A demographic, spatially explicit patch occupancy model of metapopulation dynamics and persistence

Patch occupancy models are extremely important and popular tools for understanding the dynamics, and predicting the persistence, of spatially structured populations. Typically this endeavor is facilitated either by models from classic metapopulation theory focused on spatially explicit, dispersal-driven colonization-extinction dynamics and generally assuming perfect detection, or by more recent hierarchical site occupancy models that account for imperfect detection but rarely include spatial effects, such as dispersal, explicitly. Neither approach explicitly considers local demographics in a way that can be used for future projections. However, despite being arguably of equal importance, dispersal and connectivity, local demography, and imperfect detection are rarely modeled explicitly and simultaneously. Understanding the spatiotemporal occurrence patterns of spatially structured populations and making biologically realistic long-term predictions of persistence would benefit from the simultaneous treatment of space, demography, and detectability. We integrated these key ideas in a tractable and intuitive way to develop a demographic and spatially realistic patch occupancy model that incorporates components of dispersal, local demographic stage-structure, and detectability. By explicitly relating stage-specific abundances to measurable patch properties, biologically realistic projections of long-term metapopulation dynamics could be made. We applied our model to data from a naturally fragmented population of water voles Arvicola amphibius to describe observed patch occupancy dynamics and to investigate long-term persistence under scenarios of elevated stage-specific local extinction. Accounting for biases induced by imperfect detection, we were able to estimate: stable, and higher than observed metapopulation occupancy; high rates of patch turnover and stage-specific colonization and extinction rates (juvenile and adult, respectively); and juvenile dispersal distances (average 2.10 km). We found that metapopulation persistence in the presence of elevated extinction risk differed depending on which life stage was exposed, and was more sensitive to elevated juvenile rather than adult extinction risk. Predictions of persistence when dynamics are stage-specific suggest that metapopulations may be more resilient to changes in the environment than predicted when relationships are based on patch size approximations rather than population sizes. Our approach allows explicit consideration of local dynamics and dispersal in spatially structured and stage-structured populations, provides a more detailed mechanistic understanding of metapopulation functioning, and can be used to investigate future extinction risk under biologically meaningful scenarios.