Kenneth G. Boykin ¹ and Robert A. Deitner ¹

1 New Mexico Cooperative Fish and Wildlife Research Unit, New Mexico State University, Las Cruces

Introduction

The Southwest Regional Gap Analysis Project (SWReGAP) created predictive suitable habitat models for 819 terrestrial vertebrate species within the five southwestern states of Arizona, Colorado, Nevada, New Mexico, and Utah. The necessity of capturing the data and coordinating the effort between five states led to the development of a regional database system that collected wildlife habitat relationships, created habitat models, and provided for model modification. The objective was to provide a dynamic mapping application that met objectives of the SWReGAP project and potential end users. Database design was further driven by the need for documenting model attributes and model modifications.

The number of modeled species compounded both the effort and complexity needed for the database and modeling process. The database system had to work with wildlife habitat relationships developed from 13 core datasets including land cover, elevation, aspect, slope, distance to hydrological features (springs, streams, and lakes), 10-class landform, and soils. The system also had to identify all species ranges to 8-digit hydrologic cataloging units (HUCS) and mountain ranges for specific species. A standard set of state and regional references, available peer-reviewed articles, and grey literature had to be captured.

The database system required several additional functional characteristics. The database had to allow multiple users to edit the database simultaneously, provide users with common biological terminology for use in model creation and editing, generate reports of model inputs, and provide a method to compile the data into a modeling language for use in habitat model production. The database system was created in Microsoft Access (Microsoft Corporation 2002) using Structured Query Language (SQL) and three linked databases. The system also uses ArcGIS (ESRI 2005) to produce spatial representations of input data. User interaction with the dataset was through a FrontEnd (Boykin et al. 2006 ) suite of programs that managed the data in the DataStore (Boykin et al. 2006 ), compiled GIS code, and executed the code.

The FrontEnd database provided a method for creating and editing model inputs in plain language with user-friendly forms that provide an interface for model creation. Data were manipulated through standard SQL statements directed to the DataStore which provided the repository for data. This design allowed multiple FrontEnd instances to run simultaneously depositing information into one common DataStore. This allowed the greatest flexibility because the FrontEnd is independent of the DataStore.

Habitat modelers logged into project machines remotely and created and modified models. The FrontEnd also created ERDAS graphical model (gmd) (Leica 2003 ) files to provide the connection between the dataset and creation of GIS Models. The GISEngine (Boykin et al. 2006) was created to link to the DataStore such that batch modeling jobs could be created and monitored. To facilitate the modeling process, two model input datasets were created for the project with a coarse resolution (240-m) for model review and final resolution (30-m) datasets. The 240-m dataset derived models were generated within 5 minutes whereas 30-m dataset derived models were generated in 1-3 hours.

SWReGAP Application

SWReGAP model creation by 19 habitat modelers was facilitated by our database system. Our database system also provided a method for experts to review range information, habitat data, and resulting models for each species. A map of each species’ range and a report describing the background and model attributes for each species were created directly from the database. The 240-m datasets provided the spatial model representation for review. Reviews were conducted in workshops or through the habitat modeling website with appropriate documentation. Over 80 reviewers from state and Federal wildlife agencies, university biologists, NGOs, and biological consultants evaluated 680 of 819 modeled species with a total of 1023 reviews received. Reviews identified models errors and additional references and data to enhance models. This information was provided back to habitat modelers to modify the models based on this review in context with the regional habitat of the species.

The resultant models represent the first regional habitat models for terrestrial vertebrate species at this resolution for the American Southwest. The database and associated GIS tools provide end-users functionality and the ability for model modification. The database also provided a consistent guideline for model development with the recognition that each species and model was different.

Future Applications

Models should be considered dynamic and the database provides two methods for model modification of the predicted habitat suitability models (suitable, unsuitable) developed for SWReGAP. The Erdas graphic model (.gmd) (Leica 2003) file can be modified in Erdas Imagine (Leica 2003) changing input correlates or weighting correlates if desired. Thus both the database and gmd file provide opportunity to incorporate habitat quality or preference.

Both methods also allow incorporation of other datasets that end-users may have specific to the area in question. The 8-digit HUC ranges can overpredict species ranges and finer scale HUCs (14-digit) may provide more accurate range delineation but increase delineation difficulty. Vegetation structure is an important habitat factor and the SWReGAP land cover legend provided partial structure (e.g. woodland, forest, etc.), but finer structure or succession datasets were not available, nor were microhabitat features. F actors such as prey sources, competition, or predation were also not included. Climate datasets (e.g. temperature, precipitation) were considered, but were not used because of incomplete species knowledge. Landscape metrics such as patch size, distance to habitat patches, and habitat edge were not included because of study scale and incomplete species knowledge, though patch size was captured in the database as a variable and could be included in future modeling efforts. These landscape metrics could also be used in a post-processing step with current models outputs.. Patch size information was available for single populations but limited throughout the entirety of the species’ range. Bat habitat models would benefit from mines and caves datasets but regional datasets were incomplete. State references range from detailed, authoritative compilations on reptiles and amphibians (Degenhardt et al. 1996) (Hammerson 1999), mammals (Fitzgerald et al. 1994) and birds (Andrews and Righter 1992), while others were dated general works. Furthermore future habitat modeling can include additional regional datasets (e.g. Level IV Ecoregions) with the database or GMD files.

The database is currently being used to modify existing models specifically for Clark County Nevada through a collaborative effort between New Mexico State University and the Environmental Protection Agency. Using the database, species models for 37 species will be revisited and modified as appropriate and new models generated for the area of interest to the Multi-species habitat conservation plan within Clark County. Additionally, working with the New Mexico Department of Game and Fish, the database was used to modify a model for mountain lion to reflect primary, secondary, and tertiary ranges of the mountain lion in New Mexico.

The database is the focus of ongoing workshops designed specifically for state wildlife agencies in 2007. Workshops are focused on setting up agency modeling platforms and providing the expertise to run models as needed with modifications. The next challenge for database use will include porting the database over into a web friendly platform such as MySQL (http://www.mysql.org/). This transition may provide added usability, as state agencies are limited because of software and funding.

Literature Cited

Andrews, R., and R. Righter. 1992. Colorado birds. Denver Museum of Natural History, Colorado.

Boykin, K.G., B.C. Thompson, R.A. Deitner, D. Schrupp, D. Bradford, L. O’Brien, C. Drost, S. Propeck-Gray, W. Rieth, K. Thomas, W.

Kepner, J. Lowry, C. Cross, B. Jones, T. Hamer, C. Mettenbrink, K.J. Oakes, J. Prior-Magee, K. Schulz, J. J. Wynne, C. King, J.

Puttere, S. Schrader, and Z. Schwenke. 2006. Predicted Animal Habitat Distributions and Species Richness. Chapter 3 in J.S. Prior-

Magee, ed. Southwest Regional Gap Analysis Final Report. U.S. Geological Survey, Gap Analysis Program, Moscow, ID. Available on-line at: http://fws-nmcfwru.nmsu.edu/swregap/.

Degenhardt, W.G., C.W. Painter, and A.H. Price. 1996. The amphibians and reptiles of New Mexico. University of New Mexico Press, Albuquerque.

Environmental Systems Research Institute, Inc. 2005 . ArcGis (Version 9.1), Redlands, California.

Fitzgerald, J.P., C.A. Meaney, and D.M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History, Denver and University Press of Colorado, Niwot, Colorado.

Hammerson, G.A. 1999. Amphibians and reptiles of Colorado. 2 nd edition. University Press of Colorado, Niwot and Colorado Division of Wildlife, Denver.

Leica Geosystems GIS and Mapping LLC. 2003. ERDAS IMAGINE (Version 8.7). Heerbrugg, Switzerland.

Microsoft Corporation. 2002. Microsoft Office Access 2002. Microsoft Corporation.

Return to Table of Contents