Scripts and Data: When and how data integration improves occupancy estimates for low-density and elusive species

Understanding the distributions and habitat associations of species of conservation concern is critical for informing effective management. This project investigates whether supplementing structured marsh bird surveys with participatory-science data can enhance inference on occupancy and detection probabilities across Pennsylvania, USA — where numerous marsh bird species are classified as species of greatest conservation need. We employed Bayesian multi-species integrated data occupancy models (fit in NIMBLE) to estimate state-wide marsh bird occupancy using four data sources: systematic structured surveys (WPC), Pennsylvania Game Commission surveys (PGC), volunteer solicited surveys (Vol), and participatory-science data (eBird). Results indicate that combining broad-scale participatory data with at least one broad-scale structured survey performs similarly to fully integrated models, suggesting a cost-effective monitoring foundation for wildlife managers working with limited resources. The main analysis focused on integrating four data sources (PGC, Vol, eBird, and WPC), and a secondary analysis compared utilizing WPC data as an informative prior in the detection process (respectively integrated and inform/vauge).