Modeling three-dimensional space use and overlap in birds

Abstract

How animals use space has fundamental behavioral and ecological implications. Utilization distributions are among the most common methods for quantifying space use and have advanced our knowledge of animal ecology in a variety of ways. However, until recently, they were limited to 2 spatial dimensions (2D), despite the fact that most taxa use their environments in all 3 dimensions (3D). We (1) created 3D utilization distributions via a multivariate kernel density estimator, (2) adapted 2 overlap indices for use with 3D data, (3) estimated the minimum sample sizes required for accurate estimation of territory size, and (4) assessed these methods using data from American Redstarts (Setophaga ruticilla) during their nonbreeding season in Jamaica. We found that, compared to 3D methods, 2D methods overestimated individual (pairwise) spatial overlap by 3% in scrub habitat and 4% in mangrove habitat. Similarly, 2D methods overestimated total (all neighbors combined) spatial overlap by 9% in scrub and 12% in mangrove habitat. This indicates that American Redstarts may partition territorial space in all 3 spatial dimensions. Moreover, using overlap indices, we found that American Redstarts may avoid areas of overlap, possibly to limit agonistic interactions with neighbors. Although 3D methods require larger sample sizes (80–110 locations) than 2D (40–70 locations), we argue that modeling animal space use in 3D is more realistic and will enhance understanding of niche differentiation, interspecific and intraspecific competition, habitat selection and use, and wildlife conservation.