodland
Lesser Blackwinged Plover
Vanellus lugubris
Uncommon
Woodland
Grassland
Stierling's
Calamonastes
stierlingi
Common
Natal Night jar
Caprimulgus
natalensis
Southern Grey Tit
Parus afer
COMMON
Batis
Barred Warbler
VEGI
Threatened
Woodland
Indeterminate
Coastal forest
Monitoring
Waterbodies
Rare
VEG2
Rare
Vulnerable
Common
Endemic
Grassland
Waterbodies
Shrubland
Grassland
Woodland
Common
Near endemic
Shrubland
Common
Breeding endemic
Grassland
Shrubland
Thickbilled
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Serinus flaviventris
Common
Near endemic
Shrubland
Grassland
Cercomela sinuata
Common
Endemic
Shrubland
Grassland
Parisoma layardi
Common
Endemic
Shrubland
Sicklewinged
Chat
Layard's Titbabbler
Degraded grassland
Degraded shrub land
Urban-residential
Common
Endemic
Near-threatened
Shrubland
Grassland
Uncommon
Endemic
Indeterminate
Grassland
Degraded grassland
Redeyed Bulbul
Pycnonotus
Common
Near endemic
Grassland
Shrubland
Woodland
Abdim's
Ciconia abdimii
Common
Grassland
Dryland agriculture
Pasture
Glareola nordmanni
Locally common
Grassland
Dryland agriculture
Waterbodies
Falco vespertinus
Uncommon
Grassland
Lesser Kestrel
Falco naumanni
Monitoring
Grassland
Thicket
Pallid Harrier
Circus macrourus
Common
Rare
Woodland
Dryland agriculture
Near-threatened
Grassland
Dryland agriculture
Woodland
Whitewinged Flufftail
Sarothrura ayresi
Rare
Endangered
Waterbodies
Eastern
Falco amurensis
Common
Monitoring
Grassland
Dryland agriculture
Urban-residential
White Stork
Ciconia ciconia
Common
Grassland
Dryland agriculture
Woodland
Lesser Grey Shrike
Lanius minor
Common
Thicket
Woodland
Shrubland
Siskin
Stork
Blackwinged
Pratincole
Redfooted
Redfooted
Kestrel
Kestrel
symonsi
Dryland agriculture
Spizocorys fringillaris
Western
Pseudochloroptila
Plantations
Botha's Lark
Drakensberg
nigricans
Non-breeding
endemic
European Marsh Harrier
Circus aeruginosus
Rare
Waterbodies
Gullbilled Tem
Gelochelidon
Rare
Coastal
Broadbilled Sandpiper
Limicola falcinellus
Rare
Coastal
Mongolian Plover
Charadrius mongolus
Common
Coastal
Softplumaged Petrel
Pterodroma mollis
Ocean
Crab Plover
Dromas ardeola
Common
Rare
Pintado Petrel
Daption capense
Common
Ocean
Subantartic Skua
Catharacta antartica
Common
Coastal
nilotica
VEG4
Woodland
Melierax canorus
Lark
VEG3
Grassland
Anthus hoeschi
Goshawk
Yellow Canary
high
Endemic
RED DATA
Mountain Pipit
Pale Chanting
Nonbreeding
low
DISTRIBUTION
Coastal
Urban-residential
Pasture
SCORE
Axis I
COMMON
SPECIES
ABUNDANCE
DISTRIBUTION
RED DATA
VEGI
VEG2
VEG3
Wandering Albatross
Diomedea exulans
Uncommon
Ocean
Bartailed Godwit
Limosa lapponica
Common
Coastal
Blackbrowed Albatross
Diomedea melanophris
Common
Ocean
Lemonbreasted
Serinus citrinipectus
Rare
Woodland
Grassland
Dryland agriculture
Pseudhirundo
Common
Woodland
Waterbodies
Dryland agriculture
Coastal
VEG4
Human
low
Canary
Greyrumped
Swallow
European Storm Petrel
Hydrobates pelagicus
Common
Caspian Tern
Hydroprogne
Common
Ayres' Eagle
Hieraaetus ayresii
Rare
Purplebanded
high
Nonhuman
low
Sunbird
caspia
Monitoring
Rare
Coastal
Waterbodies
Woodland
Plantations
Thicket
Coastal forest
Rivers
Urban-residential
Nectarinia bifasciata
Common
Woodland
Subantartic Skua
Catharacta antartica
Common
Coastal
Shy Albatross
Diomedea cauta
Common
Mangrove Kingfisher
Halcyon senegaloides
Uncommon
Vulnerable
Coastal forest
Lesser Flamingo
Phoeniconaias
Locally common
Near-threatened
Waterbodies
Thickbilled
Lark
minor
Galerida magnirostris
Common
Endemic
Shrubland
Grassland
Dryland agriculture
Common
Endemic
Shrubland
Grassland
Degraded grassland
Prinia jlavicans
Common
Near endemic
Shrubland
Thicket
Urban-residential
Redeyed Bulbul
Pycnonotus nigricans
Common
Near endemic
Grassland
Shrubland
Woodland
Burchell's Courser
Cursorius rufus
Uncommon
Endemic
Shrubland
Degraded grassland
Degraded shrubland
Clapper Lark
Mirafra apiata
Common
Endemic
Grassland
Shrubland
Dryland agriculture
Greater Kestrel
Falco rupicoloides
Locally common
Grassland
Shrubland
Degraded woodland
Larklike
Emberiza impetuani
Common
Shrubland
Grassland
Thicket
Abdim's Stork
Ciconia abdimii
Common
Grassland
Dryland agriculture
Pasture
Fairy Flycatcher
Stenostira scita
Common
Grassland
Thicket
Urban-residential
Grey Wagtail
Forest
Waterbodies
Chat
Blackchested
Prinia
Bunting
Motacilla cinerea
Rare
Rufousbellied
Heron
Butorides rufiventris
Rare
Pinkthroated
Longclaw
Monitoring
Near-endemic
Endemic
Endemic
Waterbodies
Macronyx ameliae
Uncommon
Vulnerable
Grassland
Waterbodies
Black Coucal
Centropus bengalensis
Uncommon
Monitoring
Waterbodies
Grassland
Broadbilled Sandpiper
Limicola falcinellus
Rare
Gullbilled Tern
Gelochelidon
Rare
Natal Night jar
Caprimulgus natalensis
Rare
Crab Plover
Dromas ardeola
Rare
Batis fratrum
Locally common
Eurystomus glaucurus
Locally common
Woodland
Thicket
Melierax canorus
Common
Near endemic
Shrubland
Woodland
Parusafer
Common
Endemic
Shrubland
Grassland
Batis
Broadbilled
Pale Chanting
Roller
Goshawk
Southern Grey Tit
nilotica
Urban-residential
Ocean
Cercomela sinuata
Sicklewinged
Woodards'
high
griseopyga
Coastal
Coastal
Vulnerable
Grassland
Waterbodies
Coastal
Endemic
Indeterminate
Coastal forest
Plantations
Degraded shrub land
Urban-residential
Dryland agriculture
Waterbodies
SCORE
Axis 1
COMMON
SPECIES
ABUNDANCE
DISTRIBUTION
Mountain Pipit
Anthus hoeschi
Common
Breeding endemic
Grassland
Shrubland
Yellow Canary
Serinus jlaviventris
Common
Near endemic
Shrubland
Grassland
Common
Endemic
Common
Endemic
Near-threatened
Shrubland
Grassland
Indeterminate
Grassland
Degraded grassland
Shrubland
Near-threatened
Grassland
Grassland
Grassland
Thicket
Layard's Titbabbler
Parisoma layardi
Drakensberg Siskin
Pseudochloroptila
Botha's Lark
Spizocorys fringillaris
Uncommon
Endemic
Rock Pipit
Anthus crenatus
Common
Endemic
Chaetops aurantius
Common
Endemic
Eupodotis afraoides
Common
Orangebreasted
Rockjumper
Whitewinged Black Korhaan
symonsi
RED DATA
VEGI
VEG2
Shrubland
VEG3
VEG4
Table B-4: ADD species that scored highest and lowest in stepwise canonical correspondence analyses on axis 2 for bird assemblages. Associated ecological
information is also provided.
SCORE
Axis2
All
COMMON
SPECIES
low
Burchell's Starling
Lamprotornis
Bennett's Woodpecker
Campethera bennettii
Thickbilled
Pachycoccyx
high
Cuckoo
australis
DISTRIBUTION
Common
Endemic
RED DATA
Locally common
VEGI
VEG2
Woodland
Thicket
Rare
Thicket
Woodland
Glaucidium perlatum
Common
Woodland
Thicket
Redbilled
Bubalornis niger
Nomadic
Woodland
Thicket
Barred Owl
Glaucidium capense
Locally common
Swallowtailed Bee-eater
Merops hirundineus
Rare
Purple Widowfinch
Vidua purpurascens
Locally common
Whitebreasted
Coracina pectoralis
Rare
Redcrested Korhaan
Eupodotis ruficrista
Common
Gullbil1ed Tern
Gelochelidon
Rare
Coastal
Cuckooshrike
nilotica
Monitoring
Rare
Woodland
Woodland
Monitoring
Woodland
Thicket
Woodland
Near endemic
Monitoring
Grassland
Woodland
Broadbilled Sandpiper
Limicola falcinellus
Rare
Coastal
Pintado Petrel
Daption capense
Common
Ocean
Mongolian Plover
Charadrius mongolus
Common
Whitechinned
Petrel
Procellaria aequinoctialis
Common
Softplumaged
Petrel
Pterodroma mollis
Common
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
Blackbrowed Albatross
Diomedea melanophris
Common
Ocean
Pomarine Skua
Stercorarius pomarinus
Common
Coastal
Cory's Shearwater
Calonectris diomedea
Common
Ocean
Bennett's Woodpecker
Campethera bennettii
Locally common
Woodland
Pearlspotted
Glaucidium perlatum
Common
Woodland
Thicket
Burchell's Starling
Lamprotornis
Common
Woodland
Thicket
OvamboSparrowhawk
Accipiter ovampensis
Rare
Woodland
Plantations
Thickbilled
Pachycoccyx
Rare
Thicket
Woodland
Woodland
Thicket
Coastal
Monitoring
Ocean
Summer
low
Owl
Cuckoo
australis
audeberti
VEG3
Woodland
Pearlspotted Owl
Buffalo Weaver
audeberti
ABUNDANCE
Endemic
Monitoring
Redbilled Buffalo Weaver
Bubalornis niger
Nomadic
Longtoed Plover
Vanellus crassirostris
Rare
Monitoring
Waterbodies
Purple Widowfinch
Vidua purpurascens
Locally common
Monitoring
Woodland
Thicket
Shrubland
VEG4
SCORE
Axis2
high
COMMON
SPECIES
ABUNDANCE
DISTRIBUTION
RED DATA
VEGI
Rare
Woodland
Monitoring
Grassland
Barred Owl
Glaucidium capense
Locally common
Redcrested Korhaan
Eupodotis ruficrista
Common
Broadbilled Sandpiper
Limicola falcinellus
Rare
Coastal
Gullbilled Tern
Gelochelidon
Rare
Coastal
Pintado Petrel
Daption capense
Common
Ocean
Mongolian Plover
Charadrius mongolus
Common
Pied Mannikin
Spermestes fringilloides
Rare
Indeterminate
Coastal forest
Procellaria
aequinoctialis
Common
Monitoring
Ocean
Whitechinned
Petrel
nilotica
Near endemic
VEG2
VEG3
Woodland
Shrubland
Urban-residential
Dryland agriculture
Coastal
Pterodroma
mollis
Common
African Black Oystercatcher
Haematopus
moquini
Common
Coastal
Lesser Blackbacked Gull
Larus fuscus
Uncommon
Coastal
Fleshfooted Shearwater
Puffinus carneipes
Common
Ocean
Burchell's Starling
Lamprotornis
Woodland
Thicket
Pearlspotted Owl
Glaucidium perlatum
Common
Woodland
Thicket
Longtailed Shrike
Corvinella melanoleuca
Common
Thicket
Woodland
Swallowtailed Bee-eater
Merops hirundineus
Rare
Woodland
Softplumaged
Petrel
Ocean
Winter
low
high
australis
Common
Endemic
Redcrested Korhaan
Eupodotis ruficrista
Common
Near endemic
Lesser Grey Shrike
Lanius minor
Common
Non-breeding
Jameson's Firefinch
Lagonosticta
Purple Roller
Burntnecked Eremomela
Monitoring
Grassland
Woodland
Shrubland
Thicket
Woodland
Shrubland
Locally common
Woodland
Thicket
Coracias naevia
Common
Woodland
Thicket
Eremomela usticollis
Common
Woodland
Paradise Whydah
Vidua paradisaea
Common
Thicket
Gullbilled Tern
Gelochelidon
Rare
Coastal
Yellow Wagtail
Motacilla flava
Uncommon
Waterbodies
Pintado Petrel
Daption capense
Common
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
Yellownosed Albatross
Diomedea chlororhynchos
Common
Whitechinned
Procellaria
Common
Petrel
rhodopareia
nilotica
aequinoctialis
Arctic Skua
Stercorarius parasiticus
Common
Bartailed Godwit
Limosa lapponica
Common
Bittern
Eotaurus stellaris
Rare
Curlew
Numenius arquata
Common
endemic
Ocean
Monitoring
Ocean
Coastal
Coastal
Vulnerable
Waterbodies
Coastal
Woodland
Degraded grassland
VEG4
SCORE
Axis2
Passerine
COMMON
SPECIES
ABUNDANCE
low
Whitebreasted
Coracina pectoralis
Rare
Burchell's Starling
Lamprotornis
Common
Redbilled
Bubalornis niger
Nomadic
Vidua purpurascens
Locally common
Longtailed Shrike
Corvinella melanoleuca
Common
Redbilled
Prionops retzii
Common
Cuckooshrike
Buffalo Weaver
Purple Widowfinch
high
Helmetshrike
high
Woodland
Monitoring
Threatened
Hippolais olivetorum
Uncommon
Non-breeding
Mirafra apiata
Common
Endemic
Redheaded Weaver
Anaplectes rubriceps
Common
Jameson's Firefinch
Lagonosticta
Locally common
Pied Mannikin
Spermestes jringilloides
House Crow
Corvus splendens
Locally common
Zoothera guttata
Uncommon
Starling
endemic
Thicket
Woodland
Thicket
Thicket
Woodland
Woodland
Degraded woodland
Grassland
Shrubland
Woodland
Thicket
Indeterminate
Coastal forest
Urban-residential
Endangered
Coastal forest
Alien
Dryland agriculture
Thicket
Shrubland
Grassland
Thicket
Plantations
Grassland
Emberiza impetuani
Common
Near-endemic
Brown Robin
Erythropygia
signata
Common
Endemic
Barratt's Warbler
Bradypterus barratti
Common
Endemic
European Sedge Warbler
Acrocephalus
Rare
Waterbodies
Grey Cuckooshrike
Coracina caesia
Uncommon
Woodland
Campethera bennettii
Locally common
Rare
Pearlspotted
Owl
audeberti
Alien
Urban-residential
Larklike Bunting
Pachycoccyx
Vulnerable
Forest(NC)
Woodland
Monitoring
Common
Barred Owl
Glaucidium capense
Locally common
Swallowtailed Bee-eater
Merops hirundineus
Rare
Redcrested Korhaan
Eupodotis ruficrista
Common
Purple Roller
Coracias naevia
Common
Grey Lourie
Corythaixoides
Whiteheaded Vulture
Trigonoceps occipitalis
Uncommon
Rare
Giant Eagle Owl
Bubo lacteus
Uncommon
Monitoring
GullbiIled Tern
Gelochelidon
nilotica
Coastal forest
Forest
Glaucidium perlatum
concolor
Dryland agriculture
Waterbodies
Common
Common
Cuckoo
Dryland agriculture
Urban-residential
Sturn us vulgaris
Bennett's Woodpecker
VEG4
Woodland
Ploceus xanthops
Thickbilled
VEG3
Thicket
Woodland
Golden Weaver
schoenobaenus
VEG2
Woodland
Rare
Spotted Thrush
VEGI
Woodland
Clapper Lark
rhodopareia
RED DATA
Endemic
Olivetree Warbler
European
Nonpasserine
low
australis
DISTRIBUTION
Rare
Thicket
Woodland
Woodland
Thicket
Woodland
Woodland
Near endemic
Monitoring
Common
Rare
Grassland
Woodland
Woodland
Thicket
Woodland
Thicket
Woodland
Woodland
Coastal
266
Shrubland
Urban-residential
Waterbodies
SCORE
Axis2
COMMON
SPECIES
ABUNDANCE
DISTRIBUTION
RED DATA
Broadbilled Sandpiper
Limicola jalcinellus
Rare
Coastal
Mongolian Plover
Charadrius mongolus
Common
Coastal
Pintado Petrel
Daption capense
Common
VEGI
VEG3
VEG4
Degraded shrubland
Dryland agriculture
Ocean
Whitechinned
Petrel
Procellaria aequinoctialis
Common
Softplumaged
Petrel
Pterodroma
Common
Ocean
mollis
VEG2
Monitoring
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
Whitebellied
Storm Petrel
Fregetta grallaria
Rare
Ocean
Blackbellied
Storm Petrel
Fregetta tropica
Common
Ocean
Pomarine Skua
Stercorarius pomarinus
Common
Coastal
Burchell's Starling
Lamprotornis
Common
Bennett's Woodpecker
Campethera bennettii
Thickbilled
Pachycoccyx
Breeding
low
Cuckoo
Nonbreeding
low
audeberti
Woodland
Endemic
Locally common
Thicket
Woodland
Rare
Monitoring
Thicket
Woodland
Pearlspotted Owl
Glaucidium perlatum
Common
Woodland
Thicket
Redbilled
Bubalornis niger
Nomadic
Woodland
Thicket
Coracina pectoralis
Rare
Burchell's Courser
Cursorius nifus
Uncommon
Swallowtailed Bee-eater
Merops hirundineus
Rare
Barred Owl
Glaucidium capense
Locally common
Rare
Woodland
Purple Widowfinch
Vidua purpurascens
Locally common
Monitoring
Woodland
Thicket
Pied Mannikin
Spermestes jringilloides
Rare
Indeterminate
Coastal forest
Urban-residential
House Crow
Corvus splendens
Locally common
Alien
Endemic
Buffalo Weaver
Whitebreasted
high
australis
Cuckooshrike
Woodland
Monitoring
Endemic
Shrubland
Degraded grassland
Woodland
Knysna Woodpecker
Campethera notata
Locally common
Near-threatened
Coastal forest
Thicket
Mangrove Kingfisher
Halcyon senegaloides
Uncommon
Vulnerable
Coastal forest
Rivers
Bittern
Botaurus stellaris
Rare
Vulnerable
Waterbodies
Green Barbet
Stactolaema
Locally common
Spotted Thrush
Zoothera guttata
Uncommon
Cape Cormorant
Phalacrocorax
Common
African Black Oystercatcher
Haematopus
Cape Gannet
Morus capensis
Common
Breeding endemic
Coastal
Olivetree
Hippolais olivetorum
Uncommon
Non-breeding
Woodland
Aquila nipalensis
Uncommon
Warbler
Steppe Eagle
olivacea
capensis
moquini
Dryland agriculture
Urban-residential
Endemic
Breeding endemic
Common
Vulnerable
Forest (AlC)
Endangered
Coastal forest
Ocean
Coastal
endemic
Woodland
267
Degraded woodland
Urban-residential
SCORE
Axis2
high
VEGI
VEG2
Rare
Forest
Waterbodies
Locally common
Grassland
Woodland
Waterbodies
Thicket
Woodland
Shrubland
Waterbodies
Coastal
Woodland
Urban-residential
COMMON
SPECIES
ABUNDANCE
Grey Wagtail
Motacilla cinerea
Bluecheeked Bee-eater
Merops persicus
Lesser Spotted Eagle
Aquila pomarina
Uncommon
European Bee-eater
Merops apiaster
Common
Redshank
Tringa totanus
Rare
Dusky Lark
Pinarocorys
Lesser Flamingo
Phoeniconaias
nigricans
minor
DISTRIBUTION
RED DATA
Monitoring
Uncommon
Locally common
Near-threatened
Waterbodies
Icterine Warbler
Hippolais icterina
Common
Woodland
Catharacta antartica
Common
Coastal
Wandering Albatross
Diomedea exulans
Uncommon
Ocean
Pintado Petrel
Daption capense
Common
Whitewinged Fluff tail
Sarothrura ayresi
Rare
Petrel
Pterodroma mol/is
Falco vespertinus
Uncommon
Whitechinned
Procel/aria aequinoctialis
Common
Petrel
Plantations
Urban-residential
Ocean
Endangered
Waterbodies
Ocean
Common
Western Redfooted Kestrel
Grassland
Monitoring
Woodland
Ocean
Diomedea melanophris
Common
Ocean
Wilson's Storm Petrel
Oceanites oceanicus
Common
Ocean
Yellownosed Albatross
Diomedea chlororhynchos
Common
Ocean
Burchell's Courser
Cursorius rufus
Uncommon
Endemic
Olivetree Warbler
Hippolais olivetorum
Uncommon
Non-breeding
Heuglin's Robin
Cossypha heuglini
Common
Thicket
Woodland
Urban-residential
Striped Cuckoo
Clamator levaillantii
Locally common
Woodland
Thicket
Urban-residential
Common
Woodland
Thicket
Urban-residential
Common
Grassland
Woodland
Dryland agriculture
Degraded woodland
Blackbrowed
Albatross
VEG4
Woodland
Subantartic Skua
Softplumaged
VEG3
Human
low
Corythaixoides
Grey Lourie
Chestnutbacked
high
Finchlark
Eremopterix
concolor
leucotis
Monitoring
endemic
Shrubland
Degraded grassland
Woodland
Degraded woodland
Degraded shrubland
Bushveld Pipit
Anthus caffer
Common
Woodland
Flappet Lark
Mirafra rufocinnamomea
Common
Woodland
Degraded woodland
Dryland agriculture
White Helmetshrike
Prionops plumatus
Common
Woodland
Thicket
Urban-residential
Sabota Lark
Mirafra sabota
Common
Woodland
Shrubland
Degraded woodland
Pintado Petrel
Daption capense
Common
Wilson's Storm Petrel
Oceanites oceanicus
Common
Whitechinned
Procel/aria
aequinoctialis
Common
Monitoring
Ocean
Pterodroma
macroptera
Common
Vulnerable
Ocean
Petrel
Greatwinged Petrel
Near endemic
Ocean
Ocean
Dryland agriculture
Plantations
SCORE
Axis2
Nonhuman
low
high
VEGI
VEG2
VEG3
Indeterminate
Coastal forest
Urban -resi den tiaI
Dryland agriculture
SPECIES
ABUNDANCE
Pied Mannikin
Spermestes jringilloides
Rare
Subantartic Skua
Catharacta antartica
Common
Coastal
Fleshfooted Shearwater
Puffinus carneipes
Common
Ocean
Ocean
Blackbrowed Albatross
Diomedea melanophris
Common
Southern Giant Petrel
Macronectes
Common
House Crow
Corvus splendens
Locally common
Bennett's Woodpecker
Campethera bennettii
Locally common
Common
giganteus
DISTRIBUTION
RED DATA
COMMON
Ocean
Alien
Urban-residential
Woodland
Burchell's Starling
Lamprotornis
australis
Thickbilled Cuckoo
Pachycoccyx
audeberti
Endemic
Pearlspotted Owl
Glaucidium perlatum
Common
Woodland
Thicket
Redbilled Buffalo Weaver
Bubalornis niger
Nomadic
Woodland
Thicket
Rare
Monitoring
Woodland
Thicket
Thicket
Woodland
Whitebreasted Cuckooshrike
Coracina pectoralis
Rare
Purple Widowfinch
Vidua purpurascens
Locally common
Monitoring
Woodland
Barred Owl
Glaucidium capense
Locally common
Rare
Woodland
Monitoring
Grassland
Woodland
Swallowtailed Bee-eater
Merops hirundineus
Rare
Redcrested Korhaan
Eupodotis ruficrista
Common
Gul1billed Tern
Gelochelidon
Rare
Coastal
nilotica
Woodland
Near endemic
Broadbilled Sandpiper
Limicola jalcinellus
Rare
Coastal
Mongolian Plover
Charadrius mongolus
Common
Coastal
Softplumaged
Pterodroma mollis
Common
Ocean
Petrel
Arctic Skua
Stercorarius parasiticus
Common
Coastal
Yel1ownosed Albatross
Diomedea chlororhynchos
Common
Ocean
Whitebellied
Fregetta grallaria
Rare
Ocean
Pomarine Skua
Stercorarius pomarinus
Common
Coastal
Cory's Shearwater
Calonectris diomedea
Common
Ocean
Blackbel1ied Storm Petrel
Fregetta tropica
Common
Ocean
Storm Petrel
Thicket
Woodland
Shrubland
VEG4
The following table lists the birds that were found in the CR survey but not found in the
ADD survey, and birds found in the ADD survey but not found in during the CR survey. ADD
birds in bold denote species found in the province before 1970 by Clancey (1964).
Table D-l: Avian species recorded in KwaZulu-Natal during the CR survey but not found in the
ADD survey, and species recorded during the ADD survey but not found in the CR survey. ADD
birds in bold denote species found in the province before 1970 by Clancey (1964).
DATA
SPECIES
COMMON
ABUNDANCE
DISTRIBUTION
RED DATA
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
CR
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
ADU
Spheniscus demersus
Pachyptila belcheri
Neophron percnopterus
Necrosyrtes monachus
Macheiramphus alcinus
Pernis apivorus
Buteo augur
Melierax metabates
Falco cuvierii
Falco dickinsoni
Coturnix adansonii
Ardeotis kori
Calidris ruficollis
Larus sabini
Sterna dougallii
Sterna anaethetus
Macrodipteryx vexillaria
Merops nubicoides
Coracias spatulata
Oenanthe pileata
Saxicola rubetra
Luscinia luscinia
Sylvia communis
Apalis melanocephala
Ficedula albicollis
Anthus trivialis
Prionopsscopij'rons
Eurocephalus anguitimens
Nectarinia venusta
Anthreptes reichenowi
Uraeginthus granatinus
Amadina fasciata
Serinus albogularis
Diomedea chrysostoma
Accipiter ovampensis
Circus aeruginosus
Sarothrura ayresi
Rynchops flavirostris
Columba livia
Cuculus gularis
Glaucidium capense
Campethera notata
Mirafra apiata
Mirafra ruddi
Spizocorys fringillaris
Coracina pectoralis
Parus afer
Prinia flavicans
Lamprotornis australis
Buphagus africanus
Anthus hoeschi
Jackass Penguin
Slenderbilled Prion
Egyptian Vulture
Hooded Vulture
Bat Hawk
Honey Buzzard
Augur Buzzard
Dark Chanting Goshawk
African Hobby Falcon
Dickinson's Kestrel
Blue Quail
Kori Bustard
Rednecked Stint
Sabine's Gull
Roseate Tern
Bridled Tern
Pennantwinged Night jar
Carmine Bee-eater
Rackettailed Roller
Capped Wheatear
Whinchat
Thrush Nightingale
Whitethroat
Blackheaded Apalis
Collared Flycatcher
Tree Pipit
Chestnutfronted Helmetshrike
Whitecrowned Shrike
Yellowbellied Sunbird
Bluethroated Sunbird
Violeteared Waxbill
Cutthroat Finch
Whitethroated Canary
Greyheaded Albatross
Ovambo Sparrowhawk
European Marsh Harrier
Whitewinged Flufftail
African Skimmer
Feral Pigeon
African Cuckoo
Barred Owl
Knysna Woodpecker
Clapper Lark
Rudd's Lark
Botha's Lark
Whitebreasted Cuckooshrike
Southern Grey Tit
Blackchested Prinia
Burchell's Starling
Yellowbilled Oxpecker
Mountain Pipit
Locally common
Uncommon
Rare
Rare
Rare
Rare
Rare
Locally common
Uncommon
Rare
Rare
Rare
Rare
Rare
Locally common
Rare
Locally common
Common
Uncommon
Common
Rare
Uncommon
Locally common
Locally common
Rare
Rare
Uncommon
Common
Locally common
Uncommon
Common
Uncommon
Common
Rare
Rare
Rare
Rare
Rare
Common
Locally common
Locally common
Locally common
Common
Uncommon
Uncommon
Rare
Common
Common
Common
Locally common
Common
Endemic
Globally nearthreatened
Extinct South Africa
Vulnerable
Indeterminate
Vulnerable
Endangered
Indeterminate
Vulnerable
Endemic
Vulnerable
Near endemic
Globally nearthreatened
Near endemic
Alien
Endemic
Endemic
Endemic
Endemic
Globally endangered
Endangered
Rare
Globally near
threatened
Endemic
Near endemic
Endemic
Critical
Indeterminate
Breeding endemic
Vulnerable
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement