A simulation study of the age‐structured spatially explicit dynamic N‐mixture model

Knowledge of age‐specific movement and vital rates is important for understanding metapopulation dynamics yet difficult to obtain without capturing/marking individual animals. The development of dynamic N‐mixture models allows for the inference of recruitment and apparent survival while accounting for imperfect detection in count data of unmarked populations. Recent studies have further developed dynamic N‐mixture models to account for age structures or movement among local populations; however, there has yet to be a dynamic N‐mixture model that simultaneously accounts for both age structure and movement despite the fact that natural populations are composed of individuals of different ages with different movement and vital rates. In this study, I developed a dynamic N‐mixture model that allows different movement and vital rates between age classes while accounting for imperfect detection in age‐structured count data. I then conducted a simulation study to evaluate the inferential performance of the model while considering different local abundances, number of sites, and detection probabilities. The simulation study showed that the model could provide unbiased estimates of adult‐related parameters under a high detection probability, but bias was found for young‐related parameters regardless of detection probability. The bias in young‐related parameters also tended to be lower when local abundance was lower, probably due to more frequent extinction‐recolonization events in these populations. Overall, the results indicated that cautions should be taken when using dynamic N‐mixture models alone. However, these models may be useful sub‐models under integrated modeling frameworks, and thus improve our understanding of metapopulation dynamics.