Simultaneous estimation of occupancy and detection probabilities: an illustration using Cincinnatian brachiopods

Abstract

Preservation in the fossil record is never perfect in the sense that we cannot sample all individuals of a given population in time and space. Incomplete detection (i.e., preservation and modern-day sampling of fossils) often affects estimates of other paleobiological parameters of interest, such as occupancy and turnover. Here, I simultaneously model the occupancy and detection probability of taxa, teasing apart the zeros in data that reflect true absences and those that imply non-detection of taxa that were actually present in the space and time of interest. Occupancy modeling, an approach first developed in population ecology, can easily incorporate covariates of interest, such as sampling effort and habitat variables. I use a data set of brachiopod taxa from the Paleozoic to illustrate the utility of this approach for paleontological questions. I demonstrate a range of models, including those that allow colonization between time intervals and those that incorporate facies as site covariates. I also suggest how future data collection can be improved so that process- and sampling-oriented approaches such as occupancy modeling can be applied with ease to paleobiological settings to answer important paleoecological and evolutionary questions.