home

Great Lakes Regional Aquatic GAP

The Great Lakes states began a regional Aquatic GAP project in 2001 to be completed in 2007 in three states.  The Ohio Aquatic GAP pilot project has been in progress since early 2000 (see separate status report for Ohio Aquatic GAP in this section).  The objectives of the regional project are to develop a riverine aquatic gap analysis for all eight states in the Great Lakes Region.  Projects are planned sequentially with new projects starting up when existing projects are nearing completion.  Active partners in the new projects are the Michigan Department of Natural Resources (MDNR), Wisconsin Department of Natural Resources (WDNR), New York State Department of Environmental Conservation (NY DEC), and U.S. Fish and Wildlife Service (USFWS)-Region 5.

Stream classification:  In mid-2002, new statewide projects started in Michigan, New York, and Wisconsin.  The statewide projects are adopting many of the protocols from the Aquatic GAP pilot studies in Missouri, Ohio, and South Dakota.  These methods include classifying streams using the Valley-Segment Type (VST) classification based upon channel, riparian zone, total catchment area, and other hydrogeomorphic features.

Before classifying streams, thematic data layers must be acquired and processed.  In 2002, map layers including surficial geology, elevation (30 m National Hydrography Dataset [NHD]), and hydrography (1:100,000) were obtained for Michigan, New York, and Wisconsin.  In Wisconsin, various automated machine language (AML) programs were acquired and tested in 2002 for processing data to determine stream order, sinuosity, gradient, and other geomorphic features across the basin.  Corrections were made to the NHD in Michigan and Wisconsin to address flow direction coding errors, disconnected reaches, and primary/secondary flow codes.

The USGS Office of Ground Water reviewed the Darcy model, a groundwater-flow model used to help predict stream temperature for the VST classification.  Modifications and improvements were recommended.  The revised Darcy model will be used in 2003 in Michigan, New York, and Wisconsin to predict the relative importance of groundwater in streams and categorize streams as being cold, cool, or warm water.

Animal modeling: There are over 300 species of fish in the Great Lakes (GL) Basin as well as many species of freshwater mussels, crayfish, and aquatic insects.  An Oracle^TM database (Central Database) is in development and is planned to serve available aquatic species occurrence and abundance data for the regional project (which covers riverine ecosystems but not the actual Great Lakes).  A prototype for serving data was developed in 2002 and is running successfully.  The Integrated Taxonomic Information System (ITIS) codification and naming system for fish species is being used for standardization across the region.  Currently, over 150,000 sampling sites in four states are being quality-assured before being entered into the Central Database.

In Michigan, the project is being coordinated with ongoing work at MDNR’s Institute for Fisheries Research.  Fish-sampling data have been acquired, including 79,961 records at 8,620 sites for presence/absence of species.  Fish abundance data are available for an additional 2,000 sites.  These data have been loaded into the Central Database, and a significant number (1,100) of additional records were added by hand.  Sources of habitat affinity data to be entered into a database for GL Aquatic GAP have been identified.

The New York project has acquired a very extensive database of fish occurrence and distribution for the entire state from the NY DEC.  The database includes more than 15,000 georeferenced samples (each an assemblage at a particular site) from 1988 through the present, most of which are verified by experts.  The historic database (1900-87) consists of more than 100,000 samples and includes extensive data from the Biological Surveys of the 1920s and ‘30s, conducted by watershed throughout the state.  Those data are also georeferenced but must be processed through the quality-control program.  Additional acquisitions from the DECs extensive aquatic invertebrate and water-quality databases dating back to the mid-1970s are planned.

In Wisconsin, the WDNR biology database, which includes fish-species occurrence data, is being developed as part of another ongoing project with contributions from the Gap Analysis project.  Updated locations for fish-species data were obtained from WDNR and loaded into the WDNR Biology Database in 2002.  This Oracle^TM database includes over 16,000 different site visits where fish records have been collected.  The database includes data for approximately 130 fish species that were collected as far back as 1880, with over 82% of the samples collected between 1970 and 2002.  Additional information from WDNR was obtained to improve the location information for more than 18,000 site visits for fish-species sampling from 1945 to 1995.

Coastal GAP Pilot Project: A pilot project to develop a coastal gap analysis for the Great Lakes also began in 2002.  Two areas are being proposed for pilot work in 2003-04 in western Lake Erie and eastern Lake Ontario.  The initial development process includes acquiring and reviewing the local and regional data availability for habitat within the coastal zone.  Assessment of data quality and extent indicates that available databases are sufficient to develop methods for successful completion of the GL Coastal Gap Analysis Pilot Project.  The near-shore region of large water bodies like the Great Lakes can be difficult to sample, particularly in high-energy areas.  The limited data collected from those areas are being gathered and assessed in Year One.  We are also making progress on development of an effective habitat classification system.  Research components including examination of methods to characterize and model coastal habitats and their relationships to the fish and other inhabitants began in Year One and will continue in subsequent years.  Environmental databases containing information on Great Lakes coastlines, bathymetry, coastal geology and geomorphic units, and some coastal aquatic substrata have been collected.  Data and information about circulation systems, exposure, and other habitat features are being acquired with the assistance and cooperation of many agencies and individuals.  Data ownership and distribution issues must be resolved to complete database acquisition and application.  Through meetings and conversations with biologists at the NY DEC, the project has acquired much of the available fish occurrence data for nearshore areas of Lake Ontario’s eastern basin.

Reporting and data distribution: A Fact Sheet was started in 2002 and is planned for completion in the first half of 2003 and for publication later in the year.  A Web site for the project was established at http://www.glsc.usgs.gov/GLGAP.htm with links to the home pages for the Ohio and Wisconsin Aquatic GAP projects and to the National GAP Web page.

Outreach and meetings: In October 2002, the regional team attended an Aquatic GAP training meeting in Missouri, hosted at the USGS Columbia Environmental Research Center by the Missouri Resource Assessment Partnership.  In November 2002, the Wisconsin Aquatic GAP project was presented and discussed with stakeholders and cooperators at the WDNR Fish and Habitat Annual Section Meeting.  In December 2002, the Wisconsin project personnel attended the Midwest Fisheries meeting in Iowa and provided an overview of the Aquatic Gap Analysis project.  Stakeholder meetings are planned in 2003 in all states with active projects.  The Coastal Pilot study team also participated in a USFWS workshop in December 2002.

A daylong session entitled “Biodiversity Conservation in the Great Lakes Region” is being planned for the annual meeting of the International Association for Great Lakes Research to be held in Chicago from June 22-26, 2003.  Planned presentations will include an overview of the National Gap Analysis Program by Mike Jennings, presentations from the New York and Upper Midwest terrestrial GAP projects, and five presentations from the Great Lakes Regional Aquatic GAP describing projects in Wisconsin, Michigan, Ohio, and the coastal pilot study.  Other invited abstracts came from The Nature Conservancy, the U.S. Environmental Protection Agency, the Ontario Ministry of Natural Resources, Ohio Sea Grant, and The Nature Conservancy Canada.

Hawaii Aquatic GAP

HI-GAP has initiated an Aquatic Gap Analysis project and is working with local, state, and federal agencies to complete a statewide aquatic species distribution data set.  Our approach to modeling vertebrate and macroinvertebrate distributions is based on a multivariate analysis of geomorphology and environmental variables.  A meeting with representatives from local freshwater agencies was held in March 2002, where strong support for the project and proposed mapping methodology was expressed.  The dramatic topography of Hawaii required a revision of standard methods for capturing a stream’s physical attributes so that an increased level of detail could be captured.  These adjustments enabled us to capture changes in habitats throughout the stream network.  In order to achieve our modeling goals and objectives, several new tools are in the process of being developed for the project.

The first geomorphologic attribute the advisory group identified as a critical component to species modeling was the classification of waterfalls.  Most aquatic biologists believe waterfalls are one of the major physical attributes defining a species’ range in the stream continuum.  To identify waterfalls, the Hawaii Gap Analysis Project has combined methods created by The Nature Conservancy’s (TNC) Freshwater Initiative and new methods developed at the Hawaii Natural Heritage Program.  In combination, these methods or tools identify the location of waterfalls and rate each in comparison with other waterfalls in a single watershed and amongst other waterfalls on a single island.  Waterfalls are identified based on user-defined parameters.  Height information is derived from the USGS 10-meter Digital Elevation Model (DEM).  The results are then used to identify the maximum, minimum, and average height of each waterfall within each watershed and for each individual island.  This information is added to a geodatabase containing all physical attributes for each island.

The second major morphological variable identified as significant for modeling species distribution was change in slope.  The State of Hawaii has a unique topography with dramatic elevation changes over short distances, making slope indices implemented in other states inapplicable.  For example, in Hawaii, a single stream can go from the headwaters at 3,000 feet to the mouth of the stream at sea level in less than four miles.  After rigorous experimentation with several approaches to slope modeling, the advisory committee chose TNC’s Freshwater Initiative slope tool as most applicable to Hawaii’s needs.  The tool was then used to identify slope changes in the stream continuum based on parameters defined by a group of aquatic biologists.  So far, the slope tool has been applied to half of the islands in the Hawaiian Island chain.  When combined, the slope and waterfall identification tools have successfully defined the analysis units for this project.  The geomorphologic information contained in the analysis units along with the habitat affinity database will be used to produce a species distribution mapping model.

All data collected for this project is being stored in a customized geodatabase designed specifically for the Hawaii Aquatic Gap Analysis Project.  The geodatabase contains all of the physical attributes derived for each analysis unit as well as all habitat affinity data provided by the State of Hawaii Division of Aquatic Resources.  The common data structure and spatial aspect of the geodatabase will allow us to use an iterative approach to the modeling effort, giving us the chance to fine-tune the model.

Over the next year the Hawaii Aquatic Gap Analysis Project will continue to collect all necessary physical attributes for the main Hawaiian Islands.  In the upcoming months we will be experimenting with species distribution modeling on Kauai.  Extensive field surveys will be conducted following the initial species distribution modeling results to determine the accuracy of the model.  Based on our results, changes will be made to the structure of the model to increase the accuracy of predicting species distributions.

Lower Missouri River Basin Aquatic GAP

a. Iowa

The statewide coverage of reaches, including an area in NW Iowa where additional stream segments were added to match the density of surrounding quad sheets, has been completed.  Unique segment IDs were added to each reach, based on a combination of the reach code and internal ID.  An ArcInfo AML from MoRAP was run on the coverage to append some separate NHD table information to the linework.  Each of the 57 hydrologic unit codes (HUCs) will have to be subset from this main coverage to be further processed to code primary or secondary (braid or loop) channels.  Disconnected stream segments will also be attached if they are judged to be incorrect.  Topographic maps will be used to analyze the disconnected segments.  Currently, 5 HUCs (watersheds) have been subset, but the reach processing has not begun.  The stream reaches (NHD) for four watersheds were made available through IRIS (http://madagascar.gis.iastate.edu/iris).

Much of the existing fish sampling data have been obtained from numerous state and federal agencies and academic institutions as well as from published literature.  The data have been entered into a relational database designed by the Missouri Aquatic GAP Project but customized for Iowa Aquatic GAP.  The fish database currently contains 4,160 community fish samples dating from 1926-2002, with a total of 40,196 species occurrence records.

Future plans: We will continue to subset the HUCs and begin to process the reaches within each HUC.  As a HUC is completed, it will be sent to MoRAP for further processing.  They will generate values for 10 variables, and the unique combination of those variables will create the valley-segment type variable.  Those 11 variables will be attached to the reaches, and the HUC will be sent back to us.  The reach information will be used, along with the biological sampling data, to generate predictive models for fish species.

Upon completion of the valley-segment characterization by MoRAP, work will continue on attributing the NHD reaches with stewardship information and physical and biological characteristics.  Completion of this phase and the completion of species habitat descriptions will allow species prediction to commence.  As data layers and species habitat descriptions are developed, they could be made accessible through IRIS.

The remaining sampling data will be collected and entered into the fish database.  Stream reach locations will be determined for each sample collected.  These data will then be used to generate statewide distribution maps for each species on a watershed-by-watershed basis, using 10-digit hydrologic units for widely distributed species and 12-digit hydrologic units for narrowly distributed species.  Once maps for all fish species are completed, they will be sent out for professional review.

Development and Use of the Iowa Rivers Information System (IRIS)

The database created within ArcView 3.2 containing variables describing certain physical features of stream reaches in Iowa is complete, with very few exceptions.  The database is represented as shapefiles of streams for each of the 57 HUC 8-level watersheds for the state.  During 2002 we added four new variables to the previous list: gradient, public land, tier/range/section (T/R/S), and 24K topographic quad name.  The public land information indicates whether or not the reach flows through public land designated by the Iowa GAP stewardship data.  T/R/S and 24K quad name information was obtained from Iowa DNR NRGIS coverages.  Gradient was calculated within ArcView using an extension from the ESRI ArcScripts page and a digital elevation model grid.  We also added a new table to the collection, similar to the percentage of GAP land cover within 90 meters of a particular reach segment.  It shows land cover percentage using the National Land Cover Dataset (NLCD) available from USGS; we added this information because we have reaches that fall outside the Iowa border.  The exceptions mentioned previously include two watersheds for which gradient has not yet been calculated, one watershed for which GAP land cover percentage has not been calculated, and ten watersheds for which stream order has not been completed due to missing reaches.  The IRIS ArcIMS data protocol has changed, so that reach information is no longer provided through an Access database table.  The reach information for all watersheds is supplied directly as shapefiles.

We are continuing work on the Web interface for IRIS (http://madagascar.gis.iastate.edu/iris) using ESRI's ArcIMS technology.  Currently users are able to view, query, and interact with IRIS data through a limited set of traditional GIS tools.  Tools include the ability to zoom in and out, find specific reaches, and classify reaches according to IRIS attributes.  Additional data layers have been added as they become available, and links to metadata or information about the different data layers now exist.  An additional layer allowing users to view USGS real-time stream gauge information has been added.  These data are currently being stored in an ArcSDE data layer.  Scripts were written to extract real-time gauge data from the USGS Web site every three hours, then update the layer in the database.

Future plans: We are investigating providing the reach data through an Arc Spatial Database Engine interface directly to the IMS page.  When GAP provides land cover data for the states surrounding Iowa, we will add that information to our database.  The gradient and land cover variables missing for certain watersheds will be calculated.  Stream ordering for the 10 watersheds will be done as time permits, but will have to be done by hand.  Assistance will be given to the improvement of the IMS interface process and the point creation for biological sample sites as requested.

A variety of display and reporting features is planned.  Mapping tools will be developed to select adjoining reaches up- or downstream from a point selected by the user and to download data for use in a GIS.  Advanced reporting options will also be developed.  A user will be able to perform a spatial or attribute query, generate a map of the selected features, and output a report that includes the map and attributes of the selected records.

b. Kansas

The Kansas Aquatic GAP Project has made substantial progress in the past year and a half.  This project is part of the regional Aquatic GAP effort in the lower Missouri River basin; however, we are including a portion of the Arkansas River basin because these drainages account for approximately half of the aquatic systems in Kansas.  Draining of aquifers and major land use changes have been well documented, even before the turn of the century (Mead 1986).  Because of these changes and the resulting loss of biodiversity in Kansas, we have received enthusiastic support from numerous cooperators in the state.

Two major steps are completed or near completion.  The first step of formatting the stream network data layers has been completed.  This base layer identifies stream valley segments, which are specific reaches delineated by stream confluences.  This data layer will be used for the finest scale of species modeling.  Numerous habitat descriptors have been attached to these valley segment habitat units, including stream size, gradient, location in the watershed, and proximity to other waterbodies.  The second step of compiling biological data for both fishes and mussels is near completion, and most records are stored in a relational database along with the habitat information.  To date we have over 2,000 collection sites in Kansas.  These data have been compiled, and species distribution maps have been constructed and are available on our Web page (www.ksu.edu/aquaticgap).  Currently, we are adding additional data from a variety of sources (e.g., museums, field notes of local biologist, etc.).  Because much of our data included instream habitat measurements, we are also in the process of calibrating our GIS layers by comparing them with on-the-ground measurements.

Mead, J.R.  1986.  A dying river.  Transactions of the Kansas Academy of Sciences 14:111-112.

c. Missouri

Aquatic ecological classification: An 8-level Aquatic Ecological Classification Hierarchy was developed in cooperation with The Nature Conservancy’s Freshwater Initiative Program.  Statistical methods for using biophysical data to classify aquatic ecological units were developed for levels 4-7 in the hierarchy, and those levels were subsequently mapped with a GIS.  Methods for preprocessing the 1:100,000 NHD were also developed to make it suitable for further classifications procedures.

The following products were created: ArcView shapefiles and ARC INFO coverages for Aquatic Subregions of Missouri, Ecological Drainage Units (EDU) of Missouri, Aquatic Ecological Systems (AES) of Missouri, and Valley Segment Types of Missouri

Written descriptions of the preprocessing and classification procedures for each of these levels in the hierarchy

A suite of Arc Macro Language scripts to automate many steps in the Valley Segment classification process

Predictive distribution modeling: We developed methods of integrating our species occurrence records and attributes from our Valley Segment Coverage into a single database and then performing Classification and Regression Tree analyses to develop predictive models for each species.  We also developed statistical methods for identifying undersampled watersheds and for more objectively correcting the geographic range of species.  These methods appear superior to the more subjective professional review process, provided enough collection data are actually available to help drive the revision process.

The following products were generated: A relational database of existing collection records for fish, mussels, crayfish, and snails within Missouri, containing nearly 8,000 records dating from 1900 to 1999 and including state, federal, and global rankings of all species.  Each record is geographically linked to the 1:100,000 National Hydrography Dataset, allowing users to query and display within a GIS the specific stream reaches in which an individual species has been collected or view all species collected within a single stream reach.  Each record is also geographically linked to the Missouri 1:24,000 12-digit Hydrologic Unit (HU) coverage.  This allows users to query and display within a GIS the geographic range of each species throughout Missouri by 12-, 10-, or 8-digit HU, based on actual sampling data.  It also allows users to query and display all species that have actually been collected within a single HU.

A species occurrence database by 8- (for mussels) or 10-digit HU, which incorporates revisions made by taxonomic experts to the geographic range maps that were produced using only actual occurrence records.  This database is also geographically linked to the Missouri 8- and 10-digit HU coverages and thus allows users to query and display within a GIS the professionally-reviewed geographic range of each species or all species occurring within a given unit based on both actual data and professional judgment.

Endemism Database which categorizes each species (except for snails) according to levels of endemism corresponding to Aquatic Ecological Unit Classification.  These categories reflect how restricted the overall geographic range of a species is and also allows us to identify which species are most distinctive within a given Ecological Unit.

General habitat-affinity descriptions extracted from existing literature for all fish, mussels, and crayfish with associated citations.

Region-specific predictive distribution models for all fish, mussels, and crayfish.  Models were constructed primarily through the use of Classification and Regression Tree analyses that analyzed the occurrence records of each species with respect to attributes attached to our Valley Segment Coverage.  For species with limited occurrence records we had to rely on more subjective model development procedures using the habitat-affinity information extracted from the literature or through the use of contingency tables for individual predictor variables, which were then qualitatively examined to identify species-habitat associations.

1:100,000 statewide predictive distribution maps for all fish, mussel, and crayfish species.  These maps show, within the geographic range of each species in Missouri, all of the individual NHD stream reaches in which a species would likely be found under natural conditions.  Unlike “terrestrial GAP projects” we are unable to predict present-day distributions because of our inability to accurately account for how the numerous and interactive effects of human-induced alterations specifically affect the distribution of riverine biota.

A 1:100,000 statewide hyperdistribution coverage, allowing users to query and display within a GIS the predicted distribution of any fish, mussel, or crayfish species throughout Missouri.  It also allows users to select individual reaches to see all of the fish, mussel, and crayfish species predicted to occur in that reach.  Users can generate and display statistics pertaining to richness, endemism, and species of conservation concern across the state or for any region or watershed of interest.

A suite of SAS programs to integrate and reorganize the species occurrence data and the attributes in the Valley Segment coverage required to generate species-specific databases in a format suitable for Classification and Regression Tree analysis.

Identifying conservation gaps and targets:

In addition to the traditional gap analysis process we have developed a method to identify conservation gaps and prioritize conservation opportunities at multiple spatial scales (i.e., Ecological Drainage Unit, Aquatic Ecological System, and Valley Segment Type) by assessing the biophysical distinctiveness and conservation status of our ecologically-defined units at multiple spatial scales.

The following products were generated: Conservation ranks for EDUs based on professionally reviewed biological data (based on richness, endemism, G1-G3 species statistics).  These ranks indicate relative importance of each EDU within each Aquatic Subregion with regards to conserving aquatic biodiversity in Missouri.

We are waiting for the final hyperdistribution database so we can attribute our AESs with the appropriate biological data and then conduct a similar assessment that will also incorporate a stewardship assessment.  This assessment will show conservation gaps and also the relative conservation status of AES types within each EDU.

We calculated over 60 land cover and land use statistics for each individual AES polygon across the state in an effort to condense this list into a meaningful set of statistics that could distinguish the relative environmental quality of each unit.  Because of the high degree of correlation among most of these variables we were able to condense this list to just 8 relative uncorrelated variables.  These include %Forest, %Wetland, %Urban, Population Change, Density of Mines, Density of Point Source Discharges, Density of Confined Animal Feeding Operations, and the Degree of Fragmentation caused by impoundments.  These will be used as the core set of variables for our conservation status/threats assessment.  An additional variable used in this analysis will be the number of exotic species.

The final assessment will reveal conservation gaps of the dominant VSTs for each AES type within each EDU.  This assessment is analogous to assessing the relative stewardship of vegetation classes by Landtype Associations within each Ecological Subsection (as defined by Bailey’s Ecological Classification System).

d. Nebraska

The Nebraska Aquatic Gap Analysis Project commenced in August 2002 with a training session with the Missouri Aquatic GAP Project team at MoRAP in Columbia, Missouri.  All NHD basin coverages relevant to Nebraska have been acquired.  An initial test basin has been edited and submitted to MoRAP for review and approval to commence further processing.  The braided course of the Platte River poses a challenge for consistent but comprehensive processing of basins and confluences.  We are studying alternate strategies for representing the Platte within the MoRAP protocol.  We have started to identify sources of specimen records.

Ohio Aquatic GAP

Stream classification: The Ohio Aquatic Gap Analysis Project is using many of the protocols used for the pilot study in Missouri (MoRAP) to classify streams in the state.  These methods include classifying streams using the Valley-Segment Type Classification (VST).  Ohio Aquatic GAP completed VST classification in 2001.  Fish species were linked to occurrence in specific VSTs for data analysis and animal modeling in 2002.  Wetlands are being classified separately based on hydrology and vegetation.  Inland lakes and the Great Lakes are not included in the project.

Aquatic animal modeling: Sample point maps of 150 fish species were completed and released on the Ohio-GAP Web site for expert review in July 2001.  These data include maps of native and introduced fish species that reproduce in Ohio streams.  Final corrections based on expert reviews were completed for all 150 fish species in 2002.  Fish distribution points associated with specific VSTs were used to model potential species distributions for 150 fish species using the Genetic Algorithm for Rule-set Production (GARP) (Stockwell and Peterson 1999) desk-top version software (Scachetti-Pereira 2002).  In 2003, fish-species models will be combined to produce a map of Ohio with a probability-like distribution showing fish-species diversity patterns.  In the first half of 2003, analyses to take into account some factors such as land use and dams and how these factors limit the accuracy of predictions of fish-species distributions are planned.  Gap analysis of Ohio fish species is planned for completion by the end of September 2003.

Completion of 80% of the crayfish database was a priority in 2002.  The crayfish database at The Ohio State University Museum of Biodiversity contains a total of 4,251 records (sites), 80% of which are from Ohio.  Additional work needs to be done in the first half of 2003 at the Cleveland Museum of Natural History.  When completed there will be about 5,000 records from both museums in the database.  Distributions of 88 species of freshwater mussels were mapped statewide in 2002.  Expert review will be completed in 2003, and modeling is planned for 2004.

Development of a database of fish and amphibian distribution in wetlands was started in 2002 and will continue in 2003.  Known distributions of 16 fish species and 13 amphibian species were mapped in 2002.  Modeling the potential distributions of fish, amphibians, reptiles, and birds in wetlands is planned for completion in 2004.

Analysis: Fish-species distribution models were developed using GARP modeling software.  GARP implements four rule-types to build species prediction models: atomic, logistic regression, bioclimatic envelope, and negated bioclimatic envelope.  GARP was used to generate 1,000 models of potential distributions for each fish species.  A different set of presence points was used to build and test each model, thus providing good cross-validation of the models.  Twenty of the “best” models for each species were chosen, selecting models that minimize omission and commission errors.  Omission errors ranged from 0 to 22%, and commission errors ranged from 1 to 66% in selected models of 150 fish species.  Error rates for narrowly distributed fish species typically were substantially lower than those for broadly distributed ones.

Reporting and data distribution: In spring 2003, fish distribution maps and valley segment attributes are planned for publication and release on the Ohio-GAP Web site (http://oh.water.usgs.gov/ohgap/ohgap.html) as well as on CD-ROM.  A manuscript discussing the Aquatic GAP project will be prepared in the summer of 2003 and published in 2004. 

Literature cited:

Scachetti-Pereira, R.  2002.  DesktopGarp.  Accessed December 16, 2002, at URL http://beta.lifemapper.org/desktopgarp/.

Stockwell, D., and D. Peterson.  1999.  The GARP modeling system: Problems and solutions to automated spatial prediction.  International Journal of Geographical Information Science 13:143-158.

Southeast Aquatic GAP

a. Alabama

We are currently spatially referencing all historical and recent collection data for the Tallapoosa Basin.  Efforts to categorize land use/land cover are under way; a Level I Anderson classification has been competed for the basin.  We are in the process of delineating watersheds above sampling sites and compiling landscape-level data for each.  We will develop faunal models using distribution data and landscape metrics.  The first models will be developed for six fishes that are considered “at-risk” by the U.S. Fish and Wildlife Service.  Models will be used to make decisions relative to status of the “at-risk” species.  Hierarchical models developed by the Georgia Aquatic GAP Project have already shown promise for identification of conservation strategies for aquatic species.         

b. Georgia

We have completed spatial referencing of all historical (pre-1995) and current fish, crayfish, and mussel sampling locations using records provided by the Georgia Department of Natural Resources (GADNR), Georgia Museum of Natural History, USGS, Auburn University, and the University of Georgia (UGA).  We are in the process of delineating watersheds for each sampling location.  We also have developed geomorphic channel classifications for Flint River Basin stream segments in cooperation with GADNR and UGA and are in the process of combining these with hydrologic and fish population response models.  These models will be used by GADNR to examine the effect of various river regulation and water use scenarios and develop streamflow management policies. 

Upper Missouri River Basin Aquatic GAP

Status:

We completed an aquatic gap analysis as part of the terrestrial GAP of South Dakota and have expanded the aquatic analysis to the Upper Missouri River Basin (UMRB).  States and provinces included in the study area are Alberta, Saskatchewan, Montana, Wyoming, North Dakota, South Dakota, Minnesota, and Iowa.  In coordination with all state, federal, and international agencies the following base layer data sets have been acquired: stream network (NHD or an equivalent for the provinces), geology layer, 80% of the stewardship layer, 80% of the land cover layer, a complete DEM (except for Alberta), hydrologic units (except North Dakota), and fish data for the UMRB.  We are attributing stream segments with ten physical habitat affinities (temperature, stream size, flow regime, channel gradient, size discrepancy, floodplain interaction, geology, elevation, stream connectivity, and groundwater input).  We have completed attributing physical habitat features to stream reaches with the above habitat affinities.  We are currently working on attributing groundwater input (80% complete), floodplain reach (90% complete), and flow regime for reaches in Canada.  The transfer of South Dakota Aquatic GAP data from RF3 to NHD data is about 95% complete.  We have produced a 30 m land cover map from Landsat 7 TM data for Canada.  We are working with agencies in North Dakota to complete the 10-digit hydrologic units for North Dakota.  Fish location data have been attributed to the stream reaches, except for Canada and the Missouri River in North Dakota. 

Analysis:

Gap analysis of aquatic diversity has been completed for South Dakota.  We modeled distributions of 116 fish species across South Dakota and assessed biodiversity of fish species and habitat in relation to land conservation.  We found that the fishes of South Dakota have more protection than the terrestrial animals.  A few areas proposed for the protection of mammals have the potential to provide additional protection for many aquatic species.  Our Web page located at http://wfs.sdstate.edu/sdgap/aquaticgap.htm has links to fish distributions by 10-digit hydrological units, fish habitat affinities, and fish-habitat models by stream reach for South Dakota.  Gap analysis of the UMRB valley segments should begin in April of 2003.

Future plans:

We plan to complete attributing physical habitat features and fish locations to individual valley segments by the end of March.  We will then conduct a quality check and begin fish-habitat modeling.  We are working with The Nature Conservancy to delineate ecoregional drainage units based upon ecoregions and major drainages to further classify fish distributions. 

Upper Tennessee River Basin Aquatic GAP

In 2001, researchers from the Department of Fisheries and Wildlife Sciences and the Conservation Management Institute of Virginia Tech began an aquatic gap analysis of the upper Tennessee River basin (UTRB), which is shared by Virginia, Tennessee, North Carolina, and Georgia.  Most of the efforts so far have been directed at assembling available GIS coverages on biota, land and water use, and physical landscape features.  These coverages are nearly complete, and we are beginning to develop models to predict species occurrence.  Sophistication and precision of models will vary with data availability.  For example, for poorly sampled species we may develop only qualitative models (low precision) from relational databases of natural-history information from the literature.  In contrast, for well sampled species we may develop more quantitative models based on logistic regression or discriminant analyses. 

A main research focus is to develop more powerful protocols to assess threats to aquatic biota.  We anticipate that merely knowing ownership of lands adjacent to aquatic habitats will not be adequate to assess protective status or level of threat.  Thus, we intend to develop an integrative protocol for assessing a wide array of threats to stream biota.  Threats vary in scope of origin (nonpoint vs. point source), frequency of occurrence (accidental spill vs. permitted effluent), and severity (heavy metal contamination vs. nutrient enrichment).  Moreover, most threats to aquatic biota emanate from outside the aquatic environment and traverse aquatic networks at varying rates.  Through the use of geographic information systems, site-specific data, and conceptual models, we will evaluate aquatic sites for overall levels of threat.  Data layers contributing to this assessment include point-source pollution, transportation corridors, historic spill locations, and areas of rapid development.  We are conducting a meta-analysis to evaluate the approaches currently used to assess threats to aquatic biodiversity.  This analysis will help us design a small set of trial protocols to apply to the UTRB.  Ultimately, we will develop a framework to organize and rank site-specific or watershed-specific threats to biota.  As is typical for gap analyses, this threat coverage will be integrated with the biotic coverages to identify priorities for conservation efforts.