A GEODATABASE OF HISTORICAL AND CONTEMPORARY OCEANOGRAPHIC DATASETS

A GEODATABASE OF HISTORICAL AND CONTEMPORARY OCEANOGRAPHIC DATASETS
A GEODATABASE OF HISTORICAL AND
CONTEMPORARY OCEANOGRAPHIC DATASETS
FOR INVESTIGATING THE ROLE OF THE PHYSICAL
ENVIRONMENT IN SHAPING PATTERNS OF
SEABED BIODIVERSITY IN THE GULF OF MAINE
M.E. Greenlaw, J.A. Sameoto, P. Lawton, N.H. Wolff, L.S. Incze, C.R.
Pitcher, S.J. Smith and A. Drozdowski
Science Branch, Maritimes Region
Fisheries and Oceans Canada
Biological Station
531 Brandy Cove Road, St. Andrews, NB
E5B 2L9
2010
Canadian Technical Report of Fisheries and Aquatic
Sciences 2895
Fisheries and Oceans
Canada
Pêches et Océans
Canada
Canadian technical report of
fisheries and aquatic sciences
Technical reports contain scientific and technical information that contributes to existing knowledge but
which is not normally appropriate for primary literature. Technical reports are directed primarily toward a
worldwide audience and have an international distribution. No restriction is placed on subject matter and the series
reflects the broad interests and policies of the Department of Fisheries and Oceans, namely, fisheries and aquatic
sciences.
Technical reports may be cited as full publications. The correct citation appears above the abstract of each
report. Each report is abstracted in Aquatic Sciences and Fisheries Abstracts and indexed in the Department’s
annual index to scientific and technical publications.
Numbers 1-456 in this series were issued as Technical Reports of the Fisheries Research Board of Canada.
Numbers 457-714 were issued as Department of the Environment, Fisheries and Marine Service, Research and
Development Directorate Technical Reports. Numbers 715–924 were issued as Department of Fisheries and the
Environment, Fisheries and Marine Service Technical Reports. The current series name was changed with report
number 925.
Technical reports are produced regionally but are numbered nationally. Requests for individual reports will
be filled by the issuing establishment listed on the front cover and title page. Out-of-stock reports will be supplied
for a fee by commercial agents.
Rapport technique canadien des
sciences halieutiques et aquatiques
Les rapports techniques contiennent des renseignements scientifiques et techniques qui constituent une
contribution aux connaissances actuelles, mais qui ne sont pas normalement appropriés pour la publication dans un
journal scientifique. Les rapports techniques sont destinés essentiellement à un public international et ils sont
distribués à cet échelon. Il n’y a aucune restriction quant au sujet; de fait, la série reflète la vaste gamme des intérêts
et des politiques du ministère des Pêches et des Océans, c’est-à-dire les sciences halieutiques et aquatiques.
Les rapports techniques peuvent être cités comme des publications complètes. Le titre exact parait audessus du résumé de chaque rapport. Les rapports techniques sont résumés dans la revue Résumés des sciences
aquatiques et halieutiques, et ils sont classés dans l’index annuel des publications scientifiques et techniques du
Ministère.
Les numéros 1 à 456 de cette série ont été publiés à titre de rapports techniques de l’Office des recherches
sur les pêcheries du Canada. Les numéros 457 à 714 sont parus à titre de rapports techniques de la Direction
générale de la recherche et du développement, Service des pêches et de la mer, ministère de l’Environnement. Les
numéros 715 à 924 ont été publiés à titre de rapports techniques du Service des pêches et de la mer, ministère des
Pêches et de l’Environnement. Le nom actuel de la série a été établi lors de la parution du numéro 925.
Les rapports techniques sont produits à l’échelon régional, mais numéroté à l’échelon national. Les
demandes de rapports seront satisfaites par l’établissement auteur dont le nom figure sur la couverture et la page du
titre. Les rapports épuisés seront fournis contre rétribution par des agents commerciaux.
Canadian Technical Report of
Fisheries and Aquatic Sciences 2895
2010
A Geodatabase of Historical and Contemporary Oceanographic Datasets for Investigating
the Role of the Physical Environment in Shaping Patterns of Seabed
Biodiversity in the Gulf of Maine
by
M.E. Greenlaw, J.A. Sameoto1, P. Lawton, N.H. Wolff 2, L.S. Incze2,
C.R. Pitcher3, S.J. Smith1 and A. Drozdowski1
Fisheries and Oceans Canada, Science Branch, Maritimes Region, St Andrews Biological
Station, 531 Brandy Cove Road, St. Andrews, New Brunswick, Canada E5B 2L9
1
Bedford Institute of Oceanography, Science Branch, Maritimes Region, Fisheries and Oceans
Canada, 1 Challenger Drive, Dartmouth, Nova Scotia, Canada B2Y 4A2
2
University of Southern Maine, Aquatic Systems Group, 350 Commercial St. Portland, Maine,
United States of America 04101
3
CSIRO Marine & Atmospheric Research, Marine Biodiversity Hub, PO Box 120, Cleveland,
Queensland, Australia 4163
This is the two hundred and ninety-third Technical Report
of the Biological Station, St. Andrews, NB
ii
© Her Majesty the Queen in Right of Canada, 2010.
Cat. No. Fs 97-6/2895E ISSN 0706-6457 (print version)
Cat. No. Fs 97-6/2895E-PDF ISSN 1488-5379 (online version)
Correct citation for this publication:
Greenlaw, M.E., Sameoto, J.A., Lawton, P., Wolff, N.H., Incze, L.S., Pitcher, C.R.,
Smith, S.J and Drozdowski, A. 2010. A geodatabase of historical and contemporary
oceanographic datasets for investigating the role of the physical environment in shaping patterns
of seabed biodiversity in the Gulf of Maine. Can. Tech. Rep. Fish. Aquat. Sci. 2895: iv + 35 p.
iii
TABLE OF CONTENTS
Abstract .......................................................................................................................................... iv
Résumé........................................................................................................................................... iv
Introduction ..................................................................................................................................... 1
CoML Cross-Project Synthesis Description ............................................................................ 1
Methods........................................................................................................................................... 2
CoML Biodiversity Analysis Cross-Project Methods ............................................................. 2
Source Biological Datasets ...................................................................................................... 2
NOAA Benthic Database Smith-Mcintyre Grab Samples ............................................... 2
NEFSC Bottom Trawl Survey .......................................................................................... 5
Selection of Time Periods ........................................................................................................ 7
Geodatabase Creation and Data Access .................................................................................. 7
XYZ Text File Creation and Data Access ............................................................................... 8
Interpolation Methods .............................................................................................................. 9
Optimal Estimation ........................................................................................................... 9
Spline Interpolation ........................................................................................................ 10
Physical Oceanographic Datasets .......................................................................................... 10
Bathymetry and Derivatives ........................................................................................... 10
Substrate ......................................................................................................................... 10
Bottom Stress.................................................................................................................. 10
Stratification ................................................................................................................... 11
Benthic Irradiance .......................................................................................................... 11
Chlorophyll ..................................................................................................................... 12
Sea Surface Temperature ................................................................................................ 12
Benthic Temperature ...................................................................................................... 12
Benthic Salinity .............................................................................................................. 13
Dissolved Oxygen .......................................................................................................... 13
Bottom Nutrients ............................................................................................................ 13
Phosphate ................................................................................................................ 13
Silicate ..................................................................................................................... 13
Nitrate ...................................................................................................................... 14
Discussion ..................................................................................................................................... 30
Acknowledgements ....................................................................................................................... 32
References ..................................................................................................................................... 32
iv
ABSTRACT
Greenlaw, M.E., Sameoto, J.A., Lawton, P., Wolff, N.H., Incze, L.S., Pitcher, C.R.,
Smith, S.J and Drozdowski, A. 2010. A geodatabase of historical and contemporary
oceanographic datasets for investigating the role of the physical environment in shaping patterns
of seabed biodiversity in the Gulf of Maine. Can. Tech. Rep. Fish. Aquat. Sci. 2895: iv + 35 p.
The management and conservation of the marine environment and marine resources
increasingly requires the synthesis of spatial data from a range of physical and biological features
across a variety of scales. The work phase of compiling these data layers for a particular project
can be time intensive and costly. However, once these datasets are compiled and processed to
generate continuous spatial layers, the preservation of these data in a common georeferenced
format can facilitate their use in future work; in particular for spatial planning, decision making
and ecosystem-based management. A comprehensive suite of physical, chemical and biological
layers (30 layers) for the Gulf of Maine area have been compiled within a single geodatabase,
and in xyz format, using publicly-available U.S. and Canadian oceanographic data sources. The
primary driver for this effort was the Census of Marine Life’s cross-project that involved three
regions (Great Barrier Reef, Gulf of Mexico and Gulf of Maine) who investigated the role of
physical variables in predicting patterns of biodiversity in seabed assemblages. In this report we
provide background methods (including formal metadata), issues faced in compiling geospatial
resources, and caveats for subsequent usage for those datasets used in the Gulf of Maine portion
of the Census project.
RÉSUMÉ
Greenlaw, M.E., Sameoto, J.A., Lawton, P., Wolff, N.H., Incze, L.S., Pitcher, C.R.,
Smith, S.J and Drozdowski, A. 2010. A geodatabase of historical and contemporary
oceanographic datasets for investigating the role of the physical environment in shaping patterns
of seabed biodiversity in the Gulf of Maine. Can. Tech. Rep. Fish. Aquat. Sci. 2895: iv + 35 p.
La gestion et la conservation du milieu marin et des ressources marines exigent de plus
en plus que l’on procède à une synthèse des données spatiales concernant plusieurs
caractéristiques physiques et biologiques à diverses échelles. L’étape de travail consistant à
compiler toutes ces couches de données peut être chronophage et coûteuse. Cependant, une fois
les données compilées et traitées de manière à générer des couches spatiales continues, la
conservation de ces données sous une forme géoréférencée commune peut en faciliter
l’utilisation dans des travaux futurs, particulièrement pour ce qui est de la planification spatiale,
de la prise de décision et de la gestion écosystémique. On a compilé une série complète de
couches de données physiques, chimiques et biologiques (30 couches) sur la région du golfe du
Maine en une seule base de données géographiques, en format xyz, en utilisant des sources de
données océanographiques des États-Unis et du Canada accessibles au public. Le principal
facteur motivant cet effort était le projet conjoint du Census of Marine Life [inventaire de la vie
marine] avec trois autres régions (le récif de la Grande Barrière, le golfe du Mexique et le golfe
du Maine), projet qui avait pour objectif d’examiner le rôle des variables physiques aux fins de la
prévision du profil de la biodiversité des assemblages du fond océanique. Dans le présent
rapport, nous présentons la méthodologie contextuelle (notamment les métadonnées officielles),
les problèmes rencontrés lors de la compilation de ces ressources géospatiales et les mises en
garde relativement à une utilisation ultérieure.
1
INTRODUCTION
The management and conservation of the marine environment and marine resources
increasingly requires the synthesis of spatial data from a range of physical and biological features
across a variety of scales. The work phase of compiling these data layers for a particular project
can be time intensive and costly. However, once these datasets are compiled and processed to
generate continuous spatial layers, the preservation of these data in a common georeferenced
format can facilitate their use in future work; in particular for spatial planning, decision making
and ecosystem-based management (Fisher and Rahel 2004, Gee 2007, Wood and Dragicevic
2007). In conducting a recent investigation on the role of physical environmental variables in
predicting biodiversity composition of benthic and demersal fish and invertebrate assemblages in
the Gulf of Maine, we compiled 31 oceanographic layers and seabed characteristics. These layers
were then preserved in xyz format and as raster grids in an Environmental Systems Research
Institute (ESRITM) file geodatabase. In this report we provide background methods (including
formal metadata), issues faced in compiling these geospatial resources, and caveats for
subsequent usage.
CoML Cross-Project Synthesis: Physical Surrogates for Predicting Seabed Biodiversity
In 2000 the International Census of Marine Life (CoML) began a global effort to assess
and explain the diversity, distribution and abundance of marine life. This marine research
initiative included 14 ―
field projects‖ covering different marine regions, habitats and functional
groups of organisms within the global ocean (e.g. the Abyssal Plains, Coral Reefs, Zooplankton
etc; www.coml.org). One of these field projects, the Gulf of Maine Area program (GOMA) was
selected as the CoML’s Regional Ecosystem project and focuses on the biodiversity of marine
life in the Gulf of Maine (GOM [grey area, Figure 1]). The GOM is located on the eastern North
American continental shelf between 47° and 39°N latitude and covers approximately 250K km2
(see Incze et al. (2010) for an overview of the region and its marine biodiversity).
One GOMA initiative has been to participate in a cross-project synthesis with two other
regions, the Great Barrier Reef and Gulf of Mexico, to characterize how physical factors affect
species distribution and abundance patterns in contrasting ecosystems. The distribution and
abundance of marine species and assemblages has been of fundamental interest to science and of
considerable importance to management and conservation. For most marine species, such
information is severely lacking, partly due to the great expense and time required for ship-based
biological surveys. To deal with this problem, methods of generalization are required. Since
many benthic organisms are strongly associated with specific habitat characteristics, the CoML
project focused on the use of physical environmental variables (e.g. substrate, benthic
temperature, nutrient concentrations) to predict the spatial pattern in seabed assemblages. The
analysis required the compilation of numerous oceanographic datasets from the GOM. These
data were determined to also be important for ecosystem-based studies in general, including
those underway with an Ecosystem Research Initiative (ERI; http://www.dfompo.gc.ca/science/publications/fiveyear-plan-quinquennal/index-eng.html#a3_2)
being
conducted by Fisheries and Oceans Canada (DFO). Therefore, as a prerequisite to the statistical
analyses for the seabed diversity project, oceanographic datasets compiled for two major time
periods corresponding to the years the biological data were sampled ([1956–1968] and [1996–
2007]) were preserved as a collection. These physical datasets, from numerous data sources,
2
were processed to derive continuous layers for the GOM study area. Biological datasets were
assembled from several publicly-available databases as well. However, in this report we only
provide information on data processing steps and provide links to the authoritative database
providers for each oceanographic dataset. The benthic biological datasets are not included in the
compilation as it is anticipated that future studies would proceed by accessing the most up-todate data directly from these biological databases.
METHODS
CoML Cross-Project Methods
All three regions involved in this project collated broad-scale biological survey datasets
comprised of site-by-species abundance data collected from trawls, benthic sleds, and
grabs/cores (Figure 1), as well as site-by-physical datasets comprised of available physical
variables.
For the GOMA project three large benthic datasets were accessed. These data were
originally acquired by federal agencies in Canada (Fisheries and Oceans Canada (DFO)) and the
United States (Northeast Fisheries Science Center (NEFSC), National Marine Fisheries Service
(NMFS), National Oceanographic and Atmoshpheric Administration (NOAA)).
These datasets represent the three most extensive benthic datasets in the GOM region: the
NEFSC Bottom Trawl Surveys, DFO Ecosystem Surveys, and the NEFSC Benthic Database
(Smith-McIntyre Grab samples). Each dataset was filtered to only include taxa identified to
genus or species.
Oceanographic habitat variables for the time period of each biological dataset were
collected and included: bathymetry and derivatives, seabed current stress, sediment
characteristics, benthic irradiance, nutrients, temperature, salinity and chlorophyll. The
oceanographic habitat variables along with the biological data were analyzed using Gradient
Forest (a modified version of a Random Forests analysis (Breiman 2001)) to identify important
environmental variables influencing the distribution and abundance of benthic species in the
GOM. A more complete summary of the Gradient Forest statistical methods is provided by Ellis
et al. (2010 In Prep).
Source Biological Datasets
NOAA Benthic Database Smith-McIntyre Grab Samples
The NOAA Benthic Database was accessed prior to its release to the Ocean
Biogeographic Information System (OBIS, http://www.iobis.org/), although, by this time
virtually all quality control measures had been completed. Only Smith-McIntyre Grab samples
were selected that occurred between 1956–1968, inclusive, in the GOM Area (Theroux and
Wigley 1998). This time period was selected to include the majority of the data collected from
Smith-McIntyre Grab samples, collected between 1956–1985 (Figure 2). The Smith-McIntyre
spring-loaded grab sampled 0.1 m2 of bottom area and had a volume of approximately 15 L.
Sampling primarily took place during summer months with the majority of collections made in
July and August. Samples (n = 618) were distributed across all years, 1956–1968, although 85%
were collected from 1957–1960 (Figure 2). Analyses included 315 species from 478 stations that
coincided with all the physical variables, including between 1 and 25 replicates per location. The
main taxa in this dataset included Arthropoda (177) and Mollusca (162), although annelids,
3
echinoderms, bryozoans, sipunculids, cnidarians, hemichordates and baccilariophytans were also
present with the number of taxa ranging from 1 to 28. The geographic distribution of samples in
the Gulf of Maine excluded the western Scotian Shelf and the Bay of Fundy (Figure 1).
a
b
c
Figure 1. Distribution of biological samples in the Gulf of Maine (grey shaded area denotes the
geographic region of interest for CoML’s GOMA program) (a) Smith-McIntyre Grab Samples
collected by Theroux and Wigley (1998) between 1956–1968 (b) DFO Winter and Summer
Ecosystem Surveys stations from 2000–2007 (c) NMFS Benthic Trawl Surveys from 1997–2007.
250
150
100
50
0
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
Number of Samples
200
Year
Figure 2. Number of Smith-McIntyre grab samples collected per year (Theroux and Wigley
1998).
4
Data collected during the DFO Ecosystem Surveys were exported from the DFO
Groundfish database for trawls conducted on the Scotia Shelf, in the Bay of Fundy or on Georges
Bank from 1996–2007, inclusive. The DFO Annual Ecosystem Survey samples demersal fish
and has increasingly been recording data on invertebrate species. The Scotian Shelf surveys were
first implemented in 1970 and occur in the summer (July) while the Georges Bank surveys began
in 1986 and take place in Feb–March (Clark 2010).
The survey uses a stratified random sampling design where stratification is primarily
based on depth, and the allocation of the number of stations per stratum is proportional to the
variance in catch of haddock (Clark, 2010). The Scotian Shelf summer survey contains 48 strata
(Figure 3) distributed among four depth zones (< 92 m, 93 – 183 m, > 184 m, and mixed) to a
maximum of 731 m along the shelf edge (Frank 2004). The Georges Bank spring survey is
stratified among 3 depth zones (< 93 m, 93 – 182, and mixed). The same trawl gear, the Western
IIA, was used for surveys during the 1996 to 2007 period. However, irregularities in the rigging
of the trawl in the 2004 summer survey made direct comparisons of catch for some species
questionable and this survey was excluded from the analysis (Clark 2010). The trawl gear uses a
small mesh codend liner (19mm) capable of retaining forage and small, non-commercial species.
At each sampling station within each stratum, a standard bottom tow defined to be a 30 minute
haul on bottom at 3.5 knots was conducted. This results in an area swept of 0.0404 km 2 for a
standard tow (Shackell and Frank 2003). More details on the surveys can be found in Chadwick
et al. (2007).
The original data were exported from a dataset published on the Ocean Biogeographic
Information System, OBIS (Clark and Branton 2007) for the period from 1996–2007, and
included 1032 trawls on the Scotian Shelf and 1299 trawls on Georges Bank (Figure 2). The
spring survey included 112 fish species comprised of 99 genera, and 49 invertebrate species
comprised of 49 genera.
The most frequent species in the Scotian Shelf summer survey dataset included:
Melanogrammus aeglefinus (Haddock), Clupea harengus (Herring), Merluccius bilinearis
(Silver Hake), Squalus acanthias (Spiny Dogfish), Illex illecebrosus (Short-finned Squid),
Myoxocephalus octodecemspinosus (Longhorn Sculpin), Pseudopleuronectes americanus
(Winter Flounder), Pollachius virens (Pollack), Placopecten magellanicus (Sea Scallop), and
Limanda ferruginea (Yellowtail Flounder).
In the Georges Bank spring survey dataset the most frequent species included:
Melanogrammus aeglefinus (Haddock), Clupea harengus (Herring), Myoxocephalus
octodecemspinosus (Longhorn Sculpin), Limanda ferruginea (Yellowtail Flounder), Merluccius
bilinearis (Silver Hake), Scomber scombrus (Atlantic Mackerel), Illex illecebrosus (Short-finned
Squid), Leucoraja ocellata (Winter Skate), Gadus morhua (Cod), and Leucoraja ocellata
(Winter Skate)
With both datasets analyses were performed with and without data collected before 2000.
These data were excluded as some shrimps, crabs and scallops were only recorded since 1999
(Tremblay et al. 2007, Clark 2010). For the Scotian Shelf summer survey dataset invertebrates
included in the analysis were: Cancer borealis, Cancer irroratus, Chaceon quinquedens,
Chionocetes opilio, Homarus americanus, Hyas araneus, Hyas coarctatus, Lithodes maja,
Pandalus borealis, Pandalus montagui, Illex illecebrosus and Placopecten magellanicus.
Abundances of Pandalus borealis and Pandalus montagui were determined by dividing by the
average species weight as only the weight was reliably measured. For the Georges Bank spring
5
survey dataset the invertebrate taxa included in the analysis were only Homarus americanus and
Illex illecebrosus.
The final analysis only included taxa that were recorded in > 5% of samples. Therefore,
rare species were not included in the analysis—a more detailed description of the reasoning and
implications of this decision be found in Ellis et al. (2010 In Prep). Samples were also removed
that did not have corresponding physical data in any one of the physical variables. The Random
Forest analyses included 81 species in the late winter-early spring survey, and 95 species in the
summer survey.
NEFSC Bottom Trawl Survey
NEFSC of NMFS conducts trawl surveys in the GOM in both the fall and spring using a
stratified random sampling design. The fall surveys began in 1963 and sample depths from 27 to
365 m (Despres-Patanjo et al. 1988). The spring survey series began in 1968. In 1972 the
geographic coverage of the surveys were extended to inshore areas landward of the 27 m isobath.
The stratified random sampling assures a fairly uniform distribution of stations throughout the
survey areas with an average allocation per seasonal survey of 350 stations. Strata are delineated
by depth (Figure 3). Stations were allocated to strata in proportion to strata area and were
randomly assigned to specific locations within strata (Azarovitz 1981). The Yankee 36 Bottom
Trawl used for the survey sweeps 0.0334 km2 during a standard 30 min tow at 3.8 knots, which
is slightly less than the area swept in the DFO Ecosystem survey.
Data from the fall and spring surveys were exported (and treated separately) for the
1996–2007 time period. This included 2001 tows that were done in the fall (average of 164 per
year, min. 152, max. 192) and 1975 tows that were done in spring (average of 165 per year, min
156, max 182). The fall surveys generally take place in October (1643 tows) but occasionally
occurs in September (226 tows) and December (132 tows), while the spring survey usually takes
place in April (1723 tows) with occasional tows in March (238 tows) and May (14 tows). The
data used are available as an OBIS dataset (NOAA's National Marine Fisheries Service
Northeast Fisheries Science Center 2005), that was clipped to the GOM area and filtered to only
include taxa identified to genus or species. Since the inception of these surveys in 1963, species
identifications have increased to include more invertebrate data, and our selection of survey data
from 1996 onwards was in part due to the species identifications for fish (fall: 146 species, 124
genera; spring: 98 species, 88 genera) having reaching 100%. Invertebrates in the dataset
included 33 species and 29 genera in the fall and 26 species and 23 genera in the spring. Data on
invertebrate species has increasingly been collected and recorded for this survey. Data from the
fall and spring surveys were treated separately for the 1996–2007 time period. Samples were also
removed that did not overlap with corresponding physical data in any one of the physical
variables.
6
a
b
Figure 3. (a) The DFO Ecosystem Survey strata for Georges Bank and the Scotian Shelf in the
GOM area (grey area). Labels indicate strata names (b) The NMFS Bottom Trawl Survey strata
in the GOM area. Labels indicate strata names.
7
Selection of Time Periods
Time periods of the biological and corresponding physical data were selected based on
those years that had abundant data available for each of the biological datasets selected. The
Smith-McIntyre grab samples included data from 1956–1985, however most data was collected
from 1956–1968 and therefore only those years were included in the analyses. Corresponding
physical data for the 1956-1968 time period was largely present, as most time series datasets for
oceanographic variables began in the 1930s; excluding silicate which only began being recorded
in 1961. Satellite information was only available for the contemporary time period as the
AVHRR and SeaWifs satellites only began collecting data in the early to mid-1990s. However, it
was decided to use the contemporary satellite for the 1956–1968 time period as no other data
were available. It was also determined that the 1956–1968 time period did not have enough
nitrate and silicate data to warrant creating layers.
The time periods selected for the DFO Ecosystem Surveys and the NEFSC Bottom Trawl
Surveys datasets were to be from 1996–2007; however, it was decided to run analyses not
including years where invertebrates had not been rigorously recorded in the DFO Ecosystem
Surveys (before the year 2000). Oceanographic datasets were largely available for this time–
period, although satellite sea surface temperate and chlorophyll only began to be sampled in
1997, and dissolved oxygen was not available for this period from the GOM Region Nutrient and
Hydrographic database.
Geodatabase Creation and Data Access
The physical environmental variables were preserved in an ESRI (Environmental
Systems Research Institute, Redlands California) file geodatabase that can be accessed by
contacting Michelle Greenlaw (St. Andrews Biological Station, [email protected]). A geodatabase is a database designed to store, query and manipulate geographic
information and spatial data. All data are stored in raster format (coordinate system: World
Geodetic System of 1984, Universal Transverse Mercator Zone 19 [WGS84 UTM Zone 19]).
Each layer also has an associated metadata file which includes the data source, description,
purpose, supplementary information and contact information for the layer. Metadata can be
accessed by opening the layer in ESRI’s ArcCatalog and using the Metadata tab (Figure 4).
8
Figure 4. A view of ESRI®’s ArcCatalog showing the geodatabase in the left panel, expanded to
show each physical layer. The right panel is a view of the metadata for the benthic temperature
layer (1956–1968).
XYZ Text File Creation and Data Access
Physical environmental variables were also exported in xyz format (three column format
including latitude, longitude then physical variable) that can be accessed by contacting Michelle
Greenlaw (St. Andrews Biological Station, [email protected]). This format is
easily imported by all Geographic Information Systems (GIS) software and is more accessible to
certain non-GIS software programs. The files are in the World Geodetic System of 1984
(WGS84) coordinate system. The xyz files do not have associated metadata therefore this report
or the geodatabase will serve as metadata for those files.
9
Interpolation Methods
Optimal Estimation
The program OAX version 5.1 was used to interpolate continuous data layers for a
number of point-based oceanographic datasets. This program, developed in the early 1990’s at
DFO (He et al. 2003), applies the method of optimal interpolation to estimate the values of
variables at specified points in space and time. Each interpolated point is calculated using a
nearest neighbour algorithm, where the weighted average is taken for a specified number of data
points that are closest to the interpolated point. Unlike many other interpolation techniques,
OAX allows the user to interpolate in four dimensions: longitude (x), latitude (y), depth (z), and
time (t).
To run OAX, the program and three additional files are required; a grid, deck, and data
file. The grid file is a text file containing a list of grid points. Each point is where an estimated
data value will be calculated. Optimally estimated benthic data layers were created using a grid
file derived from the United States Geological Survey (USGS) North American 15 arc-second
Digital Elevation Model clipped to the GOM extent (Roworth and Signell 1998). This grid was
resampled to coarser resolutions when required due to the spatial distribution of sample coverage
of the oceanographic datasets.
The deck file is a text file that specifies the dependent variable to be estimated, the
independent variables (i.e. x, y, z, t), the data file to be interpolated, the grid file, the number of
nearest neighbour data points to use, the global scales, and the statistical model to be used. The
global scales are the overall scales for the independent variables that are used to weight the
independent variables and define a distance calculation. This distance enables the selection of the
closest number of nearest neighbours from which a value will be interpolated for each grid point.
The number of nearest neighbours was a function of the spread of the data points for each
interpolated layer. For all runs of OAX, the statistical model was the estimated mean.
The data file is a text file that contains the dataset to be interpolated. All data in both the
grid and data files must be in a Cartesian coordinate system. Data was reprojected from the
geographic coordinate systems NAD 83 or WGS 84 to the projected coordinate system WGS 84
UTM Zone 19. This resulted in equivalent units (meters) in all spatial directions (x, y, and z);
which is a requirement for OAX to perform its mathematical calculations.
To derive seasonal data layers, input data were restricted to correspond to the following
day-of-year limits:
Season
Day-of-Year Limits
Winter
1–90
Spring
91–181
Summer
182–273
Fall
274–365
The code for OAX 5.1 was compiled for windows and can be run through the MS-DOS
command window. Additional information and documentation on OAX can be found online at
http://www2.mar.dfo-mpo.gc.ca/science/ocean/coastal_hydrodynamics/Oax/oax.html.
10
Spline Interpolation
The ESRI ArcGIS Spatial Analyst extension was used for spline interpolations of
substrate and stratification samples as these layers did not require interpolation in 4 dimensions
(x, y, z and t). Parameters for each specific interpolation are listed in the description of the
physical environmental dataset. Spline interpolation is a method that estimates values using a
mathematical function that minimizes overall surface curvature, resulting in a smooth surface
that passes exactly through the input point. Spline was chosen as the interpolation method to
ensure the fitted surface passes through the input points, as many of the substrate and
stratification points provided were taken at the same time as the biological samples. This method
would ensure the accuracy of the estimate at those points. At the points where an actual sample
of substrate or stratification was not available, the spline method has an average accuracy in
comparison with other interpolation methods. The average relative mean absolute error and
relative root mean square error were 40% and 30% when tested, respectively (Li and Heap
2008).
Physical Oceanographic Datasets
Bathymetry and Derivatives
A USGS digital elevation model (Roworth and Signell 1998) at a resolution of 15 arcseconds or 397 m (Figure 5), was used to intersect biological samples with bathymetry
information and to calculate slope (maximum change in depth in the 8 surrounding grid cells,
degrees), aspect (degrees from north), bathymetric position index (BPI, unitless), and benthic
complexity (maximum change in slope in the 8 surrounding grid cells, degrees). The data unit of
the bathymetry layer was in meters.
Substrate
Sediment values were extracted from the standing stock of USGS (Poppe et al. 2005) and
Canadian Geological Survey (Geological Survey of Canada 2009) database records (US: 1955–
2004, Can: 1964–2003). These were combined into one point layer of sediment samples. Fields
in the data included percent sand, percent mud, and percent gravel from point samples (Figure 6).
Point data were interpolated to a 6000 m resolution raster grid using the spline method. The
spline parameters used were: Tense, 40 Weight, and a 3 point average. Resolution of the output
layers was determined by the density of the original data.
Bottom Stress
Bottom stress (Figure 7) was calculated from the frictional velocity maps of Drozdowski
and Hannah (2010), which captured the effect of waves and modeled tidal bottom currents. Their
friction velocity calculation was done on a grid covering the GOM and Scotian Shelf and was the
synthesis of 3 earlier data products: (1) the high resolution bathymetry data (0.25 min for the
GOM and 0.4 min for the Scotian Shelf), (2) 42 year hindcast of the wave height and period data
(Swail and Cox 2000), and (3) the near-bottom tidal currents obtained from a combination of 3D
model and 2D tidal model results of Hannah et al. (2001) and Han and Loder (2003). The final
calculation of friction velocity was performed using the sediment transport model
SEDTRANS96 (Li and Amos 2001). Results of the 90th percentile significant wave height and
period were used, which have the interpretation of representing moderate to large wave events
that occur 10% of the time (~one month per year).
11
Frictional velocity was converted to bottom stress using the formula (Condie and Webster
1997): bottom stress = (Bottom frictional velocity) 2 * water density where, water density =
1027.5 kg/m3. The output resolution of the raster data layer was 952 m.
Bottom stress with only the influence of tides (Figure 7) was calculated using frictional
velocity in m2s-2 exported from Gulf of Maine Ocean Observation System (GOMOOS) Nowcast
Forecast System (Xue et al. 2005). Frictional velocity was multiplied by water density (1027.5)
to calculate bottom stress. Bottom stress in the GOM is driven mostly by tidal flow therefore; the
largest source of temporal variability is from the lunar cycle. A layer of bottom stress was
calculated for each complete month (to capture lunar cycle) rather than a year or more since
monthly data captures the majority of temporal and spatial variability in the GOM. The frictional
velocity vector data received was calculated every 3 hours from August 1, 2008 to August 31,
2008 (248 time layers). After converting the vector data to scalar data, summary statistics were
calculated for each model grid point as follows (n = 248): Mean; STD Dev; Min.; Max.; Range
(Max-Min). These summary statistics were conserved in a GIS polygon layer. The data were
then converted to a raster at a resolution of 3800 m.
Stratification
Stratification layers (Figures 8 and 9) were calculated using density point samples from
the DFO Hydrographic (Climate) database (Gregory 2004) as the density difference between the
surface (0 m) and 50 m (Helbig and Higdon 2009). Density values were exported at 0 and 50 m
depth for the time periods 1956–1968 and 1996–2007 and from both January to December
(yearly average layer) and May to September (summer average layer). Points were used only if
they had a corresponding value at both 0 and 50 m depth. The resulting points were interpolated
to 2500 m grids using the spline method with the parameters: Tense, 40 Weight, 3 point average.
Sampling effort and its monthly distribution varied between the two time periods. The
1956–1968 (Figure 8) time-period included 8654 more density samples than the 1996–2007
time-period (Figure 9). From 1956–1968, most samples were collected in August (14% of
samples), while September had the least sampling effort (4.5% of samples). During this time
period, 1968 was the most sampling intensive year, comprising 15% of all samples, while 1961
was the least sampling intensive year with 2% of all samples. For the 1996–2007 time period,
April had the most samples (15% of samples), while December had the fewest (0.9% of
samples). Most samples were collected in 1998 (13% of samples), while 2003 had the least
number of samples (5% of samples). The resolution of the output layers was determined by the
original data density.
Benthic Irradiance
SeaWiFS K490 data (Figure 10) were downloaded for the time–period of 1997–2008,
from the NOAA OceanColor ftp download site (NASA 2007) at an 8000 m resolution. K490
indicates the turbidity of the water column; how visible light in the blue–green region of the
spectrum penetrates within the water column. Monthly composites were used to calculate a value
corresponding to the month each benthic sample was taken. These values were included in the
formula below (as K490_monthN) used to calculate a seasonal benthic irradiance:
12
Benthic Irradiance = Cos ((LAT-offsetN) / 180 * π) * Exp (K490_monthN * Depth)
Where offsetN is the position of the sun for month N:
Month
OffsetN
January
-21.2
February -12.2
March
0.0
April
12.3
May
21.2
June
24.5
July
21.2
August
12.2
September
0.0
October
-12.3
November -21.2
December -24.5
Chlorophyll
Chlorophyll (CHL, mg m-3) data (Figure 10) were acquired from the Satellite
Oceanography Laboratory, University of Maine (PI: Andrew Thomas) who downloads and
processes Sea-viewing Wide Field-of-View Sensor (SeaWiFS) CHL data. These data included
monthly composites for the time-period of 1997–2008 and were used to calculate a yearly
average and range at 855 m and 1119 resolution, respectively.
Sea Surface Temperature
Sea surface temperature (SST, degrees C) data (Figure 10) were acquired from the
Satellite Oceanography Laboratory, University of Maine (PI: Andrew Thomas), who downloads
and processes SST data from the Advanced Very High Resolution Radiometer (AVHRR)
satellite. These data include monthly composites for the 1997–2008 time-period. These data were
then used to calculate the yearly average and seasonal range over the time-period. The data were
available at a resolution of 972 m.
Benthic Temperature
Benthic temperature layers were derived for the time-periods 1956–1968 (Figure 11), and
1996–2007 (Figure 12). The 1956–1968 time period used data exported from the DFO
Hydrographic (Climate) database for the entire water column (Gregory 2004) (n = 691 703 data
records). The 1996–2007 time period used data from two sources: the NEFSC (Mountain et al.
2004) and the DFO Hydrographic (Climate) database. The NEFSC data included data 10 m from
bottom taken with thermometers or Conductivity–Temperature–Depth recorders (CTD). This
included data from the NEFSC Bottom Trawl Surveys, Marine Resources Monitoring
Assessment and Prediction (MARMAP), Ecosystem Monitoring (EcoMon), and Global Ocean
and Ecosystem Dynamics (GLOBEC) programs. Areas in the GOM that were not covered by the
NEFSC data were filled in with data from the DFO Hydrographic (Climate) Database.
Hydrographic database data were filtered to within 30 m of bottom depth. Duplicate records
between NEFSC and BIO data were identified and removed. The final dataset consisted of 18132
data records. For both time periods, data were present for all years and all seasons with the
majority of the data in summer and the least amount of data in the winter.
13
Optimal estimation routines were used to derive seasonal average temperature layers for
each time period at a 6 km resolution, using 15, and 10 nearest neighbours for the 1956–1968
period, and 1996–2007 period, respectively. These data were then converted into a raster grid.
From the seasonal averages a total average and seasonal range was calculated. The data units of
these layers are degrees Celsius. The resolution of the output layers was determined by the
original data density.
Benthic Salinity
Salinity layers (psu) were created for two time periods: 1956–1968 (Figure 13) and
1996–2007 (Figure 14). For the 1956–1968 salinity layer, data were extracted from the DFO
BioChem database of biological and chemical oceanographic data (Gregory and Narayanan
2003, Fisheries and Oceans Canada 2006). Data were filtered to within 20 m of bottom and
optimal estimation routines were used to derive an annual average layer at a resolution of 40 km
using the 6 nearest neighbours. A total of 408 points were used. There was not enough data
within each season to calculate seasonal averages or an overall seasonal range.
For the 1996–2007 time period, two sources of data were used, the NEFSC dataset and
the DFO Hydrographic (Climate) Database. Where there were spatial gaps in the NEFSC data,
data from the DFO Hydrographic (Climate) Database were used. Data from NEFSC were filtered
to within 10 m of bottom while data from the DFO Hydrographic (Climate) Database were
filtered to within 30 m of bottom. Duplicates between the two datasets were removed and the
total number of salinity records was 14297. Each season had a minimum of 2000 data records.
Optimal estimation was used to derive seasonal benthic salinity layers using the 10 nearest
neighbors at a resolution of 6 km. The seasonal layers were then used to create total average and
seasonal range benthic salinity data layers. The resolution of the output layers was determined by
the original data density.
Dissolved Oxygen
Dissolved oxygen (Figure 15) was exported from the DFO BioChem Database for the
time period 1956–1968. Values were filtered to within 20 m off bottom for a total of 330
dissolved oxygen records. Optimal estimation was used to derive an annual average benthic
dissolved oxygen layer using the 6 nearest neighbours at a resolution of 40 km. There was not
enough data within each season to calculate seasonal averages or an overall seasonal range.
Dissolved oxygen data were not available from the GOM Region Nutrient and Hydrographic
database (Rebuck et al. 2009) for the 1996–2007 time period. Data are in micromoles per litre.
Bottom Nutrients
Phosphate: Phosphate (µm) was exported from the DFO BioChem Database for the time
period of 1956–1968. Values were filtered to within 20 m off bottom for a total of 196 phosphate
records. Optimal estimation was used to derive an annual average benthic phosphate layer using
the 6 nearest neighbours at a resolution of 40 km (Figure 16). There was not enough data within
each season to calculate seasonal averages or an overall seasonal range.
Phosphate data from 1996–2007 were exported from the GRAMPUS and Hydrographic
Databases (Rebuck et al., 2009) for a total of 52 833 observations, from the entire water column.
Optimal estimation was used to derive an annual average benthic phosphate layer using the 15
nearest neighbours at a resolution of 6 km (Figure 17). The spatial coverage of the data was not
sufficient for seasonal layers to be derived.
Silicate: Silicate data (µm) from 1996–2007 were exported from GRAMPUS for a total
of 53 500 observations, from the entire water column. Optimal estimation was used to derive an
14
annual average silicate layer using the 15 nearest neighbours at a resolution of 6 km (Figure 18).
The spatial coverage of the data was not sufficient for seasonal layers to be derived.
Nitrate: Nitrate data (µm) from 1996–2007 were also exported from GRAMPUS for a
total of 53 500 observations, for the entire water column. Optimal estimation was used to derive
an annual average nitrate layer using the 15 nearest neighbours at a resolution of 6 km (Figure
19). The spatial coverage of the data was not sufficient for seasonal layers to be derived.
Table 1. Physical environmental variables with a GOM-scale coverage used for analysis of the
three regional biological datasets referred to in text. (SST = Sea Surface Temperature, K490 =
Mean Diffuse Attenuation Coefficient, Stratification = density difference between 0m and 50m
depth, HD = DFO Hydrographic (Climate) Database, USGS = United States Geological Service,
SeaWifs = Sea-viewing Wide Field-of-view Sensor, CGS = Canadian Geological Survey,
NEFSC = Northeast Fisheries Science Centre).
Variable
Geodatabase
Name
Units
Aspect
ASPECT
degree
397
USGS
Benthic Current Stress with Wind and Tidal Influences
BOTSTR_WT
newtons meters -2
952
Model
-2
Resolution (m)
Data Source
Benthic Current Stress with only tidal influence
BOTSTR_T
newtons meters
3800
Model
Average Benthic Temperature 1956–1968
Average Benthic Temperature 1996–2007
BT_AVG56
degree C
6000
HD
BT_AVG96
degree C
6000
NEFSC/HD
Seasonal Range of Benthic Temperature 1956–1968
BT_RG56
degree C
6000
HD
Seasonal Range of Benthic Temperature 1996–2007
BT_RG96
degree C
6000
NEFSC/HD
-3
Average Sea Surface Chlorophyll
CHL_AVG
mg m
855
SeaWifs
Seasonal Range of Sea Surface Chlorophyll
CHL_RG
mg m-3
1119
SeaWifs
Benthic Complexity
COMPLEXITY
degree
397
USGS
Depth
DEPTH
meters
397
USGS
Benthic Dissolved Oxygen 1956–1968
DO_AVG56
μM
Gravel
GRAVEL
percent
6000
USGS/CGS
Average K490
K490_AVG
none
8000
SeaWifs
Seasonal Range of K490
K490_RG
none
8000
SeaWifs
Mud
MUD
percent
6000
USGS/CGS
Benthic Nitrate 1996–2007
NIT_AVG96
μM
40000
GRAMPUS
Benthic Phosphate 1956–1968
PHOS_AVG56
μM
40000
Biochem
Benthic Phosphate 1996–2007
PHOS_AVG96
μM
6000
Benthic Salinity 1956–1968
SAL_AVG56
psu
40000
Benthic Salinity 1996 – 2007
SAL_AVG96
psu
6000
HD/NEFSC
Seasonal Range of Benthic Salinity 1996–2007
SAL_RG96
psu
6000
HD/NEFSC
Sand
SAND
percent
6000
USGS/CGS
Benthic Silicate
SIL_AVG96
μM
6000
GRAMPUS
Slope
SLOPE
degree
397
USGS
Average SST
SST_AVG
degree C
972
SeaWifs
Seasonal Range of SST
SST_RG
degree C
972
SeaWifs
Stratification from 1956–1968
STRAT56
none
2500
HD
Stratification from 1996–2007
STRAT96
none
2500
HD
Summer Stratification from 1956–1968
STRAT_SUM56
none
2500
HD
Summer Stratification from 1996–2007
STRAT_SUM96
None
2500
HD
40000
Biochem
GRAMPUS
Biochem
15
+
,.
KiIomo .....
United Slain
,·.x-;. y
..
~
~
/"-~
" ~r\j!1:
"
+
I'_~
ry}f J
B~
Figure 5. Depth and derivatives created from the USGS 15 arc-second digital elevation model of
the GOM.
16
Figure 6. Substrate layers of percent mud, percent gravel, and percent sand (from the USGS and
CGS sediment samples). Insert maps in each figure show the density of points used to create the
layers.
17
Figure 7. Bottom stress modeled with wind and tidal influence, and bottom stress modeled with
only tidal influence. Note that the scales of the two bottom stress layers span over different
ranges to emphasize areas of high stress, but at a different magnitude when wind is not included.
Original data to create inset maps of bottom stress data density were not available.
300
Number of Samples
450
400
350
300
250
200
150
100
50
0
250
200
150
100
50
Number of Samples
1968
1967
1966
1965
1964
1963
1962
1961
1960
1959
400
350
300
250
200
150
100
50
0
December
November
October
September
August
July
June
May
April
March
Febuary
Month
1958
Year
400
350
300
250
200
150
100
50
0
January
Number of Samples
Year
1957
1956
1968
1967
1966
1965
1964
1963
1962
1961
1960
1959
1958
1957
0
1956
Number of Samples
18
Month
Figure 8. Top: Yearly average and summer stratification for the 1956–1968 time-period. Inset
maps in each figure show the density of points used to create the layers. Middle: Number of
samples per year to create the stratification and summer stratification layers, respectively.
Bottom: Number of samples per month to create the stratification and summer stratification
layers, respectively.
70000
30000
60000
25000
Number of Samples
Number of Samples
19
50000
40000
30000
20000
10000
0
20000
15000
10000
5000
0
Year
40000
35000
30000
25000
20000
15000
10000
5000
0
December
October
November
September
July
Month
August
June
May
April
March
January
Number of Samples
40000
35000
30000
25000
20000
15000
10000
5000
0
February
Number of Samples
Year
Month
Figure 9. Top: Yearly average and summer stratification for the 1996–2007 time-period. Inset
maps in each figure show the density of points used to create the layers. Middle: Number of
samples per year to create the stratification and summer stratification layers, respectively.
Bottom: Number of samples per month to create the stratification and summer stratification
layers, respectively.
20
Figure 10. Top: Average SeaWifs K490 from 1997–2008. Middle and bottom: Average and
range of chlorophyll (CHL) and sea surface temperature (SST) in the GOM from 1997–2008.
21
Number of Samples
50
40
30
20
10
0
Year
5000
4500
J
Number of Samples
4000
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
3500
3000
2500
2000
1500
1000
500
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 11. Top: The average and range of benthic temperature from 1956–1968. Inset maps
show the density of data used to create that layer. Middle and Bottom: The number of samples
used to create the benthic temperature map by year and by day of year, respectively.
22
Number of Samples
2500
2000
1500
1000
500
0
Year
240
220
J
Number of Samples
200
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
180
160
140
120
100
80
60
40
20
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 12. Top: The average and range of benthic temperature from 1996–2007. Inset maps
show the distribution of points used to create that layer. Middle and Bottom: The number of
samples used to create the benthic temperature map by year and by day of year, respectively.
Number of Samples
23
90
80
70
60
50
40
30
20
10
0
Year
25
J
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
Number of Samples
20
15
10
5
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 13. Top: Average benthic salinity from 1956–1968. Inset map shows the distribution of
points used to create that layer. Middle and Bottom: The number of samples used to create the
benthic salinity map by year and by day of year, respectively.
Number of Samples
24
2000
1800
1600
1400
1200
1000
800
600
400
200
0
Year
160
J
Number of Samples
140
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
120
100
80
60
40
20
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 14. Top: Average and range of benthic salinity from 1996–2007. Inset maps show the
density of points used to create that layer. Middle and Bottom: The number of samples used to
create the benthic salinity map by year and by day of year, respectively.
Number of Samples
25
100
90
80
70
60
50
40
30
20
10
0
Year
25
Winter
J F M
Spring
A M J
Summer
J A S
Fall
O N D
Number of Samples
20
15
10
5
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 15. Top: Average benthic dissolved oxygen from 1956–1968. Inset map shows the
distribution of points used to create that layer. Middle and Bottom: The number of samples used
to create the benthic oxygen map by year and by day of year, respectively.
Number of Samples
26
50
45
40
35
30
25
20
15
10
5
0
Year
16
J
Number of Samples
14
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
12
10
8
6
4
2
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 16. Top: Average benthic phosphate from 1956–1968. Inset map shows the distribution
of points used to create that layer. Middle and Bottom: The number of samples used to create the
benthic phosphate map by year and by day of year, respectively.
27
Number of Samples
7000
6000
5000
4000
3000
2000
1000
0
Year
700
J
Number of Samples
600
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
500
400
300
200
100
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 17. Top: Average benthic phosphate from 1997–2007. Inset map shows the distribution
of points used to create that layer. Middle and Bottom: The number of samples used to create the
benthic phosphate map by year and by day of year, respectively.
28
Number of Samples
7000
6000
5000
4000
3000
2000
1000
0
Year
700
J
Number of Samples
600
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
500
400
300
200
100
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 18. Top: Average benthic silicate from 1996–2007. Insert map shows the distribution of
samples used to create that layer. Middle and Bottom: The number of samples used to create the
benthic silicate map by year and by day of year, respectively.
29
Number of Samples
7000
6000
5000
4000
3000
2000
1000
0
Year
700
J
Number of Samples
600
Winter
F M
A
Spring
M J
Summer
J A S
O
Fall
N D
500
400
300
200
100
0
1
31
61
91 121 151 181 211 241 271 301 331 361
Day of Year
Figure 19. Top: Average benthic nitrate 1996–2007. Inset map shows the distribution of points
used to create that layer. Middle and Bottom: The number of samples used to create the benthic
nitrate map by year and by day of year, respectively.
30
DISCUSSION
A comprehensive suite of oceanographic layers for the Gulf of Maine Area has been
compiled within a single geodatabase, and in xyz format, using publicly–available U.S. and
Canadian oceanographic data sources. The primary driver for this effort was the Census of
Marine Life’s cross-project synthesis on the role of physical variables in predicting patterns of
distribution and diversity in seabed assemblages. Using Grdient Forest (multivariate Random
Forests) statistical analyses (Ellis et al. 2010 In Prep), the relative importance of different
environmental variables in their predictive capacity relative to observed species distributions in
three large regional biological survey datasets was evaluated; however these results are outside
the scope of this report (see: Pitcher et al. (In Prep)).
The preservation of the oceanographic datasets ensures quick and relatively easy access
in a common and standard format, thus facilitating their use in future studies. Although large
regional spatial physical datasets, similar to the ones presented in this report, have previously
been collated for specific research projects on the Scotian Shelf/Western GOM (Day and Roff
2000, Roff et al. 2003, Kostylev 2004, Greene et al. 2010) they have not generally been made
available in a geodatabase or other accessible formats (data from Greene et al. 2010 is available,
but only includes a few of the variety of layers assembled for our analysis). Currently, many
oceanographic physical layers developed for specific research programs are located on local
computers or are only publicly available as images in reports. The relative inaccessibility of
these types of data is currently a limitation to conducting ecosystem-based research in support of
an ecosystem approach to management.
The compiled datasets presented in this report represent what was deemed the best
available data for the GOM for the given time periods (1956–1968 and 1996–2007). However,
there were data issues when compiling both physical and biological datasets. No single
invertebrate dataset, for the contemporary time period, covered the extent of our study area; the
GOM. DFO’s Ecosystem Surveys and the NMFS groundfish trawl surveys represent a source of
invertebrate data; however many species of invertebrates had to be removed from the analyses
when it was determined they were not consistently recorded for the 1996–2007 time period
(Tremblay et al. 2007). DFO is currently attempting to consistently count all invertebrate species
during the Ecosystem Surveys, however it will be many years before this dataset represents a
useful invertebrate time-series (Clark 2010). Invertebrate datasets from smaller scale sampling
efforts, within specific basins or banks, were available for the contemporary time-period, but
could not be used because of their limited spatial extent.
The resolution of the oceanographic layers varied significantly (450 m–6 km). For those
datasets that were interpolated, the spatial scale at which these layers were produced represent
what was considered the highest reasonable resolution given the quantity and distribution of the
data. For those data collected through point-base sampling methods (e.g. CTD, biological survey,
etc), the seasonal distribution of data was patchy, with most sampling occurring in the summer
months. Other data were limited temporally by the timeframe of particular programs (e.g.
SeaWiFS; 1997 to present).
Many of the oceanographic layers derived from point-based sampling methods were
interpolated using the OAX optimal estimation program. Unfortunately, this program does not
provide any information about the inherent variance of the averages produced as layers over the
temporal and spatial scales, which would have helped to determine layer suitability for
subsequent analyses.
31
There were a number of data gaps for the oceanographic data. Sediment carbonate was
not recorded in either the USGS or Canadian Geological Survey (CGS), samples but would be of
interest for future biodiversity analyses since it was found to be influential in determining
biodiversity patterns in other regions (Pitcher et al. In Prep). Benthic current stress was also an
important factor, in the GOM and in the Great Barrier Reef (Pitcher et al. In Prep), however the
spatial extent was limited and did not cover the mid and upper Bay of Fundy, Western Scotian
Shelf, or Nantucket Shoals. In the CoML Cross-Project analysis, biological samples had to be
removed from these areas due to the lack of benthic current stress data, limiting the study from
analyzing one of the highest known current stress areas, the Bay of Fundy (Wildish et al. 1986).
Some environmental data, such as SST and CHL, were available through DFO but were
not accessible in a format that could be converted to a GIS raster. These data therefore had to be
acquired from the Satellite Oceanography Laboratory (University of Maine). The issue of GIS
compatibility of remote sensing data is being investigated by DFO's BIO Remote Sensing Unit.
Furthermore, when available, satellite CHL data should not be used in nearshore regions and
other highly turbid areas like the Bay of Fundy, as these products are known to be a combination
of CHL and turbidity. This factor did not end up being important in our analysis, as Bay of
Fundy points were removed due to the lack of benthic current stress data available for that
region.
The low density of nutrient data from 1956–1968 prevented us from producing layers of
nitrate or silicate for this time period. Phosphate data were also limited, but enough observations
were available to produce a layer gridded at a very large extent of 40 km. Finally, the original
density of the depth samples used to create the USGS’s digital elevation model of the GOM was
not available.
The compilation of layers described in this report provides a significant amount of
oceanographic data for the benthic and sea-surface environment of the GOM for historical and
contemporary time periods. As management of marine resources shifts towards an ecosystem
approach, the need for spatial data is expected to increase. The storage of the 31 data layers
presented here in a geodatabase with associated metadata should facilitate their use in future
work.
32
ACKNOWLEDGEMENTS
This work was supported in part by the Science sector of the Department of Fisheries and
Oceans through the Ecosystem Research Initiative. Peter Lawton and Roland Pitcher would also
like to acknowledge funding received from the Alfred P. Sloan Foundation, through the
Synthesis Program of the International Census of Marine Life, which facilitated workshops and
collaboration between authors.
We would like to acknowledge those who have aided in the creation of this technical
report and subsequent analyses, including the many scientists that have generously provided us
data: Charles Hannah and Brian Petrie for their help with oceanographic layers and the OAX
optimal estimation program; Huijie Xue for bottom stress data exports from the GOOMOS
Nowcast Forecast System; Don Clark for providing advice on the proper use of the DFO
Ecosystem Surveys data; John Tremblay for his technical report and candid advice on using
invertebrate data from the DFO Ecosystem Surveys; Richard Karsten and David Wildish for
their information on high current stress areas in the GOM; David Mountain for bottom
temperature data; Scott Ryan for advice on DFO’s nutrient data.
We would also like to thank the two internal DFO reviewers (Glen Harrison and Jerry
Black) the library reviewer (Charlotte McAdam) and Fred Page for their thoughtful and
constructive comments.
REFERENCES
Azarovitz, T.R. 1981. A brief historical review of the Woods Hole laboratory trawl survey time
series. In Bottom trawl surveys/releves au chalut de fond. Edited by W.G. Doubleday and D.
Rivard. pp. 62-67.
Breiman, L. 2001. Random forests. Mach. Learn. 45(1): 5-32.
Chadwick, E.P., Broodie, W., Colbourne, E., Clark, D., Gascon, D., and Hurlbut, T. 2007.
History of annual multi-species trawl surveys on the Atlantic coast of Canada. AZMP
Bulletin(6): 68.
Clark, D. 2010. Maritimes region ecosystem surveys. Edited by M. Greenlaw, St. Andrews.
Clark, D., and Branton, B. 2007. DFO Maritimes research vessel trawl surveys
http://www.iobis.org/OBISWEB/DisplayMetaData.jsp?content=maps_and_metadata/237.html.
Condie, S.A., and Webster, I.T. 1997. The influence of wind stress, temperature and humidity
gradients on evaporation from reservoirs. Water Resour. Res. 33(12): 2812-2822.
Day, J., and Roff, J.C. 2000. Planning for representative marine protected areas: A framework
for Canada's oceans. World Wildlife Fund, Toronto, ON. 156 p.
Despres-Patanjo, L.I., Azarovitz, T.R., and Byrne, C.J. 1988. Twenty-five years of fish surveys
in the northwest Atlantic: The NMFS northeast fisheries science center bottom trawl survey
program. Mar. Fish. Rev. 50(4): 69-71.
33
Drozdowski, A., and Hannah, C.G. 2010. Bottom stress maps of the eastern North American
continental shelf: Contributions from tidal currents combined with large wave events. Can. Tech.
Rep. Hydrogr. Ocean Sci. No. xxxx. In Press.
Ellis, N., Smith, S.J., and Pitcher, C.R. 2010 In Prep. Gradient forests: Calculating importance
gradients on physical predictors.
Fisher, W.L., and Rahel, F.J. 2004. Geographic information systems in fisheries American
Fisheries Society, Bethesda, Maryland. 275 p.
Fisheries and Oceans Canada. 2006. Biochem: Database of biological and chemical
oceanographic data. Version 8 (2005). [Online]. Available from http://www.meds-sdmm/dfompo.gc.ca/biochem/Biocheme.htm.
Frank, K.T. 2004. The Scotian Shelf groundfish trawl survey. AZMP Bulletin 4: 23-25.
Gee, K. 2007. Marine spatial planning: A theoretical overview [Electronic Version]. Accessed
March 24, 2010, Available from
http://www.plancoast.eu/files/berlinConference/1stDay/PlanCoast_MSPKiraGee.pdf
Geological Survey of Canada. 2009. Expedition database [Online Database]. Accessed August
2009, Available from http://gdr.nrcan.gc.ca/ed/index_e.php.
Greene, J.K., Anderson, M.G., Odell, J., and Steinberg, N. 2010. The northwest Atlantic marine
ecoregional assessment: Species, habitats and ecosystems. Phase one. The Nature Conservancy,
Eastern U.S. Division, Boston, MA.
Gregory, D. 2004. Climate: A database of temperature and salinity observations for the
northwest Atlantic. Can. Sci. Advis. Sec. Res. Doc. 2004/075.
Gregory, D., and Narayanan, S. 2003. Biochem: A national archive for marine biology and
chemistry data. AZMP Bulletin 3: 11-17 p.
Han, G.Q., and Loder, J.W. 2003. Three-dimensional seasonal-mean circulation and
hydrography on the eastern Scotian Shelf. J. Geophys. Res.-Oceans 108(3136): 21.
Hannah, C.G., Shore, J.A., Loder, J.W., and Naimie, C.E. 2001. Seasonal circulation on the
western and central Scotian Shelf. J. Phys. Oceanogr. 31(2): 591-615.
He, I., Hendry, R., and Boudreau, G. 2003. OAX demonstration and test case. Available from
http://www2.mar.dfo-mpo.gc.ca/science/ocean/coastal_hydrodynamics/Oax/oax.html2008].
Helbig, J., and Higdon, J. 2009. A preliminary investigation of two potential stability data
products for AZMP. AZMP Bulletin 8: 40-44.
Incze, L.S., Lawton, P., Ellis, S.L., and Wolff, N.H. 2010. Biodiversity knowledge and its
application in the Gulf of Maine area. In Life in the world's oceans: Diversity, distribution and
abundance. Edited by A.D. McIntyre. Oxford: Blackwell Publishing Ltd. In Press. pp. 43-63.
34
Kostylev, V.E. 2004. Habitat management template for Scotian Shelf habitat mapping. Report to
the Maritimes Regional Advisory Process.
Li, J., and Heap, A. 2008. A review of spatial interpolation methods for environmental scientists.
Geoscience Australia. 23: 137 p.
Li, M.Z., and Amos, C.L. 2001. SEDTRANS96: The upgraded and better calibrated sedimenttransport model for continental shelves. Comput. Geosci. 27(6): 619-645.
Mountain, D.G., Taylor, M.H., and Bascunan, C. 2004. Revised procedures for calculating
regional average water properties for northeast fisheries science cruises. Northeast Fisheries
Science Center Reference Document 04-08.
NASA. 2007. Oceancolor forum [Online Database]. Accessed August 2009, Available from
http://oceancolor.gsfc.nasa.gov/.
NOAA's National Marine Fisheries Service Northeast Fisheries Science Center. 2005. Northeast
fisheries science center bottom trawl survey data [Online Database]. Accessed August 2009,
Available from
http://www.iobis.org/OBISWEB/DisplayMetaData.jsp?content=maps_and_metadata/1435.html.
Pitcher, R., Lawton, P., Ellis, N., Smith, S.J., Incze, L.S., Wei, C.-L., Greenlaw, M.E., Wolff,
N.H., Sameoto, J.A., and Snelgrove, P.V.R. In Prep. The role of physical environmental
variables in shaping patterns of biodiversity composition in seabed assemblages.: 20.
Poppe, L.J., Williams, S.J., and Paskevich, V.F. 2005. USGS east-coast sediment analysis:
Procedures, database, and GIS data. USGS Open-File Report 2005-1001.
Rebuck, N.D., Townsend, D.W., and Thomas, M.A. 2009. Gulf of Maine region nutrient and
hydrographic database [Online Database]. Accessed October 2009, Available from
http://grampus.umeoce.maine.edu/nutrients/#Data.
Roff, J.C., Taylor, M.E., and Laughren, J. 2003. Geophysical approaches to the classification,
delineation and monitoring of marine habitats and their communities. Aquat. Conserv.: Mar.
Freshwat. Ecosyst. 13(1): 77-90.
Roworth, E., and Signell, R.P. 1998. Construction of a digital bathymetry for the Gulf of Maine
[Online Publication]. Accessed August 2009, Available from http://pubs.usgs.gov/of/1998/of98801/.
Shackell, N., and Frank, K.T. 2003. Marine fish diversity on the Scotian Shelf, Canada. Aquat.
Conserv.: Mar. Freshwat. Ecosyst. 13: 305-321.
Swail, V.R., and Cox, A.T. 2000. On the use of NCEP-NCAR reanalysis surface marine wind
fields for a long-term north Atlantic wave hindcast. J. Atmos. Ocean. Technol. 17(4): 532-545.
35
Theroux, R.B., and Wigley, R.L. 1998. Quantitative composition and distribution of the
macrobenthic invertebrate fauna of the continental shelf ecosystem of the northeastern United
States. NOAA Technical Report NMFS 140: 240 p.
Tremblay, M.J., Black, G.A.P., and Branton, R.M. 2007. The distribution of common decapod
crustaceans and other invertebrates recorded in annual ecosystem surveys of the Scotian Shelf
1999 - 2006, Dartmouth. Can. Tech. Rep. Fish. Aquat. Sci. 2762: 78 p.
Wildish, D.J., Peer, D.L., and Greenberg, D.A. 1986. Benthic macrofaunal production in the Bay
of Fundy and the possible effects of a tidal power barrage at economy point - Cape Tenny. Can.
J. Fish. Aquat. Sci. 43(12): 2410-2417.
Wood, L.J., and Dragicevic, S. 2007. GIS-based multicriteria evaluation and fuzzy sets to
identify priority sites for marine protection. Biodiversity Conserv. 16(9): 2539-2558.
Xue, H., Shi, L., Cousins, S., and Pettigrew, N.R. 2005. The GOMOOS Nowcast Forecast
System,. Cont. Shelf. Res. 25(17): 2122-2146.
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