Abstract
Obtaining unbiased estimates of wildlife distribution and abundance is an important objective in research and management. Occupancy and N‐mixture abundance models, which correct for imperfect detection, are commonly used for this purpose. Fitting these models in a Bayesian framework has advantages but doing so can be challenging and time‐consuming for many researchers.
We developed an R package, ubms, which provides an easy‐to‐use, formula‐based interface for fitting occupancy, N‐mixture abundance and other models in a Bayesian framework using Stan. The package also provides tools for visualizing parameter effects, calculating residuals, assessing goodness‐of‐fit and comparing models.
We demonstrate the use of ubms by fitting an N‐mixture model to ruffed grouse Bonasa umbellus count data from drumming surveys conducted at roadside points sampled on five occasions annually during 2013–2015. To demonstrate the functionality of ubms, we used survey site as a random effect, and occasion date and per cent aspen cover at each site as covariates of detection and abundance respectively. The top‐ranked model included a positive effect of per cent aspen on grouse abundance.
ubms has the potential to greatly increase the range of users who will be able to rigorously assess species distribution and abundance while correcting for imperfect detection in a Bayesian framework.