Home-range estimation within complex restricted environments: importance of method selection in detecting seasonal change

摘要

Estimating the home ranges of animals from telemetry data can provide vital information on their spatial behaviour, which can be applied by managers to a wide range of situations including reserve design, habitat management and interactions between native and non-native species. Methods used to estimate home ranges of animals in spatially restricted environments (e.g. rivers) are liable to overestimate areas and underestimate travel distances by including unusable habitat (e.g. river bank). Currently, few studies that collect telemetry data from species in restricted environments maximise the information that can be gathered by using the most appropriate home-range estimation techniques. Simulated location datasets as well as radio-fix data from 23 northern pike (Esox lucius) were used to examine the efficiency of home-range and travel estimators, with and without correction for unusable habitat, for detecting seasonal changes in movements. Cluster analysis most clearly demonstrated changes in range area between seasons for empirical data, also showing changes in patchiness, and was least affected by unusable-environment error. Kernel analysis showed seasonal variation in range area more clearly than peripheral polygons or ellipses. Range span, a linear estimator of home range, had no significant seasonal variation. Results from all range area estimators were smallest in autumn, when cores were least fragmented and interlocation movements smallest. Cluster analysis showed that core ranges were largest and most fragmented in summer, when interlocation distances were most variable, whereas excursion-sensitive methods (e.g. kernels) recorded the largest outlines in spring, when interlocation distances were largest. Our results provide a rationale for a priori selection of home-range estimators in restricted environments. Contours containing 95% of the location density defined by kernel analyses better reflected excursive activity than ellipses or peripheral polygons, whereas cluster analyses better defined range cores in usable habitat and indicate range fragmentation.