Spatially realistic dynamic N‐mixture models

Dispersal is a complex ecological process that can be driven by the abiotic environment, density‐dependent processes and species mobility. Traditionally, dispersal has been studied with capture‐based data that are often limited in space and time. More recently, spatial dynamic N‐mixture models (i.e. N‐mixture models accounting for dispersal among local populations and changes in abundance over time) allow the study of dispersal with count data. However, current spatial dynamic N‐mixture models focus mainly on distance‐based dispersal, limiting our ability to evaluate realistic patterns and drivers of dispersal at relevant spatial and temporal scales.

We developed a spatially realistic dynamic N‐mixture model that allows dispersal to be driven by environmental conditions and population density. Using simulations, we evaluated the inferential performance of this model when large proportions of sites are unsurveyed and the survey period is short, situations that are typical of real‐world studies. We then used two case studies to illustrate the applications of this model, one representing a species with limited mobility in fragmented habitat (Chiricahua Leopard Frogs, Lithobates chiricahuensis , in isolated ponds) and the other representing a species with great mobility in continuous heterogeneous habitat (Common Nighthawks, Chordeiles minor , in grasslands).

Our simulation study showed that the model provides valid inference even when the number of surveyed sites is small () or when the survey period is short for dynamic models (5 years), indicating that the model could have great practical value in real‐world studies. The case study of Chiricahua Leopard Frogs showed that the model is capable of revealing pond‐type‐specific dispersal and predicting dispersal among ponds. The case study of Common Nighthawks showed the ability of the model to reveal the effect of primary productivity on dispersal and to map dispersal across the study area.

Our effort demonstrates the potential of using count data to gain insights about complex yet realistic dispersal processes. The model that we introduce is flexible and adaptable for a broad range of situations in population and conservation ecology.