Evaluating individual variability in species‐habitat relationships using an integrated step-selection analysis

Date:

More information here

Variability among individuals in their habitat use and resource-selection patterns has important implications for ecology, conservation and behaviour. There have been substantial improvements in the scope of species-habitat modelling since the proliferation of radio-tracking technology. Integrated step-selection analyses allow researchers to model both movement and habitat selection processes using a 2-step approach where movement parameters are estimated in the absence of habitat selection, then updated once habitat selection is quantified using Conditional Logistic Regression. Recent studies have shown how random effects could be used to model individual variability when conducting an integrated step-selection analysis, but existing examples largely focus on quantifying variability in habitat selection rather than variability in movement parameters. In this study, we explore several different approaches for quantifying among-individual variability in the movement parameters using an acoustic telemetry dataset collected from red snappers (n=35) along the coast of North Carolina, USA. We consider movement and habitat use at individual and population scales as a function of reef class and distance to reef edge, comparing results from fitting models to individuals to those from a mixed-effects model. We detected substantial variation in individual coefficients with both approaches, but estimates for individuals were shrunk back towards the population mean when using random effects. We discuss the potential benefits of the mixed-effect approach and outline the steps needed to update the movement parameters when conducting a mixed-effects step-selection analysis.