Dynamic N-occupancy models: estimating demographic rates and local abundance from detection-nondetection data
Please contact the first author for questions about the code or data: Sam Rossman (rossmans@msu.edu)
Occupancy modeling is a widely used analytical technique for assessing species distributions and range dynamics. However, occupancy analyses frequently ignore variation in abundance of occupied sites, even though site abundances affect many of the parameters being estimated (e.g., extinction, colonization, detection probability). We introduce a new model (“dynamic N‐occupancy”) capable of providing accurate estimates of local abundance, population gains (reproduction/immigration), and apparent survival probabilities while accounting for imperfect detection using only detection/nondetection data. Our model utilizes heterogeneity in detection based on variations in site abundances to estimate latent demographic rates via a dynamic N‐mixture modeling framework. We validate our model using simulations across a wide range of values and examine the data requirements, including the number of years and survey sites needed, for unbiased and precise estimation of parameters. We apply our model to estimate spatiotemporal heterogeneity in abundances of barred owls (Strix varia) within a recently invaded region in Oregon (USA). Estimates of apparent survival and population gains are consistent with those from a nearby radio‐tracking study and elucidate how barred owl abundances have increased dramatically over time. The dynamic N‐occupancy model greatly improves inferences on individual‐level population processes from occupancy data by explicitly modeling the latent population structure.
Barred Owl Case Study: Contains data and code for analysis of barred owl population dynamics
Simulation Study: Contains code to run model simulations as well as simulations results and code to query results
See the following subdirectories for data and metadata: Barred Owl Case Study, Simulation Study
See the following subdirectories for data and metadata: Barred Owl Case Study, Simulation Study