Chapter 10 - Using Euclidean Distances to Assess Nonrandom Habitat Use

Abstract

This chapter presents a multivariate technique for evaluating habitat use based on euclidean distances. The technique uses a multivariate analysis of variance to compare euclidean distances between animal locations and habitats of interest to expected distances derived from random locations. The procedure uses the animal as the sampling unit; thus individual variation among animals is not assumed to be constant, does not suffer from a unit sum constraint, allows for comparison among meaningful groups, and works at multiple spatial scales, making the test similar to compositional analysis. However, habitat types that are not used by the animal do not result in undefined value, as they do in compositional analysis. Statistical tools for evaluating habitat use are important for understanding habitat ecology of wildlife species. Habitat-use metrics generally rely on some type of comparison between habitat use and habitat availability, and many approaches are multinomial. The chapter also discusses desirable characteristics of a habitat metric and suggests a simple method for analysis of habitat-use data that incorporates desirable characteristics. The procedure is suitable whenever a multinomial approach is employed. As a tool for assessing habitat use, the animal should be used as the sampling unit, permit hypothesis testing among meaningful groups, work at multiple spatial scales, and allow for the nonindependence of habitat proportions.