| Summary: | Imprecision and misclassification of land cover types are two issues commonly encountered in habitat selection studies using satellite land cover classifications and telemetry data. Here, the utility of broad land cover types is explored in a study of habitat selection by woodland caribou (Rangifer tarandus caribou). Broad land cover types have potential to reduce the misclassification error associated with finer land cover types, while remaining relevant to the factors influencing habitat selection in the species of interest. Lichen abundance and snow accumulation are two factors important in explaining the selection of land cover types by woodland caribou in winter, and they are used here to predict the probability of occupation by caribou of land cover types in three regions of the boreal forest in Eastern Canada. Land cover types were initially categorized using Landsat EOSD land cover data, and field surveys were conducted to measure terrestrial and arboreal lichen abundance in each land cover type. The relative accumulation of snow was modeled for land cover types using documented patterns of snow distribution in the boreal forest as well as data collected in the Greater Gros Morne Ecosystem, Newfoundland, and the Côte-Nord region, Quebec. Subsequently, land cover types were collapsed into three (dense forest, sparse-open forest, and non forest) that reflected differences in lichen abundance and snow accumulation while reducing misclassification errors. Resource selection functions were estimated using logistic regression where GPS and Argos satellite telemetry data existed for caribou in the Greater Gros Morne Ecosystem and Middle Ridge regions of Newfoundland and the Côte-Nord region of Quebec. In all regions, telemetry-monitored caribou selected nonforested areas, where lichen abundance was high and snow accumulation was low, more than expected by chance. The similarities in selection of non-forested areas across regions despite variation in landscape composition indicates that there are congruencies both in the factors influencing winter habitat selection and in the relative value of land cover types on a given landscape. These findings support the argument that resource selection functions with parameters based on broadly defined land cover types are applicable among different regions of caribou occurrence and are therefore a valuable tool for understanding patterns of space use in caribou throughout the boreal forest.
|
|---|