Abstract
Spatial capture–recapture (SCR) models have emerged as a robust method to estimate the population density of mobile animals. However, model evaluation has generally been based on data simulated from simplified representations of animal space use. Here, we generated data from animal movement simulated from a mechanistic individual‐based model, in which movement emerges from the individual's response to a changing environment (i.e., from the bottom‐up), driven by key ecological processes (e.g., resource memory and territoriality). We drew individual detection data from simulated movement trajectories and fitted detection data sets to a basic, resource selection and transience SCR model, as well as their variants accounting for resource‐driven heterogeneity in density and detectability. Across all SCR models, abundance estimates were robust to multiple, but low‐degree violations of the specified movement processes (e.g., resource selection). SCR models also successfully captured the positive effect of resource quality on density. However, covariate models failed to capture the finer scale effect of resource quality on detectability and space use, which may be a consequence of the low temporal resolution of SCR data sets and/or model misspecification. We show that home‐range size is challenging to infer from the scale parameter alone, compounded by reliance on conventional measures of “true” home‐range size that are highly sensitive to sampling regime. Additionally, we found the transience model challenging to fit, probably due to data sparsity and violation of the assumption of normally distributed inter‐occasion movement of activity centers, suggesting that further development of the model is required for general applicability. Our results showed that further integration of complex movement into SCR models may not be necessary for population estimates of abundance when the level of individual heterogeneity induced by the underlying movement process is low, but appears warranted in terms of accurately revealing finer scale patterns of ecological and movement processes. Further investigation into whether this holds true in populations with other types of realistic movement characteristics is merited. Our study provides a framework to generate realistic SCR data sets to develop and evaluate more complex movement processes in SCR models.