ScienceBase Catalog
Assessing Wildlife Vulnerability to Energy Development

Information derived from USGS Science for WLCI 2010 Annual Report.

Catalog Item:
  Created by: on Thu Mar 01 11:41:59 MST 2012
  Last Updated by: Auto contactType updater. on Thu Jan 17 16:20:45 MST 2013


Scope and Methods

The Assessing Wildlife Vulnerability to Energy Development (AWVED) research task was established to help prioritize the management, monitoring, and research needs of Wyoming’s SGCN, which are listed in Wyoming’s Comprehensive Wildlife Conservation Strategy (CWCS; Wyoming Game and Fish Department, 2005). The first step in this multi-year process was to develop Wyoming-specific range map for terrestrial vertebrate SGCN, which was completed in FY2009. The second step was to develop detailed distribution models for all species that refine where they are most likely to occur within their ranges, which was completed in FY2010. The next step (currently ongoing) is to develop maps of current and potential future energy development and assess how that development coincides with the predicted distribution for each species.

In May 2008, representatives of State and Federal entities met to coordinate range mapping and modeling of Wyoming SGCN and accepted the approach for range mapping and distribution modeling developed by the Wyoming Natural Diversity Database (WYNDD) as the standard. Distribution models were generated by statistically extrapolating the environmental characteristics of locations where species have been documented to occur to other areas potentially suitable for occupation (for example, Elith and others, 2006; Greaves and others, 2006; Phillips and others, 2006; Guisan and Thuiller, 2007). The basic components of creating these “environmental niche models” are occurrence data collection and processing, environmental data collection and processing, and model generation and validation. We compiled a dataset of approximately 260,000 individual records for 159 species and stored them in a geodatabase that was queried as needed for analysis and modeling. Observations varied greatly in their quality, and were not of equal value for constructing niche models, so we computed a point quality index for each record (see Keinath and others, 2010), which we used to filter data prior to modeling. We minimized the impact of autocorrelation artifacts arising from non-uniform sampling across the area of interest (Jimenez-Valverde and Lobo, 2006, Johnson and Gillingham, 2008) by using target-group background data for model building (Phillips and others, 2009). We further used a multi-pass filtering technique to construct a minimally-biased modeling dataset for each species (see Keinath and others, 2010). Environmental data layers used in modeling generally fell within six categories: climate, hydrology, land cover, landscape structure, substrate, and terrain (see Appendix 2 in Keinath and others, 2010, for explanation variables). Maximum Entropy methods were used to identify pertinent predictor variables for each species and to generate distribution models (for example, Phillips and others, 2006; Phillips and Dudik, 2008), as they have been consistently shown to be among the most accurate and robust algorithms for constructing niche models from opportunistically collected data, particularly with small sample sizes (Graham and Elith, 2005; Hijmans and Graham, 2006; Graham and others, 2008; Wisz and others, 2008). To avoid biases associated with any one validation technique, we evaluated models quantitatively and qualitatively using multiple methods, including prediction accuracy based on ten-fold cross-validation, statistics derived from receiver operating characteristic analyses, evaluations of input data quality, and the expert opinion of biologists regarding how well final models reflected their understanding of species’ distributions (for example, Fielding and Bell, 1997; Freeman and Moisen, 2008).


  • Focus conservation attention on the most vulnerable species before they become imperiled by assessing relative risks from energy related disturbances based on geospatial estimates of exposure and evaluation of biological sensitivities.

Data Owner: U.S. Geological Survey
Contact: Doug Keinath, Matthew J Kauffman


Files stored in ScienceBase: (Download Attached Files)

thumbnail Wildlife_vulnerability.jpg 51 KB image/jpeg
thumbnail Wildlife_vulnerability_map.jpg 147 KB image/jpeg
Additional Information

Great Northern Landscape Conservation Cooperative
LC MAP - Landscape Conservation Management and Analysis Portal
Wyoming Landscape Conservation Initiative

Related Items
Accessibility | FOIA | Privacy | Policies and Notices
Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
URL: Page Contact Information:
ScienceBase Version: 2.54.1 (2014/04/14 10:27 build#v2.54-16-g3b098f3)  |  Feedback  |  Support  |  Twitter