You are here:
SMALL MAMMALS: CONSEQUENCES OF STOCHASTIC DATA VARIATION FOR MODELING INDICATORS OF HABITAT SUITABILITY FOR A WELL-STUDIED RESOURCE
Citation:
Jorgensen*, E E. SMALL MAMMALS: CONSEQUENCES OF STOCHASTIC DATA VARIATION FOR MODELING INDICATORS OF HABITAT SUITABILITY FOR A WELL-STUDIED RESOURCE. Elsevier (ed.), ECOLOGICAL INDICATORS 1(4):313-321, (2002).
Description:
Increasingly, models of physical habitat variables (i.e. vegetation, soil) are utilized as indicators of small mammal habitat suitability or quality. Presumably, use of physical habitat models indicating habitat suitability or quality would be improved and enhanced by the extensive amount of research that has been conducted for these species. However, current knowledge of small mammal habitat association is based mostly upon site specific empirical observation, not the quantitative data that are the foundation of habitat modeling. Small mammal data are affected by technique-related capture variability. Existing data do not demonstrate that fine spatial and temporal variability associated with technique substantially affects habitat models. Small mammal abundance also exhibits ecologically important high spatial and temporal variability. Microhabitat spatial variability is a poor predictor of trap use compared to larger spatial scale phenomena in mesic and xeric biomes. Habitat suitability has been modeled with accuracy ranging between 80 and 93% for 14 species. This accuracy is achieved by filtering out stochastic temporal variability that was not related to physical habitat indicators. Inclusion of stochastic temporal variability degrades model performance, especially in ecosystems where predictive vegetation and substrate variables are essentially constant. Development of indicator mathematics and biology are equally incorporated when conceiving and developing indicators.