Monitoring populations across broad spatio-temporal extents often necessitate the use of indices of abundance which contain observation error and whose relationship with abundance is generally unknown. Recent statistical advances allow the estimation of observation error and high concordance among indices from different sampling methods may suggest that such trends accurately reflect trends in abundance. Broad-scale data for many mesopredator mammals are often available from count- and harvest-based sampling methods. However, harvest-based indices may also be influenced by variation in hunter/trapper numbers which may obscure their relationships with abundance. We compared statewide trends from six mesopredators mammals in Illinois, raccoon (Procyon lotor), skunk (Mephitis mephitis), opossum (Didelphis virginiana), coyote (Canis latrans), red fox (Vulpes vulpes), and gray fox (Urocyon cinereoargenteus), using multiple decades of data. We used data from three count-based methods (roadkill and spotlight surveys and archery deer hunter observations) and annual trapper harvest surveys to calculate 4–6 indices per species. We then used Bayesian state-space models to estimate each index while accounting for observation error. Count-based indices were concordant for raccoon and opossum but not for skunk. Annual trapper harvest was concordant with count-based indices for the three canid species but not raccoon, opossum, and skunk. Harvest-based indices that accounted for changes in trapper numbers generally produced greater concordance among count- and harvest-based indices although concordance was lowest for opossum and skunk. We had generally greater estimates of observation error for count-based indices than for harvest-based indices. Because the degree of index concordance may be difficult to determine a priori, we suggest there is value in using data from multiple sampling methods to evaluate mesopredator trends. We recommend data be collected in a manner that allows managers to account for sources of observation error (e.g., imperfect detection, inter-observer variability).
|Original language||English (US)|
|Title of host publication||Midwest Fish and Wildlife Conference 2020|
|State||Published - 2020|