THE PROSPECTS FOR SPREAD AND IMPACTS OF REMOVAL ERAGROSTIS LEHMANNIANA by

THE PROSPECTS FOR SPREAD AND IMPACTS OF REMOVAL ERAGROSTIS LEHMANNIANA by
THE PROSPECTS FOR SPREAD AND IMPACTS OF REMOVAL
OF ERAGROSTIS LEHMANNIANA NEES
by
Theresa Marie Mau-Crimmins
Copyright © Theresa Marie Mau-Crimmins 2005
A Dissertation Submitted to the Faculty of the
SCHOOL OF NATURAL RESOURCES
In Partial Fulfillment of the Requirements
For the Degree of
DOCTOR OF PHILOSOPHY
WITH A MAJOR IN RENEWABLE NATURAL RESOURCES
In the Graduate College
THE UNIVERSITY OF ARIZONA
2005
2
THE UNIVERSITY OF ARIZONA
GRADUATE COLLEGE
As members of the Dissertation Committee, we certify that we have read the dissertation
prepared by Theresa M. Mau-Crimmins
entitled The Prospects for Spread and Impacts of Removal of Eragrostis lehmanniana
Nees
and recommend that it be accepted as fulfilling the dissertation requirement for the
Degree of Doctor of Philosophy.
Guy R. McPherson_______________________________ Date:___3/31/05_____
H. Randy Gimblett_______________________________ Date: ___3/31/05_____
Steven Archer
_______________________________ Date: ___3/31/05_____
John Kupfer
_______________________________
Date: ___3/31/05_____
Andrew Comrie ________________________________ Date: ___3/31/05_____
Final approval and acceptance of this dissertation is contingent upon the candidate’s
submission of the final copies of the dissertation to the Graduate College.
I hereby certify that I have read this dissertation prepared under my direction and
recommend that it be accepted as fulfilling the dissertation requirement.
Guy R. McPherson_____________________3/31/05____________________________
Dissertation Director:
Date
3
STATEMENT BY AUTHOR
This dissertation has been submitted in partial fulfillment of requirements for an
advanced degree at The University of Arizona and is deposited in the University Library
to be made available to borrowers under rules of the Library.
Brief quotations from this dissertation are allowable without special permission,
provided that accurate acknowledgment of source is made. Requests for permission for
extended quotation from or reproduction of this manuscript in whole or in part may be
granted by the copyright holder.
SIGNED: Theresa M. Mau-Crimmins
4
ACKNOWLEDGEMENTS
A great deal of credit goes first to my fabulous husband, Michael Crimmins, for
providing the love and support I needed to complete this degree. He provided a
wonderful sounding board for my thoughts and had major positive influence over this
document’s content. Great thanks also are in order for my parents (both sets!), my sisters,
and my friends for their encouragement and listening ears.
Many thanks are in order for my committee, Drs. Randy Gimblett, Steve Archer,
Andrew Comrie, John Kupfer, and especially my advisor, Dr. Guy McPherson. Their
input and guidance at various times throughout my career at the University of Arizona
helped to shape this study and will have lasting impacts.
I was fortunate to perform my field experiment at three locations in southeastern
Arizona and several kind individuals made this possible. Thanks to Coronado National
Memorial staff and Chief of Resources Barbara Alberti for their on-site support of my
study. Mark Heitlinger and Dr. Mitch McClaran made it possible for me to replicate my
study at the Santa Rita Experimental Range. Finally, Dave Harris and Barbara Clark of
The Nature Conservancy considerately allowed the experiment to take place on the Three
Links Farm. Additionally, Arturo Baez and Mark Carson of the Campus Agricultural
Center, provided facilities support of the greenhouse portion of the study.
Many individuals graciously donated their time to assist with field data collection,
including Barbara Alberti, Jeff Balmat, Kristen Beaupre, Mike Crimmins, Andy
Hubbard, Sara Jensen, John Kupfer, Melissa Mauzy, Jeff McGovern, Guy McPherson,
Sheila Merrigan, Katie Nasser, Meg Quinn, Sarah Studd, Amy Tendick, Jason Welborn,
Marcela Yepez, and several volunteers from Coronado National Memorial. Meg Quinn
provided plant identification expertise. Very special recognition is order for the generous
members of Majestic Management: Dennis Crimmins, Jim Nunnold, John Connors, Steve
Britz, Gordon Ruttan. These individuals repeatedly traveled from Michigan to Arizona,
on their own time and money, to assist with fieldwork. I owe them beer wages for
decades to come!
I am especially grateful to my wonderful co-workers at the Sonoran Desert
Network Inventory & Monitoring office. Kristen Beaupre and Debbie Angell provided
invaluable data management support and Melissa Mauzy played a very important role in
field and greenhouse data collection. Andy Hubbard served as a second mentor to me,
reading multiple drafts and providing priceless guidance on data analysis and reporting
techniques. Andy also provided plenty of comic relief.
Several friends offered great intellectual contributions, especially Erika Geiger
and Heather Schussman. Our collaboration will continue far beyond this document.
5
Administrative support was provided by Valery Catt, Cheryl Craddock, Anne Hartley,
Dee Simons, and Cecily Westphal.
This research was funded in part by the University of Arizona Space Grant
Fellowship Program, Center for Invasive Plant Management Award Number ESA000011, and a grant from T&E, Inc.
Theresa Mau-Crimmins
6
DEDICATION
To Mike, who provided the love and support to make this possible, and to the
wonderful members of Majestic Management: Dennis Crimmins, Jim Nunnold,
John Connors, Steve Britz, and Gordon Ruttan.
7
TABLE OF CONTENTS
ABSTRACT...................................................................................................................8
CHAPTER 1. INTRODUCTION ................................................................................10
Explanation of Problem ...........................................................................................10
Approach..................................................................................................................11
Organization of the Dissertation ..............................................................................12
CHAPTER 2. PRESENT STUDY...............................................................................15
Summary ..................................................................................................................21
REFERENCES ............................................................................................................24
APPENDIX A. CAN AN INVADING SPECIES’ DISTRIBUTION BE PREDICTED
USING DATA FROM ITS INVADED RANGE? LEHMANN LOVEGRASS
(ERAGROSTIS LEHMANNIANA) IN THE SOUTHWESTERN UNITED
STATES.......................................................................................................................29
APPENDIX B. MODELING FUTURE POTENTIAL DISTRIBUTIONS OF
ERAGROSTIS LEHMANNIANA (LEHMANN LOVEGRASS) IN ARIZONA,
USA..............................................................................................................................62
APPENDIX C. COMMUNITY-LEVEL AND SEED BANK RESPONSE TO THE
REMOVAL OF A NONNATIVE PERENNIAL GRASS AT THREE SITES IN
SOUTHEASTERN ARIZONA, USA .........................................................................95
8
ABSTRACT
Non-indigenous invasive species are a major threat to native species diversity and
ecosystem function and have been called the single worst threat of natural disaster of this
century. Eragrostis lehmanniana Nees (Lehmann lovegrass), a tufted perennial
bunchgrass native to southern Africa, is one such problematic species in Arizona, USA.
This dissertation research is a mix of predictive modeling and field experiments designed
to inform management decisions based on greater understanding of this nonnative
species, with emphasis on the potential for spread and the impacts of removal.
The modeling studies in this dissertation aimed to predict the potential
distribution of E. lehmanniana in the southwestern United States under current and
potential future climate conditions. The first portion of study addressed a common
assumption in predictive modeling of nonnative species: data from the species’ native
range are necessary to accurately predict the potential distribution in the invaded range.
The second portion of this study predicted the distribution of E. lehmanniana under 28
different climate change scenarios. Results showed the distribution of E. lehmanniana
progressively shrinking in the southeastern and northwestern portions of the state and
increasing in the northeastern portion of the state with increasing temperatures and
precipitation. Key shifts occurred under scenarios with increases in summer and winter
precipitation of 30% or more, and increases in summer maximum and winter minimum
temperatures of at least 2oC.
The field experiment served as a pre-eradication assessment for E. lehmanniana
and indicates how semi-desert grassland communities in southeastern Arizona may
9
respond to the removal of this species. This study suggested that plant community
response to removal of an introduced species is mediated by precipitation variability
(timing and amount), local site history, and edaphic conditions. The response observed on
a site previously farmed for decades was to subsequently become dominated by other
nonnative annual species. However, the two other sites with histories of livestock grazing
responded more predictably to the removal, with an increase in annual ruderal species (2
to 10 times the amount of annual cover recorded on control plots).
10
CHAPTER 1. INTRODUCTION
Explanation of the Problem
Exotic species have been called the “single most formidable threat of natural
disaster of the 21st century” (Schnase et al. 2002), causing economic impacts estimated at
more than $100 billion annually (Pimentel et al. 1999). Non-indigenous invasive species
are a major threat to native species diversity and ecosystem function and can impact
native species and ecosystems in potentially irreversible ways (Chapin et al. 1996, Mack
and D’Antonio 1998, Wilcove et al. 1998). The continued introduction of nonnatives is
expected to increase with increases in international trade (Office of Technology
Assessment 1993).
Anticipating the impacts of introduced species is difficult, as species’
distributions are expected to change under predicted climate change (Watson et al. 1996).
Climate plays a large role in determining the distribution of plant species and
communities (Huntley and Webb 1988, Wright et al. 1993). Significant increases in
precipitation and temperature as well as shifts in precipitation seasonality are expected
for the southwestern United States in the coming decades (IPCC 2001), suggesting major
shifts in community and species distributions. The structure and diversity of ecosystems
are expected to undergo considerable changes (IGBP 1988, Watson et al. 1996),
potentially having detrimental impacts on ecosystem stability (Mooney 1997). Such
reshuffling could favor nonindigenous species, especially those with the capacity to
disperse rapidly and compete strongly for resources.
11
Managing nonnative species is resource-intensive, requiring a long-term
commitment (Hiebert 1997). Effective management strategies must be based on thorough
knowledge of introduced species’ modes and rates of spread, potential and known effects,
and control methods. Byers et al. (2002) identify a disconnect between science and
management, and managers are not receiving the information they need to make
informed decisions. Scientists can help managers make sound decisions by providing
them answers to questions such as:
•
Are nonnative species still spreading to new regions?
•
Will spread of nonnative species continue at the same rate under an altered
climate?
•
What events follow the control or eradication of a nonnative species?
This study attempts to address this research gap by exploring these questions for a
nonnative perennial grass in southern Arizona, Eragrostis lehmanniana Nees (Lehmann
lovegrass).
Approach
This dissertation research is a mix of predictive modeling and field experiments
designed to inform management decisions based on greater understanding of nonnative
species, with emphasis on the potential for spread and the impacts of removal. The target
species for these studies is Eragrostis lehmanniana Nees, a tufted perennial bunchgrass
native to southern Africa that was brought to Arizona, USA in the 1930s to counteract
low plant cover and highly eroded soils resulting from decades of overgrazing and
12
drought. As part of a far-reaching search for drought-tolerant plants, several accessions of
E. lehmanniana were brought to Arizona (Crider 1945). In Arizona, E. lehmanniana has
been associated with decreased plant and animal species richness (Cable 1971, Bock et
al. 1986) and plant species diversity (Geiger and McPherson 2004). This apomictic grass
has also been linked to alteration of ecosystem processes (Cable 1971, Williams and
Baruch 2000). The modeling studies in this dissertation aim to predict the potential
distribution of E. lehmanniana in the southwestern United States, under current and
potential future climate conditions. The field experiment serves as a pre-eradication
assessment for E. lehmanniana and indicates how semi-desert grassland communities in
southeastern Arizona may respond to the removal of this species. A second objective of
the field study was to characterize the seed banks at these sites; this was accomplished
through a greenhouse study of seed bank samples.
Organization of the Dissertation
The research presented in this dissertation consists of three separate, but related,
studies. Each study is presented as a separate paper in the appendix and is ready for
submission to a journal for consideration of publication. Literature reviews for each study
are found within respective papers.
Appendix A, titled, “Can an invading species’ distribution be predicted using data
from its invaded range? Lehmann lovegrass (Eragrostis lehmanniana) in the
southwestern United States,” was co-authored with Heather Schussman and Erika Gieger
and has been submitted to Ecological Modelling. Schussman and Geiger assisted with
13
original idea formulation and provided extensive assistance in reviewing and revising
drafts of this paper. This study addressed a common assumption in predictive modeling
of nonnative species: data from the species’ native range are necessary to accurately
predict the potential distribution in the invaded range. The purpose of this study was to
explore differences in distribution predicted from two different sets of input data.
Appendix B, titled, “Modeling future potential distributions of Eragrostis
lehmanniana (Lehmann lovegrass) in Arizona, USA” was prepared for submission to
Diversity and Distributions. In this study, I explored the potential effects of climate
change on the distribution of E. lehmanniana in the southwestern United States using
ecological niche modeling. The Genetic Algorithm for Rule-set Prediction, a well-tested
ecological niche modeling tool (Stockwell and Peters 1999, Peterson et al. 2001, Peterson
et al. 2002, Oberhauser and Peterson 2003, Peterson and Shaw 2003), was used to predict
the distribution of E. lehmanniana under 28 climate change scenarios. I will be sole
author of this manuscript.
Appendix C, titled, “Community-level and seed bank response to the removal of
Eragrostis lehmanniana (Lehmann lovegrass) at three sites in southeastern Arizona,” was
prepared for submission to Ecological Monographs. This study serves as a preeradication assessment for E. lehmanniana and indicates how semi-desert grassland
communities in southeastern Arizona may respond to the removal of this species. A
second objective of this study was to characterize the seed banks at these sites, which was
accomplished through a greenhouse study of seed bank samples. This study will be co-
14
authored by Dr. Guy McPherson, who played a large role in research question
formulation and study design.
15
CHAPTER 2. PRESENT STUDY
The methods, results, and conclusions of this dissertation are distributed in the
appended papers. Together, the three studies address gaps in our understanding of
nonnative plant species, specifically regarding the potential for spread and impacts of
their removal. This applied research is directly applicable for land management decisionmaking.
Appendix A – “Can an invading species’ distribution be predicted using data from its
invaded range? Lehmann lovegrass (Eragrostis lehmanniana) in the southwestern United
States”
Ecological niche modeling and climate-matching have gained momentum recently
for predicting potential invasions (Hoffman 2001, Peterson and Vieglais 2001, Welk et
al. 2002, Peterson 2003). These methods are based on the assumption that a species’
ecological niche can be described as the n-dimensional hypervolume of environmental
conditions under which it is able to maintain populations without immigration (Grinnell
1917). Almost exclusively, species’ invasions outside of their native ranges have been
predicted with this method using locations where it is confirmed to exist, or known
presence points, from their native ranges. However, using locations from the native range
assumes the same factors determine the distribution of the species in the invaded range.
Depending on the species’ introduction history, this assumption may not be appropriate.
In addition, acquiring data from native ranges is often very expensive and time-
16
consuming and is not always feasible. Finally, the scale at which species’ potential
invaded-range distributions can be modeled using native-range data is often too coarse
for specific management decisions. The purpose of this study was to investigate whether
invaded-range models could work as well or better than native-range models in the case
of a purposely-introduced species.
We used confirmed presence points from the native range, southern Africa, and
the invaded range, the southwestern United States, to predict the potential distribution of
the perennial bunchgrass Eragrostis lehmanniana Nees (Lehmann lovegrass), in its
invaded range in the U.S. The two models showed strong agreement for the area
encompassed by the presence points in the invaded range, and offered insight into the
overlapping but slightly different ecological niche occupied by the introduced grass in the
invaded range. Regions outside of the scope of inference showed less agreement between
the two models. Eragrostis lehmanniana was selected via seeding trials before being
planted in the United States and therefore represents an isolated genotype from the
native-range population. The results of this study demonstrate that geographic
distributions of invading species built on points occupied in the invaded range may
perform as well or better than those developed from the native range, at least for the
region encompassed by confirmed presence points in the invaded range. In addition,
predictions based on invaded-range points can offer insight into the environmental
conditions tolerated by the invader and inconsistencies in the ecological niche between
native to invaded ranges. Models created from locations in both the invaded and native
17
ranges can lead to a more complete understanding of an introduced species’ potential for
spread, especially in the case of anthropogenic selection.
Appendix B – “Modeling future potential distributions of Eragrostis lehmanniana
(Lehmann lovegrass) in Arizona, USA”
Climate plays a large role in determining the distribution of plant species and
communities (Huntley and Webb 1988, Wright et al. 1993). Significant increases in
precipitation and temperature as well as shifts in the seasonality of precipitation are
expected for the southwestern United States in the coming decades (IPCC 2001),
suggesting major shifts in community and species distributions. The structure and
diversity of ecosystems are expected to undergo considerable changes (IGBP 1988,
Watson et al. 1996), potentially having detrimental impacts on ecosystem stability
(Mooney 1997). Such reshuffling could favor nonindigenous species, especially those
with the capacity to disperse rapidly and compete strongly for resources.
Because E. lehmanniana has been observed to invade native communities in
southern Arizona and disperse widely without the aid of disturbance (Cable 1971, Anable
et al. 1992), this species can be expected to spread into new areas with a changing
climate. Future spread predictions can facilitate early detection and control measures,
thereby keeping recently established populations in check. In this study I explored the
potential effects of climate change on the distribution of E. lehmanniana in Arizona using
ecological niche modeling. In addition to the prediction maps, one benefit of this
18
approach was that model parameters were compared directly with key limiting ecological
parameters known to affect populations of the species in question.
Based on predictions of popular global-scale general circulation models, 28
climate-change scenarios were created for this study by modifying existing long-term
averages of climatic variables. Using the Genetic Algorithm for Rule-set Prediction
model (GARP; Stockwell and Peters 1999), I modeled the ecological niche for E.
lehmanniana. This niche, characterized by a set of rules, was then projected onto each of
the 28 climate change scenarios to represent potential habitat under potential future
conditions. Future scenarios show the distribution of E. lehmanniana progressively
shrinking in the southeastern and northwestern portions of the state and increasing in the
northeastern portion of the state with increasing temperatures and precipitation. Key
shifts occur under scenarios with increases in summer and winter precipitation of 30% or
more, and increases in summer maximum and winter minimum temperatures of at least
2oC. Spread to new environments will be limited by the ability of viable E. lehmanniana
seed to reach new habitats.
Appendix C – “Community-level and seed bank response to the removal of Eragrostis
lehmanniana (Lehmann lovegrass) at three sites in southeastern Arizona”
Often, the response to the discovery of an exotic species with detrimental impacts
is control or eradication, based in part on the success of other large-scale eradication
efforts (Myers et al. 2000). However, eradication may not lead to the recovery of the
affected system, as some species alter systems so greatly that they are shifted to a
19
different stable state (Westoby et al. 1989, Zavaleta et al. 2001). Alternatively, the effect
of such a management action could actually exacerbate the problem. For example,
efforts to control and reduce a nonnative plant species could initiate unexpected and
undesirable feedbacks, such as favoring other nonnative species or drastically reducing
plant cover to the detriment of wildlife and soil resources.
Zavaleta et al. (2001) argue for pre-eradication assessments to ascertain that a
removal will yield the expected results and minimize unwanted effects. Assessments can
help determine a community’s response to the removal of a nonnative species, especially
when the nonnative species is dominant. Such assessments should address direct and
indirect effects of removing the nonnative species, including response of both plants and
animals in the system. A second component of a pre-eradication assessment is an
assessment of the potential for reestablishment by members of the native community.
Seed bank studies provide some insight into a site’s potential for recovery when the
exotic species being removed is a plant. The response of animals to the removal of E.
lehmanniana was beyond the scope of this study, and not included.
This study was a pre-eradication assessment of Eragrostis lehmanniana Nees
(Lehmann lovegrass), a perennial bunchgrass that has been associated with decreased
plant and animal species richness (i.e., Cable 1971, Bock et al. 1986) and plant species
diversity (Geiger and McPherson 2004). In addition to decreased species richness, E.
lehmanniana has been associated with the alteration of ecosystem processes (Cable 1971,
Bock et al. 1986, Williams and Baruch 2000), modification of community composition
20
(Anable et al. 1992), and changes in fire regimes (Ruyle et al. 1988, Burquez and
Quintana 1994, Biedenbender and Roundy 1996).
The results of this study indicate how semi-desert grassland communities in
southeastern Arizona may respond to the removal of this species. This was addressed
through a removal experiment of E. lehmanniana at three sites dominated by the
nonnative grass. Field studies offer the best representation of the effects of a wide-scale
eradication effort, as study sites are subject to natural variations in temperature and
precipitation and reflect its disturbance history. Because seed banks represent, in part, a
site’s potential for recovery following removal of aboveground vegetation, a second
objective of this study was to characterize the seed banks at these sites. This was
accomplished through a greenhouse study of seed bank samples.
The results of this study suggest that the response of a site to the removal of a
dominant nonnative grass varies between sites. In this study, a site with a history of
intensive agriculture showed no strong response to the removal of a dominant nonnative
species. Two other sites with a history of livestock grazing demonstrated similar strong
responses to the removal, with large increases in native plant cover, increases in species
richness and no evidence of “new” nonnative species replacing removed species. The
findings from the latter two sites are consistent with other experiments that removed
dominant nonnative plants (Farnsworth and Meyerson 1999, Morrison 2002). The seed
banks of these three sites indicate that the potential for the nonnative plant to recover
from seed is limited, as the seed bank declines dramatically without the input of seed
rain.
21
Community composition of the treated plots at the three sites varied considerably,
and reflected common species near each site. Seed-bank studies revealed a small number
of additional species not observed in the aboveground vegetation for each site. These
patterns suggest that removing the nonnative plant from a site would result in the site’s
conversion to an ecosystem dominated by native plants. However, if the goal is
restoration of the site to a pre-invasion grassland, it may be advisable to undertake
restoration measures in conjunction with removal efforts. Seeds of native perennial
species may promote transition from an annual-dominated site to a more diverse,
perennial-dominated community.
Summary
Nonnative species introductions have been taking place as long as humans have
inhabited the planet (Mack 2001), some with devastating impacts. Some of the most
problematic and widespread plant introductions have been grasses in ecosystems
worldwide (Whisenant 1990, D’Antonio and Vitousek 1992, Weber 1997, White et al.
1997, D’Antonio et al. 1998). Introduced grasses directly affect resource availability
(Gordon et al. 1989, Williams and Hobbs 1989), alter fire frequency (D’Antonio and
Vitousek 1992, Brooks et al. 2004), and impact nitrogen and carbon cycling (Wedin and
Tilman 1990, Fisher et al. 1994, Johnson and Wedin 1997). The findings of these three
studies have specific implications for communities already invaded or expressing the
potential for invasion by Eragrostis lehmanniana, but also demonstrate important
22
findings that are generalizable to grasslands worldwide. Overall, the main findings can be
summarized in the following points:
•
Geographic distributions of Eragrostis lehmanniana built on points occupied in
the invaded range may perform as well or better than those developed from the
native range, at least for the region encompassed by confirmed presence points in
the invaded range.
•
Eragrostis lehmanniana is not likely to disappear under a changing climate. This
plant apparently is still spreading to available habitat, and likely will continue to
expand its range as long as seed dispersal vectors facilitate its spread to new
habitats. However, major range shifts are not predicted to occur until average
summer maximum and winter minimum temperatures increase by at least two
degrees and summer and winter precipitation increase by at least 30%.
•
The response of a site to the removal of a dominant nonnative grass varies
between sites, based on land use history, edaphic conditions, and seasonal
precipitation patterns. A site with a history of intensive agriculture responded
little to the removal of E. lehmanniana, but two other sites with similar histories
of livestock grazing shifted to communities dominated by native species.
The three studies presented herein underscore the complexities of nonnative
species introductions. Effective management solutions require a thorough understanding
of invading species’ impacts and likelihood of persistence. These studies demonstrate the
tremendous need for improved awareness and attention to nonindigenous species, as the
23
rate of introductions is predicted to increase into the future (National Research Council
2002).
24
REFERENCES
Anable, M.E., M.P. McClaran, and G.B. Ruyle. 1992. Spread of introduced Lehmann
lovegrass Eragrostis lehmanniana Nees. in Southern Arizona, USA. Biological
Conservation 61:181-188.
Biedenbender, S.H. and B.A. Roundy. 1996. Establishment of native semidesert grasses
into existing stands of Eragrostis lehmanniana in Southeastern Arizona.
Restoration Ecology 4:155-162.
Bock, C.E., J.H. Bock, K.L., Jepson, J.C. and Ortega. 1986. Ecological effects of planting
African lovegrasses in Arizona. National Geographic Research 2:456-463.
Brooks, M.L., C.M. D’Antonio, D.M. Richardson, J.B. Grace, J.E. Keeley, J.M.
DiTomaso, R.J. Hobbs, M. Pellant, and D. Pyke. 2004. Effects of invasive alien
plants on fire regimes. BioScience 54:677-688.
Burquez, A. and M. Quintana. 1994. Islands of biodiversity: ironwood ecology and the
richness of perennials in a Sonoran Desert biological reserve. Pages 9-27 in G.P.
Nabhan and J.L. Carr (Editors), Ironwood: An Ecological and Cultural Keystone
of the Sonoran Desert, Occasional Papers in Conservation Biology, Occasional
Paper No. 1,Washington, DC.
Byers, J.E., S. Reichard, J.M. Randall, I.M. Parker, C.S. Smith, W.M. Lonsdale, I.A.E.
Atkinson, T.R. Seastedt, M. Williamson, E. Chornesky, and D. Hayes. 2002.
Directing research to reduce the impacts of nonindigenous species. Conservation
Biology 16:630-640.
Cable, D.R. 1971. Lehmann lovegrass on the Santa Rita experimental range, 1937-1968.
Journal of Range Management 24:17-21.
Chapin, F.S., III., H.L. Reynolds, C.M. D'Antonio and V.M. Eckhart. 1996. The
functional role of species in terrestrial ecosystems. Pages 403-430 in B. Walker
and W. Steffan, eds. Global Change in Terrestrial Ecosystems. Cambridge:
Cambridge University Press.
Cox, J. R. and G.B. Ruyle. 1986. Influence of climatic and edaphic factors on the
distribution of Eragrostis lehmanniana Nees in Arizona, USA. Journal of the
Grassland Society of South Africa 3: 25-29.
Crider, F.J., 1945. Three introduced lovegrasses for soil conservation. United States
Department of Agriculture Circular No. 730.
25
D’Antonio, C.M. and P.M. Vitousek. 1992. Biological invasions by exotic grasses, the
grass/fire cycle and global change. Annual Review of Ecology and Systematics
23:63-87.
D’Antonio, C.M., H.R. Flint, M. Mack, D. Hitchcock, and P.M. Vitousek. 1998. The
response of native species to removal of invasive exotic grasses in a seasonally
dry Hawaiian woodland. Journal of Vegetation Science 9:699-712.
Farnsworth, E.J. and L.A. Meyerson. 1999. Species composition and inter-annual
dynamics of a freshwater tidal plant community following removal of the invasive
grass, Phragmites australis. Biological Invasions 1:115-127.
Fisher, M.J., I.M. Rao, M.A. Ayarza, C.E. Lascano, J.I. Sanz, R.J. Thomas, and R.R.
Vera. 1994. Carbon storage by introduced deep-rooted grasses in the South
American savannas. Nature 371:236-238.
Geiger, E.L. and G.R. McPherson. 2004. Maintaining grasslands: the interaction of fire,
climate, and nonnative species. Ecological Society of America Annual Meeting,
Portland, OR, July 31 – Aug 6.
Gordon, D.R., J.M. Welker, J.W. Menke, and K.J. Rice. 1989. Competition for soil water
between annual plants and blue oak (Quercus douglasii) seedlings. Oecologia
79:533-541.
Grinnell, J., 1917. Field tests of theories concerning distributional control. American
Naturalist 51:115-128.
Hiebert, R.D. 1997. Prioritizing invasive plants and planning for management. Pages
195-212 in J.O. Luken and J.W. Theiret, eds. Assessment and management of
plant invaisions. New York: Springer-Verlag.
Hoffman, M.H. 2001. The distribution of Senecio vulgaris, capacity of climatic range
models for predicting adventitious ranges. Flora 196:395-403.
Huntley, B. and T. Webb III. 1988. Vegetation History. Kluwer Academic Publishers:
Dordrecht.
IGBP 1988. Toward an understanding of global change. Washington DC: National
Academic Press.
International Panel on Climate Change. 2001. Climate Change 2001: Summary for
Policymakers. A Report of Working Group I of the Intergovernmental Panel on
Climate Change. Cambridge: Cambridge University Press.
26
Johnson, N.C. and D.A. Wedin. 1997. Soil carbon, nutrients and mycorrhizae during
conversion of dry tropical forest to grassland. Ecological Applications 7:171-182.
Mack, R.N. 2001. Motivations and consequences of the human dispersal of plants. Pages
23-34 in J.A. McNeely, ed. The great reshuffling: human dimensions of invasive
alien species. IUCN The World Conservation Union.
Mack, M.C. and C.M. D’Antonio. 1998. Impacts of biological invasions on disturbance
regimes. Trends in Ecology and Evolution 13:195-198.
Mooney, H.A. 1997. Ecosystem function of biodiversity: the basis of the viewpoint.
Pages 341-354 in T.M. Smith, H.H. Shugart and F.I. Woodward, eds. Plant
Functional Types: Their Relevance to Ecosystem Properties and Global Change.
IGBP Book Series 1. Cambridge: Cambridge University Press.
Morrison, J.A. 2002. Wetland vegetation before and after experimental purple loosestrife
removal. Wetlands 22:159-169.
Myers, J.H., D. Simberloff, A.M. Kuris, and J.R. Carey. 2000. Eradication revisited:
dealing with exotic species. Trends in Ecology and Evolution 15:316-320.
National Research Council. 2002. Predictiong Invasions of Nonindigenous Plants and
Plant Pests. Washington, DC: National Academy Press.
Oberhauser, K. and A.T. Peterson. 2003. Modeling current and future potential wintering
distributions of eastern North American monarch butterflies. Proceedings of the
National Academy of Sciences 100:14063-14068.
Office of Technology Assessment. 1993. Harmful non-indigenous species in the United
States, OTAF-565. Washington, D.C.: U.S. Government Printing Office,
September 1993.
Pearson, R.G. and T.P. Dawson. 2003. Predicting the impacts of climate change on the
distribution of species: are bioclimatic envelope models useful? Global Ecology
& Biogeography 12:361-371.
Pimentel, D., Lach, L., Zuniga, R. and Morrison, D., 1999. Environmental and Economic
Costs Associated with Non-indigenous species in the United States. Available online: http://www.news.cornell.edu/releases/Jan99/species_costs.html (Accessed
August 2004).
Peterson, A.T. 2003. Predicting the geography of species’ invasions via ecological niche
modeling. The Quarterly Review of Biology 78:419-433.
27
Peterson, A.T. and J. Shaw. 2003. Lutzomyia vectors for cutaneous leishmaniasis in
Southern Brazil: ecological niche models, predicted geographic distributions, and
climate change effects. International Journal for Parasitology 33:919-931.
Peterson, A.T. and D.A. Vieglais. 2001. Predicting species invasions using ecological
niche modeling: New approaches from bioinformatics attack a pressing problem.
BioScience 51:363-371.
Peterson, A.T., V. S<nchez-Cordero, J. Sober\n, J. Bartley, R.W. Buddemeier and A.G.
Navarro-Sighenza. 2001. Effects of global climate change on geographic
distributions of Mexican Cracidae. Ecological Modelling 144:21-30.
Peterson, A.T., M.A. Ortega-Huerta, J. Bartley, V. S<nchez-Cordero, J. Sober\n, R.H.
Buddemeier, and D.R.B. Stockwell. 2002. Future projections for Mexican faunas
under global climate change scenarios. Nature 416:626-629.
Ruyle, G.B., B.A. Roundy, and J.R. Cox. 1988. Effects of burning on germinability of
Lehmann lovegrass. Journal of Range Management 41:404-406.
Schnase, J.L., J.A. Smith, T.J. Stohlgren, S., Graves, and C. Trees. 2002. Pages 122-124
in Biological invasions: A challenge in ecological forecasting. IEEE International
Geoscience and Remote Sensing Symposium, 24th Canadian Symposium.
Stockwell, D.R.B. and D. Peters. 1999. The GARP modeling system: problems and
solutions to automated spatial prediction. International Journal of Geographic
Information Systems 13:143-158.
Watson, R.T., M.C. Zinyowera, and R.H. Moss. 1996. Climate Change 1995. Impacts,
adaptations and mitigation of climate change: scientific-technical analysis.
Cambridge: Cambridge University Press.
Weber, E.F. 1997. The alien flora of Europe: a taxonomic and biogeographic review.
Journal of Vegetation Science 8:565-572.
Wedin, D.A. and D. Tilman. 1990. Species effects on nitrogen cycling: a test with
perennial grasses. Oecologia 84:433-441.
Welk, E., K. Schubert, and M.H. Hoffman. 2002. Present and potential distribution of
invasive garlic mustard (Alliaria petiolata) in North America. Diversity and
Distributions 8:219-233.
Westoby,M., B. Walker, and I. Noy-Meir. 1989. Opportunistic management for
rangelands not at equilibrium. Journal of Range Management 42:266-274.
28
Whisenant, S. 1990. Changing fire frequencies on Idaho’s Snake River plains: ecological
and management implications. Proceedings Symposium on cheatgrass invasion,
shrub dieoff and other aspects of shrub biology and management. USFS Gen.
Tech. Rep. INT-276, pp. 4-10. US Govt Publications Office.
White, T.A., B.D. Campbell, and P.D. Kemp. 1997. Invasion of temperate grasslands by
a subtropical annual grass across an experimental matrix of water stress and
disturbance. Journal of Vegetation Science 8:847-854.
Wilcove, D.S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying the
threats to imperiled species in the United States. Bioscience 48:607-615.
Williams, D.G. and Z. Baruch. 2000. African grass invasion in the Americas: ecosystem
consequences and the role of ecophysiology. Biological Invasions 2:123-140.
Williams, K. and R. Hobbs. 1989. Control of shrub establishment by springtime soil
water availability in an annual grassland. Oecologia 81:62-66.
Wright, H., J. Kutzbach, T. Webb III, W. Ruddiman, F. Street-Perrot and P. Bartlein, eds.
1993. Global climates since the last glacial maximum. Minneapolis: University of
Minnesota Press.
Zavaleta, E.S., R.J. Hobbs, and H.A. Mooney. 2001. Viewing invasive species removal in
a whole-ecosystem context. Trends in Ecology and Evolution 16:454-459.
29
APPENDIX A
CAN AN INVADING SPECIES’ DISTRIBUTION BE PREDICTED USING DATA
FROM ITS INVADED RANGE? LEHMANN LOVEGRASS (ERAGROSTIS
LEHMANNIANA) IN THE SOUTHWESTERN UNITED STATES
THERESA M. MAU-CRIMMINS, HEATHER R. SCHUSSMAN, AND ERIKA L.
GEIGER
(In review at Ecological Modelling)
30
Abstract
Predictions of species invasions are often made using information from their native
ranges. Acquisition of native range information can be very costly and time-consuming
and in some cases may not reflect conditions in the invaded range. Using information
from the invaded range can enable models to be built more rapidly and at finer
geographic resolutions than using information from a species’ native range. We used
confirmed presence points of the perennial bunchgrass Eragrostis lehmanniana Nees,
(Lehmann lovegrass) from its native range in southern Africa, and its invaded range, the
southwestern United States, to predict the potential distribution, in its invaded range in
the U.S. The two models showed strong agreement for the area encompassed by the
presence points in the invaded range and offered insight into niche differentiation in the
invaded range. Regions outside of Arizona showed less agreement between the two
models. Eragrostis lehmanniana was selected via seeding trials before being planted in
the United States and therefore represents an isolated genotype from the native
population. Models built using confirmed presence points from the invaded range can
provide insight into how the selected genotype is expressed on the landscape and
considers influences not present in the native range. Models created from locations in
both the invaded and native ranges can lead to a more complete understanding of the an
introduced species’ potential for spread, especially in the case of anthropogenic selection.
31
Keywords: Genetic Algorithm for Rule-set Prediction; distribution modeling, Eragrostis
lehmanniana; Invasive plants; Invaded-range models.
1. Introduction
Non-indigenous invasive species are a major threat to native species diversity and
ecosystem function, causing economic impacts estimated at more than $100 billion
annually (Pimentel et al. 1999). Invasive species have been called the “single most
formidable natural disaster threat of the 21st century” (Schnase et al. 2002). Early
detection of invaders is critical: predictive models can enable early detection by focusing
research efforts on areas most likely to be impacted.
Ecological niche modeling and climate-matching has gained momentum recently for
predicting potential invasions (Hoffman 2001, Peterson and Vieglais 2001, Welk et al.
2002, Peterson 2003). These approaches are based on the assumption that a species’
ecological niche can be described as the n-dimensional hypervolume of environmental
conditions under which it is able to maintain populations without immigration (Grinnell
1917). Almost exclusively, species’ invasions outside of their native ranges have been
predicted with this method using locations where it is confirmed to exist (known presence
points) from their native ranges. However, using locations from the native range assumes
the same factors will determine the distribution range of the species from the invaded
locations. Depending on the species’ introduction history, this assumption may not be
32
appropriate (Hierro et al. 2005). In addition, acquiring data from native ranges is often
very expensive and time-consuming and not always feasible. Finally, the scale at which
species’ potential invaded-range distributions can be modeled using native-range data is
often too coarse for specific management decisions.
Because of the drawbacks associated with making predictions for an invasion using
native range information, the question of whether confirmed presences in a species’
invaded range can be used to successfully predict its eventual distribution is fertile
ground for research. A recent study tested whether the invaded-range distribution of
Lythrum salicaria (Purple loosestrife) in North America could be predicted accurately
from known presence points in the invaded range (Welk 2004). The results of this study
suggested that such an approach could be successfully implemented only 100-150 years
post-introduction, arguing that this amount of time was necessary to amass confirmed
presence locations representing the full range of climatic values to which the invader was
adapted (Welk 2004). The purpose of the present study was to investigate whether
invaded-range models could work as well or better than native-range models in the case
of a purposely-introduced species.
To explore this idea, we selected Eragrostis lehmanniana, a perennial bunchgrass with a
unique introduction history to the southwestern United States. In the 1930s, several
accessions of E. lehmanniana were brought from southern Africa to Arizona to
counteract low plant cover and highly eroded soils resulting from decades of overgrazing
33
and drought (Crider 1945). Seed from a single cultivar (A-68), selected for its droughttolerance and high seed production, was produced and distributed widely in Arizona and
neighboring states from the 1930s through the 1980s (Cox and Ruyle 1986). Geographic
range predictions for E. lehmanniana made using invaded-range points may be superior
to those using native-range points because E. lehmanniana populations in the U.S.
represent a subset of genetic variation present in African populations (Schussman 2002).
Further, very little genetic change is expected because the individuals brought to Arizona
and used to produce seed for erosion control were apomictic (Burson and Voigt 1996).
Apomictic reproduction allows for the production of seeds that are genetically identical to
the maternal plant. Given its introduction history and reproductive biology, we believe
that E. lehmanniana represents an ideal candidate for investigating the variability in
models predicted using native and invaded-range information for species with little
genetic variability.
We built two models to predict the range of E. lehmanniana to explore differences in
distribution predicted using two different sets of input data. The first model predicted the
potential distribution of E. lehmanniana in the United States using the ecological niche
predicted from its native range in South Africa; the second model predicted the potential
distribution of E. lehmanniana in the U.S. using the ecological niche predicted using
environmental data from areas known to be invaded within the United States. The results
of these models were then compared to independent test points in the invaded range.
34
2. Materials and methods
2.1 Distribution data
We collected 350 point locations of E. lehmanniana within South Africa, Namibia,
Botswana, and Lesotho from the Southern African Botanical Diversity Network’s
PRECIS database, the Natal Herbarium, the C.E. Moss Herbarium, and Kruger National
Park. To minimize spatial autocorrelation, we randomly selected a subset of 200 welldistributed points. The native-range study area for predictions and sensitivity analysis
was restricted to a rectangular grid containing southern Africa, from 16oS to 36oS and
11oE to 38oE.
We obtained over 1,000 localities of species’ occurrences within Arizona and New
Mexico from several sources including the Santa Rita Experimental Range, Bureau of
Land Management, The Nature Conservancy, U.S. Department of Defense, U.S. Fish and
Wildlife Service, U.S. Forest Service, U.S. Geological Survey, and the U.S. National
Park Service. We randomly selected 100 of the 1000 tightly-clustered presence points to
minimize spatial autocorrelation. The invaded range study area was restricted to the
southwestern United States, from 23o N to 51o N and 66o W to 126o W, then subsetted to
the region roughly bounded by the state of Arizona, from 106o W to 115o W and 31o N to
37o N.
2.2 Environmental data
35
The base environmental data consisted of 19 global geographic coverages. Elevation,
slope, aspect, flow direction, flow accumulation, and topographic index (U.S. Geological
Survey 2001) were generalized to 0.1-degree resolution from 0.01-degree datasets.
Climate data averaged for the period 1961-1990 were resampled from 0.5-degree datasets
to 0.1-degree datasets; these data included mean annual precipitation; maximum,
minimum, and mean annual temperatures; wet days; vapor pressure; solar radiation; and
frost days (International Panel on Climate Change 2004). We created grids representing
seasonal precipitation by summing monthly precipitation averages for the period 19611990 (IPCC 2004). Seasons were defined as winter (DJFM), spring (AMJ), summer
(JAS), and autumn (ON) in the U.S. and winter (JAS), spring (ON), summer (DJFM), and
autumn (AMJ) in southern Africa (Cox et al. 1988). Season definitions differ from the
convention of even three-month seasons to capture the unique seasonality of precipitation
and temperature in Arizona and southern Africa. Datasets representing soil texture in the
upper soil horizon (top 1-m) were downloaded from the Oak Ridge National Laboratory
Distributed Active Archive Center (Post and Zobler 2000) and resampled from 0.5degree to 0.1-degree datasets.
2.3 Modeling
The Genetic Algorithm for Rule-Set Prediction (Stockwell and Noble 1992, Stockwell
and Peters 1999) is a niche-based model receiving wide application. GARP is an iterative
artificial intelligence-based approach that includes several inferential tools. This model
has proven successful at predicting species’ potential distributions under a wide variety of
36
conditions (Peterson and Cohoon 1999, Peterson et al. 1999, Peterson et al. 2001,
Peterson et al. 2002a, Peterson et al. 2002b, Peterson et al. 2002c, Godown and Peterson
2000, Sánchez-Cordero and Martinez-Meyer 2000, Peterson 2001, Feria and Peterson
2002, Stockwell and Peterson 2002a, Peterson 2002b).
The GARP model predicts species’ environmental niches by identifying non-random
relationships between environmental characteristics of known presence localities in
comparison with the entire study region. Known presence points are divided into training
and test data sets. GARP works in an iterative process of rule selection, evaluation,
testing, and incorporation or rejection to create a rule-set that best represents the
environmental conditions under which the species is found (Peterson et al. 1999). First,
GARP chooses a method from a set of possibilities (e.g. logistic regression, bioclimatic
rules) and applies it to the data. Then, a rule is generated and its accuracy is evaluated via
test points intrinsically re-sampled from both the known study region and from the study
region as a whole. The change in predictive accuracy from one iteration to the next is
used to select among rules for the final model. Rules may change in ways similar to the
ways DNA mutates, hence, the name “genetic algorithm.” As implemented here, the
algorithm runs either for 1,000 iterations or until convergence. The final rule-set, or
ecological niche model, is then projected onto a digital map as the species’ potential
geographic distribution and imported into ArcView 3.2 (ESRI 1999) using the Spatial
Analyst extension for visualization.
37
2.4 Native and Invaded Range Modeling
The general steps for the modeling were:
1. Build models using native-range (southern African) datasets, select subset of 10
best models
2. Project rules onto native range (southern Africa) and invaded range (southwestern
U.S.)
3. Build models using invaded-range (southwestern U.S.) datasets, select subset of
10 best models
4. Project rules onto invaded range (southwestern U.S.) and native range (southern
Africa)
Specifically, we produced 300 replicate models of E. lehmanniana’s ecological niche
using the subset of training points for southern Africa (native range). For each model,
points were randomly split into two equally sized training and testing datasets of
available occurrence points. To choose the best subset of the 300 native range models, we
adopted a best-subsets selection procedure (Anderson et al. 2003, Peterson et al. 2003).
Following this method, we selected the best subset of models by eliminating all models
that had non-zero omission error based on independent test points, calculated the median
area predicted present among these zero-omission points, and then identified the ten
models closest to the overall median area predicted. These ten models were summed to
create a final output grid of model agreement, ranging from 0 (areas not predicted present
by any of the ten models) to 10 (areas predicted present by all ten models). The rule sets
38
from these ten models were projected onto the same environmental grids for the United
States (invaded range) to generate predictions of E. lehmanniana distribution. This
procedure was repeated using the environmental base layers and unique presence points
collected within the southwestern U.S. and then projected onto both the South African
and U.S environmental grids to determine how well GARP could predict E.
lehmanniana’s ecological niche within southern Africa and the southwestern United
States and Mexico.
2.5 Model Testing and Visualization
We tested the predictive power of the models by overlaying extrinsic test data and
tallying observed correct predictions. The proportion of the total study extent predicted as
present (occupied by the species) multiplied by the number of extrinsic test data points is
used as a random expectation of successful prediction points if no non-random
association existed between prediction and test points (Peterson and Shaw 2003).
Following Peterson and Shaw (2003), we implemented a chi-squared test (1 df) to test the
significance of the departure from random expectations. We used 147 randomly selected
extrinsic points to test the native range models, and 183 randomly selected extrinsic
points to test the invaded range models.
To identify environmental dimensions important for defining E. lehmanniana’s
geographic potential, we used a series of sequential jackknife manipulations in which all
possible combinations of a reduced set (e.g., N-1) of N environmental coverages were
39
used to generate native range models. We assessed model quality by exploring
correlation between variable inclusion and omission error (Peterson and Cohoon 1999).
Variables that were positively correlated with improvement in avoiding omission error
were considered to be most important in defining E. lehmanniana’s environmental niche.
Using the subsets of known presence points in Arizona and South Africa selected for the
models, a correlation-based principal components analysis (PCA) was performed on the
environmental variables associated with the points using PRIMER v5 (PRIMER-E Ltd.
2001). Prior to analysis, skewed variables were transformed using the log or square-root
transformation.
3. Results
3.1 Model Predictivity and Influential Variables
Ecological niche models developed in this study were highly predictive of the distribution
of E. lehmanniana based on random subsets. All of the best-subsets models were
statistically significant when compared with random expectations (X2 tests, df = 1, all
best-subsets models P < 0.001).
Jackknife manipulations of the different environmental coverages (Peterson and Cohoon
1999) suggested that the following variables were critical in constituting the ecological
niche of E. lehmanniana in its native range: spring, summer, and fall precipitation;
radiation; silt; annual minimum temperature; annual number of freezing days; and
40
elevation. In the invaded range, slope, aspect, fall precipitation, and elevation appeared to
be influential in defining the plant’s niche.
3.2 Model Output Comparison and Visualization
The two approaches used to predict the geographic distribution of E. lehmanniana in the
United States produced results with rather dissimilar patterns, except for the region
surrounding the known presence points (Figure 1). The results of the two methods had
strongest agreement for the region surrounding the known presence points in the U.S.,
centered on Arizona. The strongest disagreements occurred in Texas, California, and
Mexico, regions not represented by invaded range points. Models generated from nativerange points predict much greater distributions for these areas than models created using
invaded-range points. Areas in the northeastern and southwestern portions of Arizona
were more strongly predicted by the native-range model (i.e., possessing higher values on
the final model output grid), and the central portion of the state was more strongly
predicted by the invaded-range model. The two models also showed strong disagreement
for portions of Mexico.
Using the invaded range rules to predict E. lehmanniana’s distribution in its native range
provides insight into the different rules being formulated in the two regions. When rules
generated from invaded range points are projected onto the native range, a very small set
of the known distribution is predicted. The area predicted appears in the south-central
portion of South Africa, characterized by relatively warm, dry conditions similar to
41
Arizona. Figure 2 displays the area predicted occupied by E. lehmanniana using points
from the native range (top) and from the invaded range (bottom).
The first four components from the PCA of environmental variables for confirmed
presence points in the native and invaded ranges accounted for 80.9% of the variance
(36.4%, 20.7%, 15.4%, and 8.5%, respectively; Table 1). Component 1 loaded most
highly on temperature variables. Precipitation variables contributed the highest loadings
to Component 2. Components 3 and 4 are a mix of soil variables, radiation, and
precipitation variables. Separation of known presence points in Arizona from known
presence points in southern Africa occur primarily along the second axis (Figure 3).
4. Discussion
Predictions of species invasions are most commonly made using information from their
native ranges (Hoffman 2001, Peterson and Vieglais 2001, Welk et al. 2002, Peterson et
al. 2003, Peterson et al. 2004). Acquisition of native range information can be very costly
and time-consuming, and in some cases infeasible, involving extensive literature searches
of sources in different languages. Records must be geographically referenced, which is
rare for many herbaria. Peterson et al. (2003) reported spending 2 months obtaining
native-range records, versus 1 hour obtaining invaded-range records for a study similar to
this one.
42
Another advantage of invaded-range models is that they usually can be built using finerscale input data layers. To build models in the native range and project the resulting rules
onto the invaded range requires the same environmental layers for the two regions.
Currently, most environmental layers available on a global scale are coarse, on the order
of 0.1-degree to 1-degree cells. Limiting models to only one continent enables finer-scale
datasets specific to that region to be included in the models.
Few studies have explored the performance of invaded- versus native-range datasets for
predicting invaded-range distributions. These that have, suggested that species will
occupy an overlapping but different set of environmental conditions in an invaded range
than in their native range (Malanson et al. 1992). Kriticos and Randall (2001) suggested
that a species’ invaded range is predicted best using information from a second invaded
range; they reasoned that in invaded ranges the fundamental niche may be more fully
realized than in the native range where the species is constrained by competition and
dispersal barriers. Welk (2004) tested the use of invaded range points to predict the
distribution of the invasive Lythrum salicaria in North America and suggested a method
combining the use of invaded and native range information would be most insightful.
E. lehmanniana constitutes a special case for invasion prediction, given its unusual
introduction scenario. Seed traceable to a limited number of apomictic individuals was
widely distributed and planted in the southwestern U.S., leading to an invading
population with low genetic variability and few mechanisms to evolve. For these reasons,
43
it seems logical to build predictive models using both native and invaded range points to
explore the potential distribution of E. lehmanniana. Building models from the species’
introduced range, or a combination of both native and introduced ranges, may be the most
appropriate approach for purposely-introduced plants. This is especially suitable when
specific traits were selected for during the introduction, for plants with limited genetic
variation due to asexual reproduction, or for plants where a limited number of individuals
were introduced.
4.1 Model visualization and interpretation
Predictions made of the distribution of E. lehmanniana in the U.S. show similar patterns
for Arizona and the region surrounding known presence points in the U.S.; the largest
differences occur in Texas, California, and New Mexico as well as Mexico. Models
created using native range points predict much greater distributions for these areas than
models created using invaded range points. Spatial coverage of known presence points
used in this study come from sites in Arizona, where confirmed presence points are most
readily available. Because of the extent of the known presence points used to build and
test the models, inference is approximately limited to the state of Arizona. Currently,
Arizona is the state in the U.S. most heavily impacted by E. lehmanniana. Within
Arizona, areas in the northeastern and southwestern portions of the state are more
strongly predicted (i.e., possessing higher values on the final model output grid) by the
native range model, and the central portion of the state is more strongly predicted by the
invaded range model.
44
The disagreement in the two models’ output could be due to a paucity of points in the
invaded range, as known presence points used in this study represent only Arizona, where
confirmed presence points are most readily available. However, the disagreement in the
two models’ output is more likely due to differences in the ecological niches realized by
the native-range populations and U.S. populations of E. lehmanniana. The results of the
PCA depict the range of environmental conditions occupied by E. lehmanniana in both
its native and invaded ranges. Points in Arizona occupy an overlapping but slightly
different location in multivariate space than those in E. lehmanniana’s native range
(Figure 3), which coincides with the slight disagreement in the two models for Arizona.
This suggests that the niche occupied by E. lehmanniana in Arizona is narrower and
slightly different than that occupied in its native range. This conclusion is supported by
the projection of invaded range niche rules onto the native range. Using invaded-range
rules, only a small portion of the entire distribution is represented, presumably
representing only a portion of the ecological niche available in its native range (Figure 2).
The invaded-range model predicts occupation under conditions of higher temperature and
lower total precipitation as well as lower precipitation in all seasons than the native-range
model, which argues against E. lehmanniana’s occupation of Texas and California.
Eragrostis lehmanniana was originally planted in portions of Texas and New Mexico
with the A-68 accession, but disappeared within five years of planting (Cox and Ruyle
1986). Experts in Texas, New Mexico, and California all report that though E.
45
lehmanniana has been observed—and in some cases is intentionally planted—in these
states, it does not appear to substantially encroach on native grasslands relative to other
species (B. Carr and J. Bergan, The Nature Conservancy, Texas Chapter; F. Miller, New
Mexico State Weed Coordinator; B. Rice, The Nature Conservancy Wildland Invasive
Species Team; J. DiTomaso, University of Californa-Davis; D. Johnson, California
Invasive Plant Council, pers. comms.; Barkworth and Capels 2004). These observations
suggest that northern New Mexico, California, and Texas are on the edge of the
ecological niche that the A-68 E. lehmanniana accession occupies, and that the potential
distribution map created using the invaded-range model is most plausible. No information
was available regarding the status of E. lehmanniana in Mexico.
The second axis of the PCA is dominated by precipitation variables including summer
precipitation, spring precipitation, and annual precipitation, all decreasing up the axis.
Based on the partial overlap for the two sets of points along the second axis, it appears
that both temperature and precipitation, particularly seasonality of precipitation, is a key
variable in determining the distribution of U.S. E. lehmanniana populations. Because the
A-68 accession was selected to be especially drought-tolerant, the U.S. E. lehmanniana
may not compete as well under more mesic conditions, as in western Texas and
California. However, the potential for E. lehmanniana to dominate grasslands in these
areas may exist if the conditions in Texas and California grasslands were to become drier.
46
Based on field observations by experts, it appears that the model constructed using
invaded-range points more accurately represents the potential distribution for E.
lehmanniana in Arizona. In addition, building two separate models suggests some
difference in environmental niches occupied by the same species in the two locations,
suggesting genetic variations in the two populations. The results of this study suggest that
models built using both native- and invaded-range points can provide insight into how the
selected genotype is expressed on the landscape. In addition, this study demonstrates that
invaded-range models may work as well or better than native-range models, especially in
the case of a purposely-introduced species. Therefore, species with little genetic variation
and well-known introduction histories may be best modeled with invaded range points,
minimizing the costs and time involved, as well as increasing model accuracy.
Experiments are the best way to test predictions made by models. Predictive models can
offer insight into the potential spread of invasive species, but do not consider factors
affecting species distributions such as biotic interactions. Mechanistic studies addressing
these factors are necessary to augment the predictions made by correlative models. In
addition, incorporation of predictive model output with invasibility models (Shea and
Chesson 2002) could offer insight into areas within the predicted habitat that are most
threatened by invasion.
5. Conclusions
47
The results of this study demonstrate that geographic distributions of invading species
built on points occupied in the invaded range may perform as well or better than those
developed from the native range, at least for the region encompassed by confirmed
presence points in the invaded range. In addition, predictions made using invaded-range
points can offer insight into the environmental conditions tolerated by the invader and
inconsistencies in the ecological niche between native to invaded ranges. The sole use of
invaded-range points to make invaded-range predictions may be the most appropriate
method for modeling distributions of intentionally introduced plants, especially when the
introduction involved intense selection of the plant, as in the case of Eragrostis
lehmanniana. Intense selection, asexual reproduction, and limited introduction numbers
all function to create a nonnative population with an environmental niche that represents
a subset of the native range niche. Since not all invaders can be predicted to represent a
subset of the native range and most likely occupy an overlapping but different niche in
the invaded range, it is likely best to use information from the plants’ native and invaded
ranges.
6. Acknowledgments
This research was supported by the Center for Invasive Plant Management grant ESA 000011,
T&E, Inc, and the University of Arizona/NASA Space Grant. The South African National
Biodiversity Institute is thanked for the use of data from the National Herbarium, Pretoria (PRE)
Computerized Information System (PRECIS). Kruger National Park, Natal Herbarium, and C.E.
Moss Herbarium also kindly provided South African data. The authors thank the following
organizations for providing United States data: Santa Rita Experimental Range; The Nature
48
Conservancy, Arizona Chapter; U.S. National Park Service; U.S. Bureau of Land Management;
U.S. Department of Defense; U.S. Fish and Wildlife Service; U.S. Forest Service; and the U.S.
Geological Survey. M. Crimmins, R. Gimblett, A. Hubbard, and G. McPherson provided useful
suggestions on the manuscript.
49
References
Anable, M.E., McClaran, M.P., and Ruyle, G.B., 1992. Spread of introduced Lehmann
lovegrass Eragrostis lehmanniana Nees. in Southern Arizona, USA. Biol. Conserv.,
61:181-188.
Anderson, R.P., Lew, D. and Peterson, A.T., 2003. Evaluating predictive models of
species’ distributions: criteria for selecting optimal models. Ecol. Modelling, 162:211232.
Austin, M.P. 1994. Data capability, Sub-project 3, Modeling of Landscape Patterns and
Processes Using Biological Data. Division of Wildlife and Ecology, Commonwealth
Scientific and Industrial Research Orientation, Canberra.
Barkworth, M.E. and Capels, K.M., 2004. Intermountain Herbarium. Available on-line:
http://herbarium.usu.edu/webmanual (Accessed August 2004).
Biedenbender, S.H. and Roundy, B.A., 1996. Establishment of native semidesert grasses
into existing stands of Eragrostis lehmanniana in southeastern Arizona. Resto. Ecol.,
4:155-162.
Bock, C.E., Bock, J.H., Jepson, K.L. and Ortega, J.C., 1986. Ecological effects of
planting African lovegrasses in Arizona. Nat. Geog. Res., 2:456-463.
50
Burquez, A. and Quintana, M., 1994. Islands of biodiversity: ironwood ecology and the
richness of perennials in a Sonoran Desert biological reserve. In: G.P. Nabhan and J.L.
Carr (Editors), Ironwood: An Ecological and Cultural Keystone of the Sonoran Desert,
Occasional Papers in Conservation Biology, Occasional Paper No. 1,Washington, DC,
pp. 9-27.
Burson, B.L. and Voigt, P.W., 1996. Cytogenetic relationships between the Eragrostis
curvula and Eragrostis lehmanniana complexes. Intl. J. Plant Sci., 157:632-637.
Cable, D.R., 1971. Lehmann lovegrass on the Santa Rita experimental range, 1937-1968.
J. Range Manage., 24:17-21.
Cox, J. R. and Ruyle, G.B., 1986. Influence of climatic and edaphic factors on the
distribution of Eragrostis lehmanniana Nees in Arizona, USA. J. Grassld Soc. S. Afr., 3:
25-29.
Cox, J.R., Ruyle, G.B., Fourie, J.H., and Donaldson, C., 1988. Lehmann lovegrass—
central South Africa and Arizona, USA. Rangelands, 10:53-55.
Crider, F.J., 1945. Three introduced lovegrasses for soil conservation. United States
Department of Agriculture Circular No. 730.
51
Feria, T.P. and Peterson, A.T., 2002. Using point occurrence data and inferential
algorithms to predict local communities of birds. Diversity Distrib., 8:49-56.
Geiger, E.L. and McPherson, G.R., 2004. Response of semi-desert grasslands dominated
by nonnative perennial grass to fire season. In: Proceedings of Biodiversity and
Management of the Madrean Archipelago II, in press.
Godown, M.E. and Peterson, A.T., 2000. Preliminary distributional analysis of U.S.
endangered bird species. Biodiversity and Conservation 9:1313-1322.
Grinnell, J., 1917. Field tests of theories concerning distributional control. Am. Nat.,
51:115-128.
Hierro, J.L., J.L. Maron, and R.M. Callaway. 2005. A biogeographical approach to plant
invasions: the importance of studying exotics in their introduced and native range.
Journal of Ecology 93:5-15.
Hoffman, M.H., 2001. The distribution of Senecio vulgaris, capacity of climatic range
models for predicting adventitious ranges. Flora, 196:395-403.
52
Intergovernmental Panel on Climate Change. 2004. Climate data archive. IPCC, Geneva,
Switzerland. Available on-line: http://www.ipcc.ch/ (Accessed June 2004).
Kriticos, D.J. and Randall, R.P., 2001. A comparison of systems to analyze potential
weed distributions. In: R.H. Groves, F.D. Panetta, and J.G. Virtue (Editors), Weed Risk
Assessment, CSIRO Publishing, Collingwood, Victoria, Australia, pp. 61-79.
Malanson, G.P., Westman, W.E., and Yan, Y.L., 1992. Realized versus fundamental
niche functions in a model of chaparral response to climatic change. Ecol. Modelling,
64:261-277.
McClaran, M.P. and Anable, M.E., 1992. Spread of introduced Lehmann lovegrass along
a grazing intensity gradient. J. Appl. Ecology, 29:92-98.
Peterson, A.T., 2001. Predicting species’ geographic distributions based on ecological
niche modeling. Condor, 103:599-605.
Peterson, A.T., 2003. Predicting the geography of species’ invasions via ecological niche
modeling. Q. Rev. Biol., 78:419-433.
Peterson, A.T. and Cohoon, K.P., 1999. Sensitivity of distributional prediction algorithms
to geographic data completeness. Ecol. Modelling, 117:159-164.
53
Peterson, A.T. and Shaw, J. 2003. Lutzomyia vectors for cutaneous leighmaniasis in
Southern Brazil: ecological niche models, predicted geographic distributions, and climate
change effects. Int. J. Parasitol., 33:919-931.
Peterson, A.T. and Vieglais, D.A., 2001. Predicting species invasions using ecological
niche modeling: New approaches from bioinformatics attack a pressing problem.
BioScience, 363-371.
Peterson, A.T., Ball, L.G. and Cohoon, K.P., 2002a. Predicting distributions of tropical
birds. Ibis, 144:E27-E32.
Peterson, A.T., Papes, M. and Kluza, D.A., 2003. Predicting the potential invasive
distributions of four alien plant species in North America. Weed Sci., 51:863-868.
Peterson, A.T., Scachetti-Pereira, R. and Hargrove, W.W., 2004. Potential geographic
distribution of Anoplophora glabripennis (Coleoptera: Cerambycidae) in North America.
Am. Midl. Nat., 151:170-178.
Peterson, A.T., Soberòn, J. and Sánchez-Cordero, V., 1999. Conservatism of ecological
niches in evolutionary time. Science, 285:1265-1267.
54
Peterson, A.T., Stockwell, D.R.B. and Kluza, D.A., 2002c. Distributional prediction
based on ecological niche modeling of primary occurrence data. In: J.M. Scott, P.J.
Heglund, M.L. Morrison, J.B. Haufler, M.G. Raphael, W.A. Wall, F.B. Samson (Editors),
Predicting Species Occurrences: Issues of Accuracy and Scale, Island Press, Washington,
DC, pp. 617-623.
Peterson, A.T., Sánchez-Cordero, V., Soberòn, J., Bartley, J., Buddemeier, R.W. and
Navarro-Sighenza, A.G., 2001. Effects of global climate change on geographic
distributions of Mexican Cracidae. Ecol. Modelling, 144:21-30.
Peterson, A.T., Ortega-Huerta, M.A., Bartley, J., Sánchez -Cordero, V., Soberòn, J.,
Buddemeier, R.H. and Stockwell, D.R.B., 2002b. Future projections for Mexican faunas
under global climate change scenarios. Nature, 416:626-629.
Pimentel, D., Lach, L., Zuniga, R. and Morrison, D., 1999. Environmental and Economic
Costs Associated with Non-indigenous species in the United States. Available on-line:
http://www.news.cornell.edu/releases/Jan99/species_costs.html (Accessed August 2004).
Post, W. M., and Zobler, L., 2000. Global Soil Types, 0.5-Degree Grid (Modified
Zobler). Data set. Available on-line: http://www.daac.ornl.gov from Oak Ridge National
Laboratory Distributed Active Archive Center, Oak Ridge, Tennessee, U.S.A (Accessed
June 2004).
55
Ruyle, G.B., Roundy, B.A., and Cox, J.R., 1988. Effects of burning on germinability of
Lehmann lovegrass. J. Range Manage., 41:404-406.
Sánchez -Cordero, V. and Martinez-Meyer, E., 2000. Museum specimen data predict
crop damage by tropical rodents. Proc. Nat. Acad. Sci. USA, 97:7074-7077.
Schnase, J.L., Smith, J.A., Stohlgren, T.J., Graves, S. and Trees, C., 2002. Biological
invasions: A challenge in ecological forecasting. IEEE International Geoscience and
Remote Sensing Symposium, 24th Canadian Symposium, pp. 122-124.
Schussman, H.R. 2002. Genetic variation in the introduced apomictic grass, Eragrostis
lehmanniana Nees, on the Santa Rita Experimental Range in southeastern Arizona, USA.
M.S. Thesis, University of Arizona, Tucson.
Shea, K. and Chesson, P., 2002. Community ecology theory as a framework for
biological invasions. Trends Ecol. Evol., 17:170-176.
Stockwell, D.R.B. and Noble, I.R., 1992. Induction of sets of rules from animal
distribution data: a robust and informative method of data analysis. Math. Comput.
Simul., 33:385-390.
56
Stockwell, D.R.B. and Peters, D., 1999. The GARP modeling system: problems and
solutions to automated spatial prediction. Int. J. Geog. Inf. Sys., 13:143-158.
Stockwell, D.R.B. and Peterson, A.T., 2002a. Controlling bias in biodiversity data. In: J.
M. Scott, P.J. Heglund, M.L. Morrison, J.B. Haufler, M.G. Raphael, W.A. Wall, F.B.
Samson (Editors), Predicting Species Occurrences: Issues of Accuracy and Scale, Island
Press, Washington, DC, pp. 537-546.
Stockwell, D.R.B. and Peterson, A.T., 2002b. Effects of sample size on accuracy of
species distribution models. Ecol. Modelling, 148:1-13.
U.S. Geological Survey, 2001. HYDRO1k elevation derivative database. USGS,
Washington, D.C. Available on-line: http://edcdaac.usgs.gov/gtopo30/hydro/ (Accessed
January 2004).
Welk, E., 2004. Constraints in range predictions of invasive plant species due to nonequilibrium distribution patterns: Purple loosestrife (Lythrum salicaria) in North
America. Ecol. Modelling, In press.
Welk, E., Schubert, K. and Hoffman, M.H., 2002. Present and potential distribution of
invasive garlic mustard (Alliaria petiolata) in North America. Diversity Distrib., 8:219233.
57
Williams, D.G. and Baruch, Z., 2000. African grass invasion in the Americas: ecosystem
consequences and the role of ecophysiology. Biol. Inv., 2:123-140.
Zacharias, P.J.K., 1990. Acock’s Notes: Key Grasses of South Africa. Grassland Society
of Southern Africa, Howick, 77 pp.
58
Table 1. Four-component solution generated by principal components analysis for
predicting distribution of Eragrostis lehmanniana in Arizona, USA.
Variable
Max annual temperature
Annual temperature
Min annual temperature
Summer precipitation
Frost days
Fall precipitation
Elevation
Aspect
Winter precipitation
Wet days
Spring precipitation
Slope
Silt
Sand
Radiation
Annual precipitation
Clay
Eigenvalue
Explained variance (%)
Cumulative variance explained (%)
PC1
0.351
0.332
0.272
-0.145
-0.211
-0.213
-0.173
-0.006
-0.252
-0.314
-0.246
-0.214
-0.229
0.241
0.259
-0.276
-0.170
PC2
-0.086
-0.205
-0.298
-0.469
0.266
-0.256
0.126
-0.025
0.236
-0.272
-0.384
0.162
0.209
-0.094
0.005
-0.327
-0.163
PC3
0.131
0.214
0.269
-0.112
-0.354
-0.227
-0.321
-0.012
0.150
-0.031
-0.002
0.208
0.393
-0.417
-0.310
-0.023
0.283
PC4
-0.175
-0.101
-0.014
-0.011
0.047
0.365
-0.023
0.042
-0.371
-0.124
-0.166
-0.132
0.121
-0.280
0.364
-0.277
0.566
6.19
36.4
36.4
3.51
20.7
57.1
2.62
15.4
72.5
1.44
8.5
80.9
59
a
b
Figure 1. Eragrostis lehmanniana potential distribution using native range data (a) and
introduced range data (b). Known presence points are black dots. Increasingly dark
shades of gray indicate greater model agreement.
60
a
b
Figure 2. Eragrostis lehmanniana potential distribution using native range data (a) and
introduced range data (b). Known presence points are black dots. Increasingly dark
shades of gray indicate greater model agreement.
PC2
61
PC1
Figure 3. Principal components analysis plot of environmental conditions associated with
known Eragrostis lehmanniana presence points in Arizona, USA and southern Africa.
Triangles represent known points in Africa; crosses represent known points in Arizona.
62
APPENDIX B
MODELING FUTURE POTENTIAL DISTRIBUTIONS OF ERAGROSTIS
LEHMANNIANA (LEHMANN LOVEGRASS) IN ARIZONA, USA
THERESA M. MAU-CRIMMINS
(To be submitted to Diversity and Distributions)
63
Abstract
Climate plays a large role in determining the distribution of plant species and
communities (Huntley and Webb 1988). Significant increases in precipitation and
temperature as well as shifts in the seasonality of precipitation are expected for the
southwestern United States in the coming decades, suggesting major shifts in community
and species distributions. The structure and diversity of ecosystems are expected to
undergo considerable changes (IGBP 1988, Watson et al. 1996), potentially having
detrimental impacts on ecosystem stability (Mooney 1997). Such reshuffling could favor
nonindigenous species, especially those with the capacity to disperse rapidly and compete
strongly for resources. In this study I explored the potential effects of climate change on
the distribution of Eragrostis lehmanniana (Lehmann lovegrass), a perennial warmseason bunchgrass, in the southwestern U.S. using ecological niche modeling.
Based on predictions of popular global-scale general circulation models, 28
climate-change scenarios were created by modifying existing long-term averages of
climatic variables in Arizona and western New Mexico. Using the Genetic Algorithm for
Rule-set Prediction model (GARP; Stockwell and Peters 1999), I modeled the ecological
niche for E. lehmanniana. This niche, characterized by a set of rules, was then projected
onto each of the 28 climate change scenarios to represent potential habitat within Arizona
under potential future conditions. Future scenarios show the distribution of E.
lehmanniana progressively shrinking in the southeastern and northwestern portions of
Arizona and increasing in the northeastern portion of the state with increasing
temperatures and precipitation. Key shifts occurred under scenarios with increases in
64
summer and winter precipitation of 30% or more, and increases in summer maximum and
winter minimum temperatures of at least 2oC. Spread to new environments will be limited
by the ability of viable E. lehmanniana seed to reach new habitats.
Introduction
Eragrostis lehmanniana, a tufted perennial bunchgrass native to southern Africa,
was brought to Arizona, USA in the 1930s to counteract low plant cover and highly
eroded soils resulting from decades of overgrazing and drought. This plant is now one of
the most dominant and problematic nonnative species in the southwestern U.S. Following
intentional seeding to 69,000 ha in Arizona, E. lehmanniana spread to an additional
79,000 ha, invading recently cleared lands and undisturbed native grasslands alike (Cox
and Ruyle 1986, McClaran and Anable 1992). The ability to produce copious droughttolerant seed and to survive long dry periods makes this grass an effective competitor
(Sumrall et al. 1991, Abbott and Roundy 2003). E. lehmanniana also produces abundant
lignin-rich biomass, and may be associated with changes in fire regime and nutrient
cycling (Cable 1971, Biedenbender and Roundy 1996, Williams and Baruch 2000).
Accordingly, sharp declines in animal species richness have been associated with E.
lehmanniana-dominated plots (Cable 1971, Bock et al. 1986, Medina 1988). It is unclear
whether this plant is still spreading to new locations in the U.S. (Cox et al. 1988).
Climate plays a large role in determining the distribution of plant species and
communities (Huntley and Webb 1988). Significant increases in precipitation and
temperature are expected for the southwestern United States in the coming decades
65
(IPCC 2001), potentially altering ecosystem stability by causing major shifts in
community and species distributions and hence, the structure and diversity (IGBP 1988,
Watson et al. 1996, Mooney 1997). Increases in annual average temperature have been
documented at 1o-2 o C in the western United States since the late 1940s (Dettinger et al.
1995), concurrent with earlier onset of phenological events (Cayan et al. 2001).
Additionally, documented increases in carbon dioxide concentrations and availability of
nitrogen and other nutrients directly affect photosynthesis and elicit a variety of responses
in plants (Holland et al. 1999). Such changes could favor nonindigenous species (Dukes
and Mooney 1999, Smith et al. 2000), especially those with the capacity to disperse
rapidly and compete strongly for resources.
Because E. lehmanniana has been observed to encroach into native communities
in southern Arizona and disperses seed widely via wind and water, this species can be
expected to spread into new areas with a changing climate. Future spread predictions can
facilitate early detection and control measures, thereby keeping newly established
populations in check. Here I explore the potential effects of climate change on the
distribution of E. lehmanniana in the southwestern U.S. using ecological niche modeling.
In addition to the prediction maps, one benefit of this approach is that model parameters
can be compared directly with key limiting ecological parameters known to affect
populations of the species in question.
66
Methods
Niche-based modeling is one approach for predicting species’ invasion potential
that has recently gained popularity (Hoffman 2001, Peterson and Vieglais 2001, Welk et
al. 2002, Peterson 2003). This method is based on the belief that a species’ ecological
niche can be described as the n-dimensional hypervolume of environmental conditions
under which it is able to maintain populations without immigration (Hutchinson 1957).
One example of a model applying this concept is the Genetic Algorithm for Rule-set
Prediction (GARP). The GARP model has been applied to predict changes in species’
distributions following climate change in several cases (Peterson et al. 2001, Peterson et
al. 2002, Oberhauser and Peterson 2003, Peterson and Shaw 2003). Advantages of this
model over other approaches are its use of several inferential tools, theoretically
increasing its predictive ability over any one method independently (Stockwell and Peters
1999); the ability to incorporate information from both categorical and continuous
variables (Stockwell and Peterson 2002); and its ability to work with relatively small
sample sizes (Stockwell and Peterson 2002).
Input Data
Eragrostis lehmanniana seeded throughout Arizona, New Mexico, and Texas can
be traced to a single plant introduced from South Africa (Crider 1945). Recent research
has shown that genetic variation in U.S. E. lehmanniana populations does exist, though it
is unknown whether this variation is the result of the U.S. population undergoing
mutation and genetic drift or due to genotypic variability present in the introduced
67
individual (Schussman 2002). Regardless, the U.S. population may occupy a slightly
different environmental niche than populations in its native range due to this isolation
(Mau-Crimmins et al. submitted). Accordingly, predictions about future E. lehmanniana
distributions were made in this study using known presence points in the U.S., a
departure from convention (Peterson and Vieglais 2001, Peterson 2003, Peterson et al.
2003). Traditionally, species’ potential distributions outside of their native ranges are
predicted using data from the species’ native ranges (Hoffman 2001, Peterson and
Vieglais 2001, Welk et al. 2002, Peterson 2003). This study relies on invaded-range data
to make future distribution predictions, which could potentially misrepresent the species’
environmental niche.
Unique localities of species’ occurrences within Arizona and western New
Mexico were obtained from several sources including the Santa Rita Experimental
Range, Bureau of Land Management, The Nature Conservancy, U.S. Department of
Defense, U.S. Fish and Wildlife Service, U.S. Forest Service, U.S. Geological Survey,
and the U.S. National Park Service, resulting in nearly 500 localities. Due to tight
clustering of presence points and the potential for spatial autocorrelation between them, a
subset of 100 evenly distributed points was randomly selected. The remainder of the
points was retained for extrinsic model testing.
The base environmental data consisted of 13 geographic coverages obtained at 1km resolution covering Arizona and western New Mexico. Elevation was obtained from
the U.S. Geological Survey (2001); slope and cosine of aspect were derived from this
grid. Additional coverages included aspects of climate averaged for the period 1980 to
68
1997 encompassing mean monthly precipitation, solar radiation, and average maximum
and minimum annual temperatures (DayMet grids; Thornton et al. 1997, Thornton et al.
2000); and soil characteristic data and depth to bedrock (Miller and White 1998). From
the monthly precipitation data, I calculated grids representing mean summer precipitation
(June, July August) and mean winter precipitation (December, January, February,
March). Season definitions were designed to capture the unique seasonality of
precipitation in Arizona. Soil characteristic data were averaged across the soil layers up
to 1 m in depth to estimate the percent sand, clay, and silt in the soil profile.
Niche Modeling
All modeling in this study was carried out on a desktop implementation of GARP
(http://www.lifemapper.org/desktopgarp). The Genetic Algorithm for Rule-set Prediction
models species’ environmental niche by identifying non-random relationships between
environmental characteristics of known presence localities versus those within the overall
study region. Occurrence points are divided evenly into training and test data sets. GARP
is designed to work based on presence-only data; absence information is included in the
modeling via sampling of pseudoabsence points from the set of pixels where the species
has not been detected. GARP works in an iterative process of rule selection, evaluation,
testing, and incorporation or rejection to produce a heterogeneous rule-set characterizing
the species’ ecological requirements (Peterson et al. 1999). First, a method is chosen from
a set of possibilities (e.g. logistic regression, bioclimatic rules), and it is applied to the
data. Then, a rule is developed and predictive accuracy (sensu Stockwell and Peters
69
1999) is evaluated via training points intrinsically re-sampled from both the known study
region and from the study region as a whole. The change in predictive accuracy from one
iteration to the next is used to evaluate whether a particular rule should be incorporated
into the model (rule-set). Rules may evolve by a number of means that mimic DNA
evolution: point mutations, deletions, and crossing over. As implemented here, the
algorithm runs either for 1,000 iterations or until convergence. The final rule-set, or
ecological niche model, is then projected onto a digital map as the species’ potential
geographic distribution and imported into ArcView 3.2 (ESRI 1999) using the Spatial
Analyst extension for visualization.
To reduce environmental layers to just those that provide the highest predictive
accuracy, I used a jackknife manipulation (Tukey 1958). I ran multiple iterations of
models, omitting each data layer systematically. I then calculated correlations between
inclusion of each data layer in the model (coded binarily) and omission error (percentage
of extrinsic test presence data not predicted as present) to detect data layers that
contribute negatively to model performance when evaluated based on independent test
data. Correlations of r > 0.05 have been suggested to be indicative of data layers that
detract from model quality (Peterson et al. 2003); no data layers were correlated at this
level. Therefore all layers were retained for further analyses.
Using the subset of 100 unique species occurrence points for Arizona and New
Mexico, I produced 300 replicate models of Eragrostis lehmanniana’s ecological niche
based on random 50-50 splits of available occurrence points. To choose the best subset of
native range models, I adopted a best-subsets selection procedure (Anderson et al. 2003,
70
Peterson et al. 2003a). Following this method, I selected the best subset of models by
eliminating all models that had non-zero omission error based on independent test points,
calculated the median area predicted present among these zero-omission points, and then
identified the ten models closest to the overall median area predicted. These ten models
were summed to create a final output grid, ranging from 0 (areas not predicted present by
any of the ten models) to 10 (areas predicted present by all ten models). For area
calculations, pixels predicted by at least 9 of the best-subsets models were used.
Climate Change Projections
General circulation models (GCMs) predict changes in temperatures from –0.5 to
+3oC and increases in precipitation –0.5 to +3 mm/day by 2020 for the southwestern
U.S. (IPCC 2001). However, Arizona’s summer monsoon season is not well captured by
these models, introducing uncertainty in regional precipitation predictions (Southwest
Regional Assessment Group 2000.). In addition, the spatial resolution of GCMs is very
coarse in comparison to the scale of this study, resulting in unrealistic “jumps” in
neighboring pixel values when GCM predictions are applied to current conditions.
Because of these limitations, I created 28 climate-change scenarios within the range
suggested by GCMs for the southwestern United States (Table 1). Precipitation increases
were represented as a percent of the current value, to represent increases with greater
precision than GCM predictions. The local fine-scale climate datasets were modified to
reflect each of these scenarios, resulting in smoother, more realistic future scenarios than
that from GCMs.
71
Ecological niche models developed with GARP can be projected onto modeled
future landscapes to predict potential distributions. Once the current distribution map was
completed, the ordered series of if-then statements comprising this model were applied to
each of the future climate scenarios to generate a series of future distribution predictions.
Testing Predictive Power of Ecological Niche Models
Tests of model predictive power can be accomplished by overlaying extrinsic test
data and tallying observed correct predictions. The proportion of the total study extent
predicted as present (occupied by the species) multiplied by the number of extrinsic test
data points is used as a random expectation of successful prediction points if no nonrandom association existed between prediction and test points (Peterson and Shaw 2003).
Following Peterson and Shaw (2003) and using the 316 extrinsic test points held out from
the model, I implemented a chi-squared test (1 df) to test the significance of the departure
from random expectations.
To identify environmental dimensions important for defining E. lehmanniana’s
geographic potential, I conducted a series of sequential jackknife manipulations in which
all possible combinations of a reduced set (i.e., N-1) of N environmental coverages were
used to generate native-range models. I assessed model quality by exploring correlations
between variable inclusion and omission error (Peterson and Cohoon 1999). Variables
that were positively correlated were considered to be most important in defining E.
lehmanniana’s environmental niche.
72
Model Output Visualization
Current Distribution Prediction
To explore relationships between known presence points and predicted presence
points in multivariate space, a correlation-based principal components analysis (PCA)
was performed on the environmental variables associated with the known presence points
and pixels within the study area using PRIMER v5 (PRIMER-E Ltd. 2001). Prior to
analysis, variables were transformed using the log or square-root transformation if
distributions were skewed.
Results
Tests of Model Predictivity
Ecological niche models developed in this study were highly predictive of
distributional phenomena based on random subsets. All of the best-subsets models were
highly statistically significant in comparison with random expectations (X2 tests, df = 1,
all best-subsets models P < 0.001).
Jackknife manipulations of the different environmental coverages (Peterson and
Cohoon 1999) suggested summer precipitation, winter minimum temperature, elevation,
summer maximum temperature, solar radiation, and clay content were the critical
variables, in decreasing magnitude, constituting the ecological niche of E. lehmanniana.
These variables correlated positively with improvement in avoiding omission error
(Peterson and Cohoon 1999, Oberhauser and Peterson 2003).
73
Model Output and Visualization
Current Distribution Prediction
Approximately 110,200 km2 are predicted as ‘present,’ or highly likely to host E.
lehmanniana under current conditions (predicted present by at least 9 out of 10 bestsubset models). Models of current E. lehmanniana distribution consistently strongly
predict presence across the southeastern portion of the state and up into the northwest
portion of the state (Figure 1).
PCA on the environmental variables for confirmed presence points in the native
and invaded ranges yielded a 4-component solution accounting for 81.4% of the variance
in the raw dataset (Table 2; Figure 3). Component 1 loads highly on elevation, total
precipitation, summer precipitation, summer maximum temperature, and winter
minimum temperature. The highest loadings for component 2 were winter precipitation,
radiation, sand, and winter minimum temperature. Component 3 loads highly on silt,
clay, and slope; component 4 loads most highly on sand and rock depth.
Future Climate Change Predictions
Understanding both the location and amount of area predicted to be invaded by an
invasive species is important for planning and management decisions. It is key to know
both whether E. lehmanniana is expected to spread to new areas as well as whether it will
expand or shrink in distribution.
Future scenarios show E. lehmanniana’s distribution progressively shrinking in
the southeastern and northwestern portions of the state and increasing in the northeastern
74
portion of the state with increasing temperatures and precipitation (Figure 2). The amount
of land area predicted to be potential habitat for E. lehmanniana does not increase
drastically under the different scenarios, but does change spatially (Table 3). For
scenarios 1-20, the change in total area predicted present by at least 9 of the 10 best
subsets models increases up to 10.7% (117.4 ha) from the predicted current distribution.
Key shifts occur under scenarios 21-28 (Figure 4), where overlap with the current
predicted distribution drops sharply from near 80% down to 30%. These scenarios are
characterized by increases in summer and winter precipitation of 30% or more and
increases in summer maximum and winter minimum temperatures of at least 2oC.
Discussion
Byers et al. (2002) call for a prioritized approach to invasive species control,
recognizing limited resources available to land managers. E. lehmanniana is widely
recognized in Arizona and New Mexico as a problem plant, nullifying the need for
prioritization. Instead, emphasis should be placed on understanding regions most
vulnerable to new infestations under current and future conditions. Predictions can
facilitate early detection and control measures, thereby keeping new infestations in check
(Moody and Mack 1988).
Current Distribution Prediction
Studies of E. lehmanniana in the southwestern U.S. have suggested that elevation,
summer precipitatation, winter minimum temperature, and soil texture play key roles in
75
the distribution of this species (Anderson et al. 1957, Cable 1971, Cox and Martin 1984,
Cox and Ruyle 1986, Cox et al. 1988, Anable 1990). My model projections support these
findings. Based on the jackknife of the environmental layers, summer maximum and
winter minimum temperatures, summer precipitation, elevation, incoming solar radiation,
and clay in the upper horizons have the greatest influence on predicting the distribution of
E. lehmanniana. However, winter precipitation, slope, aspect, average annual
precipitation, and depth to bedrock also play a role in E. lehmanniana’s presence.
The current potential distribution for E. lehmanniana depicts the influence of
elevation in the model; mountaintops were excluded from predictions. The northeastern
and southwestern corners of the state are not predicted to fall within the ecological niche
of E. lehmanniana. The southwestern portion of the state is characterized by higher
winter minimum and summer maximum temperatures and lower precipitation values than
the ranges of these variables predicted to be present by the model. The northeastern
portion of the state appears to be predicted as inappropriate habitat primarily due to low
winter minimum temperatures. In this prediction, approximately 110,200 km2 are
predicted as ‘present,’ or highly likely to host E. lehmanniana under current conditions.
This figure is nearly 75 times as great as that proposed by Cox and Ruyle (1986), who
stated that E. lehmanniana appeared to have expanded as far as possible within Arizona.
My model predicts areas much farther north and west than Cox and Ruyle (1986)
predicted, in areas both colder, wetter, and drier than expected in their study.
Because several environmental variables together shape the presence and
distribution of E. lehmanniana, it is perhaps best to interpret its niche in multivariate
76
space. The first axis of the PCA on the environmental variables for confirmed and
predicted presence points is that of decreasing values (from left to right) for winter
minimum temperature and summer maximum temperature, roughly equally weighted,
and elevation, total precipitation, and summer precipitation, roughly equally weighted,
with the opposite trend. The second axis is dominated by sand and radiation, increasing
up the PC2 axis, and winter minimum temperature and winter precipitation decreasing up
the axis. Of the environmental conditions represented within the study area, E.
lehmanniana is known and predicted to occupy regions characterized by lower winter
minimum and summer maximum temperatures and winter precipitation as well as higher
elevation, summer precipitation, total precipitation, and sandier soils (Figure 3).
The multivariate plot also offers insight into model performance. From this plot, it
is possible to see that the model does not predict conditions far outside of the set of
conditions known to be occupied by E. lehmanniana currently. Therefore, regions
predicted as current habitat but not yet occupied by E. lehmanniana are the regions most
likely to host E. lehmanniana.
Climate Change Predictions
For scenarios 1-20, the amount of area predicted to host E. lehmanniana does not
increase a great deal; in addition, the predicted distribution continues to overlap by at
least 80% for all of these scenarios. This suggests that changes of a magnitude of 1oC and
+25% precipitation are unlikely to have dramatic impacts on the distribution of E.
lehmanniana in Arizona.
77
Large changes in the predicted distribution of E. lehmanniana occur under
scenarios 20-28, where changes in summer and winter temperatures range from 0.5 to
3oC and changes in summer and winter precipitation range from 0 to 125% (Figure 2).
However, the distributions predicted by scenarios 24-28 do not vary considerably, and the
amount of area predicted as likely habitat by these models is actually less than that
predicted by the current model. E. lehmanniana is predicted to move further upslope and
to increasing latitudes under warmer and wetter conditions. Areas predicted as potential
habitat under the current distribution model, in the southeastern and northwestern
portions of the state, become too warm to fall within the environmental niche predicted
for E. lehmanniana by this model. Expanding the extent of the study area may show that
E. lehmanniana has the potential to greatly expand its range into neighboring states under
these possible future conditions.
Limitations of the Study
Three important limitations of niche-based models are the lack of accounting for
biotic interactions, dispersal, and evolutionary change (Pearson and Dawson 2003). I now
address each of these potential shortcomings with the present model of E. lehmanniana.
Biotic interactions
Correlative models such as niche-based or bioclimatic envelope models have been
criticized for not capturing species interactions (Davis et al. 1998). Competitive
78
interactions can limit the geographic distribution realized by a species; models built on
known presence points under these conditions can underestimate the environmental niche
that the target species has the ability to occupy (Peterson et al. 2002). Projecting this
underestimated niche onto future climate scenarios would compound underprediction.
Conversely, several studies have shown complex relationships between species, whereby
removal of one member has rippling, unpredicted effects through the ecosystem (e.g.,
Silander and Antonovics 1982). Environmental niche models that do not account for
these relationships can lead to incorrect predictions of future species distributions
(Pearson and Dawson 2003).
Current knowledge of E. lehmanniana in the U.S. suggests that this plant is
minimally constrained by interactions with other species. Instead, it has been observed to
spread aggressively into native grasslands and shrublands from seeded areas, independent
of disturbance (Anable et al. 1992). Cable (1971) predicted the spread of E. lehmanniana
into appropriate habitat would be limited only by the ability of viable seed. Biotic
interactions operate at a local scale, on the order of meters to kilometers (Willis and
Whittaker 2002, Pearson and Dawson 2003), the scale of the model presented in this
paper. However, because E. lehmanniana apparently moves across the landscape
independently of other species, this criticism of correlative models does not pertain to the
present model.
79
Dispersal
Niche-based models assume the ability of species to follow their habitat across the
landscape under a changing climate. Whether E. lehmanniana will move to new locations
under increases in temperature and precipitation will depend primarily on whether viable
seed is spread to these regions. Previous studies have estimated spread of E. lehmanniana
at 6-10 m/yr and up to 175 m/yr (Kincaid et al. 1959, McClaran and Anable 1992), citing
its ability to produce copious wind- and water-dispersed seed.
E. lehmanniana is widespread across the southwestern portion of Arizona and
seems to still be spreading north and west across the state (D. Robinett, pers. comm.
2001), actively occupying native stands as well as areas cleared by grazing, fire, or
drought (Sumrall et al. 1991, Anable et al. 1992). Therefore, it is likely that E.
lehmanniana will be able to spread with a changing climate, moving to new locations at
higher elevation and latitude. Areas receiving frequent seed introductions and near roads
and trails are most likely to become invaded first, with populations spreading from these
loci (Moody and Mack 1988). Remote locations, far from existing E. lehmanniana
populations and from roads and trails, are least likely to become populated due to lower
probability of seed introduction. In addition, locations separated from existing invasions
by barriers, such as the region to the north of the Grand Canyon, are also less likely to
become quickly invaded.
80
Evolutionary change
Under the pressures of climate change, some species evolve rapidly, selecting
traits enabling their survival (Woodward 1990, Thomas et al. 2001). The ability of some
species to quickly evolve is an important factor impacting species’ distributions not
addressed by niche models. The introduction history of E. lehmanniana to the U.S. is
well known; seed planted across Arizona can be traced to a single plant (Crider 1945).
Recent research has shown that genetic variation in U.S. populations does exist, though it
is unclear whether this variation is the result of genetic mutations following introduction
or due to genotypic plasticity in the original maternal plant (Schussman 2002).
Regardless, these findings demonstrate that E. lehmanniana may possess the ability to
adapt to new environments quickly. The future distributions presented in this paper may
therefore be conservative.
Caveats
It is important to interpret the future-distribution scenarios presented here as first
approximations and not precise simulations of E. lehmanniana’s future distribution for
several reasons. First, two key assumptions are made with environmental niche models
that, if not met, can drastically impact the results. One assumption is that the species is in
equilibrium with its environment and that the presence points used to build the model
accurately capture the species’ niche. Next, it is assumed that the environmental variables
used to build the model are the parameters that have primary influence on the species’
distribution. Additionally, this model does not account for biotic interactions, which
81
could have great influence on the species’ future distribution. Finally, the climate change
models employed in this study are not necessarily representative of future conditions;
large uncertainty surrounds future climate conditions. Therefore, the current and future
predictions are “best guesses” for where E. lehmanniana could likely remain and spread
but should not be interpreted literally.
Consideration of additional variables, such as land use and land cover, could
improve model predictions. In addition, incorporation of disturbance regimes could
highlight regions expected to become invaded most quickly under future climate
conditions.
Conclusion
Eragrostis lehmanniana, a troublesome perennial grass occupying continuous,
dense stands in the southwestern U.S., is not likely to disappear under a changing climate.
Predicted to potentially occupy over 110,000 km2 in Arizona, this plant apparently is still
spreading to available habitat. It likely will continue to expand as long as seed dispersal
vectors facilitate its spread to new habitats. Under changing climatic conditions, E.
lehmanniana is expected to expand its range to the north and upslope. Its projected future
range will contract in the southern and central portions of the state as conditions become
too warm. However, major range shifts are not predicted to occur until average summer
maximum and winter minimum temperatures increase by at least 2oC and summer and
winter precipitation increases by at least 30%.
82
The current model does not account for E. lehmanniana’s ability to adapt to new
environmental conditions, which could allow E. lehmanniana to occupy more area than
predicted here. In addition, biotic interactions are not considered; these can have
profound impacts on species’ distributions. Therefore, predictions presented in this paper
should be considered “best guesses,” but not interpreted literally.
83
References
Abbott, L.B. and B.A. Roundy. 2003. Available water influences field germination and
recruitment of seeded grasses. Journal of Range Management 56:56-64.
Anable, M.E. 1990. Alien plant invasion in relation to site characteristics and
disturbance: Eragrostis lehmanniana on the Santa Rita Experimental Range,
Arizona 1937-1989. M.S. Thesis, University of Arizona, Tucson.
Anable, M.E., M.P. McClaran, and G.B. Ruyle. 1992. Spread of introduced Lehmann
lovegrass Eragrostis lehmanniana Nees. in Southern Arizona, USA. Biological
Conservation 61:181-188.
Anderson, D., L.P. Hamilton, H.G. Reynolds, and R. R. Humphrey. 1957. Reseeding
desert grassland ranges in southern Arizona. Arizona Agricultural Experimental
Station. Bull. No. 249.
Anderson, R.P., D. Lew, and A.T. Peterson. 2003. Evaluating predictive models of
species’ distributions: criteria for selecting optimal models. Ecological Modelling
162:211-232.
Biedenbender, S.H. and B.A. Roundy. 1996. Establishment of native semidesert grasses
into existing stands of Eragrostis lehmanniana in Southeastern Arizona.
Restoration Ecology 4:155-162.
Bock, C.E., J.H. Bock, K.L. Jepson, and J.C. Ortega. 1986. Ecological effects of planting
African lovegrasses in Arizona. National Geographic Research 2:456-463.
Byers, J.E., S. Reichard, J.M. Randall, I.M. Parker, C.S. Smith, W.M. Lonsdale, I.A.E.
Atkinson, T.R. Seastedt, M. Williamson, E. Chornesky, and D. Hayes. 2002.
Directing research to reduce the impacts of nonindigenous species. Conservation
Biology 16:630-640.
Cable, D.R. 1971. Lehmann lovegrass on the Santa Rita experimental range, 1937-1968.
Journal of Range Management 24:17-21.
Cayan, D.R., S.A. Kammerdiener, M.D. Dettinger, J.M. Caprio, and D.H. Peterson. 2001.
Changes in the onset of spring in the western United States. Bulletin of the
American Meteorological Society 82:399-415.
Cox, J.R. and M.H. Martin. 1984. Effects of planting depth and soil texture on the
emergence of four lovegrasses. Journal of Range Management 37:204-205.
84
Cox, J.R. and G.B. Ruyle. 1986. Influence of climatic and edaphic factors on the
distribution of Eragrostis lehmanniana Nees. in Arizona, USA. Journal of the
Grassland. Society of South Africa 3:25-29.
Cox, J.R., G.B. Ruyle, J.H. Fourie, and C. Donaldson. 1988. Lehmann lovegrass B
Central South Africa and Arizona, USA. Rangelands 10:53-55.
Crider, F.J., 1945. Three introduced lovegrasses for soil conservation. United States
Department of Agriculture Circular No. 730.
Davis, A.J., L.S. Jenkinson, J.L. Lawton, B. Shorrocks, and S. Wood. 1998. Making
mistakes when predicting shifts in species range in response to global warming.
Nature 391:783-786.
Dettinger, M.D., M. Ghil, and C.L. Keppenne. 1995. Interannual and interdecadal
variability in United States surface-air temperatures, 1910-87. Climatic Change
31:35-66.
Dukes, J. and H. Mooney. 1999. Does global change increase the success of biological
invaders? Trends in Ecology and Evolution 14:135-139.
Hoffman, M.H. 2001. The distribution of Senecio vulgaris, capacity of climatic range
models for predicting adventitious ranges. Flora 196:395-403.
Holland, E.A., F.J. Dentener, .H. Braswell, and J.M. Sulzman. 1999. Contemporary and
pre-industrial global reactive nitrogen budgets. Biogeochemistry 46:7-43.
Huntley, B.J. and T. Webb. 1988. Vegetation History. Dordrecht: Kluwer Academic
Publishers.
Hutchinson,G.E. 1957. Concluding remarks. Cold Spring Harbor Symposium on
Quantitative Biology 22:415-427.
IGBP 1988. Toward an understanding of global change. Washington DC: National
Academic Press.
International Panel on Climate Change. 2001. Climate Change 2001: Summary for
Policymakers. A Report of Working Group I of the Intergovernmental Panel on
Climate Change. Cambridge: Cambridge University Press.
Kincaid, D.R., G.A. Holt, P.D. Dalton, and J.S. Tixier. 1959. The spread of Lehmann
lovegrass as affected by mesquite and native perennial grasses. Ecology 40:738742.
85
Leemans, R. and P. Halpin. 1992. Global change and biodiversity. Pages 254-255 in B.
Groombridge, ed. Biodiversity 1992: Status of the Earth’s Living Resources.
London: Chapman & Hall.
Mau-Crimmins, T.M., H.S. Schussman, and E.L. Geiger. Can an invading species’
distribution be predicted using data from its invaded range? Lehmann lovegrass
(Eragrostis lehmanniana) in the southwestern United States. Ecological
Modelling. In review.
McClaran, M.P. and M.E. Anable. 1992. Spread of introduced Lehmann lovegrass along
a grazing intensity gradient. Journal of Applied Ecology 29:92-98.
Medina, A.L. 1988. Diets of scaled quail in southern Arizona (USA). Journal of Wildlife
Management 52:753-757.
Miller, D.A. and R.A. White. 1998. A conterminous United States multi-layer soil
characteristics data set for regional climate and hydrology modeling. Earth
Interactions 2:1-26. Available on-line at http://EarthInteractions.org (accessed
May 2003).
Moody, M.E. and R.N. Mack. 1988. Controlling the spread of plant invasions: the
importance of nascent foci. Journal of Applied Ecology 25:1009-1021.
Mooney, H.A. 1997. Ecosystem function of biodiversity: the basis of the viewpoint.
Pages 341-354 in T.M. Smith, H.H. Shugart and F.I. Woodward, editors. Plant
Functional Types: Their Relevance to Ecosystem Properties and Global Change.
IGBP Book Series 1. Cambridge: Cambridge University Press.
Oberhauser, K. and A.T. Peterson. 2003. Modeling current and future potential wintering
distributions of eastern North American monarch butterflies. Proceedings of the
National Academy of Sciences 100:14063-14068.
Pearson, R.G. and T.P. Dawson. 2003. Predicting the impacts of climate change on the
distribution of species: are bioclimatic envelope models useful? Global Ecology
& Biogeography 12:361-371.
Peterson, A.T. 2003. Predicting the geography of species’ invasions via ecological niche
modeling. The Quarterly Review of Biology 78:419-433.
Peterson, A.T. and K.P. Cohoon. 1999. Sensitivity of distributional prediction algorithms
to geographic data completeness. Ecological Modelling 117:159-164.
86
Peterson, A.T. and J. Shaw. 2003. Lutzomyia vectors for cutaneous leighmaniasis in
Southern Brazil: ecological niche models, predicted geographic distributions, and
climate change effects. International Journal for Parasitology 33:919-931.
Peterson, A.T. and D.A. Vieglais. 2001. Predicting species invasions using ecological
niche modeling: New approaches from bioinformatics attack a pressing problem.
BioScience 363-371.
Peterson, A.T., M.A. Ortega-Huerta, J. Bartley, V. S<nchez-Cordero, J. Sober\n,, R.H.
Buddemeier, and D.R.B. Stockwell. 2002. Future projections for Mexican faunas
under global climate change scenarios. Nature 416:626-629.
Peterson, A.T., M. Papes, and D.A. Kluza. 2003. Predicting the potential invasive
distributions of four alien plant species in North America. Weed Science 51:863868.
Peterson, A.T., V. S<nchez-Cordero, J. Sober\n, J. Bartley, R.W. Buddemeier, A.G.
Navarro-Sighenza. 2001. Effects of global climate change on geographic
distributions of Mexican Cracidae. Ecological Modelling 144:21-30.
Schussman, H.R. 2002. Genetic variation in the introduced apomictic grass, Eragrostis
lehmanniana Nees, on the Santa Rita Experimental Range in southeastern
Arizona, USA. M.S. Thesis, University of Arizona, Tucson.
Silander, J.A. and J. Antonovics. 1982. Analysis of interspecific interactions in a coastal
plant community – a perturbation approach. Nature 298:557-560.
Smith, S.D., T.E. Huxman, S.F. Zitzer, T.N. Charlet, D.C. Housman, J.S. Coleman, L.K.
Fenstermaker, J.R. Seemann, and R.S. Nowak. 2000. Elevated CO2 increases
productivity and invasive species success in an arid ecosystem. Nature 408:79-82.
Southwest Regional Assessment Group. 2000. Preparing for a changing climate: The
potential consequences of climate variability and change. Tucson, AZ: University
of Arizona, Institute for the Study of Planet Earth.
Stockwell, D.R.B. and D. Peters. 1999. The GARP modeling system: problems and
solutions to automated spatial prediction. International Journal of Geographic
Information Science 13:143-158.
Stockwell, D.R.B. and A.T. Peterson. 2002. Effects of sample size on accuracy of species
distribution models. Ecological Modelling 148:1-13.
87
Sumrall, L.B., B.A. Roundy, J.R. Cox, and V.K. Winkel. 1991. Influence of canopy
removal by burning or clipping on emergence of Eragrostis lehmanniana
seedlings. International Journal of Wildland Fire 1:35-40.
Thomas, C.D., E.J. Bodsworth, R.J. Wilson, A.D. Simmins, Z.G. Davies, M. Musche,
and L. Conradt. 2001. Ecological and evolutionary processes at expanding range
margins. Nature 411:577-581.
Thornton, P.E., S.W. Running, and M.A. White. 1997. Generating surfaces of daily
meteorology variables over large regions of complex terrain. Journal of
Hydrology 190:214-251.
Thornton, P.E., H. Hasenauer, and M.A. White. 2000. Simultaneous estimation of daily
solar radiation and humidity from observed temperature and precipitation: An
application over complex terrain in Austria. Agricultural and Forest Meteorology
104:255-271.
Tukey, J. 1958. Bias and confidence in not quite large samples. Annals of Mathematical
Statistics 29: 614.
Watson, R.T., M.C. Zinyowera, and R.H. Moss. 1996. Climate Change 1995. Impacts,
adaptations and mitigation of climate change: scientific-technical analysis.
Cambridge: Cambridge University Press.
Welk, E., K. Schubert, and M.H. Hoffman. 2002. Present and potential distribution of
invasive garlic mustard (Alliaria petiolata) in North America. Diversity and
Distributions 8:219-233.
Williams, D.G. and Z. Baruch. 2000. African grass invasion in the Americas: ecosystem
consequences and the role of ecophysiology. Biological Invasions 2:123-140.
Willis, K.J. and R.J. Whittaker. 2002. Species diversity – scale matters. Science
295:1245-1248.
Woodward, F.I. 1990. The impact of low temperatures in controlling the geographical
distribution of plants. Philosophical Transactions of the Royal Society of London
B 326:585-593.
88
Table 1. Future climate scenarios used to predict potential distribution of Eragrostis
lehmanniana in Arizona, USA. Negative values represent a decrease; positive values
represent an increase.
Scenario
Change in
Summer
Precipitation (%)
Change in Winter
Precip (%)
Change in Summer
Maximum
Temperature (oC)
Change in Winter
Minimum
Temperature (oC)
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
-10
-10
-5
-5
0
0
0
0
0
5
5
10
10
10
10
10
10
20
25
25
30
30
50
75
75
100
100
125
-10
-10
-5
-5
0
10
25
0
25
5
5
10
0
10
0
10
10
20
0
0
30
30
50
75
75
100
100
125
0.5
1
0.5
1
0.5
0.5
0.5
1
1
0.5
1
0.5
0.5
1
1
1.5
2
1.5
0.5
1
2
3
2
2
3
2
3
3
0.5
1
0.5
1
0.5
1
1
1
0.5
0.5
1
0.5
1
1
0.5
1.5
2
1.5
1
0.5
2
3
2
2
3
2
3
3
89
Table 2. Four-component solution generated by principal components analysis for
predicting distribution of Eragrostis lehmanniana in Arizona, USA.
Variable
Winter minimum temperature
Silt
Sand
Elevation
Cosine of Aspect
Clay
Winter precipitation
Total precipitation
Summer precipitation
Summer maximum temperature
Slope
Rock depth
Radiation
Eigenvalue
Explained variance (%)
Cumulative variance explained (%)
PC1
0.367
-0.076
0.100
-0.415
0.004
-0.125
-0.227
-0.405
-0.406
0.432
-0.187
0.147
-0.194
PC2
-0.332
-0.243
0.349
0.255
-0.041
-0.268
-0.468
-0.228
-0.074
-0.200
-0.272
0.163
0.392
PC3
0.091
0.483
-0.125
-0.095
-0.183
0.447
-0.069
0.029
0.144
0.075
-0.406
0.384
0.395
PC4
0.149
0.022
0.562
-0.066
0.013
-0.330
0.379
0.293
0.223
0.037
-0.040
0.506
-0.097
PC5
0.079
-0.222
-0.025
-0.045
-0.942
-0.062
0.056
0.029
0.008
0.048
0.186
-0.040
0.094
4.61
35.5
35.5
2.19
16.9
52.3
1.62
12.5
64.8
1.13
8.7
73.5
1.02
7.9
81.4
90
Table 3. Area occupied and spread of Eragrostis lehmanniana under climate change
scenarios (Table 1) in Arizona, USA.
Scenario
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
Area
predicted
(km2)
106,186
114,880
109,787
107,649
111,914
115,682
117,297
112,316
107,257
114,606
117,897
118,778
117,672
119,561
108,073
122,204
122,250
121,619
121,932
112,487
123,864
128,213
114,574
112,049
109,645
98,133
102,458
98,813
Change Relative to
Current Prediction
%
km2
-4,004
-3.6
4,690
4.3
-403
-0.4
-2,541
-2.3
1,724
1.6
5,492
5.0
7,107
6.4
2,126
1.9
-2,933
-2.7
4,416
4.0
7,707
7.0
8,588
7.8
7,482
6.8
9,371
8.5
-2,117
-1.9
12,014
10.9
12,060
10.9
11,429
10.4
11,742
10.7
2,297
2.1
13,674
12.4
18,023
16.4
4,384
4.0
1,859
1.7
-545
-0.5
-12,057
-10.9
-7,732
-7.0
-11,377
-10.3
Overlap with Current
Prediction
2
km
%
98,491
89.4
94,033
85.3
102,759
93.3
88,963
80.7
105,953
96.2
96,220
87.3
95,892
87.0
92,728
84.2
95,245
86.4
104,443
94.8
95,778
86.9
100,694
91.4
97,921
88.9
98,936
89.8
97,179
88.2
94,578
85.8
91,111
82.7
91,548
83.1
99,636
90.4
99,019
89.9
81,878
74.3
77,098
70.0
64,758
58.8
55,300
50.2
42,766
38.8
38,017
34.5
33,875
30.7
33,555
30.5
91
Figure 1. Predicted current distribution of Eragrostis lehmanniana in Arizona and
western New Mexico using ten best models. Pixels shaded dark grey are predicted
present by at least 9 out of 10 models; medium grey is predicted present by 6-8 models;
light grey is predicted present by 3-5 models. Black dots represent known presence
locations.
92
Figure 2. Predicted E. lehmanniana under various climate change scenarios. 2a displays
o
prediction for scenario 1 (10% decrease in precipitation; 0.5 C increase in temperature);
o
2b displays scenario 14 (10% increase in precipitation; 1 C increase in temperatures); 2c
o
displays scenario 21 (30% increase in precipitation; 2 C increase in temperatures); 2d
o
displays scenario 23 (50% increase in precipitation; 2 C increase in temperatures); 2e
o
displays scenario 25 (75% increase in precipitation; 3 C increase in temperatures); 2f
o
displays scenario 28 (125% increase in precipitation; 3 C increase in temperatures).
Darker shades of grey indicate greater model coincidence in predicting presence for E.
lehmanniana.
a
b
c
d
e
f
93
PC2
Figure 3. Principal components analysis for predicting distribution of Eragrostis
lehmanniana in Arizona. Black triangles represent all conditions available in study area.
Dark grey squares are pixels predicted present by the model. Light grey circles are known
presence points.
PC1
94
Figure 4. Spatial overlap of predicted future distributions of Eragrostis lehmanniana
with predicted current distribution in Arizona, USA.
100
90
Overlap with current distribution (%)
80
70
60
50
40
30
20
10
0
5 10 3 12 19 20 14 1 13 15 6
7 11 9 16 2
Scenario
8 18 17 4 21 22 23 24 25 26 27
95
APPENDIX C
COMMUNITY-LEVEL AND SEED BANK RESPONSE TO THE REMOVAL OF
A NONNATIVE PERENNIAL GRASS AT THREE SITES IN SOUTHEASTERN
ARIZONA
THERESA M. MAU-CRIMMINS AND GUY R. MCPHERSON
(To be submitted to Ecological Monographs)
96
Abstract
An understanding of ecosystem response to the removal of a dominant nonnative
species is critical for planning control or eradication efforts. The objective of this study
was to evaluate the response of herbaceous plants to the removal of Eragrostis
lehmanniana, a dominant perennial bunchgrass, in semi-desert grasslands. E.
lehmanniana was removed from 5 m x 5 m plots at three sites in southeastern Arizona.
Because seed banks represent, in part, a site’s potential for recovery following removal of
aboveground vegetation, a second objective of this study was to characterize the seed
banks at these sites.
This study suggests that plant community response to removal of an introduced
species is mediated by precipitation variability (timing and amount), local site history,
and edaphic conditions. The response observed on a site previously farmed for decades
was to subsequently become dominated by other nonnative annual species. However, the
two other sites with histories of livestock grazing responded more predictably to the
removal, with an increase in annual ruderal species (2 to 10 times the amount of annual
cover recorded on control plots). Treated plots at these two sites also exhibited increases
in percent native cover (1.5 to 4 times native plant cover than in control plots) and forb
cover (3 to 25 times the amount of forb cover recorded on control plots). The findings
suggest that changes may continue on these sites if E. lehmanniana were to continue to
be excluded from treatment plots, resulting in the eventual establishment of native
perennial species.
97
The seed banks of all three sites hosted several species not observed in
aboveground vegetation. However, the seed banks also indicated that the potential for the
E. lehmanniana to recover from seed is limited; the E. lehmanniana seed bank tapered off
noticeably (up to 25% less than in control plots) without the input of seed rain from
above-ground plants.
Introduction
Nonnative species introductions have been taking place as long as humans have
inhabited the planet (Mack 2001), some with devastating impacts. It is estimated that
approximately 10% of species introduced to a new environment become truly
problematic (Williamson 1996), having impacts ranging from the displacement or
replacement of native species to the complete reorganization of biotic communities.
Some of the most problematic and widespread plant introductions have been
grasses, affecting ecosystems worldwide (Whisenant 1990, D’Antonio and Vitousek
1992, Weber 1997, White et al. 1997, D’Antonio et al. 1998). Introduced grasses directly
affect resource availability (Gordon et al. 1989, Williams and Hobbs 1989), alter fire
frequency (D’Antonio and Vitousek 1992, Billings 1994, Peters and Bunting 1994), and
impact nitrogen and carbon cycling (Wedin and Tilman 1990, Fisher et al. 1994, Johnson
and Wedin 1997).
Often, the response to the discovery of an exotic species with detrimental impacts
is control or eradication, based in part on the success of other large-scale eradication
efforts (Myers et al. 2000). However, eradication may not lead to the recovery of the
98
affected system, as some species alter systems so greatly that they are shifted to a
different stable state (Westoby et al. 1989, Zavaleta et al. 2001). Alternatively, the effect
of such a management action could actually exacerbate the problem. For example,
efforts to control and reduce a nonnative plant species could initiate unexpected and
undesirable feedbacks, such as favoring other nonnative species or drastically reducing
plant cover to the detriment of wildlife and soil resources.
Zavaleta et al. (2001) argue for pre-eradication assessments to ascertain that a
removal will yield the expected results and minimize unwanted effects. Such assessments
can help determine a community’s potential response to the removal of a nonnative
species, especially when the nonnative is dominant. Ideally, such assessments should
address direct and indirect effects of removing the nonnative species, including response
of both plants and animals in the system. A second component of a pre-eradication
assessment is an assessment of the potential for reestablishment by members of the native
community. Seed bank studies provide some insight into a site’s potential for recovery
when the exotic species being removed is a plant (Rice 1989). Seeds of many perennial
grasses are short-lived, requiring seed additions to restore vegetation.
This study is a pre-eradication assessment for Eragrostis lehmanniana Nees
(Lehmann lovegrass), a perennial bunchgrass that has been associated with decreased
plant and animal species richness (i.e., Cable 1971, Bock et al. 1986) and plant species
diversity (Geiger and McPherson 2004). In addition to decreased species richness, E.
lehmanniana has been associated with the alteration of ecosystem processes (Cable 1971,
Bock et al. 1986, Williams and Baruch 2000), modification of community composition
99
(Anable et al. 1992), and changes in fire regimes (Ruyle et al. 1988, Burquez and
Quintana 1994, Biedenbender and Roundy 1996).
The results of this study sought to address how semi-desert communities in
southeastern Arizona would respond to the removal of this species. This was addressed
through a removal experiment of E. lehmanniana at three sites dominated by this plant.
Field studies offer the best representation of the effects of a wide-scale eradication effort,
as study sites are subject to natural variations in temperature and precipitation and reflect
its disturbance history. Because seed banks represent, in part, a site’s potential for
recovery following removal of aboveground vegetation, a second objective of this study
was to characterize the seed banks at these sites in a greenhouse study.
Methods
Species information
Eragrostis lehmanniana was introduced from Africa and officially released by the
U.S. Soil Conservation Service (currently the NRCS) to stabilize soils and provide cattle
forage (Cox et al. 1984). This species spreads aggressively to dominate perennial grass
composition, establishes in new areas without disturbance, and produces copious amounts
of seed that disperse by wind and water (Anable et al. 1992). In the 50 years following its
introduction, this grass occupied twice the area to which it was originally sown (Cox and
Ruyle 1986), and is predicted to spread to areas far beyond its current range (MauCrimmins et al. submitted). E. lehmanniana seed planted throughout Arizona, New
Mexico, and Texas can be traced to a single plant (Crider 1945). Recent research has
100
shown that genetic variation in U.S. E. lehmanniana populations does exist, though it is
unknown whether this variation is the result of the U.S. population undergoing mutation
and genetic drift or due to genotypic variability present in the introduced individual
(Schussman 2002).
Site descriptions
This experiment was conducted at three locations in southeastern Arizona: the
Santa Rita Experimental Range (SRER), Coronado National Memorial (CORO), and the
Three Links Farm (TLF, Figure 1). At all study locations, E. lehmanniana was the
dominant plant species.
The Santa Rita Experimental Range is 40 km southeast of Tucson, AZ (31.80oN,
110.83oW; elevation 1,200 m). The pasture selected for this study, the “airstrip”
exclosure in pasture 34, has been excluded from cattle grazing for nearly 25 years (M.
Heitlinger, pers. comm. 2005), and an arson fire burned the site in June of 1994
(Biedenbender and Roundy 1996). The soil is a Comoro sandy loam (thermic Typic
Torrifluvent), comprised of recent alluvium weathered from granitic rocks. The soil
varies in depth to 60 cm with a moderately acid pH of 6.2-6.9 (Hendricks 1985, USDA
NRCS 2004). Annual precipitation is 56.5 cm (Western Regional Climate Center 2005).
Average daytime temperatures are 18oC and often reach 42oC in June and July. Average
winter minimum temperatures are 3oC in January and February (WRCC 2005).
The second study site was located within the Montezuma Allotment at Coronado
National Memorial near Hereford, Arizona (31.33oN, 110.23oW; elevation 1,500 m).
101
Cattle grazed this allotment from at least 1929 until 1990 (University of Arizona Student
Chapter, Society for Range Management 1995). The soil at this site is a Lanque sandy
loam (thermic Pachic Haplustoll), with a slightly acid to slightly alkaline pH (6.1-7.8)
and a depth of greater than 150 cm (Denny and Peacock 2000). The soils are alluvium
derived from granite (Denny and Peacock 2000). Annual precipitation at Coronado NM
is 52.5 cm (WRCC 2005). Average daytime temperatures are 16oC. Maximum
temperatures reach 32oC in June and July, and winter minimum temperatures reach
approximately 0oC in January and February (WRCC 2005).
The third study site was located on the Three Links Farms (TLF) near Cascabel,
Arizona (32.11oN, 110.31oW; elevation 1,060 m), on agricultural fields farmed for 30-40
years then abandoned in 1999 and invaded by E. lehmanniana (B. Clark, pers. comm.
2003). The soil is a saline-sodic Hantz silty clay loam (thermic Vertic Torrifluvent) with
a pH of 7.9 to 8.4 (Svetlik unpublished data). At nearby Cascabel, AZ, annual
precipitation is 35.1 cm (WRCC 2005). Average daily temperatures are 18oC, with June
and July maximum temperatures reaching 37oC. Wintertime low temperatures drop to 1oC in December and January (WRCC 2005).
The three study sites are dominated by E. lehmanniana, but they exhibit
considerable variation in annual precipitation, edaphic characteristics, and land use
history. Replicating the experiment at three different locations potentially expands the
generality of the results in addition to the site-specific findings (Willems et al. 1993,
Gurevitch and Collins 1994).
102
Field methods
The experiment used a randomized block design. Twelve experimental plots 5 m
x 5 m in size were established at TLF; fourteen plots were established at CORO in
anticipation that one or two plots might have needed to be abandoned due to high foot
traffic in this area on the US-Mexico border. Twenty-four plots were established at SRER
with the original intention of a second experimental treatment. Plot locations were
selected to maximize cover by E. lehmanniana and minimize presence of other species. A
buffer of at least 1 m separated plots. At each site, one-half of the plots were randomly
assigned the ‘removal’ treatment; the remaining plots were untreated.
In the spring (April and May) of 2003, plant cover was estimated in all plots using
the point-intercept method (Greig-Smith 1983). Five parallel transects spaced 1 m apart
were placed in each plot. Aerial cover was recorded every 10 cm along each transect,
resulting in 250 points per plot. In addition, all species observed in a plot but not
encountered using the point-intercept method also were recorded. Species were identified
and assigned native or nonnative status; annual, biennial, or perennial habit; and
graminoid, forb/herb, shrub, tree, or vine growth form according to the PLANTS
database (http://plants.usda.gov). A few species had insufficient characteristics for
identification but were clearly distinct from other species in the plots and were labeled as
morphospecies. Plant names follow the Integrated Taxonomic Information System (ITIS)
convention (http://www.itis.usda.gov).
Immediately following the plot inventories, the non-selective herbicide
glyphosate (RoundUp) was applied to the “treatment” plots to remove E. lehmanniania.
103
Prior to treatment, all non-target forbs and grasses were identified and shielded from
treatment by covering with plastic bags. The glyphosate was applied in a low-volume
broadcast application using hand-held spray equipment at a rate of 45 ml/ha. Dead E.
lehmanniana litter was not removed, as a previous study suggested E. lehmanniana does
not replace itself on herbicide-treated plots with the canopy left intact (Sumrall et al.
1991). Follow-up treatments to kill E. lehmanniana plants which subsequently
established in the “treatment” plots were applied in July and August of 2003 and again in
May and July of 2004. Plots were again sampled in the fall (August-September) of 2003,
and again in the spring and fall of 2004. From this point forward, spring seasons are
referred to as SP03 and SP04, and fall seasons are named FA03 and FA04.
In addition to measures of species cover, standing live and dead biomass and
surface litter was measured in areas adjacent to the study plots. At each site, 20 1.0 m x
0.5 m quadrats were placed in random locations within 1 m of study plots. The samples
were sorted into standing biomass by species and litter. The samples were dried for 48
hours at 68oC and weighed to the nearest 0.1 g. Biomass samples were collected in FA02,
SP03, FA03, SP04, and FA04.
Seed banks were sampled in May, July, and September of 2003 and 2004
(denoted SP03, SU03, FA03 and SP04, SU04, and FA04) in an attempt to capture peaks
in seed abundance. The emergence technique (Brown 1992) was used to assess the
composition of the soil seed banks. Four cores measuring 7 cm in diameter and 3 cm
deep were randomly located and sampled from each of the study plots. Samples were
combined by plot in a single labeled paper bag. Soil samples were stored in paper bags
104
until air dry. Each sample was sieved through a #10 mesh to remove rocks and other
debris and spread approximately 1 cm thick onto sterilized potting soil in four 100 cm2
pots. Seedling trays were kept in a greenhouse and watered daily. The greenhouse was
cooled in summer and heated in winter, but experienced considerable daily and seasonal
variation in temperature. To ensure conditions were appropriate for native species we
expected to see, seed of three common grasses (Bouteloua gracilis (Willd. ex Kunth)
Lag. ex Griffiths, Bouteloua curtipendula (Michx.) Torr., Digitaria californica (Benth.)
Henr.) were added to pots containing only the sterilized growing medium.
When seedlings could be identified, they were counted and pulled to minimize
competition within pots. The top layer of the soil samples was stirred approximately four
weeks following initiation to encourage further germination. Seed bank samples were
maintained in the greenhouse until emergence ceased, typically after six months or less.
Seedlings recorded for the four pots per plot were combined for further analysis.
To characterize basic soil properties for the three sites, soil samples consisting of
six cores 7 cm in diameter and 5 cm depth were collected across each of the three study
sites, outside of plots in July 2004. Two such samples were collected each at CORO and
SRER; one sample was taken at TLF due to the homogeneity observed at the previously
tilled site. The samples taken from each site were combined and air-dried in paper bags.
The Soil, Water, and Plant Analysis Laboratory (Tucson, AZ) analyzed the samples for
pH, cation exchange capacity, phosphorus, potassium, total organic carbon, and texture.
105
Precipitation data
Daily precipitation data for the three study sites were downloaded from the
National Climatic Data Center (NOAA 2005). Stations for CORO and SRER were
located within three km of the respective study plots. However, the closest station to TLF
was located in Cascabel, AZ, approximately 15 km northwest of the study plots at TLF.
Precipitation was totaled for each site for winter months (December-January) and
summer months (June-September) of 2003 and 2004 to capture the bimodal precipitation
pattern evident in southern Arizona. Precipitation patterns at CORO and SRER were
similar over the study period; precipitation for each winter and summer month was
compared between these two sites using a non-parametric Wilcoxon/Kruskal-Wallis
(rank sums) test to detect differences in the temporal distribution of seasonal
precipitation.
Statistical analyses: Aboveground vegetation
To test whether there was a significant treatment or site effect on the plant
communities of the study plots, I performed several multivariate analysis of variance
(MANOVA) tests with each sampling date treated as a different dependent variable.
Dependent variables included percent plant cover, relative percent native cover, relative
percent nonnative cover, relative percent grass cover, relative percent forb cover, relative
percent annual cover, relative percent perennial cover, and species richness, each treated
individually. Independent variables included site, treatment, and the site*treatment
interaction. Repeated-measures data were analyzed with MANOVA due to potential
106
autocorrelation between consecutive sampling dates. Data were transformed with the
arcsine square-root transformation prior to analysis to increase homogeneity of variances
(Sokal and Rohlf 1995). To minimize spatial autocorrelation, the Greenhouse-Geisser
adjustment was applied to “within-subjects” effects (Greenhouse and Geisser 1959,
Littell et al. 1992). MANOVA analyses were performed using SAS v8.02 (SAS Institute
2001). Because FA04 samples were not allowed to completely germinate, these data were
excluded from analyses.
Aboveground community analyses
Following the removal of rare species (contributing <5%), Bray-Curtis similarity
coefficients (Bray and Curtis 1957) were calculated for aboveground abundance data
using a square-root transformation to balance the influence of dominant and rare species
(Clarke and Warwick 2001). Analysis of similarity (ANOSIM) was then used to test for
changes in community composition following experimental treatments. Within a
particular season, communities of treated plots at a site were compared to control plots at
the same site. ANOSIM was also used to test for differences from season to season within
each site and treatment combination. ANOSIM is a non-parametric permutation
procedure based on the ranking coefficients derived from similarity matrices. ANOSIM
calculates a test statistic (R) typically ranging from 0 to 1, where R=0 when there is an
equivalent degree of similarity between and within groups, and where R=1 when all
replicates within groups are more similar than any replicates from different groups
107
(Clarke and Warwick 2001). The statistic was recomputed 5,000 times/test using a Monte
Carlo permutation (Clarke and Warwick 2001).
When differences in species assemblages were detected using ANOSIM, the
particular species responsible for differences in overall community composition were
determined using the ‘Similarity Percentages’ (SIMPER) routine. SIMPER is a technique
that identifies the species responsible for particular aspects of a multivariate plot by
considering the contribution of species to both the average similarity within groups as
well as the average dissimilarity between groups (Clarke and Warwick 2001). All
community analyses were performed using PRIMER software v5.2.8 (PRIMER-E Ltd.
2001).
Seed bank abundance
To test whether there was a significant treatment or site effect on the abundances
of germinable seedlings in the seed bank of the study plots, I performed MANOVA with
each sampling date treated as a different dependent variable. Dependent variables
included total seeds germinated, percent native seedlings, percent graminoid seedlings,
and percent perennial seedlings. Independent variables included site, treatment, and the
site*treatment interaction. Repeated-measures data were analyzed with MANOVA due to
potential correlation between consecutive sampling dates. “Total number of seedlings”
was transformed with a square-root transformation prior to analysis and all other
variables were transformed using the arcsine square-root transformation to increase
homogeneity of variances. To minimize spatial autocorrelation, the Greenhouse-Geisser
108
adjustment was applied to “within-subjects” effects (Greenhouse and Geisser 1959,
Littell et al. 1992). To test whether site or treatment had an effect on the percent of
seedlings that were E. lehmanniana, I used MANOVA. Percent of seedlings that were E.
lehmanniana was the dependent variable and independent variables included site,
treatment, and the site*treatment interaction.
Seed bank community analyses
Seedlings counted were totaled for each plot and sampling event. The percent of
the total seedlings for each plot and sampling event was then calculated by species.
Following the removal of rare species (contributing <5%), Bray-Curtis similarity
coefficients were calculated for seed bank data using a square-root transformation to
balance the influence of dominant and rare species. ANOSIM was then used to test for
changes in community composition following experimental treatments. Within a
particular season, communities of treated plots at a site were compared to control plots at
the same site. ANOSIM was also implemented to test for differences from season to
season within each site and treatment combination. When differences in species
assemblages were detected using ANOSIM, the particular species underlying differences
in overall community composition were determined using SIMPER.
Comparison of abundance and soil seed bank datasets
To compare the composition of species represented in both the seed bank and the
vegetation survey, the data were combined into one matrix and analyzed using non-
109
metric multi-dimensional scaling (NMS) based on Bray-Curtis similarity values. Rare
(contributing <5%) species we removed and similarity values were calculated for data
using a square-root transformation. NMS is a non-parametric ordination technique that
constructs a configuration of the sample data such that similar samples are located in
close proximity in ordination space (Clarke and Warwick 2001).
Biomass data
Analysis of variance (ANOVA) was used to determine whether site had an effect
on the biomass of E. lehmanniana within a single season. Abundance in kilograms per
hectare of E. lehmanniana was the dependent variable and site was the independent
variable. The Tukey-Kramer HSD test for multiple comparisons was used to control the
overall error rate. A visual estimation of residual plots indicated that distributions were
not substantially skewed. Therefore, analyses were performed on untransformed data.
Results
Aboveground vegetation statistical analyses
Over the course of the study, 65 different species were recorded within study plots
at TLF, 79 species were recorded at SRER, and 80 species were recorded at CORO.
CORO and SRER shared 33 species in common, SRER and TLF shared 22 species in
common, and only 14 species were found at both CORO and TLF. For two of the
dependent variables tested (relative percent perennial cover and species richness), there
was a significant relationship three-way interaction between site, treatment, and sampling
110
season (p #0.038). The two-way interaction between treatment and sampling season was
significant for every other variable tested (p # 0.0012; Table 1). Finally, the two-way
interaction between site and sampling season was significant for total percent plant cover,
relative percent forb cover, relative percent grass cover, and relative percent annual cover
(p # 0.015; Table 1). The interactive effects of both site and treatment with sampling
season were significant for total plant cover; relative percent forb cover, relative percent
grass cover, and relative percent annual cover (p # 0.015; Table 1).
For total plant cover, control plots exhibited 1.5 to 3 times the percent plant cover
as treated plots (Figure 2). The site exhibiting the highest percent cover fluctuated by
season. Following treatment, total percent cover dropped sharply on treated plots across
all three sites for two seasons (FA03-SP04). Total percent cover began to rebound in
FA04 (Figure 2). Treated plots hosted from 3 to 25 times the amount of forb cover as
control plots. SRER consistently produced greater forb cover than CORO; TLF exhibited
high forb cover in both seasons in 2004 (Figure 3). Relative percent forb cover showed a
dramatic increase following treatment. The two fall seasons post-treatment exhibited
higher forb cover than the post-treatment spring season (Figure 3). Control plots
exhibited nearly double the relative grass cover than treated plots. Relative grass cover at
SRER and CORO fluctuated seasonally in a similar fashion; grass cover dropped off
sharply one year post-treatment at TLF. Finally, treated plots exhibited between 2 and 10
times the amount of relative cover by annual species than control plots (Figure 4). The
three sites showed a consistent pattern over the three post-treatment sampling seasons,
with TLF hosting the greatest proportion of annuals by cover and CORO consistently
111
hosting the least. Relative percent annual cover showed dramatic increases following
treatment across the three sites, with the greatest amounts appearing in the two fall
sampling seasons (Figure 4).
The interaction of treatment and sampling season had a significant “within
subjects” effect on relative percent native cover (p = 0.004; Table 1), with treated plots
exhibiting 1.5 to 4 times the native plant cover than control plots (Figure 5). Percent
native cover was higher in both fall seasons sampled than the spring sampling event.
Control plots exhibited rather consistent relative percent native cover, whereas treated
plots fluctuated between 35% and 65% relative native cover.
The three-way interaction between site, treatment, and sampling season was
significant for relative percent perennial cover (p = 0.038; Table 1). In FA03 and FA04,
control plots at CORO and SRER were not significantly different from each other, but
exhibited significantly higher relative percent perennial cover than all other site and
treatment combinations (Figure 6). In SP04, control plots at CORO and SRER again were
not significantly different from each other, but the differentiation between these plots and
treated plots at these same sites was not as clear (Figure 6).
The three-way interaction between site, treatment, and sampling date was
significant for species richness (p = 0.014; Table 1). In both FA03 and FA04, treated
plots at both SRER and CORO showed significantly higher richness values than control
plots at these sites (Figure 7). In SP04, treated plots had significantly higher species
richness than control plots at SRER, but this was not true at CORO. No significant
112
pattern could be detected in any season for TLF. Overall, species richness increased
slightly over the course of the study (Figure 7).
Aboveground vegetation – community analyses
In the pre-treatment season (SP03), composition on control plots was not
significantly different from treatment plots within any site (Table 2). Within FA03, the
first post-treatment sampling period, treated and control plots were significantly different
at both CORO and SRER. The same pattern held through SP04 and FA04. However,
treated and control plots were not differentiated during any season at TLF (Table 2).
Results of SIMPER analyses indicate that the differences between treated and
control plots at CORO in all post-treatment seasons were due mainly to the lack of E.
lehmanniana in treated plots, as well as increases in Mollugo verticillata L., and
Calliandra eriophylla Benth. (Table 2). The differences between treated and control plots
at SRER can also be mainly attributed to the removal of E. lehmanniana; treated plots
saw increases in Mollugo verticillata, Urochloa arizonica (Scribn. & Merr.) O. Morrone
& F. Zuloaga, and Kallstroemia grandiflora Torr. ex Gray in the two fall sampling
seasons and Ambrosia confertiflora DC. and Spermolepis echinata (Nutt. ex DC.) Heller
in SP04 (Table 2). All of these species are native to Arizona; most are annuals. However,
Calliandra eriophylla is a short-statured, perennial woody plant.
No pattern was detectable within the CORO control plots, and the only significant
difference in successive seasons in treated plots at CORO occurred between the first two
seasons sampled (SP03-FA03; p < 0.001; Table 3), due to the removal of E.
113
lehmanniana. Aside from the effect of the removal treatment, SIMPER analyses
indicated increases in Mollugo verticillata, Nissolia wislizeni (Gray) Gray, and
Calliandra eriophylla, all native species, from SP03 to FA03. Also attributable to the
removal of E. lehmanniana, treated plots at CORO were significantly different from
SP03 to SP04. SIMPER indicated increases in Calliandra eriophylla as the other major
contributing species (Table 3).
All seasons were significantly different (p < 0.001) for both control and treated
plots at SRER (Table 3). From SP03 to FA03, E. lehmanniana cover increased on control
plots by 20%; in addition, Mollugo verticillata increased by 3%. In SP04, Mollugo
verticillata was no longer present, a 15% decrease in E. lehmanniana was observed, and
several species appeared in small amounts (<1% cover), including Gilia scopulorum M.E.
Jones, Spermolepis echinata, Boerhavia coccinea P. Mill., and Ambrosia confertiflora.
FA04 was characterized by the disappearance of Gilia scopulorum, which is a spring
annual forb, and the reappearance of a small amount of Mollugo verticillata and
Urochloa arizonica (1% and 0.5%, respectively). Every one of these species is
considered native to Arizona.
Treated plots at SRER exhibited different community assemblages from control
plots, and each season was consistently different from the previous. Following removal of
E. lehmanniana, treated plots saw a dramatic increase in Mollugo verticillata (27%),
Kallstroemia grandiflora (4%), and Urochloa arizonica (4%) in FA03. In SP04, none of
these annual species were observed; Ambrosia confertiflora and Spermolepis echinata
114
became more prevalent. Finally, FA04 saw resurgences in Mollugo verticillata and
Urochloa arizonica.
The spring-to-spring and fall-to-fall differences observed for both SRER control
and treated plots are consistent across the treatments. SP03 was characterized by
Erigeron divergens Torr. & Gray, Lotus greenei Ottley ex Kearney & Peebles, and
Ambrosia confertiflora. SP04 exhibited Ambrosia confertiflora in greater amounts, as
well as Gilia scropulorum, Solanum elaeagnifolium Cav., and Spermolepis echinata. The
common species were rather consistent across the two falls sampled; Mollugo verticillata,
Kallstroemia grandiflora, and Urochloa arizonica were recorded in higher amounts in
FA03. FA04 also saw the proliferation of Boerhavia coccinea (Table 3).
The within-site/treatment ANOSIM tests show on-going changes in community
composition at TLF. For both control and treated plots, all seasons were different from all
others tested (Table 3). SIMPER results show a shift in both control and treated plots
from an E. lehmanniana-dominated community in SP03 to much more bare ground and
cover of Eragrostis cilianensis (All.) Vign. ex Janchen and Sporobolus cryptandrus
(Torr.) Gray in FA03. Control plots shifted from 59% E. lehmanniana cover in SP03 to
3% E. lehmanniana cover in FA03; treated plots shifted from 62% to 5% E. lehmanniana
cover in the same time. SP04 saw increases in Salsola kali L., and Sisymbrium irio L.with
concurrent declines in Eragrostis cilianensis and Sporobolus cryptandrus. Finally, in
FA04, a surge in Salsola kali (37% and 40% increases on treated and control plots,
respectively) and increases in Eragrostis cilianensis and Bouteloua aristidoides (Kunth)
Griseb. occurred. With the exception of Sporobolus cryptandrus, all of the dominant
115
species at TLF were annual species. Additionally, all were nonnative except Sporobolus
cryptandrus and Bouteloua aristidoides, a perennial grass and a short-lived summer
grass, respectively.
The spring-to-spring differences observed in control and treated plots at TLF
(Table 3) can be attributed mainly to the sharp decrease in E. lehmanniana following
SP03. The fall-to-fall differences are mainly due to the proliferation of Salsola kali in
FA04.
Seed bank statistical analyses
Over 22,000 seedlings germinated over the course of the study period. Of these,
more than 36% (8,152) were E. lehmanniana seedlings. Table 4 summarizes total
numbers of seedlings counted by sampling season. Fifty-nine species were recorded for
SRER seed bank samples. For TLF, 38 species were identified, and 32 species were
recorded for CORO. All three native grasses seeded to sterilized growing medium
(Bouteloua gracilis, Bouteloua curtipendula, and Digitaria californica) germinated,
demonstrating that conditions within the greenhouse were appropriate for these grass
species.
There were no significant three-way interactions between site, treatment, and
sampling date (p < 0.05) for total seeds germinated, percent graminoid seedlings, or seed
bank species richness (Table 5). The two-way interaction between site and sampling date
had a significant “within subjects” effect on both total seeds germinated and seed bank
species richness (p < 0.0001; Table 5). SRER germinated 3 to 8 times as many seeds as
116
CORO throughout the course of the study. TLF showed very high germination rates in
the first two sampling seasons, but then germination rates dropped off markedly (Figure
8). Samples from CORO consistently exhibited the lowest species richness (Table 5).
Richness values for all study sites fluctuated between approximately three and nine
(Figure 9).
The two-way interaction between site and sampling season was significant for
percent graminoid seedlings (p = 0.02; Table 5); similarly, the two-way interaction
between treatment and sampling season was significant for percent graminoid seedlings
(p = 0.0009; Table 5). Samples taken from treated plots had lower percentages of
graminoid seeds for all sample dates. TLF consistently exhibited the highest percentage
of graminoid seedlings, and SRER consistently germinated the fewest grass seedlings
(Figure 10).
The three-way interaction between site, treatment, and sampling date was
significant for percent perennial seedlings (p = 0.029; Table 5). Control plots at CORO
and SRER consistently exhibited higher percentages of perennial seedlings than treated
plots at these sites, but the response was mixed for TLF (Figure 11). Similarly, the threeway interaction between site, treatment, and sampling date was also significant for
percent native seedlings (p = 0.002; Table 5). Control plots at CORO had consistently
lower proportions of native seeds than treated plots at CORO, but this pattern was not as
clear for the other sites (Figure 12).
The three-way interaction between site, treatment, and sampling date was also
significant for the percent of the seedlings that were E. lehmanniana (p = 0.003; Table 5).
117
Patterns within parks were quite consistent, with treated plots at both CORO and SRER
germinating lower percentages of E. lehmanniana than control plots on the same sites in
every season sampled (Figure 13). The response was more mixed for TLF, but the
temporal pattern was consistent between treatments.
Seed bank community composition
Within control plots at all three sites, most seasons were significantly different
from their predecessor with respect to community composition (Table 6). Likewise,
within treated plots at all three sites, most seasons were significantly different from their
predecessors. However, the pattern of seasons that were different was not consistent
within sites (Table 6).
SIMPER analyses indicate the main differences between seed bank sampling
seasons for control plots at CORO to be attributed to fluctuations in the abundances of E.
lehmanniana, Pseudognaphalium arizonicum (Gray) A. Anderb., Gnaphalium palustre
Nutt., Mollugo verticillata, and Oxalis corniculata L., all native annual forb species with
the exception of E. lehmanniana. Seed bank samples from treated plots at CORO varied
slightly in composition of the same species as for control plots, as well as Erigeron
divergens (Table 6). Differences in seed bank composition between seasons in both
control and treated plots at SRER were mainly due to fluctuations in E. lehmanniana,
Pseudognaphalium arizonicum, Mollugo verticillata, Androsace occidentalis Pursh., and
Crassula connata (Ruiz & Pavón) Berger (Table 6), also all annual forb species.
Differences between seasons in both treated and control plots at TLF were mainly
118
characterized by fluctuations in the abundances of E. lehmanniana, E. cilianensis, Oxalis
corniculata, Panicum sp., Sporobolus cryptandrus, and Sisymbrium irio (Table 6). These
species are a mix of native and nonnative species, as well as grass and forb species.
The seed banks of treated and untreated plots were not significantly different at
TLF within any season sampled, and only differed at CORO for the SU04 sampling
period (Table 7). SIMPER results indicated differences in the abundances of Mollugo
verticillata, Oxalis corniculata, Erigeron divergens, Pseudognaphalium arizonicum, and
Eragrostis intermedia A.S. Hitchc. to account for the significant differences between
samples taken from control and treated plots in SU04 at CORO (Table 7). In addition to
differences in these species, the percent of germinating seeds that were E. lehmanniana
was markedly different between the two treatments (79% for samples from control plots;
44% for samples from treated plots).
The seed banks of treated and control plots were significantly different at SRER
in SP04, SU04, and FA04 (Table 7). In all three cases, the percent of germinating
seedlings that were E. lehmanniana was dramatically lower in treated plots than in
control plots (53% in control plots vs. 32% in treated plots in SP04; 54% in control plots
vs. 17% in treated plots in SU04; 8% from control plots vs. 3% from treated plots in
FA04). Other major differences in the seed bank composition were attributable to
Mollugo verticillata (64% of seeds from treated plots vs. 32% of seeds from control plots
in SP04; 66% of seeds from treated plots vs. 33% of seeds from control plots in SP04),
Oxalis corniculata, Pseudognaphalium arizonicum, Androsace occidentalis and Crassula
connata (Table 7).
119
Comparison of abundance and soil seed bank datasets
The non-metric multi-dimensional scaling ordination illustrates significant
compositional dissimilarities between the seed bank and aboveground plant communities
at each site. Within each site, seed bank sample communities and aboveground
communities are very clustered, with very little overlap (Figure 14). This suggests that
despite the season sampled, communities germinating from seed bank samples do not
represent the aboveground communities. These plots include all sampling seasons
together; this demonstrates that there is no substantial relationship between seed bank and
aboveground plant abundances in any of the seasons sampled and no evidence that a lag
relationship exists.
At CORO, of the 80 species observed in the aboveground vegetation, 16 were
germinated from seed bank samples. An additional sixteen species were germinated from
soil samples that were not observed in aboveground vegetation. Of the species residing in
the seed bank not expressed in aboveground vegetation, all but 3 species were natives,
and all but 2 species were annuals. Five species were graminoids; the rest were forb
species.
Of the 79 species recorded in aboveground vegetation at SRER, 13 appeared in
seed bank samples. Twenty-four other species were recorded in the seed bank samples
taken from SRER. Of these, only three are known to be nonnatives. Most species
recorded were annual species; about half were graminoids and half were forb species.
120
At TLF, of the 65 species recorded in aboveground vegetation at study plots, 19 appeared
in seed bank samples. Nineteen other species were recorded in TLF seed bank samples.
Of these, all but two species were natives, and only one species was perennial. Six of the
19 species were graminoids.
Aboveground biomass from field sites
Biomass of E. lehmanniana was much greater at CORO and SRER that at TLF
throughout the course of the study (Figure 15). Abundance of E. lehmanniana at TLF
dropped off markedly following SP03. Conversely, E. lehmanniana abundance increased
(p = 0.022) at SRER and remained the same at CORO following SP03 (Figure 15).
Other species which were abundant at CORO included Calliandra eriophylla (338
kg/ha; 95% CI: 176 to 498 kg/ha averaged over 5 sampling periods) and Bouteloua
eripoda (Torr.) Torr. (8 kg/ha; 95% CI: 0 to 18 kg/ha). Aside from ERLE, abundant
species at SRER were Solanum elaeagnifolium (16 kg/ha; 95% CI: 0 to 30 kg/ha),
Kallstroemia grandiflora (10 kg/ha; 95% CI 0 to 24 kg/ha), and Mollugo verticillata (8
kg/ha; 95% CI: 0.8 to 14.8 kg/ha). The most common species at TLF were Salsola kali
(620 kg/ha; 95% CI: 246 to 744 kg/ha), Sporobolus cryptandrus (74 kg/ha; 95% CI: 26 to
104 kg/ha), and Cynodon dactylon (L.) Pers. (90 kg/ha; 95% CI: 36 to 128 kg/ha).
Soils properties
Soil pH fell within the expected range of 6.2 to 6.9 at SRER (Table 8). Soil pH
measurements for CORO were slightly lower than expected based on soil survey
121
information, and the pH measured at TLF was lower than predicted by the soil survey
(Svetlik unpublished data). Additionally, texture analysis suggested TLF was
characterized as a sand with 4.5% clay, much different from the silty clay loam suggested
by the soil survey (Svetlik unpublished data). Soil electrical conductivity (EC), total
organic carbon, and potassium were highest at TLF (Table 8). Soil pH was on average
approximately 0.5 unit lower at CORO than at SRER, which equates to approximately
three times the acidity. Values for other soil characteristics did not vary perceptibly
between SRER and CORO.
Precipitation data
Total precipitation over the study period was much less at TLF than at the other
two sites (Figure 16); CORO and SRER followed similar patterns over this period. At
TLF, seasonal precipitation totals were much lower than at CORO and SRER. SP03
precipitation measured just less than 10 cm, and FA03 precipitation was even lower, at
9.4 cm. SP04 totaled 9.5 cm and FA04 measured just over 17.8 cm. Because precipitation
was so different at TLF than at the other two sites, it was treated separately for further
analyses.
For SRER, precipitation was much more evenly distributed across winter months
in 2004 than in 2003 (Figure 17a). The same pattern was generally true for CORO, but
precipitation was not as evenly distributed across winter months in 2004 for CORO as for
SRER. Within the monsoon season (June-September), moisture was more evenly
distributed across the months at CORO than at SRER, especially in 2004 (Figure 17b).
122
December precipitation differed in event size and frequency at CORO in 2002 and
2003 (one-sided p-value = .003, from the rank-sum test). In addition, January
precipitation differed at CORO between 2003 and 2004 (one-sided p-value = .013, from
the rank-sum test). There was not strong evidence of a difference between the two years
for either February (one-sided p-value = .462, from the rank-sum test) or March (onesided p-value = .494, from the rank-sum test) between the two years.
At SRER, there was suggestive but inconclusive evidence of a difference between
2003 and 2004 January precipitation (one-sided p-value = .096, from the rank-sum test).
Differences between 2002 and 2003 precipitation at SRER were even less suggestive
(one-sided p-value = .125, from the rank-sum test). Precipitation did not differ between
2003 and 2004 for the months of February (one-sided p-value = .278, from the rank-sum
test) or March (one-sided p-value = .780, from the rank-sum test) at SRER.
June precipitation differed in event size and frequency at SRER in 2003 and 2004
(one-sided p-value = .078, from the rank-sum test). August precipitation differed at
SRER in 2003 and 2004 (one-sided p-value = .059, from the rank-sum test). However,
there was little evidence of a difference between years at SRER for July (one-sided pvalue = .970, from the rank-sum test) or September (one-sided p-value = .416, from the
rank-sum test).
There was little evidence of differences in precipitation in the month of June
between 2003 and 2004 at CORO (one-sided p-value = .394, from the rank-sum test).
Also, there was not strong evidence for a difference between the two years for July (one-
123
sided p-value = .451, from the rank-sum test), August (one-sided p-value = .122, from the
rank-sum test), or September (one-sided p-value = .400, from the rank-sum test).
Discussion
This study suggests that the post-treatment effects of removing an introduced
species are mediated by both precipitation variability (timing and amount) and local site
history and edaphic conditions. The net result of these complex interactions are
aboveground plant communities dominated in the short-term by several native species
that bear little resemblance to the membership of their corresponding seed bank.
Treatment Effect
All three study sites showed immediate responses to the removal of a dominant
nonnative species, but the magnitude and composition of the responses varied by
location. First, as expected, removing E. lehmanniana resulted in significant drops in
total plant cover in treated plots, which rebounded following the initial post-treatment
reduction. The removal also resulted in significant increases in species richness at CORO
and SRER that were sustained for the duration of the experiment. This finding is
consistent with several other removal experiments where removal of a dominant species
resulted in increases in species richness (Abul-Fatih and Bazzaz 1979, Hils and Vankant
1982, Armesto and Pickett 1985, Farnsworth and Meyerson 1999). Responses to the
complete removal of E. lehmanniana could potentially be stronger than those observed in
124
this study; treatment plots were surrounded by thick stands of E. lehmanniana, providing
ample seed source and the opportunity for the plant to vegetatively encroach.
Following removal of the dominant nonnative species, substantial increases in the
cover of annual species, forb species, and native species were observed. These increases
are not attributable to the increase in one or two species, but rather to broader shifts in
community composition. At SRER, increases in the cover of Mollugo verticillata,
Kallstroemia grandiflora, and Ambrosia spp. comprised the main differences between
treated and untreated plots in FA03 and FA04 (Table 2). These species are all native
annual forbs. However, Urochloa arizonica, a native annual graminoid, and Solanum
elaeagnifolium and Boerhavia coccinea, both native perennial forbs, also contributed to
these compositional changes. At CORO, species demonstrating the greatest increases on
treated plots were Mollugo verticillata, Nissolia wislizeni, Evolvulus nuttallianus J.A.
Schultes, and Calliandra eriophylla (Table 2). Calliandra eriophylla is a native perennial
shrub. Both Nissolia wislizeni and Evolvulus nuttallianus are native perennial forb
species. Similarly, other community-level studies have demonstrated an increase in less
common species following removal of the dominant species (Hils and Vankat 1982,
Silander and Antonovics 1982, Farnsworth and Meyerson 1999). The control and treated
communities were not significantly different at TLF.
The removal treatment had moderate to strong positive impacts on the proportion
of the seed banks of the three sites that were E. lehmanniana, suggesting the potential for
site restoration. The percent of seeds that were graminoids, the percent that were
perennial species, and the percent that were native species differed by both site and
125
treatment. The difference between treatments reflects reduced number of E. lehmanniana
seeds in the seed banks of treated plots. Removing the aboveground E. lehmanniana from
treated plots resulted in a substantial decrease in E. lehmanniana seeds germinating in
soil samples from these plots. This phenomenon is critical from a site restoration
perspective: Removing the aboveground E. lehmanniana permanently has a large impact
on the E. lehmanniana seed bank. This is in opposition to several other removal studies
of exotic species, where these exotics were observed to restore dominance of aboveground vegetation from the seed bank following a disturbance (Major and Pyott 1966,
Pyke 1990, Kyser and DiTomaso 2002). This may be due to particular life history traits
of E. lehmanniana and edaphic factors of semi-desert grasslands.
Influence of land use history
Overall TLF, a site with a very different land use history than the other two study
sites, showed very different responses from SRER and CORO for many of the variables
tested in this study, indicating that TLF is responding differently than the other two sites.
Because E. lehmanniana individuals in southeastern Arizona can be traced to seed grown
from a single accession (Crider 1945), it can be concluded that differences observed
among the study sites is not attributable to genetic differences.
At the outset of this study, TLF was dominated by E. lehmanniana (nearly 60%
absolute cover in December 2002, T. Mau-Crimmins, unpublished data), though the
plants were reduced in size compared with CORO and SRER. The small plant size is
evident in the biomass measures (Figure 15). However, following SP03, E. lehmanniana
126
nearly vanished from the site regardless of treatment. Total plant cover dropped sharply
in FA03, and the plants observed in this season were primarily E. lehmanniana and E.
cilianensis. This is likely the result of very low total precipitation measured for FA03 at
this site. SP04 saw a dramatic increase in plant cover, specifically annual forbs –
primarily Sisymbrium irio. The same pattern held in FA04, primarily due to a large
increase in Salsola kali (Table 3). Both of these species are nonnative annual forb
species. Sisymbrium irio is considered an invasive species by both the Western Society of
Weed Science and the Southern Weed Science Society (Whitson et al. 1996, Southern
Weed Science Society 1998). Salsola kali is considered invasive in several states (NRCS
PLANTS database 2005). However, it is worth noting that the cover of Sporobolus
cryptandrus, a native perennial bunchgrass, increased from 0.2% (95% CI: 0 to 0.4%) to
an average of 7% (95% CI: 3.3 to 10.5%) over the course of the study in both control and
treated plots at TLF, with 21% cover in one plot in FA04. Sporobolus cryptandrus is one
of the few native perennial grasses that is persistent in seed banks (Coffin and Laurenroth
1989). No forb species showed such patterns at TLF.
There were no discernible differences between control and treated plots for total
plant cover, species richness, percent relative native cover, percent relative forb or grass
cover, or percent relative annual or perennial cover at TLF. In addition, treated and
control plots were not significantly different on a community basis. The absence of
difference between control and treated plots at TLF for any of the dependent variables
tested as well as the community-level analyses suggests that TLF is not responding to the
removal of E. lehmanniana. Rather, TLF seems to be hosting one opportunistic weedy
127
species after another, responding to fluctuations in seasonal precipitation. This response
could be in large part due to the disturbance history of the site – the fields were
intensively farmed for decades prior to this study. The soil is very sandy and it appears to
lack aggregates and structure, which are important for soil moisture retention (Brady and
Weil 2002). In desert and semi-desert ecosystems, precipitation is the dominant
controlling factor, constraining plant growth and survivorship as well as other ecological
processes such as carbon and nitrogen fixation (Noy-Meir 1973, Cui and Caldwell 1997,
Schwinning et al. 2003, Belnap et al. 2004). In such a water-limited system, the ability
for soils to retain plant-available moisture is critical.
Electrical conductivity was at least two times higher at TLF than the other
samples. Though the salinity measurement from this site is not considered high enough to
injure plants, it is considered to be “medium salinity” and can impact the composition of
plants occupying the site (Silvertooth 2001, Bauder 2005). Soil salinity affects plants’
ability to maintain osmotic balance; the more saline the soils, the stronger the pull of
water from plants (Brady and Weil 2002). The presence of soil salts at TLF, in
conjunction with sandy texture lacking aggregates, may yield much lower plant-available
water than at the other two sites, even under adequate precipitation amounts.
One interesting and potentially encouraging pattern observed at TLF is the
steadily increasing species richness over the course of this study. Over this period,
average species richness for both control and treated plots at TLF increased from just
over 5 (95% CI: 4.5 to 6.5) to over 13 species per plot (95% CI: 12.2 to 14.3). This may
suggest a slow improvement in conditions, from a site dominated by one nonnative
128
species to a more diverse community. This pattern also may be due to the dramatic
increase in precipitation recorded in FA04 from FA03. Either way, these results suggest
that even in a semi-arid environment on a heavily impacted site, progression toward a
stable, native community may be possible and may be occurring at TLF.
It is also interesting to note that the increase in aboveground species richness was
not matched by a concomitant increase in seed bank species richness. The seed bank at
TLF, similar to the aboveground vegetation, appears to be undergoing major shifts. The
first two sampling periods germinated large quantities of seeds, the majority of which
were E. lehmanniana seedlings. The E. lehmanniana proportion of the seed bank tapered
off dramatically over the course of the study, concurrent with the drop-off in
aboveground E. lehmanniana. The relative paucity of E. lehmanniana seeds in later seed
bank samples can be explained by the lack of seed rain inputs. Later seed bank samples
exhibited relatively large increases in the abundance of Sporobolus cryptandrus and
Panicum sp. seedlings; both grass species and likely both native as there are no known
nonnative Panicum species in the area. The discrepancies between species richness of
aboveground vegetation and the seed bank suggests that many of the species observed in
aboveground vegetation arrived from off-site, rather than as a result of viable seeds
remaining latent in the seed bank. The species exhibiting the greatest consistencies
between aboveground vegetation and seed bank were Sporobolus cryptandrus, E.
lehmanniana, and E. cilianensis, all grass species. These similarities may explain the
slight overlap observed in aboveground vegetation and seed bank samples in the NMS
plot (Figure 14).
129
Eragrostis lehmanniana produces abundant seed (Allison 1998). Therefore, the
relative rarity of E. lehmanniana seeds in later seed bank samples from TLF implies a
decrease in E. lehmanniana seeds in these seed bank samples. However, E. lehmanniana
seeds require approximately six to nine months of afterripening, but can remain dormant
for much longer (Weaver and Jordan 1986, Voigt et al. 1996). One study reported only
10% germination after 88 weeks post-harvest (Hardegree and Emmerich 1993),
suggesting that a large number of viable E. lehmanniana seeds may have still remained in
the soil samples but had not yet broken dormancy. To address this unanswered question,
subsequent longer-term seed bank studies should be undertaken.
Species that were observed in the seed bank but not in aboveground vegetation
can offer some additional clues to the site’s potential for restoration. Several species
appeared in very small quantities (one or two seedlings), but the abundance of a few key
native grasses suggest a viable seed bank. Over the course of the six sampling seasons,
402 Panicum sp. seedlings were recorded and 41 Leptochloa panicea (Retz.) Ohwi ssp.
brachiata (Steudl.) N. Snow seedlings were counted. Leptochloa panicea is a native
perennial graminoid, and Panicum sp. is a graminoid, likely native due to the lack of
nonnative Panicum species in the area. The next three most commonly germinating
species in this group were also native species, Chamaesyche micromera (Boiss. ex
Engelm.) Woot. & Standl., Eriochloa acuminata (J. Presl) Kunth, and Erigeron
divergens. Two of the most common species germinating in the seed bank that were also
observed in the aboveground vegetation were native perennial grasses, Sporobolus
cryptandrus and Eragrostis intermedia. However, these seed numbers are swamped by
130
the nonnative grasses observed at the site; 4,005 E. lehmanniana seedlings were counted
over the seed bank study and 2,382 E. cilianensis seeds were recorded. These facts
suggest that though some native species are increasing in numbers and cover, the study
sites harbor many abundant nonnative species providing plenty of competition for limited
nutrients at this site.
Similarities and differences between two study sites with similar land use histories
Removal effects on plant communities and seasonal dynamics
Overall, the communities sampled at CORO and SRER responded similarly over
the course of this study. E. lehmanniana biomass followed the same seasonal patterns
(Figure 15), exhibiting peaks in the fall seasons sampled. Treated plots at both sites
exhibited increases in relative native cover following treatments. In addition, both study
sites experienced similar increases in relative forb and annual cover in the two posttreatment fall seasons. For several of the variables tested, differences between the two
study sites were not significant.
The results of the within-site/treatment ANOSIM tests reveal interesting patterns
that suggest that SRER and CORO, though similar in some ways, are responding
differently to the removal of E. lehmanniana. On control plots at CORO, each successive
season was not significantly different from the previous, indicating a stable, E.
lehmanniana-dominated community. However, each successive season is significantly
different from the previous on control plots at SRER, suggesting fluctuations in the nonE. lehmanniana portion of the community. Indeed, the non-E. lehmanniana portion of
131
these communities showed little overlap through the four seasons (Table 2). Nearly all of
the species observed in all of these seasons were native annual species, primarily forbs.
The exception is E. curvula, a nonnative perennial grass related to E. lehmanniana and
introduced from South Africa at the same time (Crider 1945).
Every successive season was significantly different from the previous on treated
plots at SRER from a community perspective, suggesting considerable changes in
community composition from one season to the next. The shifts in community
composition in treated plots are caused by many of the same species listed for control
plots at SRER above; however, in treated plots, they typically occurred in much greater
amounts.
From a community perspective, only one successive season was significantly
different from the previous on treated plots at CORO: FA03 differed from the previous
spring. This difference is explained by the removal of E. lehmanniana from these plots.
The lack of other significant changes from one season to the next indicates consistent
community composition in treated plots. Every one of the most common species recorded
is native to Arizona.
Climate and soil influences on the effects of removal
Monthly precipitation values during the monsoon seasons of 2003 and 2004
differed in both June and August at SRER, but no such differences were detectable at
CORO (Figure 17). The more consistent rainfall at CORO during this wet period may
explain the similar communities observed at CORO in FA03 and FA04. The sporadic
132
rainfall observed at SRER, especially in the summer of 2004, may explain the differences
in community composition and cover values in both treated and untreated plots at SRER.
Much of the 2004 monsoon season rainfall recorded at SRER occurred in July,
specifically in a few large events. Recent greenhouse and field studies have demonstrated
the sensitivity of native species to variable rainfall patterns (Frasier et al. 1985, Frasier
1989, Abbott et al. 1995, Abbott and Roundy 2003). High and rapid germination rates of
native species are common following initial rainfall events. However, without consistent
rainfall following germination, these species are susceptible to desiccation (Abbott and
Roundy 2003). The 2004 monsoon season rainfall at SRER exhibits the pattern of rapid
bursts of precipitation followed by long dry periods that could have impacted native
species negatively. The presence of Kallstroemia granidlfora at SRER in large quantities
in FA03 but not in FA04 supports this hypothesis. Kallstroemia granidlfora is a summer
annual that responds strongly to summer rains (Dimmitt 2000).
Between CORO and SRER, one notable difference in the soil characteristics
measured was pH, which was on average approximately three times more acidic at
CORO than at SRER. Soil pH has a large influence on root uptake of nutrients as well as
the formation of aggregate structure (Brady and Weil 2002). In addition, pH is a major
determinant of the species which can grow at a site (Brady and Weil 2002). The other
notable difference in the soil characteristics measured at the two sites was potassium,
which was slightly higher in both samples at SRER than at CORO. The values for both
sites are considered to be in the low range for dryland ecosystems (Soltanpour and Follett
1999). Potassium is the third most likely nutrient to limit plant growth, after nitrogen and
133
phosphorus, and plays an important role in plants’ drought tolerance and disease
resistance (Brady and Weil 2002). However, potassium levels apparently are not highly
correlated with grassland organization or composition (Curtis 1955, Aerts et al. 1990,
Piper 1994). It appears that at least for the edaphic characteristics measured, soils are not
responsible for the differences observed between CORO and SRER in this study. Rather,
differences in the timing and amount of precipitation appear to be important mediating
factors.
Seasonal and removal effects on seed bank
As with aboveground communities, seed bank samples from CORO and SRER
also exhibit strong similarities. Fluctuations in the proportion of samples that were E.
lehmanniana seeds were very consistent across the two sites, suggesting stable, wellestablished communities at each site responding similarly to seasonal cues. Untreated
plots at both sites germinated much higher proportions of E. lehmanniana seeds than
treated plots, responding to the consistent aboveground inputs of these plots. The two
sites also exhibited similar seasonal patterns in total seeds germinated and species
richness, though the magnitude of the fluctuations was greater at SRER.
On a community basis, seed bank samples from treated plots were significantly
different from control plots at SRER on three occasions and once at CORO in the second
year of the study (Table 6). These differences were in part due to the smaller numbers of
E. lehmanniana seeds germinating in samples from treated plots at these sites, but in most
cases, other species, typically native species, were responsible for more of the
134
dissimilarity than E. lehmanniana. This demonstrates the potential for a strong response
on behalf of native species residing in the seed bank to the removal of E. lehmanniana on
these sites.
Correspondence between plant communities and seed banks
Species that appeared in the seed banks at SRER but not in aboveground
vegetation were mainly annual species, many of which occurred in very small numbers.
The most abundant seed bank species, by far, was Crassula connata, a mat-forming
succulent annual, with more than 3,500 individuals germinating. Also abundant were
Androsace occidentalis, Oxalis corniculata, and Gnaphalium palustre, with 445, 394,
and 112 seedlings, respectively. Each of these species is a native annual forb. Several
other native annual forbs were abundant in both the seed bank and the aboveground
vegetation; these included Mollugo verticillata, Pseudognaphalium arizonicum,
Eriastrum diffusum (Gray) Mason, Erigeron divergens, and Spermolepis echinata. Seeds
of native grasses generally do not have persistent seed banks (Rabinowitz 1981, Coffin
and Laurenroth 1989, Kinucan and Smeins 1992), so it is not surprising that their seeds
did not appear in the seed bank. The presence of native annuals is encouraging, as these
species could mitigate erosion and provide a safe site for native perennial establishment
following nonnative removal.
At CORO, the most common species germinating in the seed bank but not
appearing in aboveground vegetation was Pseudognaphalium arizonicum. Oxalis
corniculata, Sisymbrium irio, and Veronica peregrina L. were also common. All of these
135
are native annual forbs, with the exception of Sisymbrium irio, which is not native to the
continent. The most abundant species in the seed bank that also appeared in the
aboveground vegetation at CORO included Muhlenbergia sp., Pseudognaphalium
arizonicum, and Oxalis corniculata. Similar to at SRER, the lack of perennial grass
species is not surprising, but the presence of native annual species is heartening.
Inconsistencies between aboveground vegetation and seed banks are often
observed (Thompson and Grime 1979, Rabinowitz 1981, Hills and Morris 1992, Warr et
al. 1993, Arkle et al. 2002). It has been suggested that species observed in seed banks but
not expressed in aboveground vegetation are a “memory” of past ecological conditions at
the site and may germinate if these conditions reappear (Templeton and Levin 1979,
Rabinowitz 1981). Another viable explanation for seed bank species not appearing in
aboveground vegetation is lack of necessary soil moisture in the field, which is one of the
most important site characteristics for germination (Winkel et al. 1991). Regardless, these
species offer insight into the possible trajectory of the site following a large-scale
removal of E. lehmanniana. The results of the E. lehmanniana removal experiment and
the seed bank study suggest that upon removal of E. lehmanniana at both CORO and
SRER, the native species latent in the seed bank begin to appear in aboveground
vegetation. E. lehmanniana also continues to reappear from seed as expected due to its
high degree of dormancy (Weaver and Jordan 1986, Voigt et al. 1996). However, the
numbers of E. lehmanniana are reduced considerably when the aboveground E.
lehmanniana seed inputs are removed.
136
Native species observed in both seed banks and aboveground vegetation were in
large part annual species. Removing E. lehmanniana from sites heavily dominated by this
species resulted in the removal of nearly all plant cover on treatment plots at CORO and
SRER. The proliferation of annual species following this disturbance could be the natural
response of plants in this area. It may be possible to expect change to continue on sites if
E. lehmanniana were to continue to be excluded from treatment plots, and result in the
eventual establishment of native perennial species. Further studies are necessary to
determine whether such a pattern would occur.
Conclusion
Although the negative impacts of introduced species are widely recognized
(Chapin et al. 1996, Mack and D’Antonio 1998, Wilcove et al. 1998), few studies are
undertaken to examine the response of a site following a large-scale eradication effort.
Eradication or control is often the preferred response to an invasion by a particularly
problematic species, but as Zavaleta et al. (2001) propose, pre-eradication assessments
are necessary to determine whether a removal will result as expected and not augment the
problem. The results of this study suggest that the response of a site to the removal of a
dominant nonnative grass varies between sites. In this study, a site with a history of
intensive agriculture showed no strong response to the removal of a dominant nonnative
species. Two other sites with a history of livestock grazing demonstrated similar strong
responses to the removal, with large increases in native plant cover, increases in species
richness and no evidence of “new” nonnative species replacing removed species. The
137
findings from the second two sites are consistent with other experiments removing
dominant nonnative species (Farnsworth and Meyerson 1999, Morrison 2002).
Community composition of the treated plots at the three sites varied considerably,
and reflected common species near each site. Seed bank studies revealed a small number
of additional species not observed in the aboveground vegetation for each site. These
patterns suggest that removing the nonnative from a site would result in the site’s
conversion to an ecosystem dominated by native plants. However, if the goal is
restoration of the site to a pre-invasion grassland, it may be advisable to undertake
restoration measures in conjunction with removal efforts. Seeds of native perennial
species may promote transition from an annual-dominated site to a more diverse,
perennial-dominated community.
The results of this experiment demonstrate that removing E. lehmanniana causes
an increase in relative percent cover of native species, annual species, and forb species.
The response varied by site, which could be the result of differences in land use history,
precipitation patterns, soil characteristics, or other edaphic factors. In order to ascertain
whether differences in responses were the result of various factors, further
experimentation modifying these variables is necessary.
Several studies have suggested that the responses observed following a species
removal could be the result of indirect effects including the concurrent soil disturbance,
breakdown of aboveground or root biomass, and transient responses in nutrient
availability (Underwood 1986, Aarssen and Epp 1990, Campbell et al. 1991). The short
time span of this study (less than two years post-treatment) precludes determination of
138
the importance of these effects. Removal of competition for resources including space,
light, or nutrients could also be responsible for the response. Further monitoring of the
study plots would demonstrate the sites’ long-term response to the removal.
139
References
Aarssen. L.W. and G.A. Epp. 1990. Neighbor manipulations in natural vegetation: a
review. Journal of Vegetation Science 1:13-30.
Abbott, L.B. and B.A. Roundy. 2003. Available water influences field germination and
recruitment of seeded grasses. Journal of Range Management. 56:56-64.
Abbott, L.B., B.A. Roundy, and S.H. Biedenbender. 1995. Seed fate of warm-season
perennial grasses. Pages 37-43 in B.A. Roundy, E.D. McArthur, J.S. Haley, and
D.K. Mann, eds. Wildland Shrub and Arid Land Restoration Symposium, 19-21
October 1993. Las Vegas, NV. General Technical Report INT-GTR-315. Ogden:
Forest Service Intermountain Research Station.
Abul-Fatih, H.A. and F.A. Bazzaz. 1979. The biology of Ambrosia trifida. L.I. Influence
of species removal on the organization of the plant community. New Phytology
83:813-816.
Aerts,R., F. Berendse, H. de Caluwe, and M. Schmitz. 1990. Competition in heathland
along an experimental gradient of nutrient availability. Oikos 57:310-318.
Allison, C. 1988. Seeding New Mexico rangeland. Circular 525. Las Cruces, NM: New
Mexico State University, College of Agriculture and Home Economics,
Cooperative Extension Service. 15 p.
Anable, M.E., M.P. McClaran, and G.B. Ruyle. 1992. Spread of introduced Lehmann
lovegrass Eragrostis lehmanniana Nees. in Southern Arizona, USA. Biological
Conservation 61:181-188.
Arkle, P.J., D.C. Malcolm, and C.J. Nixon. 2002. Role of the soil seed bank in the
restoration of a native pinewood at Glen Garry, Inverness-shire. Botanical Journal
of Scotland 54:221-236.
Armesto, J.J. and S.T.A. Pickett. 1985. Experiments on disturbance in old-field plant
communities: impact on species richness and abundance. Ecology 66:230-240.
Bauder, J. 2005. Soil Quality and Water Quality Texture Salinity Sodium Water.
http://waterquality.montana.edu/docs/irrigation/agnote73.shtml Accessed: Jan 19
2005.
Belnap, J., S.L. Phillips, M.E. Miller, and S.D. Flint. 2004. Response of desert biological
soil crusts to alterations in precipitation frequency. Oecologia 141:306-316.
140
Biedenbender, S.H. and B.A. Roundy. 1996. Establishment of native semidesert grasses
into existing stands of Eragrostis lehmanniana in Southeastern Arizona.
Restoration Ecology 4:155-162.
Billings, W.D. 1994. Ecological impacts of cheatgrass and resultant fire on ecosystems in
the Western Great Basin. Pages 12-30 in S.B. Monson and S.G. Kitchen, eds.
Proceedings—ecology and management of annual rangelands. General Technical
Report INT-GTR-313. Ogden: U.S. Department of Agriculture, Forest Service,
Intermountain Research Station.
Bock, C.E., J.H. Bock, K.L. Jepson, and J.C. Ortega. 1986. Ecological effects of planting
African lovegrasses in Arizona. National Geographic Research 2:456-463.
Brady, N.C. and R.R. Weil. 2002. The Nature and Properties of Soils. Thirteenth Edition.
Upper Saddle River: Prentice Hall.
Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of
southern Wisconsin. Ecological Monographs 27:325-349.
Brown, D. 1992. Estimating the composition of a forest seed bank: a comparison of the
seed extraction and seedling emergence methods. Canadian Journal of Botany
70:1603-1612.
Burquez, A. and Quintana, M., 1994. Islands of biodiversity: ironwood ecology and the
richness of perennials in a Sonoran Desert biological reserve. Pages 9-27 in G.P.
Nabhan and J.L. Carr, eds. Ironwood: An Ecological and Cultural Keystone of the
Sonoran Desert, Occasional Papers in Conservation Biology, Occasional Paper
No. 1,Washington, DC.
Cable, D.R. 1971. Lehmann lovegrass on the Santa Rita experimental range, 1937-1968.
Journal of Range Management 24:17-21.
Campbell, B.D., J.P. Grime, J.M.L. Mackey, and A. Jalili. 1991. The quest for a
mechanistic understanding of resource competition in plant communities: the role
of experiments. Functional Ecology 5:241-253.
Chapin, F.S., III, H.L. Reynolds, C.M. D’Antonio and V.M. Eckhart. 1996. The
functional role of species in terrestrial ecosystems. Pages 403-430 in B. Walker
and W. Steffan, eds. Global Change in Terrestrial Ecosystems. Cambridge:
Cambridge University Press.
Clarke, K. R., and R. M. Warwick. 2001. Change in Marine Communities: An
Approach to Statistical Analysis and Interpretation, 2nd Edition. Plymouth:
Primer-E, Ltd.
141
Crider, F.J., 1945. Three introduced lovegrasses for soil conservation. United States
Department of Agriculture Circular No. 730.
Coffin, D.P. and W.K. Laurenroth. 1989. Spatial and temporal variation in the seed bank
of a semiarid grassland. American Journal of Botany 76:53-58.
Cox, J.R. and M.H. Martin. 1984. Effects of planting depth and soil texture on the
emergence of four lovegrasses. Journal of Range Management 37:204-205.
Cox, J.R. and G.B. Ruyle. 1986. Influence of climatic and edaphic factors on the
distribution of Eragrostis lehmanniana Nees. in Arizona, USA. Journal of the
Grassland. Society of South Africa 3:25-29.
Cui, M. and M.M. Caldwell. 1997. A large ephemeral release of nitrogen upon wetting of
dry soil and corresponding root responses in the field. Plant Soil 191:291-299.
Curtis, J.T. 1955. A prairie continuum in Wisconsin. Ecology 36:558-566.
D’Antonio, C.M. and P.M. Vitousek. 1992. Biological invasions by exotic grasses, the
grass/fire cycle and global change. Annual Review of Ecology and Systematics
23:63-87.
D’Antonio, C.M., H.R. Flint, M. Mack, D. Hitchcock, and P.M. Vitousek. 1998. The
response of native species to removal of invasive exotic grasses in a seasonally
dry Hawaiian woodland. Journal of Vegetation Science 9:699-712.
Denny, D.W. and C.R. Peacock. 2000. Soil Survey of Coronado National Memorial,
Arizona. USGS Sonoran Desert Field Station Technical Report No. 63.
Dimmitt, M. 2000. Plant ecology of the Sonoran Desert region. Pages 127-152 in S. J.
Phillips and P. W. Comus, eds. A Natural History of the Sonoran Desert.
Tucson: Arizona-Sonora Desert Museum Press.
Farnsworth, E.J. and L.A. Meyerson. 1999. Species composition and inter-annual
dynamics of a freshwater tidal plant community following removal of the invasive
grass, Phragmites australis. Biological Invasions 1:115-127.
Fisher, M.J., I.M. Rao, M.A. Ayarza, C.E. Lascano, J.I. Sanz, R.J. Thomas, and R.R.
Vera. 1994. Carbon storage by introduced deep-rooted grasses in the South
American savannas. Nature 371:236-238.
142
Frasier, G.W. 1989. Characterization of seed germination and seedling survival during
the initial wet-dry periods following sowing. Journal of Range Management
42:299-303.
Frasier, G.W., J.R. Cox, and D.A. Woolhiser. 1985. Emergence and survival responses of
seven grasses to six wet-dry sequences. Journal of Range Management 38:372377.
Geiger, E.L. and G.R. McPherson. 2004. Maintaining grasslands: the interaction of fire,
climate, and nonnative species. Ecological Society of America Annual Meeting,
Portland, OR, July 31 – Aug 6.
Gordon, D.R., J.M. Welker, J.W. Menke, and K.J. Rice. 1989. Competition for soil water
between annual plants and blue oak (Quercus douglasii) seedlings. Oecologia
79:533-541.
Greenhouse, S. W., and S. Geisser. 1959. On methods in the analysis of profile data.
Psychometrika 32:95-112.
Greig-Smith, P. 1983. Quantitative Plant Ecology. Berkeley: University of California
Press.
Gurevitch, J. and S.L. Collins. 1994. Experimental manipulation of natural plant
communities. Trends in Ecology and Evolution 9:94-98.
Hardegree, S.P. and W.E. Emmerich. 1993. Germination response of hand-threshed
lehmann lovegrass seeds. Journal of Range Management 46:203-207.
Hendricks, D.M. 1985. Arizona Soils. Tucson: University of Arizona Press.
Hills, S.C. and D.M. Morris. 1992. The function of seed banks in Northern Forest
Ecosystems (a literature review). Forest Research Information Paper 107.
Ministry of Natural Resources, Ontario.
Hils, M.H. and J.L. Vankant. 1982. Species removals from a first-year old-field plant
community. Ecology 63:705-711.
Johnson, N.C. and D.A. Wedin. 1997. Soil carbon, nutrients and mycorrhizae during
conversion of dry tropical forest to grassland. Ecological Applications 7:171-182.
Kinucan, R.J. and F.E. Smeins. 1992. Soil seed bank of a semiarid Texas grassland under
three long-term (36 years) grazing regimes. American Midland Naturalist 128:1121.
143
Kuvlesky et al. 2002
Kyser, G.B. and J.M. DiTomaso. 2002. Instability in a grassland community after the
control of yellow starthislte (Centaurea solstitialis) with prescribed burning.
Weed Science 50:648-657.
Littell, R. C., R. J. Freund, and P. C. Spector. 1992. SAS System for Linear Models, 3rd
Edition. SAS Series in Statistical Application. Cary: SAS Institute.
Mack, R.N. 2001. Motivations and consequences of the human dispersal of plants. Pages
23-24 in J.A. McNeely, ed. The great reshuffling: human dimensions of invasive
alien species. The World Conservation Union.
Mack, M.C. and C.M. D’Antonio. 1998. Impacts of biological invasions on disturbance
regimes. Trends in Ecology and Evolution 13:195-198.
Major, J. and W.T. Pyott. 1966. Buried, viable seeds in two California bunchgrass sites
and their bearing on the definition of a flora. Vegetatio 13:253-282.
Mau-Crimmins, T.M., H.S. Schussman, and E.L. Geiger. Can an invading species’
distribution be predicted using data from its invaded range? Lehmann lovegrass
(Eragrostis lehmanniana) in the southwestern United States. Ecological
Modelling. In review.
Morrison, J.A. 2002. Wetland vegetation before and after experimental purple loosestrife
removal. Wetlands 22:159-169.
Myers, J.H., D. Simberloff, A.M. Kuris, and J.R. Carey. 2000. Eradication revisited:
dealing with exotic species. Trends in Ecology and Evolution 15:316-320.
National Research Council. 2002. Predictiong Invasions of Nonindigenous Plants and
Plant Pests. Washington, DC: National Academy Press.
Noy-Meir, I. 1973. Desert ecosystems: environment and producers. Annual Review of
Ecological Systems 4:25-51.
Peters, E.F. and S.C. Bunting. 1994. Fire conditions and pre- and postoccurrence of
annual grasses on the Snake River Plain. Pages 31-36 in S.B. Monsen and S.G.
Kitchen, eds. Proceedings—ecology and management of annual rangelands, pp.
31-36. General Technical Report INT-GTR-313. Ogden: U.S. Department of
Agriculture, Forest Service, Intermountain Research Station.
144
Piper, J.K. 1994. Composition of prairie plant communities on productive versus
unproductive sites in wet and dry years. Canadian Journal of Botany 73:16351644.
Pyke, D.A. 1990. Comparative demography of co-occurring introduced and native
tussock grasses: persistence and potential expansion. Oecologia 82:537-543.
Rabinowitz, D. 1981. Buried viable seeds in a North American tall-grass prairie: the
resemblance of their abundance and composition to dispersing seeds. Oikos
36:191-195.
Rice, K.J. 1989. Impacts of seed banks on grassland community structure and population
dynamics. Pages 211-230 in M.A. Leck, V.T. Parker, and R.L. Simpson, eds.
Ecology of Soil Seed Bank. San Diego: Academic Press.
Ruyle, G.B., B.A. Roundy, and J.R. Cox. 1988. Effects of burning on germinability of
Lehmann lovegrass. Journal of Range Management 41:404-406.
Schussman, H.R. 2002. Genetic variation in the introduced apomictic grass, Eragrostis
lehmanniana Nees, on the Santa Rita Experimental Range in southeastern
Arizona, USA. M.S. Thesis, University of Arizona, Tucson.
Schwinning, S., B.I. Starr, and J.R. Ehleringer. 2003. Dominant cold desert plants do not
partition warm season precipitation by event size. Oecologia 136:252-260.
Silander, J.A. and J. Antonovics. 1982. Analysis of interspecific interactions in a coastal
plant community – a perturbation approach. Nature 298:557-560.
Silvertooth, J.C. 2001. Saline and sodic soil identification and management for cotton.
University of Arizona Cooperative Extension Publication AZ1199.
Soltanpour, P.N. and R.H. Follett. 1999. Soil test explanation. Colorado State University
Cooperative Extension bulletin no. 0.502. Available online at:
http://www.ext.colostate.edu/pubs/crops/00502.html. Accessed 19 Jan 2005.
Sokal, R.R., and F.J. Rohlf. 1995. Biometry: The Principles and Practices of Statistics in
Biological Research, 3rd Edition. New York:W. H. Freeman & Sons,.
Southern Weed Science Society. 1998. Weeds of the United States and Canada. CDROM. Champaign: Southern Weed Science Society.
Sumrall, L.B., B.A. Roundy, J.R. Cox, and V.K. Winkel. 1991. Influence of canopy
removal by burning or clipping on emergence of Eragrostis lehmanniana
seedlings. International Journal of Wildland Fire 1:35-40.
145
Svetlik, W. No date. Soil Survey of Cochise County, Arizona, Northwestern Part.
Advance copy, unpublished data. USDA-NRCS.
Templeton, A.R. and D.A. Levin. 1979. Evolutionary consequences of seed pools.
American Naturalist 114:232-249.
Thompson, K. and J.P. Grime. 1979. Seasonal variation in the seed banks of herbaceous
species in ten contrasting habitats. Journal of Ecology 67:893-921.
Underwood, A.J. 1986. The analysis of competition by field experiments. Pages 260-268
in J. Kikkawa and D.J. Anderson, eds. Community ecology: pattern and process.
Oxford: Blackwell Scientific.
University of Arizona Student Chapter, Society for Range Management. 1995. Coronado
National Memorial inventory and grazing management alternatives for Joe’s
Spring and Montezuma grazing allotments. Unpublished report.
USDA Natural Resource Conservation Service. 2004. Soil Survey Geographic
(SSURGO) database for Santa Cruz and Parts of Cochise and Pima Counties,
Arizona. Fort Worth, TX.
Voigt, P.W., C.R. Tischler and M.M. Poverene. 1996. Seed dormancy and its alleviation
in lovegrass hybrids. Crop Science 36:1699-1705.
Warr, S.J., K. Thompson, and M. Kent. 1993. Seed banks as a neglected area of
biogeographical research: a review of literature and sampling techniques. Progress
in Physical Geography 17:329-347.
Weaver, L.C. and G.L. Jordan. 1986. Effects of selected seed treatment on germination
rates of five range plants. Journal of Range Management 38:415-418.
Weber, E.F. 1997. The alien flora of Europe: a taxonomic and biogeographic review.
Journal of Vegetation Science 8:565-572.
Wedin, D.A. and D. Tilman. 1990. Species effects on nitrogen cycling: a test with
perennial grasses. Oecologia 84:433-441.
Western Regional Climate Center. 2005. Arizona Climate Summaries.
http://www.wrcc.dri.edu/summary/climsmaz.html. Accessed: 22 Jan 2005.
Westoby,M., B. Walker, and I. Noy-Meir. 1989. Opportunistic management for
rangelands not at equilibrium. Journal of Range Management 42:266-274.
146
Whisenant, S. 1990. Changing fire frequencies on Idaho’s Snake River plains: ecological
and management implications. Pages 4-10 in Proceedings Symposium on
cheatgrass invasion, shrub dieoff and other aspects of shrub biology and
management. USFS Gen. Tech. Rep. INT-276 US Govt Publications Office.
White, T.A., B.D. Campbell, and P.D. Kemp. 1997. Invasion of temperate grasslands by
a subtropical annual grass across an experimental matrix of water stress and
disturbance. Journal of Vegetation Science 8:847-854.
Whitson, T.D. 1996. Weeds of the West. Western Society of Weed Science in
cooperation with Cooperative Extension Services, Laramie: University of
Wyoming..
Wilcove, D.S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying the
threats to imperiled species in the United States. Bioscience 48:607-615.
Willems, J.H., R.K. Peet, and L. Differences in chalk grassland structure and species
richness resulting from selective nutrient additions. Journal of Vegetation Science
4:203-212.
Williams, D.G. and Z. Baruch. 2000. African grass invasion in the Americas: ecosystem
consequences and the role of ecophysiology. Biological Invasions 2:123-140.
Williams, K. and R. Hobbs. 1989. Control of shrub establishment by springtime soil
water availability in an annual grassland. Oecologia 81:62-66.
Williamson, M. 1996. Biological Invasions. London: Chapman and Hall.
Winkel, V.K., B.A. Roundy, and D.K. Blough. 1991. Effects of seedbed preparation and
cattle trampling on burial of grass seeds. Journal of Range Management 44:171175.
Zavaleta, E.S., R.J. Hobbs, and H.A. Mooney. 2001. Viewing invasive species removal in
a whole-ecosystem context. Trends in Ecology and Evolution 16:454-459.
147
Table 1. Repeated measures MANOVA for aboveground vegetation analyses at three
sites in southeastern Arizona. The four factors are: Season = post-treatment sampling
season (Fall 2003, Spring 2004, Fall 2004), Site = site (Three Links Farm, Coronado NM,
Santa Rita Experimental Range), Treatment = herbicide treatment (E. lehmanniana
removed, control). Greenhouse-Geisser adjustment was used to correct for
autocorrelation among repeated measure factor Season.
Response
Variable
Source
DF
F value
p
Total percent plant cover
Season*Treatment
Season*Site
2, 78
4, 78
4.41
59.00
<.0001
0.015
Relative percent native cover
Season*Treatment
2, 81
5.89
0.004
Relative percent nonnative cover
Season*Treatment
2, 88
7.23
0.0012
Relative percent forb cover
Season*Treatment
Season*Site
2, 81
4, 81
6.73
59.09
0.0019
<.0001
Relative percent grass cover
Season*Treatment
Season*Site
2, 86
4, 86
7.30
62.94
0.0012
<.0001
Relative percent annual cover
Season*Treatment
Season*Site
2, 77
4, 77
11.07
11.15
<.0001
<.0001
Relative percent perennial cover
Season*Treatment*Site
4, 88
2.66
0.038
Species richness
Season*Treatment*Site
4, 86
3.33
0.014
148
Table 2. ANOSIM and SIMPER results discriminating within-season species
assemblages. R = test statistic indicating relative dissimilarity (0 –1, with 1 = identical
composition between groups). “*” denotes significant (P < 0.05) dissimilarity between
groups. Percentages reflect proportion of dissimilarity attributable to each species.
Season
SP03
FA03
Global/Pairwise
Test
Global
CORO-C v. CORO-H
SRER-C v. SRER-H
TLF-C v. TLF-H.
Global
R
0.369
-0.025
-0.044
-0.041
0.797
p
<.001*
0.49
0.84
0.671
0.001
CORO-C v. CORO-H
0.591
.002*
SRER-C v. SRER-H
TLF-C v. TLF-H.
Global
CORO-C v. CORO-H
0.946
-0.007
0.706
0.618
.001*
0.444
<.001*
.001*
SRER-C v. SRER-H
TLF-C v. TLF-H.
Global
0.709
0.048
0.705
.001*
0.26
<.001*
CORO-C v. CORO-H
0.46
.002*
SRER-C v. SRER-H
TLF-C v. TLF-H.
0.714
0.135
.001*
0.139
SP04
FA04
Species Differentiation- SIMPER
(Percent of Dissimilarity)
n/a
n/a
n/a
n/a
n/a
Eragrostis lehmanniana (37%), Mollugo
verticillata (11%), Calliandra eriophylla
(11%)
E. lehmanniana (30%), M. verticillata
(16%), Urochloa arizonica (6%)
n/a
n/a
E. lehmanniana(48%), C. eriophylla (9%)
E. lehmanniana (28%), Ambrosia
confertiflora (5%), Spermolepis echinata
(5%)
n/a
n/a
E. lehmanniana (27%), C. eriophylla
(10%), M. verticillata (6%)
E. lehmanniana (28%), M. verticillata
(8%), U. arizonica (6%), Kallstroemia
grandiflora (5%)
n/a
149
Table 3. ANOSIM and SIMPER results discriminating within-site/treatment combination
species assemblages between treatments. R = test statistic indicating relative dissimilarity
(0 –1, with 1 = identical composition between groups). “*” denotes significant (P < 0.05)
dissimilarity between groups. Percentages reflect proportion of dissimilarity attributable
to each species.
Site/
Treatment
CORO Control
SRER Control
TLF Control
CORO Herbicide
SRER Herbicide
Global/
Pairwise
Test
Global
SP03 - FA03
FA03 - SP04
SP04 - FA04
SP03 - SP04
FA03 - FA04
Global
R
0.071
0.074
0.112
-0.055
0.153
0.01
0.338
p
0.067
0.15
0.1
0.783
0.052
0.362
0.001
SP03 - FA03
FA03 - SP04
SP04 - FA04
SP03 - SP04
FA03 - FA04
Global
0.448
0.467
0.325
0.284
0.256
0.978
0.001*
0.001*
0.001*
0.001*
0.002*
0.001
SP03 - FA03
0.981
0.002*
FA03 - SP04
0.957
0.002*
SP04 - FA04
0.976
0.002*
SP03 - SP04
1
0.002*
FA03 - FA04
Global
0.993
0.337
0.002*
0.001
SP03 - FA03
FA03 - SP04
SP04 - FA04
SP03 - SP04
FA03 - FA04
Global
0.599
0.234
0.07
0.59
-0.022
0.814
0.001*
0.052
0.193
0.001*
0.449
<.0001
SP03 - FA03
1
<.0001*
FA03 - SP04
0.995
<.0001*
SP04 - FA04
0.723
<.0001*
Species DifferentiationSIMPER (Percent of Dissimilarity)
n/a
n/a
n/a
n/a
n/a
n/a
n/a
Eragrostis lehmanniana (7%), Mollugo
verticillata (5%)
E. lehmanniana (5%), M. verticillata (5%)
E. lehmanniana (4%), Gilia scopulorum (4%)
G. scopulorum (4%)
E. lehmanniana (6%), M. verticillata (5%)
n/a
E. lehmanniana (33%), Sporobolus
cryptandrus (9%), E. cilianensis (7%)
Salsola kali (11%), Eragrostis cilianensis
(9%), Sisymbrium irio (8%), Schismus
barbatus (7%)
S. kali (15%), Bouteloua aristidoides (7%), S.
barbatus (5%), S. cryptandrus (5%)
E. lehmanniana (22%), S. kali (10%), S.
cryptandrus (7%), E. cilianensis (7%), S. irio
(7%), S. barbatus (6%)
S. kali (29%), E. cilianensis I (8%), B.
aristidoides (8%), S. cryptandrus (6%)
n/a
E. lehmanniana (36%), M. verticillata (11%),
Calliandra eriophylla (10%)
n/a
n/a
E. lehmanniana (46%), C. eriophylla (10%)
n/a
n/a
E. lehmanniana (29%), M. verticillata (22%),
Kallstroemia grandiflora (8%), Urochloa
arizonica (8%)
M. verticillata (25%), K. grandiflora (9%), U.
arizonica (8%), E. lehmanniana (8%)
M. verticillata (11%), E. lehmanniana (9%),
U. arizonica (7%)
150
TLF –
Herbicide
SP03 - SP04
0.791
<.0001*
FA03 - FA04
Global
0.583
0.973
<.0001*
<.0001
SP03 - FA03
FA03 - SP04
SP04 - FA04
0.865
0.961
1
.002*
0.002*
0.002*
SP03 - SP04
0.924
0.002*
FA03 - FA04
0.989
0.002*
E. lehmanniana (28%), Ambrosia
confertiflora (6%)
M. verticillata (14%), E. lehmanniana (9%),
Boerhavia sp. (6%)
n/a
E. lehmanniana (33%), E. cilianensis (7%),
Cynodon dactylon (5%), Chloris virgata (5%)
E. cilianensis (9%), S. irio (9%), S. kali (8%)
S. kali (18%), B. aristidoides (12%)
E. lehmanniana (21%), S. irio (8%), S. kali
(7%), E. cilianensis (5%)
S. kali (29%), B. aristidoides (15%), E.
cilianensis (7%)
151
Table 4. Total seed germinated by sampling event in each of three study sites in southeastern Arizona.
CORO
SP03
SU03
FA03
SP04
SU04
FA04
SP03
SU03
0
0
0
0
0
0
1
0
Amaranthus palmeri
0
0
0
0
0
1
44
49
Androsace occidentalis
0
0
0
0
0
0
0
0
Aristida adscensionis
0
0
0
3
0
0
0
0
Aristida ternipes
0
0
0
0
0
0
0
0
Baccharis sarothroides
0
0
0
0
0
1
0
0
Bouteloua aristidoides
0
0
0
0
0
0
0
0
Bouteloua barbata
1
0
0
0
0
0
0
0
Bouteloua curtipendula
0
0
0
0
0
0
0
0
Bouteloua gracilis
0
0
0
0
0
0
0
0
Bouteloua sp.
0
0
0
0
0
0
0
0
Chamaesyce sp.
0
0
0
0
0
0
0
0
Chenopodium sp.
0
0
0
0
0
0
0
0
Chamaesyce maculata
3
0
0
0
2
0
7
0
Chamaesyce micromera
0
0
0
0
0
0
1
0
Chamaecrista nictitans
0
0
0
0
0
0
0
0
Chloris virgata
0
0
0
0
0
0
27
40
Corydalis aurea
0
0
2
0
0
0
4
1
Conyza canadensis
0
1
1
0
2
1
149
269
Crassula conniculata
0
0
0
0
0
0
0
0
Cryptantha angustifolia
0
0
0
0
0
0
0
0
Cynodon dactylon
0
0
0
0
0
0
15
1
Descurainia pinnata
0
0
0
0
0
0
0
0
Eriochloa acuminata
1
0
0
0
0
0
0
0
Eragrostis sp.
1
0
0
1
0
0
1
0
Eragrostis cilianensis
6
4
0
0
0
0
98
37
Eriastrum diffusum
1
6
4
7
3
14
5
35
Erigeron divergens
2
0
2
5
0
0
5
0
Eragrostis intermedia
61
159
80
95
283
47
465
418
Eragrostis lehmanniana
0
0
0
0
0
0
0
0
Euphorbia sp.
0
0
0
0
0
0
1
0
Evolvulus sp.
0
0
0
0
0
0
0
0
Gilia sp.
0
0
0
0
0
0
0
3
Gilia sinuata
0
0
32
0
0
0
5
3
Gnaphalium palustre
0
0
0
0
0
0
0
1
Heterotheca subaxillaris
0
0
0
0
0
0
0
0
Juniperus sp.
1
0
0
0
0
0
0
0
Laennecia coulteri
0
0
0
0
0
0
0
1
Lactuca sp.
1
0
0
1
0
0
0
0
Leptochloa panicea
SRER
FA03
SP04
0
0
131
20
0
1
0
7
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
1
0
1804
21
0
0
0
2
1
1
0
3
0
0
0
4
0
0
85
19
0
1
143
455
0
0
0
0
1
0
0
0
79
0
0
0
1
0
0
0
0
0
0
0
SU04
0
95
0
0
0
4
0
0
0
2
1
0
0
0
0
0
0
0
154
0
0
2
0
0
0
0
21
0
1868
1
1
3
0
2
0
0
0
0
0
FA04
0
106
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1174
0
0
1
0
0
1
0
30
0
73
0
3
0
17
23
0
0
0
0
0
SP03
2
0
0
0
0
0
0
0
2
0
0
0
0
9
0
4
0
0
1
0
21
1
13
90
534
0
0
79
2090
0
0
0
0
0
0
0
0
0
5
SU03
6
0
0
0
0
0
0
0
0
0
1
0
0
6
0
2
0
0
0
0
10
0
0
36
888
0
0
24
847
0
0
0
0
0
0
0
0
0
26
TLF
FA03
SP04
2
2
0
0
0
0
0
1
0
0
0
16
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
1
0
0
0
4
0
0
0
0
0
0
0
0
18
4
0
0
0
0
1
3
94
249
0
0
0
4
1
2
299
321
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
SU04
2
0
0
0
0
57
1
0
0
0
0
0
0
1
0
8
0
0
0
0
16
1
0
5
467
0
1
6
394
0
0
0
0
0
0
0
0
0
7
FA04
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
11
0
0
0
150
0
10
1
54
0
0
0
0
0
0
0
0
0
3
152
Leptochloa sp.
Lotus greenei
Lotus humistratus
Lotus sp.
Lupinus sp.
Mitracarpus breviflorus
Monolepis nuttalliana
Mollugo verticillata
Muhlenbergia sp.
Nothoscordum sp.
Nuttallanthus texanus
Oxalis corniculata
Panicum sp.
Pellaea sp.
Phacelia sp.
Physalis longifolia
Phacelia ramosissima
Plantago patagonica
Polygonum argyrocoleon
Portulaca halimoides
Portulaca pilosa
Portulaca sp.
Portulaca umbraticola
Pseudognaphalium
arizonicum
Schismus barbatus
Setaria verticillata
Sida abutifolia
Sisymbrium irio
Solanaceae
Sporobolus cryptandrus
Spermolepis echinata
Talinum aurantiacum
Triodanis perfoliata
unknown forb
unknown grass
unknown
Urochloa arizonica
Veronica peregrina
Vulpia octoflora
Woodsia sp.
0
0
0
0
0
0
0
6
0
0
0
2
0
0
0
0
0
1
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
5
0
1
0
0
0
0
0
0
0
0
0
0
12
1
0
0
27
0
0
0
0
0
5
0
2
0
0
0
0
0
0
0
0
2
0
23
0
0
0
10
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
2
0
84
0
0
0
37
0
0
0
0
0
1
0
1
0
0
3
0
0
0
0
0
0
0
5
0
0
1
16
0
0
0
0
0
9
0
0
0
0
0
0
8
18
1
0
0
1
490
0
1
11
67
0
0
0
0
0
5
0
0
0
0
0
0
0
10
0
0
0
0
114
0
1
9
25
0
1
0
0
6
0
0
0
0
0
1
0
0
0
0
0
0
0
30
0
0
42
200
0
0
0
0
1
1
0
0
0
0
0
0
0
0
1
0
0
0
692
0
0
0
55
0
0
0
1
0
0
0
0
0
0
5
0
0
4
14
0
0
0
1841
0
0
18
38
0
0
0
0
0
1
0
0
0
0
0
0
0
22
1
2
0
0
53
0
0
70
9
0
0
3
0
0
3
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
9
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
1
0
1
0
81
15
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
0
0
0
29
74
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
17
298
0
0
0
0
0
0
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
7
15
0
0
0
0
0
0
0
0
0
0
35
64
93
11
49
42
50
70
212
6
25
200
1
0
0
1
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
6
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
1
0
0
0
0
2
0
0
25
0
0
0
0
0
0
0
6
0
0
0
0
1
0
0
0
0
0
0
0
0
51
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
1
0
0
0
0
0
1
0
0
0
0
0
5
0
0
0
0
0
0
5
0
2
1
0
0
2
0
2
3
0
0
0
1
0
0
8
0
3
1
0
2
0
0
0
0
0
0
1
0
0
1
0
1
0
1
0
0
2
0
0
0
0
0
21
31
0
0
68
0
0
12
0
0
10
0
0
0
0
0
0
0
0
0
71
0
0
23
4
0
0
0
0
0
0
2
0
12
0
6
0
0
0
0
0
0
0
0
0
5
0
0
0
1
0
1
0
0
0
0
0
0
0
6
0
0
0
0
0
8
0
12
0
0
0
13
0
0
0
4
0
0
0
0
0
30
0
63
0
0
0
1
2
0
0
0
0
0
0
0
0
15
0
61
0
0
0
8
0
0
0
0
0
0
5
0
0
41
0
91
1
0
0
7
18
0
0
0
0
0
153
Table 5. Repeated measures MANOVA for seed bank analyses at three sites in
southeastern Arizona. The four factors are: Group = post-treatment sampling season
(Summer 2003, Fall 2003, Spring 2004, Summer 2004, Fall 2004), Site = site (Three
Links Farm, Coronado NM, Santa Rita Experimental Range), Treatment = herbicide
treatment (E. lehmanniana removed, control). Greenhouse-Geisser adjustment was used
to correct for autocorrelation among repeated measure factor Group.
Response
Variable
Source
DF
F value
p
Germinating seeds
Group*Site
6, 135
17.82
<.0001
Percent graminoid seedlings
Group*Treatment
Group*Site
3, 128
6, 128
5.26
3.39
0.0009
0.020
Percent perennial seedlings
Group*Treatment*Site
6, 129
2.31
0.029
Percent native seedlings
Group*Treatment*Site
6, 125
3.70
0.0024
Seed bank species richness
Group*Site
7, 148
11.09
<.0001
Percent of seedlings
that were E. lehmanniana
Group*Treatment*Site
8, 148
3.11
0.0034
154
Table 6. ANOSIM and SIMPER results discriminating within-season seedbank species
assemblages. R = test statistic indicating relative dissimilarity (0 –1, with 1 = identical
composition between groups). “*” denotes significant (P < 0.05) dissimilarity between
groups. Percentages reflect proportion of dissimilarity attributable to each species.
Site/
Treatment
Global/
Pairwise
Test
Global
SP03-SU03
R
0.173
-0.015
p
0.001
0.369
SU03-FA03
0.255
.03*
FA03-SP04
SP04-SU04
0.242
-0.017
.031*
0.554
SU04-FA04
Global
SP03-SU03
0.344
0.58
0.308
.005*
.001*
.001*
SU03-FA03
0.433
.001*
FA03-SP04
SP04-SU04
0.944
0.131
.001*
0.051
SU04-FA04
Global
SP03-SU03
SU03-FA03
0.873
0.631
0.43
0.737
.001*
.001*
.002*
.002*
FA03-SP04
SP04-SU04
0.454
0.252
.002*
.035*
SU04-FA04
Global
SP03-SU03
0.465
0.229
0.046
.002*
0.001
0.258
SU03-FA03
0.227
.003*
FA03-SP04
SP04-SU04
SU04-FA04
Global
SP03-SU03
0.526
0.146
0.144
0.766
0.084
.004*
0.098
0.071
.001**
0.07
SU03-FA03
0.774
.001*
FA03-SP04
1
.001*
CORO Control
SRER Control
TLF Control
CORO Herbicide
SRER Herbicide
Species DifferentiationSIMPER (Percent of Dissimilarity)
Pseduognaphalium arizonicum (8%), Eragrostis
lehmanniana (7%), Gnaphalium palustre (6%)
P. arizonicum (12%), Mollugo verticillata (6%),
Oxalis corniculata (6%), G. palustre (6%), E.
lehmanniana (6%)
P. arizonicum (8%), E. lehmanniana (7%), O.
corniculata (6%), Plantago patagonica (6%)
Crassula connata (7%), M. verticillata (7%)
C. connata (9%), E. lehmanniana (8%), M.
verticillata (5%), P. arizonicum (5%)
C. connata (15%), M. verticillata (11%), E.
lehmanniana (11%), P. arizonicum (7%),
Androsace occidentalis (5%)
C. connata (12%), E. lehmanniana (12%), M.
verticillata (10%), P. arizonicum (7%)
Eragrostis. cilianensis (6%)
E. cilianensis (8%), O. corniculata (8%)
E. cilianensis (5%), Panicum sp. (5%),
Sporobolus cryptandrus (5%)
Panicum sp. (5%)
Panicum sp. (9%), E. lehmanniana (6%),
Sisymbrium irio (5%), E. cilianensis (5%), S.
cryptandrus (5%)
G. palustre (9%), E. lehmanniana (8%), P.
arizonicum (7%), O. corniculata (5%)
P. arizonicum (12%), G. palustre (9%), E.
lehmanniana (7%), M. verticillata (7%), Erigeron
divergens (5%), O. corniculata (5%)
E. lehmanniana (9%), C. connata (9%), M.
verticillata (5%)
C. connata (20%), M. verticillata (17%), E.
lehmanniana (7%), P. arizonicum (7%)
155
TLF Herbicide
SP04-SU04
0.496
.001*
SU04-FA04
Global
SP03-SU03
SU03-FA03
FA03-SP04
SP04-SU04
0.989
0.611
0.665
0.889
0.239
0.054
.001*
.001*
.002*
.002*
0.032*
0.31
SU04-FA04
0.476
.004*
C. connata (5%)
M. verticillata (14%), C. connata (13%), E.
lehmanniana (6%), P. arizonicum (5%)
E. cilianensis (7%), E. lehmanniana (5%)
O. corniculata (8%), E. cilianensi (7%)
O. corniculata (4%)
S. cryptandrus (8%), Panicum sp. (7%), E.
lehmanniana (6%), E. cilianensi (6%)
156
Table 7. ANOSIM and SIMPER results discriminating within-site/treatment combination
seed bank species assemblages between treatments. R = test statistic indicating relative
dissimilarity (0 –1, with 1 = identical composition between groups). “*” denotes
significant (P < 0.05) dissimilarity between groups. Percentages reflect proportion of
dissimilarity attributable to each species.
Season
SP03
SU03
FA03
SP04
SU04
Global/Pairwise
Test
Global
CORO-C v. CORO-H
SRER-C v. SRER-H
TLF-C v. TLF-H
Global
CORO-C v. CORO-H
SRER-C v. SRER-H
TLF-C v. TLF-H
Global
CORO-C v. CORO-H
SRER-C v. SRER-H
TLF-C v. TLF-H
Global
CORO-C v. CORO-H
R
0.597
0.007
0.078
-0.004
0.659
0.078
-0.055
-0.115
0.725
-0.055
-0.013
-0.102
0.631
-0.14
p
.001*
0.329
0.13
0.468
.001*
0.22
0.834
0.857
.001*
0.697
0.519
0.857
.001*
0.92
SRER-C v. SRER-H
TLF-C v. TLF-H
Global
0.327
0.015
0.76
.001*
0.418
.001*
CORO-C v. CORO-H
0.228
.015*
SRER-C v. SRER-H
TLF-C v. TLF-H
0.45
-0.044
.001*
0.649
Species Differentiation- SIMPER
(Percent of Dissimilarity)
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
Molluga verticillata (9%), Eragrostis
lehmanniana (8%), Oxalis corniculata
(4%)
n/a
n/a
M. verticillata (7%), O. corniculata
(5%), Erigeron divergens (5%),
Pseudognaphalium arizonicum (5%),
Eragrostis intermedia (5%)
E. lehmanniana (8%), M. verticillata
(6%), Crassula connata (4%)
n/a
157
Table 8. Results of soils analyses for three E. lehmanniana-dominated semi-desert
grasslands in southeastern Arizona.
Sample
CORO 1
CORO2
SRER 1
SRER 2
TLF
pH
5.36
5.93
6.01
6.42
6.65
EC
Phosphorus
(mmohs/cm)
(ppm)
0.24
7.36
0.15
16.36
0.31
17.02
0.24
21.28
0.42
14.44
Total
Organic
Carbon (%)
0.91
0.66
0.84
0.74
1.42
Potassium
(ppm)
19.18
14.16
22.88
23.73
33.46
Nitrogen
(NH4+)
(ppm)
32.08
35.20
30.04
51.21
27.82
158
Figure 1. Location of grass removal study sites in southeastern Arizona.
159
Figure 2. Effects of site (a) and treatment (b) on total plant cover (mean±SE) before
(SP03) and after removing Eragrostis lehmanniana from three sites in southeastern
Arizona. SP = Spring, FA = Fall; sampling seasons 2003 and 2004. Filled circles =
Coronado National Memorial, filled squares = Santa Rita Experimental Range, filled
triangles = Three Links Farm. Open circles = control plots, open squares = treated plots.
*, **, and *** denote significant within season effects (p < 0.05, p < 0.01 and p < 0.001,
respectively). See Table 1 for more detailed information.
160
Figure 3. Effects of site (a) and treatment (b) on relative percent forb cover (mean±SE)
before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona. SP = Spring, FA = Fall; sampling seasons 2003 and 2004. Filled
circles = Coronado National Memorial, filled squares = Santa Rita Experimental Range,
filled triangles = Three Links Farm. Open circles = control plots, open squares = treated
plots. *, **, and *** denote significant within season effects (p < 0.05, p < 0.01 and p <
0.001, respectively). See Table 1 for more detailed information.
161
Figure 4. Effects of site (a) and treatment (b) on relative percent annual cover
(mean±SE) before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona. SP = Spring, FA = Fall; sampling seasons 2003 and 2004. Filled
circles = Coronado National Memorial, filled squares = Santa Rita Experimental Range,
filled triangles = Three Links Farm. Open circles = control plots, open squares = treated
plots. *, **, and *** denote significant within season effects (p < 0.05, p < 0.01 and p <
0.001, respectively). See Table 1 for more detailed information.
162
Figure 5. Effects of treatment on relative percent native cover (mean±SE) before (SP03)
and after removing Eragrostis lehmanniana from three sites in southeastern Arizona. SP
= Spring, FA = Fall; sampling seasons 2003 and 2004. Open circles = control plots, open
squares = treated plots. *, **, and *** denote significant within season effects (p < 0.05,
p < 0.01 and p < 0.001, respectively). See Table 1 for more detailed information.
163
Figure 6. Effects of site and treatment on relative percent perennial cover (mean±SE)
before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona. SP = Spring, FA = Fall; sampling seasons 2003 and 2004. Circles
= Coronado National Memorial, squares = Santa Rita Experimental Range. Filled
symbols = control plots, open symbols = treated plots. Dotted triangles = Three Links
Farm; treated and control plots combined due to no significant differences. See Table 1
for more detailed information.
164
Figure 7. Effects of site and treatment on species richness of aboveground vegetation
(mean±SE) before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona, represented as departure from control plots. SP = Spring, FA =
Fall; sampling seasons 2003 and 2004. Circles = Coronado National Memorial, squares =
Santa Rita Experimental Range.
.
165
Figure 8. Effects of site on total seeds germinated (mean±SE) before (SP03) and after
removing Eragrostis lehmanniana from three sites in southeastern Arizona. SP = Spring,
SU = Summer, FA = Fall; sampling seasons 2003 and 2004. Circles = Coronado National
Memorial, squares = Santa Rita Experimental Range, triangles = Three Links Farm. *,
**, and *** denote significant within season effects (p < 0.05, p < 0.01 and p < 0.001,
respectively). See Table 1 for more detailed information.
166
Figure 9. Effects of site on seed bank species richness (mean±SE) before (SP03) and
after removing Eragrostis lehmanniana from three sites in southeastern Arizona. SP =
Spring, SU = Summer, FA = Fall; sampling seasons 2003 and 2004. Circles = Coronado
National Memorial, squares = Santa Rita Experimental Range, triangles = Three Links
Farm. *, **, and *** denote significant within season effects (p < 0.05, p < 0.01 and p <
0.001, respectively). See Table 1 for more detailed information.
167
Figure 10. Effects of site (a) and treatment (b) on relative percent grass seedlings
(mean±SE) before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona. SP = Spring, SU = Summer, FA = Fall; sampling seasons 2003
and 2004. Circles = Coronado National Memorial, squares = Santa Rita Experimental
Range, triangles = Three Links Farm. Open circles = control plots, open squares = treated
plots. *, **, and *** denote significant within season effects (p < 0.05, p < 0.01 and p <
0.001, respectively). See Table 1 for more detailed information.
168
Figure 11. Effects of site and treatment on relative percent perennial seedlings
(mean±SE) before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona, represented as departure from control plots. SP = Spring, SU =
summer, FA = Fall; sampling seasons 2003 and 2004. Circles = Coronado National
Memorial, squares = Santa Rita Experimental Range, triangles = Three Links Farm.
169
Figure 12. Effects of site and treatment on relative percent native seedlings (mean±SE)
before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona, represented as departure from control plots. SP = Spring, SU =
summer, FA = Fall; sampling seasons 2003 and 2004. Circles = Coronado National
Memorial, squares = Santa Rita Experimental Range, triangles = Three Links Farm.
170
Figure 13. Effects of site and treatment on percent of seedlings that were E. lehmanniana
(mean±SE) before (SP03) and after removing Eragrostis lehmanniana from three sites in
southeastern Arizona, represented as departure from control plots. SP = Spring, SU =
summer, FA = Fall; sampling seasons 2003 and 2004. Circles = Coronado National
Memorial, squares = Santa Rita Experimental Range, triangles = Three Links Farm.
171
Figure 14. Non-metric multi-dimensional scaling (NMS plots of above-ground plant
communities and seed bank communities in a nonnative grass removal experiment in
southeastern Arizona for all sampling seasons together; a=CORO, b=SRER, c=TLF.
Green triangles represent samples of seed banks from control plots; blue triangles are
seed bank samples from treated plots. Red diamonds represent above-ground
communities in control plots; light blue squares are above-ground communities in treated
plots.
172
Figure 15. Biomass (mean±SE) of E. lehmanniana at the three study locations over the
course of the experiment before (SP03) and after removing Eragrostis lehmanniana from
three sites in southeastern Arizona. SP = Spring, SU = Summer, FA = Fall; sampling
seasons 2003 and 2004. Circles = Coronado National Memorial, squares = Santa Rita
Experimental Range, triangles = Three Links Farm. * denotes significant within season
effects (p < 0.05). See Table 1 for more detailed information.
173
Figure 16. Cumulative precipitation (a) and seasonal total precipitation (b) at three study
locations over the course of a nonnative grass removal experiment in southeastern
Arizona.
174
Figure 17. Monthly totals of winter (a) and monsoon season (b) precipitation at
Coronado National Memorial and the Santa Rita Experimental Range, two of the study
locations in a nonnative grass removal study in southeastern Arizona.
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