Integrating multiple surveys using state–space models improves inference of population trends for Illinois furbearers

Long-term population monitoring is fundamental to wildlife population management. Furbearer populations are often surveyed using multiple methods, but many estimates of their abundance, especially abundance indices, do not account for detection error, and lead to conflicting interpretations of population trends. We developed a series of Bayesian state-space models to estimate detection error in long-term survey data of northern raccoons Procyon lotor, Virginia opossums Didelphis virginiana, and striped skunks Mephitis mephitis in Illinois between 1979–2023. We examined abundance indices from four survey types, including spotlight surveys, archery deer hunter observations, trapper harvest and a roadkill index. For each species, we developed four state–space models estimating population growth rate (r) and relative abundance from each survey individually. We also developed one integrated model combining the four datasets for each species. We found that integrated models typically had better precision on estimates of r and relative abundance, whereas the precision of estimates from individual models varied. When simulating data, integrated models detected a long-term 2% growth rate (10–20% change in relative abundance) within 10 years, but individual models needed 10–40 years, depending on the survey and species. Raccoon models indicated that r was greater in counties with lower housing density and more forest cover, but for skunks, the effect of agricultural and forest cover varied by survey. Integrated models reduce uncertainty and improve precision on population estimates when only indices of abundance are available, but they are not a replacement for data on absolute abundance. These models can improve precision and help resolve conflicting long-term trends from abundance indices, especially for stable or increasing populations.