Abstract
Ecological surveys rarely achieve perfect detection of target species, and failure to account for imperfect detection can produce erroneous estimates of abundance. N-mixture models account for variation in detectability by separating the observation process from the ecological process that determines true site-level abundance, making these models theoretically well-suited to studies of inconspicuous species, such as reptiles. Multiple N-mixture model variants have been published, but little is known about their ability to provide ecologically realistic abundance estimates from real-world observation data. Given their novelty and potential for wider use, studies that help users decide which variant to use in a particular case would be valuable. If different, yet data-appropriate N-mixture model variants provide substantially incongruent abundance estimates for the same dataset, then their uncritical use in ecology is problematic. Using a dataset of reptile observations from south-eastern Zimbabwe, we compare the estimates of five N-mixture model variants. For each species, we assess congruence between the site-level abundance estimates of each variant. We then use a novel metric to assess the performance of each model variant based on the precision and ecological feasibility of its abundance estimates, accounting for goodness-of-fit. We find that model variant pairs were rarely congruent in their abundance estimates, and that model performance varies significantly according to species occupancy and detection probability. We provide a framework for the application of multiple N-mixture model variants in faunal ecology to guide analytical decision-making.