A RETROSPECTIVE OF AR-GAP: WHAT WE LEARNED DURING OUR JOURNEY
Catanzaro, Don
Center for Advanced Spatial Technologies, 12 Ozark Hall, University of Arkansas, Fayetteville, Arkansas 72701
AR-GAP was started in 1992, while National GAP was in transition from a project to a full-blown program. As a research endeavor, most state GAP projects encountered many hurdles along a winding road to completion. As National GAP moved to a program, many of these same obstacles were overcome by solutions put forward by state projects. This paper will focus on historical aspects of AR-GAP, why the AR-GAP project was successful, solutions discovered by team members, and problems and solutions discovered by other GAP teams. The intent of this paper is to put current National GAP in context with its past in order to pass on these strengths to new GAP states. Specific examples such as aquiring TM images (MRLC and processing problems), SPECTRUM, vegetation hierarchy (NVCS and others) will be used to illustrate just how far National GAP has come from its inception. In conclusion, while many solutions have been put to good use, new (and old) problems still occur in GAP projects. Some of these problems will be revisited, and new solutions will be put forth.
TECHNICAL OVERVIEW OF AR-GAP CD-ROM
Catanzaro, Don
Center for Advanced Spatial Technologies, 12 Ozark Hall, University of Arkansas, Fayetteville, Arkansas 72701
The completed AR-GAP CD-ROMs will be displayed. The final version includes all data sets and report in PDF format. Over 3,000 maps are archived, in a variety of formats: 7.5 minute quad, 100,000 tile, and state-wide. Each format has a set of maps displaying false-color CIR, tassle cap transformed, and 100-ha MMU vegetation categories. Terrestrial vertebrate maps include breeding distribution, breeding habitat, and predicted distribution at a statewide scale. Interfacing of maps and reports was done via standard Adobe products for a seamless report allowing users to read and view data effortlessly.
USING GAP ANALYSIS TO DO PRELIMINARY RESERVE DESIGN FOR A NATIONAL WILDLIFE REFUGE
Clark, Forest S., B. Slusher1, and R. L. Pressey2
1U.S. Fish and Wildlife Service, Bloomington Field Office, 620 South Walker Street, Bloomington, IN 47403
2New South Wales National Parks and Wildlife Service, P.O. Box 1967, Hurtsville, New South Wales, Australia 2220
The Kankakee River watershed in northeastern Indiana and northwestern Illinois is a landscape that historically contained the approximately 1 million acre Grand Marsh, oak savanna, prairie, and a diverse riverine system. Only fragments of these primary ecosystems remain today. The U.S. Fish and Wildlife Service (FWS) began the National Environmental Policy Act (NEPA) process to establish the proposed 30,000 acre Grand Kankakee Marsh National Wildlife Refuge (GKMNWR) during the summer of 1997. Our refuge design strategy involves a landscape approach formulated to complement existing management efforts within the watershed. The Indiana and Illinois Gap Analysis projects provided the mechanism to adopt a data-driven approach to initial reserve design for the proposed GKMNWR. While FWS trust resources drove the analysis, the resource interests of partner organizations were also considered. GAP provided the resources to move beyond the expert workshop approach. Using simple on-screen analysis, we established focus areas that formed the basis for the Environmental Assessment and the Economic Impact Study required under NEPA. We have just begun Phase II of the reserve design analysis using the C-PLAN software in conjunction with the New South Wales National Parks and Wildlife Service.
MAPPING THE VEGETATION OF MICHIGAN CIRCA 1800: MAP DEVELOPMENT AND SELECTED APPLICATIONS
Comer, Patrick J.1, D.A. Albert, H.A. Wells, B.L. Hart, J.B. Raab, D.L. Price, D.M. Kashian, R.A. Corner, and D.W. Schuen (map interpretation); T.R. Leibfreid, M.B. Austin, C.J. DeLain, L. Prange-Gregory, L.J. Scrimger, K.M. Korroch, and J.G. Spitzley (digital map production)2
1The Nature Conservancy, Western Regional Office, Boulder, CO
2Michigan Natural Features Inventory, Lansing, MI
Basic to our understanding of nature is an appreciation for the relative composition and spatial pattern of vegetation across the land's surface. Knowledge of the composition and ecological context of Michigan's vegetation, as it appeared prior to rapid and widespread European-American settlement and industrialization in the 1800s, provides an important and useful point of reference. Ecologists of the Michigan Natural Features Inventory developed a methodology to translate data and descriptive text of the General Land Office surveys (conducted in Michigan 1816-1856) into a detailed digital map that can be used by researchers, land managers, and the general public. The methodology for producing the map involved plotting the section line information onto U.S. Geological Survey topographic maps (1:24,000 scale). Vegetation type boundaries were established using topography and dominant tree species as primary sources of information. Type boundaries between each section line were interpolated using elevation lines, surface geology maps, and other historic vegetation maps, where available. Eighty-four vegetation types were distinguished and mapped from the land survey data. Major disturbances (wildfire, windthrow, etc.) and cultural features, as noted by land surveyors, were also transferred to the digital map. The historical vegetation map is utilized for spatial analysis of historical pattern, assessment of long-term vegetation trends, and for efficiently targeting field inventory for natural features.
LIFE SCIENCE GAP ANALYSIS AND CORE NATURAL HERITAGE AREA SELECTION IN ONTARIO: AN ITERATIVE APPROACH
Crins,William J., Sheila Boyd, Fiona McKay, Peter Sorrill1, and Ian Finlay2
1Ontario Ministry of Natural Resources, 300 Water St., Peterborough, Ontario K9J 8M5
2Linnet Geomatics, 1600-444 St. Mary Ave., Winnipeg, Manitoba R3C 3T1
Ontario's system of natural heritage areas is a composite of sites identified through two disciplines, earth science and life science, each with its own set of targets and approaches. The life science component focuses on representation of the full suite of naturally occurring landform/vegetation associations in each ecodistrict of the province. Five selection criteria (representation, condition, diversity, ecological considerations, special features) are applied to potential sites to identify the 'best' candidates for protection. Ontario's current gap analysis methodology identifies potential core representative areas in a parsimonious manner, maximizing residual diversity of contiguous, unrepresented landform/vegetation features at each step in an iterative selection process. This is accomplished by 1) overlaying vegetation data on landform data (surficial and bedrock geology), 2) comparing landform/vegetation associations within protected areas to those found in the entire ecodistrict, 3) applying a filter to exclude anthropogenic disturbances, and 4) selecting potential core representative areas using tallies of unrepresented landform/vegetation features (gaps). Methodological assumptions and limitations relating to scale, temporal change in vegetation, available data sets, parsimony, and the cluster-building portion of the algorithm are discussed. A future enhancement and research direction relevant to life science gap analysis in Ontario would be the development of a classification system for aquatic ecosystems that could be incorporated into the site selection method. Additional refinements would include modeling of previous landscape structure and composition, testing assumptions about land use and resource management practices adjacent to protected areas, and assessing adequacy of representation on a landscape scale over time.
GAP DATA DISTRIBUTION: AN UPDATE
Crist, Patrick
National Gap Analysis Program, Moscow, ID
The National GAP office has begun development and distribution of a CD-ROM product standardized for all states. The NM-GAP product will be demonstrated, and the data and format requirements will be described.
A GAP BIOLOGICAL DECISION SUPPORT SYSTEM FOR COUNTY PLANNING
Crist, Patrick1, Tom Kohley2
1National Gap Analysis Program, Moscow, ID
2University of Wyoming, Spatial Data Visualization Center, Laramie, WY
Local governments have nearly exclusive jurisdiction over 80% of the coterminous U.S. but until recently have received little attention or support by state and federal land management agencies in the conservation of biodiversity. This pilot project, funded by the USGS Biological Resources Division, sought to provide a simple and inexpensive tool for county land use planners to consider biological resources in the review of land use proposals. While there is tremendous need for tools to aid longer-term comprehensive planning, professional planners expressed an urgent need for quick and inexpensive tools to "raise a red flag" of potential biodiversity conflicts in the their day-to-day operations. This Biological Decision Support System operates in ESRI's ArcView 3 environment with a customized interface that allows the operator to query individual parcels for their known or predicted biotic occurrences of plant communities and terrestrial vertebrates, the state, federal, TNC, and Gap Analysis conservation status of the elements, and the predicted level and type of conflicts with a proposed land use. The system is also envisioned for "front desk" access by land owners and developers to help them avoid conflicts and identify stewardship opportunities prior to pursuing development permits.
A PRIORITIZED METHOD TO IDENTIFY UNPROTECTED AT-RISK PLANT COMMUNITIES IN THE WESTERN U.S.
Wright, R. Gerald, J. Michael Scott1, and Jesse D'Elia2
1USGS, Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho, Moscow, ID
2Department of Fish and Wildlife, University of Idaho, Moscow, ID
Prioritizing the conservation of individual plant communities at regional scales requires knowledge of their location, distribution, range, patch size, land stewardship, and patterns of human population growth rates within their range. We present the results of the first large regional analysis of these factors over a 150,000 km2 area of the western U.S. We aggregated and mapped 107 plant communities and used landownership/land stewardship maps to classify the level of protection for all parcels. Plant communities ranged in size from 16 million to 1200 ha; 65% had <10% of their total area in lands managed to protect biodiversity. From these, we selected those types that had their entire distributional range within the study area, were located in a small number of patches <2,500 ha in size, and/or were located in county census tracts with >2% population growth rates. Combining these categories, we identified and mapped the 20 plant communities considered to be most at risk in the western U.S.
ACCURACY ASSESSMENT OF THE NEW YORK GAP LAND COVER AND ECOLOGICAL COMMUNITIES MAP
DeGloria, S.D., M. Laba, S.K. Gregory, J. Fiore, and E. Hill
New York Cooperative Fish and Wildlife Research Unit and Cornell Institute for Resource Information Systems, Cornell University, Ithaca, NY 14853
A state-wide map of land cover and ecological community types was generated as a component of the New York Gap Analysis Project. Major emphasis was placed on classifying and mapping ecological communities within the framework of the National Vegetation Classification System (NVCS). The mapping process integrated spectral models derived from Landsat Thematic Mapper (TM) multi-spectral data with expert knowledge of the type and distribution of ecological communities throughout the state. After each TM scene was classified, the scenes were edge-matched and mosaicked to produce a state-wide map at 30-meter resolution. The 30-meter resolution map was spatially aggregated to a four-hectare minimum mapping unit using GigaMerge. Qualitative assessments of map accuracy are being conducted through structured interviews of local experts to isolate and correct major classification and mapping errors prior to conducting a quantitative accuracy assessment. Quantitative map accuracy assessment is based on a stratified random sample of 112 photo plots distributed throughout the state and referenced to the origin of Breeding Bird Survey routes. Land cover polygons from the spectrally classified map are overlaid on a 1600-hectare circular plot within the effective area of each 1:40,000 scale, NAPP color-infrared aerial photo. Our field team and regional experts are assigning land cover types to each polygon. Paired field- and predicted observations of land cover are analyzed using standard and fuzzy accuracy assessment methods. The nature, magnitude, and frequency of errors associated with mapping land cover and ecological community types at landscape scale will be reported.
MISSOURI'S APPROACH TOWARD LAND USE/LAND COVER DATA DEVELOPMENT
Diamond, David D., Kan He, and Taisia Gordon
Missouri Resource Assessment Partnership (MoRAP), 4200 New Haven Road, Columbia, MO
Missouri is pursuing a two-phased approach to accomplish statewide satellite data land cover classification. Phase I, which was completed last November, is a statewide classification done from dual date, scene-by-scene, ISODATA unsupervised classification. Aerial photographs, expert interpretation, and limited field surveys were used to assign categories. Though several techniques such as critical spectral band selection and urban/non-urban spatial separation dramatically improved the quality of the Phase I unsupervised classification clusters, photo interpretation, and limited accuracy of TM data. Phase II will try to improve the accuracy of the Phase I data through intensive ground survey, subsetting by ecological subsection, and efficient GIS spatial data analysis. A seamless ground-verification data processing software has been written in-house to provide a more efficient and accurate way to manage and use the large volume of ground data collected. This involved use of MS Access for initial data capture and use of MS Data Access Object and ESRI's Open Development Environment to link Access tables with Arc/INFO map utilities.
ACCURACY ASSESSMENT OF VERTEBRATE DISTRIBUTION MODELS FOR THE GAP PROGRAM IN ARIZONA
Drost, Charles A., Sarah Jacobs, and Kathryn Thomas
Colorado Plateau Field Station, Box 5614, Northern Arizona University, Flagstaff, AZ 86011
We evaluated the accuracy of distribution models for mammals, birds, reptiles, and amphibians in Arizona, developed as part of the National Gap Analysis Program (GAP). Our evaluation was based on GAP-generated lists of vertebrates for parks and similar protected areas, compared to lists compiled from field inventories of the same areas. For all major taxonomic groups, accuracy of GAP distribution models showed a positive trend in relation to size of area compared. Accuracy rates were highest for mammals and lowest for amphibians. We also noted wide differences in accuracy among smaller taxonomic divisions, which appear to relate both to problems in modeling distribution and also to relative completeness of inventory.
MODELING VEGETATION STRUCTURE USING MODERN REGRESSION TOOLS: METHODS AND APPLICATIONS
Edwards, Thomas C.,Jr.1, Gretchen G. Moisen, Tracey S. Frescino2, and Joshua J. Lawler3
1USGS Biological Resources Division, Utah Cooperative Research Unit, Utah State University, Logan, UT 84322-5210
2USDA Forest Service, Rocky Mountain Research Station, Ogden, 507 25th Street, UT 84401
3Department of Fisheries and Wildlife, Utah State University, Logan, UT 84322-5210
Region-wide cover maps, like those produced by the Gap Analysis Program, typically model vegetation as a binary response. The type is, for example, aspen or something else. Rarely are structural aspects of the type modeled. Here we demonstrate the applicability of modern regression tools for modeling vegetation structure in region-wide cover maps, thereby enhancing their applicability to management. We begin with a description of several common statistical models and their strengths and weaknesses for modeling vegetation structure. Using one of these tools (generalized additive models) as a demonstration example, we modeled and field-validated two nominal (forest presence, lodgepole pine presence) and three continuous variables (snag density, shrub cover, basal area) at a 90 m grid resolution in northern Utah. Data for model building came from the USDA Forest Service Forest Inventory and Analysis Program. Accuracy of the models was 80% and 83% for lodgepole pine and forest presence, respectively. Assessing accuracy of the continuous variables was more problematic, but 82%, 79%, and 89% of the model predictions for snag density, shrub cover, and basal area, respectively, were within 2 standard normal deviates of the field values. We close with example applications demonstrating how prediction of vegetation structure can be used to enhance wildlife modeling and wildlands management that are central to the Gap Analysis Program.
GAP LAND COVER MAPPING: CONVENTIONS, CHALLENGES AND CREATIVITY
Eve, Marlen D., James W. Merchant, and Jill Wolf
CALMIT, University of Nebraska, Room 113 Nebraska Hall, Lincoln, NE 68588-0517
Land cover mapping for the Gap Analysis Program (GAP) is conducted on a state-by-state basis. The resulting land cover maps are used extensively in completing the vertebrate modeling and the biodiversity analysis. One of the goals of the program is regional analysis of land cover, vertebrate habitat, and conservation. Therefore it is critical that each state's land cover products be consistent with products developed by their neighboring states. Many of the conventional approaches to land cover mapping do not work well for the detailed mapping of large regions required for GAP. Furthermore, the challenges of such detailed mapping vary by region, land cover, and project. This has led to a lot of creativity in the approaches that states have used to map their project area. We conducted an extensive investigation of mapping approaches and techniques used within the GAP land cover mapping community. Our findings are published on an Internet Web site at http://www.calmit.unl.edu/gapmap. This report is an overview of the conventions, challenges, and creativity demonstrated in the development of land cover mapping for GAP, and a tour of the information available on the Web site. It is hoped that this accumulation of land cover mapping information will assist in the implementation and efficiency of future mapping efforts.
SOME CONSIDERATIONS ON DEVELOPING A GAP ANALYSIS PROGRAM FOR MEXICO
Gonzalez-Rebeles, Carlos
Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX 79409-2110
A cooperative venture between the Mexican agency "Comision Nacional para el Conocimiento y Uso de la Biodiversidad" (CONABIO) and the United States Geological Survey has been funded to develop an international Gap Analysis project along the Lower Rio Grande region. Experiences generated from this study will help CONABIO evaluate approaches needed to extend Gap Analysis across all of Mexico. The development of a program for _Mexico will have to be built based on experiences in the U. S. However, different issues will also have to be considered when applying this methodology in a country with different socioeconomic and political structure and more diversified ecosystems. For example, land tenure system in Mexico (more than half of the land under common property) combined with the economic limitations affecting rural communities and poor land-use planning, have resulted in extensive impacts on the landscape. To accurately map land cover it might be necessary to categorize vegetation types at different levels of conservation status. Similarly, it may be necessary to weight vertebrate distribution estimates into different probability levels of occurrence. In addition, it may be useful to include socioeconomic factors, adding an element of risk for the determination of priorities and levels of urgency for conservation action. The scope of Gap Analysis may also need modification to comply with conservation needs in Mexico. For example, besides identifying underrepresented high-richness sites, Gap Analysis could be used to direct land use zoning to delineate appropriate sites for sustainable use practices. This will promote rural development opportunities and at the same time buffer out intensive use from critical sites. The refinement of procedures will test the flexibility of Gap Analysis, enhancing its applicability for Latin America.
DATABASE DESIGN AND OPERATION FOR GAP IMPLEMENTATION: A DEMONSTRATION OF SPECIES CONSERVATION AND MODELING SOFTWARE
Gorham, James N.
U.S. Fish and Wildlife Service, Delaware Bay Estuary Project
An early consideration in the development of vertebrate models and other data sets for the Mid-Atlantic GAP project (Delaware, New Jersey, Maryland) was developing an effective vehicle for distribution to make data available to land managers and biologists. We have designed a custom ArcView GUI which provides desktop access to GAP data sets and enables users to perform custom modeling or queries of GAP data. Our species modeling was performed using a series of data tables in a relational database and applying the results of queries of the database to spatial layers in GIS software. Structured Query Language (SQL) scripts were embedded in ArcView's Avenue programs and were used to access an Oracle database via an ODBC driver from the ArcView platform. The custom ArcView interface enables users to access maps developed for GAP, as well as directly query the database using the powerful SQL scripts developed for the modeling. Users can also customize modeling by controlling equation parameters and rerunning the models.
MAKING EFFECTIVE QUERY/ANALYSIS OF GAP DATA AVAILABLE TO MANAGERS
Gorham, James N.
U.S. Fish and Wildlife Service, Delaware Bay Estuary Project
One of the primary benefits of GAP is the compilation of large and varied data sets often previously unavailable to decision makers and land managers. The scope of data developed as a part of the Mid-Atlantic GAP project supports a wide range of questions and data queries, with the potential of providing critical information with real effects on species conservation. Effective database and query tool design is an important component of this. The Species Conservation and Modeling Software tool was developed to meet this need and can provide easy access to information important to managers to assist in conservation objectives. Examples include information on species distribution and status, evaluation of impacts of proposed site activities, identifying important sites for riparian restoration, or identifying sites for land acquisition and conservation easements.
EXTENDING GAP ON THREE CONTINENTS
Griffin, Curtice, and Dana Slaymaker
University of Massachusetts
No abstract submitted.
MODELING PREDICTED VERTEBRATE DISTRIBUTIONS IN MISSOURI BASED ON SPECIES PERCEPTIONS OF HABITAT AND LANDSCAPE
Haithcoat, Timothy1, Kelly Wetteroff, and Ron Drobney2
1University of Missouri, Department of Geography, Geographic Resources Center
2University of Missouri, School of Natural Resources, Department of Fisheries and Wildlife
We will present the Missouri methodology for creating predicted vertebrate ranges for Gap Analysis utilizing species-specific perceptions and spatial constraints based on measures of the environment and landscape within which they live. This methodology utilizes many more capabilities of GIS as well as incorporating statistical measures of landscape to refine the spatial extent of a species range. We will include discussions of scale, patch, matrix measures, landscape context and structure, and ancillary data sets as they relate to the selection and modeling of vertebrates within a GIS. We have collected statewide information on distributions and habitat preferences for species within each terrestrial taxon (including mammals, amphibians, reptiles, and birds). This database was used in association with the Missouri land cover map to determine areas of suitable habitat within the known geographical range of each species modeled. These were then combined to provide a species richness map for the state from which the gap analysis was performed.
MISSOURI GAP PRODUCT SENSITIVITY ANALYSIS WITH REGARDS TO SPATIAL AND CATEGORICAL RESOLUTION
Haithcoat, Timothy1, and Ronald Drobney2
1Geographic Resources Center, University of Missouri
2Missouri Cooperative Fish and Wildlife Research Unit, University of Missouri
The basic methodology underlying Gap Analysis is that vegetation categories are mapped via TM image analysis, related to habitat via a crosswalk table, and, if available, ancillary data are added for the distributional modeling of vertebrate bio-diversity within a user-specified MMU. This methodology makes three assump-tions: 1) The limited vegetation categories defined from the satellite image can be directly related to the needed habitats for a terrestrial species. 2) The species potential occurrence within these now defined habitats can be binarily modeled based on the inputs. 3) The classification, aggregation, and generalization of these habitat units for modeling and reporting have little effect on modeling outcomes.
These assumptions are currently being taken for granted by most people who are applying these data to management decisions. To date, there are few established guidelines detailing a state's GAP data sets' appropriate usefulness at different planning and management scales. We feel the largest issue, which has yet to be clearly examined, is resolution--both spatial and categorical. The second issue is how the modeling of the species/habitat relations is being conducted. We feel that with the use of simple geographic analysis tools the output of most models can be refined. We will refer to this method as the 'Missouri method' for lack of a better term. By incorporating more of the specifics of the species niche as well as its view of the landscape, the results of the vertebrate modeling, and therefore the final gap analysis, can potentially be significantly different.
This presentation will track these issues through the entire GAP methodology to determine how resolution changes could affect the output of the vertebrate model-ing and thus the final gap analysis. We hope to define scale-bounding criteria for the use of GAP data sets for planning and management at regional reporting levels as well as in-state mapping levels. This constitutes a methodological sensitivity analysis of GAP layers of varying resolutions (both spatial and categorical) and their associated modeling and outcomes. The aim is to define their boundaries of use so as to provide the needed quantitative and intellectual basis for informed decision making and planning. In this way we hope to establish guidelines for the appropriate use of GAP-derived data sets and initiate research which will lead to a better second-generation GAP product at regional and local levels.
AQUATIC GAP - CURRENT STATUS AND NEXT STEPS
Haverland, Pamela S., and Thomas A. Muir
US Geological Survey, Biological Resources Division, Reston, VA
The need for information on the aquatic resources of the nation has never been greater. Forty-five percent of the species currently listed under the Endangered Species Act are freshwater and coastal aquatic organisms--fish, freshwater mussels, crayfish, amphibians, and plants. This loss of aquatic biodiversity is an indicator of habitat destruction and degradation of water resources that are vital for economic growth and stability. The continued decline of aquatic biodiversity is a major concern of the state and federal fish and wildlife management agencies. Currently, there is a lack of tools to objectively characterize the loss of aquatic biodiversity, particularly at the watershed or landscape level. There is a critical need on the part of decision makers and resource managers for consistent and comparable measures of aquatic resource health to assist them in developing and evaluating management strategies. The goal of an Aquatic GAP Partnership is to characterize, at a landscape scale, aquatic resources seamlessly from freshwater to estuarine and near-shore marine environments. With this information, decision makers and resource managers at the local, state, regional, and national levels can evaluate their aquatic resources and more effectively make land and water resource decisions. Aquatic GAP is not a shift away from, or reprogramming of funds from the current terrestrial component. Aquatic GAP is designed to build on technical and practical experience of existing projects and partnerships. This presentation will review recent developments in the rapidly expanding Aquatic GAP and cover the next steps in developing an Aquatic GAP partnership.
IF YOU BUILD IT, THEY WILL COME: MAKING NBII A REALITY WITH GAP ON THE WEB
Herdendorf, Margo, Thomas Kohley, and Jeffrey Hamerlinck
Spatial Data and Visualization Center, University of Wyoming
The goal of the NBII is to provide swift user access to biological databases and information, being dedicated to the development of an electronic federation of biological data and information sources. Its success rests on a growing network of partners who share biological information. But how do you get the potential partners sold into the idea when they have to contribute valuable time to making their data accessible and documented? The Wyoming Bioinformation Node, a cooperative effort between the University of Wyoming and National GAP, has been been conducting informal NBII "outreach" using the Wyoming GAP data. WBN provides web-based tools, such as interactive mapping, to facilitate access to information. The information is provided by partners who are willing to share their data or to make portions of their data available for more exposure. By building on-line tools for displaying Wyoming Gap Analysis data, we've attracted interest from a growing number of groups who not only want to contribute their data, but are also willing to format/document their data in order to get it on-line and visible to a wider audience. Specific examples of tools that WBN has developed/participated in include the Wyoming Vertebrate Atlas, the Wyoming Plant Atlas, and the Wyoming Internet Map Server, demonstrations of which will be shown in this presentation.
HOW WELL DOES MANAGEMENT STATUS PREDICT VULNERABILITY OF REGIONAL BIOTA?
Hollander, Allan D., Frank W. Davis, and David M. Stoms
Institute for Computational Earth System Science, University of California, Santa Barbara, CA 93106
The Gap Analysis Program has translated its general principles into a pragmatic, operational approach for assessing the regional distribution and management status of plant communities and vertebrate species. The classification of management status assumes a correspondence between the kinds of land uses allowed in different management designations and the vulnerability of biota based on responses to those land uses. We want to determine whether this relatively simple model of management status is adequate for setting regional conservation priorities. Here we compare trends in distribution and abundance of birds from the Breeding Bird Survey with GAP predictions based on management status of their habitats. We also augment the GAP management classes with additional indicators of stress such as roadedness and population growth indices to test how well these refine the predictions of vulnerability. We conclude with suggestions for future analyses of vulnerability that may incorporate such indicators on a species-specific basis.
DEVELOPMENT AND USE OF A HABITAT PROFILE DATABASE APPLICATION
Holt, Eric A., and Nick C. Parker
U.S. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX 79409-2110
To map the habitat in which each vertebrate of Texas is expected to be found, the Texas Gap Analysis Project developed a Habitat Profile Database Application. After identifying the various habitat variables to be modeled, Microsoft Access was used to create the application. The application consists of (1) tables for storing the data associated with the variables, (2) forms, which serve as the data-entry interface, and (3) reports, which serve as the hardcopy media for the data. The completed application was given to recognized experts (i.e., mammalogist, ornithologist, and herpetologist) to fill out. These experts had the complete responsibility of building the habitat profiles for their selected vertebrates. As of 9 June 1998, 164 profiles (25%) have been completed. Texas Gap Analysis Project personnel reviewed submitted profiles, and corrections were made under consultation with the experts. Having experts responsible for creating the habitat profiles reduces the time required to create profiles and improves the quality of the profiles. Having GAP personnel create a custom database application for the experts for building the profiles (1) eliminates the data entry step associated with using paper data sheets, (2) ensures that all experts complete profiles using the same variables, and (3) ensures that all data are stored in the same electronic format. Although completed data sets are currently being successfully mailed on computer disks between cooperators and TX-GAP, a better method would be to allow cooperators to save data directly to our server through a Web-based Habitat Profile Database Application.
DEVELOPING A CONSERVATION RESERVE PROGRAM LANDS MASK FOR SOUTH DAKOTA GAP: USE OF MULTISPECTRAL SCANNER SUBSYSTEM DATA
Jenks, Jonathan A., Dorothy M. Dateo1, Paul M. Seevers2, Jeremy J. Higgins3, and Kenneth F. Higgins4
1Department of Wildlife and Fisheries Sciences, South Dakota State University, Brookings, SD 57007
2USGS Eros Data Center, Baltic, SD 57003
3Department of Biological and Microbiological Sciences, South Dakota State University, Brookings, SD 57007
4USGS, Biological Resources Division, South Dakota Cooperative Fish and Wildlife Research Unit, South Dakota State University, Brookings, SD 57007
The Conservation Reserve Program (CRP) was initiated under the 1985 Food Security Act to reduce soil erosion, improve air and water quality, and create wildlife habitat. In this program, land owners agree to take cropland out of production for 10 years. In return they receive annual rent plus half the cost of establishing permanent land cover, which in South Dakota is primarily non-native vegetation. The large quantity of CRP land in South Dakota (> 685,357 hectares) makes mapping natural vegetation from Thematic Mapper satellite data difficult because CRP land and native vegetation have similar spectral characteristics. We investigated identifying CRP land using multispectral scanner subsystem (MSS) data to construct a CRP 'mask' for South Dakota GAP. MSS images from two time-periods (summer 1973, 1992) for 20 counties in eastern South Dakota were analyzed using Land Analysis System and Arc/Info software. Each image was spectrally clustered and classified as vegetated or unvegetated land (primarily tilled agriculture). A model identifying potential CRP land was created by selecting pixels that were unvegetated in 1973 (before initiation of CRP), but vegetated in 1992 (potential CRP land). A preliminary accuracy assessment for Marshall County, South Dakota, indicated CRP land was identified correctly 76.4 % of the time; incorrect pixels were most often confused with agricultural lands (85.4%) and not native vegetation (1.45%). Plans are under way to assess accuracy of CRP delineation in 11 other counties in eastern South Dakota. The use of MSS data for identifying CRP land is a promising technique for identifying natural vegetation in a highly modified landscape.
THE NATURE CONSERVANCY'S AQUATIC COMMUNITY CLASSIFICATION: A FRAMEWORK FOR FRESHWATER CONSERVATION
Lammert, Mary, Jonathan Higgins1, Mark Bryer, and Dennis Grossman2
1Great Lakes Program, The Nature Conservancy
2International Headquarters, The Nature Conservancy.
In response to the crisis threatening freshwater biodiversity, The Nature Conservancy initiated the development of an aquatic community classification in 1995. The goal of the project was to create a standard means to identify and describe aquatic communities in streams and lakes, analyze their abundance and distribution, and identify aquatic conservation priorities. The classification framework describes both the biological composition and abiotic context of aquatic communities. The Conservancy recently completed its first application of the classification to the Great Lakes basin (U.S. only). In this pilot, we were faced with a lack of comprehensive biological community data and thus focused on delineating ecological units at a scale intended to predict the diversity of biological communities based on physical factors. This approach draws on the extensive literature describing habitat and assemblage interactions, and is modeled particularly after the Michigan River Inventory. Aquatic ecologists mapped macrohabitats (units of streams and lakes) in a GIS, based on interpretation of geologic, hydrographic and elevational information. Land use/land cover overlays, water quality data, and expert opinion were used to identify the best remaining examples of the macrohabitat types. With expert review, the Conservancy has designated a set of priority aquatic systems that together represent the full variety of macrohabitat types across the Great Lakes region. Methods developed in the Great Lakes basin will be applied on an ecoregional basis to identify priority sites for freshwater conservation in the rest of the country in the next three to five years under the Conservancy's new Freshwater Initiative.
APPLICATIONS OF GAP TECHNIQUES TO DESCRIBE SCALED QUAIL (CALLIPEPLA SQUAMATA) POPULATION CHANGES IN TEXAS
Leyva, Raquel I., and Nick C. Parker
U.S. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX 79409-2110
The applicability of TX-GAP data to describe the effects of long-term habitat modification on species densities has not been tested. The objective of this study is to evaluate the use of data produced by Texas-GAP to describe changes in populations of scaled quail (Callipepla squamata) in Texas based on changes in the percent of suitable habitat, temporal climatic variation, and combined effects of long-term habitat modifications and temporal climatic variation.
A Geographic Information System (GIS) will be used to facilitate the organization and management of three main databases: 1) abiotic factors (climate, soil, and topography), 2) landscape characteristics (vegetation based on current and historical changes in the land use and land cover), and 3) scaled quail population trends based on the Breeding Bird Survey. Historical climate data were acquired from the National Climatic Data Center (NCDN) and include records from 1840 to 1991. Data were converted into INFO tables and exported to ARC/INFO to generate climate coverages using TOPOGRID and KRIGING tools. Historical climate data will be used to describe possible correlation with changes in populations of scaled quail. Other coverages such as soils, land use, and vegetation are under development. The final product of this research will provide a tool for development of management plans for the species at the county and ecoregion level. This approach may also prove the usefulness of GAP data to describe the effect of long-term habitat modification on other wildlife species.
USING GAP ANALYSIS TO DEVELOP ECOSYSTEM MANAGEMENT PLANS: THE NORTHERN AND EASTERN COLORADO DESERT PLAN CASE STUDY
Lock-Dawson, Patricia A.1, Nanette Pratini2, and Kimberly Nicol3
1U.S. Bureau of Land Management, California Desert District
2University of California, Riverside
3California Department of Fish and Game, Natural Heritage Division
Currently, the U.S. Bureau of Land Management (BLM) is involved in several bioregional planning efforts in the Southern and Central California Desert. We will discuss one particular planning effort which encompasses 5.5 million acres in the Northern and Eastern Colorado (Sonoran) desert. The planning area spans two desert ecoregions and includes urban, agricultural, and natural areas. Land ownership is mixed, and the area is currently managed by various jurisdictions. The purpose of this plan is to provide for recovery of the desert tortoise, a federally listed threatened species, while simultaneously accommodating human uses throughout the planning area. We are using a coarse filter approach to species and habitat modeling which relies heavily on GIS technology. We developed a vegetation map by ground-truthing a preliminary map provided by the U.S. Geological Service's Gap Analysis Program (at UCSB). The intent of this paper is to describe our experiences utilizing gap analysis and wildlife-habitat relations modeling in a real-world context. Specifically, we will focus on the difficulty in choosing "indicator" or "umbrella" species to provide protection for entire communities and ecosystems. Rather than choose one species or suite of species, we chose to use multiple biotic and abiotic parameters to identify areas of ecological importance. We will share our methods and make recommendations for improving future efforts.
SPATIAL VARIABLE MODELING TECHNIQUES AND POSSIBLE FOREST BIRD CONSERVATION SOLUTIONS APPLIED TO A MID-ATLANTIC LANDSCAPE
McCorkle, Richard C.
U.S. Fish and Wildlife Service, Delaware Bay Estuary Project
Distribution modeling was performed for a number of forest area-dependent birds and applied to vegetative cover maps and other data layers for the state of Delaware. Various GIS spatial functions were employed to improve the accuracy of these models, and model verification was performed using data from the statewide Breeding Bird Atlas. Species considered are highly sensitive to forest fragmentation. Important forest stands were identified by overlays of individual species distributions, and the management status of these stands was evaluated with overlays of the stewardship layer. In addition, the distribution models for the most sensitive species (e.g., cerulean warbler, black-and-white warbler, northern parula, American redstart, Louisiana waterthrush, hooded warbler) were used to identify high-priority habitat restoration areas that would increase habitat suitability for these species, based on remodeling these species using 'what-if' scenarios. These scenarios included identifying potential reforestation areas that would add to existing forest stands, closing forest gaps, or providing buffers between stands and encroaching urban development. The process of identifying these potential restoration areas was automated through the use of various GIS modeling functions (e.g., selectively expanding forest into only those land use classes that are potentially available for reforestation).
CHALLENGES IN MODELING SPATIAL VARIABLES FOR FOREST AREA-DEPENDENT BIRDS IN A FRAGMENTED LANDSCAPE
McCorkle, Richard C., and James N. Gorham
U.S. Fish and Wildlife Service, Delaware Bay Estuary Project
Efforts to model forest bird distributions are confounded by the dependence of these species on both spatial and structural habitat variables, as well as forest composition. The structural variables (e.g., snags, sub-canopy strata) are the most difficult to consider in modeling efforts that depend on remotely sensed data. Landscape-level spatial variables, on the other hand, can be considered in modeling and can improve modeling accuracy. We employed a number of GIS techniques to model suitability of forest habitats for forest area-dependent species. Forest patch size (i.e., forest-interior), riparian forest width, and percent forest cover within a certain distance of each forest patch, were among the spatial parameters that were modeled. Additional GIS processes were applied to forest cover to improve forest patch selectivity and model accuracy. These modeling approaches have enabled us to integrate the statistical results of key studies of forest bird breeding distributions in the fragmented Mid-Atlantic landscape. We believe that these and other spatial variable modeling functions have greatly improved the accuracy of forest bird habitat models.
DATA, MODELS, MAPPING, AND RELIABILITY: RESULTS AND QUESTIONS FROM CANADA.
McKenney, Daniel W.
Canadian Forest Service, P.O. Box 490, 1219 Queen Street East, Sault Ste. Marie, Ontario, Canada P6A 5M7
This paper will provide an overview of some activities in Canada that would support a GAP-type program if one were in place. This includes the development of national climate surfaces, a new national Digital Elevation Model and various biological field survey data sets that have been used for spatial modeling. This work has implications for joint Canada/U.S. GAP-type projects. The framework is being applied in a collaborative bird mapping project in the Great Lakes Basin. That project is trying to map bird species' distributions (actual and potential) and abundance at relatively fine scales (approx. 1 km) across the entire Basin. In addition, some preliminary Monte Carlo tests of map reliability and sample size will be presented.
APPLYING THE NATIONAL VEGETATION CLASSIFICATION USING AERIAL VIDEOGRAPHY AND DECISION RULE MODELING
McKerrow, Alexa J., and Steven G. Williams
North Carolina Gap Analysis Project, 5123 Jordan Hall, NCSU, Box 7106, Raleigh, NC 27695-7106
The North Carolina Gap Analysis Project has been working with the North Carolina Natural Heritage Program and the Southeast Regional Office of The Nature Conservancy in developing a vegetation map for the southeastern coastal plain of North Carolina. As a part of the Southeast Biodiversity Project, the vegetation map will be used to help define conservation priorities for North Carolina. This study area was selected as the starting point for our efforts because it represents one of the most diverse areas in the state and is therefore a conservation priority for North Carolina.
To create the vegetation map, we started with a point coverage derived from plant community data from the Heritage Program, field data from the Division of Coastal Management, and aerial video interpretation. We then developed a methodology based on Slaymaker et al. (1996) that incorporates iterative decision rules in the creation of a refined vegetation map. Modifications on the methods include the use of detailed soils, National Wetland Inventory data, and the use of an interim decision rule image which allows the classifier to trace errors in the modeling process. Incorporation of the National Heritage Program data gives us the potential for identifying which small patch alliances are mapped within the final map units. The results of the vegetation mapping for the southeastern coastal plain will be presented with a discussion of the confusion matrix developed for assessing the vegetation map.
Slaymaker, D.M., K.M.L. Jones, C.R. Griffin, and J.T. Finn. 1996. Mapping deciduous forests in Southern New England using aerial videography and hyperclustered multi-temporal Landsat TM imagery. Pages 87-101 in J. Michael Scott, Timothy H. Tear, and Frank W. Davis, editors. Gap Analysis: A landscape approach to biodiversity planning.
DEVELOPING A LAND USE/LAND COVER MAP FOR THE MISSISSIPPI GAP ANALYSIS PROGRAM
Batten, Susan D., David L. Evans1, Francisco J. Vilella, and Richard B. Minnis2
1Department of Forestry, Mississippi State University, Mississippi State, MS 39762
2Cooperative Fish and Wildlife Research Unit, Mississippi State University, Mississippi State, MS 39762
Vegetation classification has been called the "heart" of Gap Analysis, as this layer is the basis of all GAP vertebrate modeling. Each state conducting Gap Analysis is responsible for developing the best methodology possible for constructing their vegetation data layer. Here we describe the methodology in use for developing Mississippi's vegetation layer. The Mississippi Gap Analysis Project has completed its first full year of work. We have primarily focused our efforts on one TM satellite image scene that best represents the state as a whole. Three spectral bands from 2 different dates of imagery from the scene were used, one taken during fall (leaf on) and one in early spring (leaf off). The 2 dates of imagery were used to accurately represent all spectral characteristics of the classes to be identified. Spectral channels used were the green, near infrared, and infrared.
The image was partitioned using the 3 forest provinces of Mississippi located within the scene of interest. Analysis indicated this reduced variance within the same species over the different soil types and topographic conditions. Each province was classified individually using a guided clustering classification. This classification was based on an existing Mississippi land cover map developed by the Stennis Remote Sensing Center (SRSC) that contained 25 broad classes (approximately Anderson class level II). Classes of agriculture, urban, transportation, and wetlands (NWI) were used directly from this product. All forested, shrub-scrub, and grassland classes were grouped together and reclassified. Recently harvested areas were delineated manually to eliminate confusion with other classes. Resultant classes were identified and labeled based on the FGDC classification system to the alliance level when possible. Most classes contain multiple FGDC alliances that could not be separated spectrally. The identity of each class was determined using georeferenced field data plots.
EXPLORATION OF MAPPING ALTERNATIVES FOR GAP ANALYSIS IN PENNSYLVANIA
Myers, Wayne, Robert Brooks, Gerald Storm, and Joe Bishop
The Pennsylvania State University, University Park, PA
Pennsylvania Gap Analysis was launched under the coarse-scale and less definitive early guidelines, whereby there was latitude for us to view each component as an opportunity for research innovation. The GAP era in Pennsylvania has been especially opportune in this regard by virtue of several collateral initiatives in the arenas of both mapping technology and promoting sustainability. The Pennsylvania GAP project has served as a synergistic bonding agent that lends higher-level coherence to a collage of such ancillary efforts. Instead of doing a single-scale land cover mapping, we have mapped different ways at different scales with legend variations as appropriate. This provides a comparative suite of landscape views that invites further multiscale landscape ecological investigation. The goal of more detailed vegetation mapping has been linked with a cooperative ecological landtype mapping thrust in Pennsylvania that supports a new generation of State Forest management plans wherein sustainability is a planning foundation. Innovative mappings have likewise become a basis for outreach in Pennsylvania. We have taken the essence of the hypercluster idea and developed it into a means of making enhanced compressed image-maps publicly available on CD-ROM as dataforms that are directly compatible with ArcView and ArcExplorer by ESRI. We have been joined by several other sponsors in this digital image outreach. We have likewise developed an echelon approach to surface data that is applicable to analyzing the spatial topology of biodiversity relative to habitats. Finally, the culmination of Pennsylvania GAP is coincident with a major conference/workshop on creating a cooperative framework for sustaining our natural diversity heritage.
AQUATIC GAP ANALYSIS OF SMALL AND MEDIUM RIVERS OF KOREA
Park, Chong-hwa1, and Sung-hak Hong2
1Professor, Graduate School of Environmental Studies, Seoul National University, Seoul, Korea
2GIS Analyst, SK C&C, Inc., Seoul, Korea
The objectives of this research are to develop fish habitat relationship models which can be used to estimate fish species richness of small and medium rivers of Korea and test the accuracy of the models. The models are based on the Aquatic Gap Analysis model of the New York Cooperative Fish and Wildlife Research Unit, and they employ three habitat factors: river size, physical habitat, and water quality of each river segment. Model I and model II are based on the water quality standard for life support of EPA and the water quality class of Korea, respectively. Test sites for this study include one urban stream and three less spoiled tributaries of the Han River. The outcomes of this research can be summarized as follows. First, the number of habitat types identified by model I and model II are nine and 14, respectively. Second, the average accuracy of three distribution maps for rare or endangered fish species is 80.6% (model I) and 81.2% (model II). Third, the accuracy of fish species richness is 94% (model I) and 95% (model II), and water quality is the most important factor affecting fish species richness. Fourth, the accuracy of the fish species list is 50.5% (model I) and 68.7% (model II), but the accuracy of less spoiled stream segments and that of polluted stream segments is 67.1% and 86.5%, respectively. Finally, it can be concluded that the overall performance of model II is better than model I at our test sites.
THE TEXAS GAP PROJECT: STATUS AND POTENTIAL
Parker, Nick C., Carlos Gonzalez-Rebeles, T. Scott Schrader, Andrea E. Ernst, Yonglun Lan, Kelly E. Allen, Eric Holt, and Sheri Haskell
Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX 79409-2120
The Texas Gap Analysis Project (TX-GAP) is part of a nationwide effort to document the spatial distribution of biodiversity and its representation in the current conservation system. The objectives of TX-GAP are: (a) to develop a map of current land cover of Texas from recent Landsat TM satellite scenes; (b) to estimate potential distribution of Texas wildlife vertebrate species; (c) to map land stewardship categorized by level of conservation; and (d) to combine the above data layers in a GIS and perform analyses of species richness relative to known levels of land conservation and management. There are 52 Landsat scenes covering the state of Texas. Pixels representing 30 x 30 m areas are classified with Spectrum software according to the Nature Conservancy vegetation classification at the alliance level. Fifty to 200 points per scene are being used to ground-truth scenes in West Texas. Aerial videography provides additional data to interpret the Landsat imagery. Seventeen scenes in West Texas have now been tentatively classified. Wildlife Habitat Relationships (WHRLs) for Texas vertebrates are being prepared based upon habitat affinities for vegetation type, soil type, precipitation, elevation, temperature, and other abiotic and biotic factors. WHRL databases are approximately 25% complete for mammals, 50% for herptiles, and 50% for birds. We anticipate completion of all vertebrate databases this year. WHRLs are used in a Geographic Information System to map the distribution of Texas' vertebrates. These maps are then used to evaluate the status of biological diversity in the state and ultimately the nation.
An initiative of the Biological Resources Division of the U.S. Geological Survey has been funded to extend TX-GAP into Mexican lands adjacent to the Lower Rio Grande (Rio Bravo) River. The study area is covered by 14 Landsat TM scene areas that span the Lower Rio Grande River plus six adjacent scene areas wholly in Mexico. This transnational Rio Grande Gap Analysis Project will cover a buffer area approximately 150 km wide to each side of the border. The project will generate valuable geographic and biological data sets to support binational efforts for conservation and land use planning, provide opportunities for biological data sharing and the potential standardization of procedures applicable in this region with common ecological characteristics.
MAPPING LAND COVER WITH 4-METER MULTISPECTRAL AIRBORNE IMAGERY AND VIDEOGRAPHY: A PILOT STUDY IN MARYLAND
Rasberry, D.A.1, and M.L. Francis2
1Maryland Department of Natural Resources
2ICF Kaiser International
Gap Analysis projects are using Landsat Thematic Mapper (TM) imagery with 30-meter resolution to create land cover maps following the National Vegetation Classification System (NVCS). While the goal is to map to the community alliance level, many projects are finding the resolution of the TM imagery to be a detriment to mapping alliances. In the eastern deciduous forests, the heterogeneous forests along with the small patch alliances prominant in the coastal plain are easily found on video but are not discernable on the imagery. Mid-Atlantic GAP (MID-A GAP) had an opportunity to work with NASA, ICF Kaiser International, and the U.S. Navy, Patuxent River Naval Air Station (NAS), to prototype methods for high spatial resolution imaging platforms which will become available commercially over the next couple of years. Two locations were chosen for the pilot study: the area surrounding the NAS and the barrier islands of Assateague, Maryland, and Chincoteague, Virginia. NASA flew transects using their Calibrated Airborne Multispectral Scanner (CAMS), a nine channel multispectral sensor with direct digital recording capability, from a Learjet. The CAMS data were resampled to 4-meter resolution, georeferenced and mosiacked at Stennis Space Center in Mississippi, and methods employed by MID-A GAP were used to classify the imagery. Results were favorable; community alliances visible in video were discernible in the imagery and sometimes identification was to individual species level. On the barrier islands, where the alliances are small and patchy, classification with the 4-meter data yielded nearly three times the number of map classes as with the 30-meter data.
MONTANA LAND COVER: THE GOOD, THE BAD, AND THE UGLY
Redmond, Roly
Montana Cooperative Wildlife Research Unit, University of Montana, Missoula
The good is that we succeeded in mapping the land cover of the fourth largest state in the Union at relatively fine resolution. The bad is that the process took more than six years, and now people want even better resolution in a smaller file. The ugly? Come and judge for yourself. We developed a two-stage, digital process to independently classify and label 33 Landsat TM scenes covering Montana. Using more than 23,000 ground reference data, nearly 4.4 million raster polygons were labeled to one of 94 land cover types. Upland cover types were mapped to a 2-ha minimum map unit (MMU) statewide. In eastern Montana, riparian and woody draw vegetation types were mapped to a 0.4 ha MMU, whereas in western Montana, a separate pixel classification was performed for riparian types. These independent classifications were edge-matched to create a virtually seamless raster database. From this, we created a statewide layer with a single-attribute land cover type. Recognizing that this data set would be too large for MT-GAP, several additional steps were carried out to reduce the output file size, including 1) regrouping the 94 cover types to 50 types that were mapped more consistently across the state, and 2) resampling the 30-m grid to 90 m. Although this file was still too large to distinguish individual raster polygons (by "region-grouping"), it did provide an adequate basis for predicting vertebrate distributions statewide. Thematic map accuracy was calculated using fuzzy sets and a bootstrap procedure which allowed misclassification probabilities to be estimated at the location of each training data point. Absolute thematic accuracy for 45 cover types averaged 61.4%, ranging from 4.4% for Western Hemlock to 93.2% for Missouri Breaks; at the "acceptable" level, mean accuracy increased to nearly 89%. Finally, we mapped mean thematic accuracy across the state to illustrate where the land cover map tends to be more or less accurate.
DIGITAL HABITAT: FROM DATABASES TO DISTRIBUTION MAPS
Hart, Melissa, and Roly Redmond
Montana Cooperative Wildlife Research Unit, University of Montana, Missoula
We describe the process used to develop a wildlife habitat relationships database and to predict vertebrate distributions for MT-GAP. Using FileMaker Pro on a Macintosh, relational databases were built to track the modeling process for 425 species. These databases include 1) general habitat relationships, plus assumptions and drawbacks inherent in each model; 2) GIS data layers used in each model, including specific land cover types and elevation zones; and 3) key references. Data then were ported to a UNIX workstation, and ARC/INFO AMLs were written to create predicted distribution grids for each species. Working at 90-m resolution, 77.25 hours of processing time were required to create 425 grids, which then occupied about 1.15 gigabytes of disk space. Accuracy of predictions was assessed using species checklists for 14 managed areas. Overall, accuracy was calculated to be 63.2% across all taxa, with higher accuracies for birds and mammals than for amphibians and reptiles. Advantages and drawbacks of the method will be outlined, along with examples and further results for MT-GAP.
OVERVIEW OF RESEARCH AT THE NATIONAL CENTER FOR ECOLOGICAL ANALYSIS AND SYNTHESIS (NCEAS)
Reichman, O.J.
National Center for Ecological Analysis and Synthesis, 735 State Street, Suite 300, Santa Barbara, CA 93101
The National Center for Ecological Analysis and Synthesis was established at the University of California, Santa Barbara, in May 1995 and is supported by NSF, the State of California, and UCSB. The mission of the Center is to advance the state of ecological knowledge through the search for general patterns and principles and to make such knowledge useful to researchers, resource managers, and policy makers. The Center focuses on synthesizing and analyzing existing data to address pertinent ecological issues and does not support the acquisition of new data. This is accomplished through various opportunities, including research and training workshops, working groups, and Fellow (sabbatical) and postdoctoral positions. To promote these activities, the Center's facility includes meeting rooms, offices, and high-performance computing facilities, and the staff assists with the logistics of organizing and hosting Center activities. A number of NCEAS projects are directly relevant to the Gap Analysis Program, including several working groups concerned with conservation planning and reserve siting. Complete information about opportunities at NCEAS are available at http://www.nceas.ucsb.edu.
DEVELOPMENT AND PILOT APPLICATION OF THE CALIFORNIA URBAN AND BIODIVERSITY ANALYSIS (CURBA) MODEL
Landis, John D., Juan Pablo Monzon, Michael Reilly, and Chris Cogan
Department of City and Regional Planning, University of California-Berkeley
The U.S. has made tremendous progress over the last 25 years in improving its air and water quality. Far less progress has been made in the areas of land and habitat conservation. One reason why is that local land planning institutions typically lack the analytical capability to coordinate their land use and habitat conservation efforts. The California Urban and Biodiversity Analysis (CURBA) Model was developed to help bridge the gap between urban land use planners, who are principally concerned with directing urban growth, and conservationists and wildlife ecologists, who are concerned with promoting environmental and ecological quality. The CURBA Model integrates three sets of data sources and modeling approaches which have heretofore been separate:
So far, CURBA Model data sets and equations have been developed for nine California counties. This paper explains the logic, calibration, and use of the CURBA Model. We begin by presenting the structure of the model, how it measures and reports habitat quality, and how it integrates Gap Analysis and other data sources. Next, we present the results of a pilot study of the use of the CURBA Model, focusing on issues of urban expansion and habitat loss in Santa Cruz County. We conclude by reviewing the model's features, strengths, and limitations.
GAP ANALYSIS FOR PLANTS: WHICH SPECIES?
Reiners, W.A., Fertig, W.F., and R.L. Hartman
Department of Botany, The University of Wyoming, Laramie, WY 82070
Plant as well as animal species may experience geographic gaps between their areas of distribution and those areas managed for species conservation. How to deal with this? One problem is that there is no standard system for modeling distributions of plants as we have for vertebrate species. A second problem is numbers. State floras number in the thousands in comparison with terrestrial vertebrate faunas numbering in the hundreds. Thus, modeling the distribution of all plant species is not practical. Which species, then, merit initial attention? Our objective was to select 200 vascular plant species (a modeling goal) representative of Wyoming's 2,400 vascular species. How should these be chosen? From the Rocky Mountain Herbarium collections we compiled a relational database of the entire state vascular flora in terms of : 1) global and TNC heritage rankings (G1-G5; S1-S5), 2) geographic distribution patterns within and around the state (endemic, etc.), 3) primary biome affinity (grassland, etc.), and 4) life form (shrub, etc.). Next, flora members were excluded which: 1) are regional exotics, 2) are ubiquitous throughout the state, and 3) have fewer than 25 known locations within Wyoming and all adjacent states (25 was set as a minimum number of locations for statistical modeling). The remaining species in each of the possible sets were counted and the 200 target species selected on a random basis in proportion to the number of species in each set. Besides providing a multivariate representation of plant species for modeling distributions, this exercise produced a very interesting phytogeographic analysis of a flora centered on Wyoming.
ROCKS AND ICE REVISITED: AN ASSESSMENT OF THE GEOGRAPHICAL AND ECOLOGICAL DISTRIBUTION OF RESERVES IN THE UNITED STATES
Scott, J.M., and R.G. Wright
US Geological Survey, Cooperative Fish and Wildlife Research Unit, University of Idaho, Moscow, ID 83844-1141
Creation of a complete network of biological reserves in a country requires that the level of protection attained with existing reserves be known before new refuges are established. This knowledge can be used to further protect biodiversity with a minimum of duplication of past efforts and the most efficient filling of gaps in the reserve network. We present the results of a study in which we mapped the occurrence of biological reserves in different physical environments across the coterminous United States. We examined the occurrence of these reserves by 500-meter elevation intervals, quintiles of soil prductivity, five-degree blocks of latitude and longitude, and ecoregions. Observed patterns of occurrence suggested uneven distribution within all these coarse-filter features. The areas with the highest level of protection were those that were least productive and/or least accessible. We discuss the implications of these findings for future siting of reserves.
DEMONSTRATION OF THE NBII METADATA STANDARD AND METAMAKER
Shin, Sharon, and Jennifer Gaines
U.S. Geological Survey, Biological Resources Division, Center for Biological Informatics and Office of Biological Informatics and Outreach
The U.S. Geological Survey (USGS) is leading a broad cooperative effort to develop the National Biological Information Infrastructure (NBII). The NBII is a distributed electronic federation of biological data through which a network of partners, cooperators, and others can find, share, and apply information. An important component of NBII is the NBII Clearinghouse (a node on the National Spatial Data Infrastructure, <http://www.nbii.gov/clearinghouse.html>) which facilitates effective searching and retrieval of biological data and information resources through the Internet. Metadata are the standardized descriptions that comprise the "card catalog" of the NBII Clearinghouse. To function properly, metadata must be consistent descriptions of data content, quality, accuracy, taxonomy, and methodology. As part of the NBII program, USGS collaborated with its partners to develop a biological metadata standard which acts as a "profile" of the Federal Geographic Data Committee's (FGDC) geospatial metadata content standard and provides additional elements particularly useful in describing biological data. MetaMaker, a metadata documentation tool developed by the USGS, enables users to prepare FGDC/NBII compliant metadata for biological and/or geospatial data. These three components--the NBII Clearinghouse, the biological metadata standard, and MetaMaker software--provide the fundamental components of biological data documentation and retrieval within the NBII program.
THE MISSOURI AQUATIC GAP PILOT PROJECT: AN OVERVIEW AND STATUS REPORT
Sowa, Scott P.
USGS Environmental and Contaminants Research Center, Columbia, MO 65211
The National Biological Service (now USGS-Biological Resources Division) initiated the Gap Analysis Program in 1988 to evaluate the conservation of biodiversity in North America. Gap Analysis initially focused on terrestrial ecosystems; however, the USGS-BRD has recently taken steps for initiating a Gap Analysis Program directed at aquatic ecosystems. In January 1997, the Missouri Resource Assessment Partnership (MoRAP) entered into an agreement with the USGS-Gap Analysis Program to conduct the first statewide pilot project for the aquatic component of Gap Analysis. Numerous objectives are being addressed in the Missouri pilot project. However, the three principal objectives are to 1) develop an objective method for identifying gaps in existing aquatic biodiversity conservation efforts and prioritizing future conservation strategies, 2) identify methods which effectively integrate the terrestrial and aquatic components of Gap Analysis, and 3) document information needs, successes, failures, major obstacles, time, costs, etc., which will assist other states with their efforts on the aquatic component of Gap Analysis. This presentation will cover 1) administration of the Missouri Aquatic GAP Project, 2) objectives and focus of the project, 3) theory and details behind the project approach, 4) some preliminary results of the project, and 5) major obstacles and future efforts.
PLANNING FOR BIODIVERSITY: APPLYING GAP ANALYSIS TO COUNTIES AND REGIONS
Stevenson, Matthew R., Frank Westerlund, Nora Camacho, Pat Iolavera, Michael Kerins, Todd Klinka, Tammy Kutzmark, Gabe Snedeker, Bob Thiel, Wood Turner, and Benj Wadsworth
University of Washington, Department of Urban Design and Planning, P.O. Box 355740, Seattle WA 98195-5740
Working with Spokane County, Washington GAP, and the Washington Department of Fish and Wildlife, students in the Department of Urban Design and Planning at the University of Washington conducted a comprehensive analysis of the distribution and conservation status of vertebrates and vegetation zones within Spokane County and designed a system of wildlife corridors and habitat reserves. The project holds great promise for application in other counties and regions because it demonstrates how Gap Analysis can be combined with other existing data sets to help ensure that county land use plans more effectively incorporate the preservation of biodiversity.
This process can be generalized and applied to any county or region in Washington State to produce potential wildlife corridor alignments and reserve locations. Students used three methodologies to demonstrate how GAP data can be used in conjunction with other data sets to: 1) determine representation for all terrestrial vertebrates predicted to occur within the study area, 2) delineate potential connections between representative areas and areas of high species richness, and 3) narrow and modify the resultant polygon sets to determine potential corridor alignments and reserve locations.
Additionally, the GAP data were used to create the ecological context within which the vertebrate analysis took place. Using the vegetation zone and stewardship data from Washington GAP, students determined the percentage of land within all vegetation zones in the study area, the percentage of land within all vegetation zones managed primarily for the conservation of biodiversity, the percentage of study area vegetation zones managed primarily for biodiversity, and finally, the new percentage of study area vegetation zones which would be managed primarily for biodiversity within any proposed system of corridors and reserves.
APPLICATIONS OF GAP DATA AND FINDINGS IN CALIFORNIA
Stoms, David M., and Frank W. Davis
Institute for Computational Earth System Science, University of California, Santa Barbara, CA 93106
CA-GAP data have been used for a surprising diversity of applications beyond their initial use in assessing the conservation status of the state's biodiversity. In this paper, we first summarize some of the more innovative uses for resource assessment and planning. Next we discuss recent research on evaluating GAP's assumptions about the land uses allowed in management status levels and how those relate to the actual level or probability of threat associated with those uses. Then we will describe how we are using data on allowable land uses, roadedness, and population growth to complement GAP status information to prioritize plant communities for conservation action. In an attempt to find measures of threat from readily available data, we evaluated roadedness and projected population growth as possible surrogates for existing impacts and potential threats. We modeled population growth projections based on some simple assumptions about where growth seems likely to occur. While types that are most heavily roaded or most susceptible to future urbanization tend to be the least well-represented in designated managed areas, the converse, that poorly represented types are the most vulnerable to stressors, is not necessarily a strong relationship. We conclude by discussing the transfer of management of the database to the California Department of Fish and Game.
STRENGTHENING CAPACITY FOR CONTINENTAL BIODIVERSITY ASSESSMENT AND PROTECTION: IDEAS FOR EXPANDING GAP'S TAXONOMIC DIMENSION AND THE CONCEPT OF RECURSIVE MAPPING
Umphrey, Gary J.1 and Alan D. Tomlin2
1Dept. of Zoology, University of Western Ontario, London, Ont., Canada N6A 5B7
2Agriculture and Agri-Food Canada - SCPFRA, 1391 Sandford St., London, Ont., Canada N5V 4T3
We have been exploring the feasibility of adapting the U.S. Gap Analysis Program in Canada, both for its specific value as a methodology for assessing, monitoring, and proactively reducing the impact of Canadian agriculture and other anthropogenic agents on biodiversity, and as a model for constructing a core geospatial framework for a Canadian national biodiversity information infrastructure. Among the numerous benefits, were Canada to initiate a GAP-based biodiversity mapping program in coordination with the U.S. and Mexico, would be a greatly enhanced capacity for continental biodiversity assessment and protection. But there are concerns about the amount of data available in Canada for developing species data layers. There is also a strong desire to more thoroughly inventory Canada's biodiversity and to incorporate biota other than vertebrates into biodiversity assessment and conservation planning. We believe that GAP offers a logical base for building a continental biodiversity inventory and provides a clear rationale for doing so. In this paper we will present some ideas on expanding GAP's taxonomic dimension. We will also outline the concept of "recursive mapping" and its potential for efficiently mapping distributions of species for which there is currently little data.
OREGON BIODIVERSITY PROJECT OVERVIEW
Vickerman, Sara E., and Keith Hupperts
Defenders of Wildlife
The Oregon Biodiversity Project was initiated in 1993 as a "GAP Implementation Program." Defenders of Wildlife administered the public-private partnership in cooperation with The Nature Conservancy, Oregon Natural Heritage Program, and more than 40 other partners from industry, academia, agencies, and conservation organizations. Three separate committees addressed project management and financing, technical and scientific issues, and implementation. The project completed all of its major products in early to mid-1998: a full-color book containing a statewide biodiversity strategy and highlighting 42 "conservation opportunity areas," a poster, CD-ROM, stewardship incentives report, and a process report. The strategy represents the first major attempt to use GAP data in statewide conservation planning. It may serve a s a model for other states, although consideration of unique circumstances is important. The conclusions and recommendations of the project will be presented, along with lessons learned from the Oregon experience. Financing, administration, timing, political sensitivity, and technical and institutional issues will be addressed.
AN EVALUATION OF SPECIES DISTRIBUTION EDGE-MATCHING FOR THE MID-ATLANTIC GAP REGION
Klopfer, Scott D., and Jefferson L. Waldon
Fish and Wildlife Information Exchange, Virginia Tech, Blackburg, VA
Regional coordination of terrestrial vertebrate species distribution mapping is integral to Gap Analysis. The Mid-Atlantic GAP projects (Delaware and Maryland), Pennsylvania GAP, Virginia GAP, and West Virginia GAP agreed to examine their distribution data simultaneously in an attempt to establish the current condition of interstate species distributions. Each state provided species distribution maps for all EMAP hexagons intersecting their states. These data were combined to create regional species distributions. Discrepancies between states with shared hexagons were quantified and used to guide further qualitative analysis. The results of this preliminary study will allow the Mid-Atlantic region to better coordinate data and concentrate mapping efforts on specific areas or taxonomic groups.
TESTING THE PERFORMANCE OF WILDLIFE HABITAT RELATIONSHIP MODELS: THEORY AND APPLICATION
Karl, J. W., J.M. Scott, and N.M. Wright
USGS/Biological Resources Division, Idaho Fish and Wildlife Cooperative Research Unit, University of Idaho, Moscow, ID 83844
Wildlife habitat relationship models have widespread application for wildlife research and management. Knowledge of the influence of model complexity and scale of application could guide future modeling efforts. Our objective was to determine how model performance corresponds with different model complexities, data resolutions, and levels of analysis. We used the Gap Analysis methodology to model the predicted habitat of 66 northern Idaho birds at three resolutions of geographic information system (GIS) data varying the complexity of the models. We tested these models against breeding bird survey data at two spatial scales (site and cover type level). Our results indicate that model performance generally increases as model complexity increases. This performance increase results from a decrease in commission errors (species predicted but not observed). Overall performance is affected little by model complexity at coarse (10 ha) data resolutions, indicating that complex models may not be necessary to achieve desired levels of performance at coarse (e.g., statewide) scales. Model performance is also higher at the cover type level than the site level. The influence of model complexity, data resolution, and level of analysis on the performance of wildlife habitat relationship models requires that specific objectives be defined prior to modeling wildlife habitats. Also, wildlife habitat relationship models designed for one application may not generalize well to other situations.
ADAPTING EASTERN VEGETATION ALLIANCES FOR USE WITH REMOTELY SENSED DATA
Yuill, Charles B.
Natural Resource Analysis Center, College of Agriculture, Forestry, and Consumer Sciences, West Virginia University, 1138 Agricultural Sciences Building, Morgantown, WV 26505-6108
The Nature Conservancy's vegetation alliances present a comprehensive framework for natural vegetation characterization in the Central Appalachians. However, many of the alliances are still somewhat poorly defined descriptively, quantitatively, and hierarchically. That is, though often similar in terms of defining characteristics (e.g., importance values of the predominate species), the roles various alliances assume in the landscape are often very different. Some are extensive components of the Appalachian landscape while others occur in only small isolated locations. We have therefore regrouped the alliances using landscape ecology criteria-mosaics, large patches, small patches, corridors, and isolated remnants. These groups have assisted us in determining mapping priorities without impacting the integrity of the alliance hierarchy. We have also established an additional working level in the TNC hierarchy. This level, which we have termed alliance groups, aggregates the alliances according to general vegetation/ecological criteria. The resulting level includes species groups such as oak dominant, mountain hardwoods, mountain conifer/hardwoods, and floodplain species, where each group includes a number of alliances. Again, the overall integrity and structure of the current TNC framework is fully maintained with such an additional level while adding a level that is necessary for mapping from satellite data with coarse spatial resolution. This coarse level represents the level at which our wildlife models are actually implemented.