In Washington State, USA, mountain goats (Oreamnos americanus) have experienced a long-term population decline. To assist management, we created annual and seasonal (summer and winter) habitat models based on 2 years of data collected from 38 GPS-collared (GPS plus collar v6, Vectronic-Aerospace GmbH, Berlin, Germany) mountain goats in the western Cascades. To address GPS bias of position acquisition, we evaluated habitat and physiographic effects on GPS collar performance at 543 sites in the Cascades. In the western Cascades, total vegetation cover and the quadratic mean diameter of trees were shown to effect GPS performance. In the eastern Cascades, aspect and total vegetation cover were found to influence GPS performance. To evaluate the influence of bias correction on the analysis of habitat selection, we created resource selection functions with and without bias correction for mountain goats in the western Cascades. We examined how well the resultant habitat models performed with reserved data (25% of fixes from 38 study animals) and with data from 9 other GPS-collared mountain goats that were both temporally and spatially independent. The statistical properties of our GPS bias correction model were similar to those previously reported explaining between 20 and 30% of the variation, however, application of bias correction improved the accuracy of the mountain goat habitat model by only 1–2% on average and did not alter parameter estimates in a meaningful, or consistent manner. Despite statistical limitations, our habitat models, most notably during the winter, provided the widest extent and most detailed models of the distribution of mountain goat habitat in the Cascades yet developed.
Required Publisher's Statement
Remote Sensing ISSN 2072-4292 www.mdpi.com/journal/remotesensing
Wallin, David O.; Wells, Adam G.; Rice, Clifford G.; and Chang, Wan-Ying, "GPS Bias Correction and Habitat Selection by Mountain Goats" (2011). Environmental Sciences Faculty Publications. 14.
Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.