Evaluation of resource selection methods with different definitions of availability

Because resource selection is of paramount importance to ecology and management of any species, we compared 6 statistical methods of analyzing resource selection data, given the known biological requirements of radiomarked Merriam's wild turkey (Meleagris gallopavo merriami) hens with poults in the Black Hills of South Dakota. A single variable, habitat, was used for the comparisons. Hen turkeys with poults are ideal for these comparisons because poults have narrow dietary protein requirements, and habitats that meet these requirements are well documented. We also tested the effects on the statistical methods of varying the habitats considered as available. Availability of habitats was defined by 100-, 200-, and 400-m-radii buffers around individual poult locations and also as the total study area. The 6 statistical techniques gave different results because of how the methods addressed variability in habitat use of individual birds and how availability was defined. These differences made identification of a "best" technique difficult. Only 1 method (Neu's) identified habitat selection patterns consistent with known requirements of poults at all levels of habitat availability. However, defining availability at the study-area level was generally superior to use of circular buffers. We recommend that during analyses of resource selection researchers include the study-area level of habitat availability based on the distribution of radiomarked animals.