Bayesian occupancy models are increasingly being used in ecology to model the spatial distribution of imperfectly detected species. These models allow researchers to study the effect that covariates have on presence of a quantity of interest, while accounting for imperfect detection and spatial dependence. In this talk, we introduce various frameworks for Bayesian occupancy models and discuss advantages and limitations of each. We present results that indicate the choice of model has a large impact on estimation of covariate effects. This work is applied to data from the citizen science project Snapshot Serengeti - a large camera-trap study in Serengeti National Park in Tanzania. We apply Bayesian occupancy models to identify drivers of spatial variability of herbivores.