Density‐dependent home‐range size revealed by spatially explicit capture–recapture

Abstract

The size of animal home ranges often varies inversely with population density among populations of a species. This fact has implications for population monitoring using spatially explicit capture–recapture (SECR) models, in which both the scale of home‐range movements σ and population densityDusually appear as parameters, and both may vary among populations. It will often be appropriate to model a structural relationship between population‐specific values of these parameters, rather than to assume independence. We suggest re‐parameterizing the SECR model usingkp= σp√Dp, wherekprelates to the degree of overlap between home ranges and the subscriptpdistinguishes populations. We observe thatkpis often nearly constant for populations spanning a range of densities. This justifies fitting a model in which the separatekpare replaced by the single parameterkand σpis a density‐dependent derived parameter. Continuous density‐dependent spatial variation in σ may also be modelled, using a scaled non‐Euclidean distance between detectors and the locations of animals. We illustrate these methods with data from automatic photography of tigersPanthera tigrisacross India, in which the variation is among populations, from mist‐netting of ovenbirdsSeiurus aurocapillain Maryland, USA, in which the variation is within a single population over time, and from live‐trapping of brushtail possumsTrichosurus vulpeculain New Zealand, modelling spatial variation within one population.Possible applications and limitations of the methods are discussed. A model in whichkpis constant, while density varies, provides a parsimonious null model for SECR. The parameterkof the null model is a concise summary of the empirical relationship between home‐range size and density that is useful in comparative studies. We expect deviations from this model, particularly the dependence ofkpon covariates, to be biologically interesting.