Estimating site occupancy rates when detection probabilities are less than one

MacKenzie and colleagues introduced the occupancy modeling framework that revolutionized how ecologists analyze presence-absence and camera-trap detection-nondetection data. They recognized that failing to detect a species at a site does not prove absence, and built a likelihood that simultaneously estimates the probability a site is occupied and the per-visit probability of detection given occupancy. Repeated surveys at each site provide the information to separate these parameters. For camera trapping, where rare cats are easily missed, this corrects the systematic underestimation that plagues naive occupancy maps. The model accommodates site and survey covariates, enabling inference about habitat associations. It is among the most influential statistical papers in wildlife ecology and underpins the majority of occupancy-based felid distribution studies.