Comparing Naïve Occupancy Versus Modeled Occupancy to Monitor Declines in Rare Species

Monitoring changes in occupancy (i.e., probability a site has at least one individual of a species) across time is considered an inexpensive alternative to monitoring changes in abundance and can be used to monitor multiple species simultaneously across a watershed. Occupancy can be measured as the proportion of sites where a species is detected during surveys (i.e., naïve occupancy), but is more commonly modeled by surveying sites multiple times to estimate detection probability and address false-positive survey errors (sites that are occupied but with no survey detections of the species). This results in an unbiased estimate of occupancy, but at the expense of more effort. The purpose of this study was to determine management implications of using naïve occupancy versus using modeled occupancy. We generated simulated data to represent monitoring a population, then compared performance of using naïve occupancy vs. modeled occupancy for detecting changes. Different sampling scenarios were compared using different values of catchability (0.05 to 0.70) and various levels of known occupancy decline (35%, 55%, and 85%). Power to detect declines in both naïve occupancy and modeled occupancy increased with higher catchability and greater declines. Naïve occupancy and modeled occupancy performed similarly when catchability was high. Modeled occupancy performed slightly better than naïve occupancy at lower catchability; however, at a catchability of 0.05, neither occupancy approach was successful at correctly estimating the correct decline. Although modeled occupancy provides more accurate estimates of species occupancy, results of our study indicate that regulatory agencies concerned with personnel constraints could likely use a naïve occupancy approach to maximize geographical coverage without sacrificing their ability to correctly assign conservation status to imperiled species.