Understanding animals' space requirements is a core goal of ecology and is a key reason for collecting relocation data. The amounts, types, and qualities of the various habitats an individual occupies will all influence its fitness. As remaining natural areas decrease in size and become increasingly fragmented and disturbed, an understanding of species' space requirements is critical for designing effective conservation strategies. The importance of quantifying space use was recognized early in the development of ecology and has led to the concepts of “home range” and “utilization distribution”. Space use, however, is a multifaceted aspect of animal behavior. Consequently, the home range concept has broadened substantially over time and a large and diverse literature on the subject now exists. This proliferation of concepts and methods has led to fundamentally different definitions of space use being conflated in the literature.
Results/Conclusions
To cut through this complexity, we show that many of these ideas cluster around two distinct notions of the utilization distribution. The range distribution describes the long-run behavior of a space-restricted movement process, is consistent with the classical definition of the home range, and is estimated by kernel density estimation (KDE), autocorrelated KDE, and minimum convex polygons. In contrast, the occurrence distribution quantifies uncertainty in the location of an individual during the observation period, is not directly related to the classical notion of the home range, and is estimated by time-series Kriging and the Brownian bridge. Using mathematical arguments, simulated data, and a multispecies tracking dataset of 20 range resident species, we demonstrate that the range and occurrence distributions quantify profoundly different aspects of animal space use. Specifically, we show that: 1) the occurrence distribution estimate is generally much smaller than the range distribution estimate on the same data, and 2) the occurrence distribution collapses to the movement path in the limit of finely sampled data, while the range distribution does not. Finally, we describe how these contrasting notions of space use can and should be used to answer different ecological questions.