PARENT SESSION
Oral Session #26: Conservation Ecology: Terrestrial.
Presiding: P. Kleintjes
Tuesday, August 6. 8:00 AM to 11:30 AM. Greenlee Meeting Room, TCC.
GIS-based model of timber rattlesnake (Crotalus horridus) denning habitat using partitioned Mahalanobis D2.
BROWNING, DAWN1, BEAUPRE, STEVEN2, DUNCAN, LYNETTE*,3, 1 NEW MEXICO COOPERATIVE FISH AND WILDLIFE RESEARCH UNIT, LAS CRUCES, NM2 DEPARTMENT OF BIOLOGICAL SCIENCES, FAYETTEVILLE, AR3 CENTER FOR STATISTICAL CONSULTING, FAYETTEVILEE, AR
ABSTRACT- Mahalanobis Distance (D2) Statistic is a multivariate statistical method that has been used to model habitat suitability for wildlife species. The output, whether standardized squared distance or probability of occurrence, represents the similarity of a given range of values with those for known use sites. We partitioned D2 into contributions from specific principal components and used only principal components corresponding to "species requirements" to determine habitat suitability. This method was applied to model denning habitat of timber rattlesnakes on the Madison County Wildlife Management Area in northwest Arkansas using slope, aspect, elevation, and 11 physical soil characteristics as descriptor variables of 39 known rattlesnake dens. Resampling methods (bootstrap and crossvalidation) were used to examine the stability of the correlation matrix and examine the effect of each den on overall D2 values. Bootstrap and crossvalidation results along with interpretation of eigenvector loadings were used to identify "species requirements" or invariant aspects of the environment across all known use sites. The probability threshold of 0.25 maximized the model gain by capturing 74.4% of the known dens in only 25.3% of the 5666 ha study area. Partitioned Mahalanobis D2 uses only data for where species are known to occur, thus circumventing costly errors in misclassifying species use with commonly used multivariate techniques that require dichotomous data sets (i.e. logistic regression, discriminant function analysis).
KEY WORDS: GIS, habitat model, partitioned Mahalanobis D2 , timber rattlesnake (Crotalus horridus)