Quantifying the conservation status and abundance trends of wildlife communities with detection-nondetection data
Matthew T. Farr, Timothy O'Brien, Charles B. Yackulic, & Elise F. Zipkin
Please contact the first author for questions about the code or data: Matthew T. Farr (matthewtfarr@gmail.com)
Effective conservation requires understanding species’ abundance patterns and demographic rates across space and time. Ideally, such knowledge should be available for whole communities, as variation in species’ dynamics can elucidate factors leading to biodiversity losses. However, collecting data to simultaneously estimate abundance and demographic rates is often prohibitively time-intensive and expensive for communities of species. We developed a “multi-species dynamic N-occupancy model” to estimate unbiased, community-wide relative abundance and demographic rates. Our model uses detection-nondetection data (e.g., repeated presence-absence surveys) to estimate both species- and community-level parameters as well as the effects of environmental factors. We conducted a simulation study that validated our modeling framework, demonstrating how and when such an approach can be valuable. Using data from a network of camera traps across tropical equatorial Africa, we then used our model to evaluate the statuses and trends of a forest-dwelling antelope community. We estimated relative abundance, rates of recruitment (i.e., reproduction and immigration), and apparent survival probabilities for each species’ local population. Our analysis indicated that the antelope community was fairly stable in this region (although 17% of populations [species-park combinations] declined over the study period), with variation in apparent survival linked more closely to differences among national parks rather than individual species’ life histories. The multi-species dynamic N-occupancy model requires only detection-nondetection data to evaluate the population dynamics of multiple sympatric species and can thus be a valuable tool for conservation efforts seeking to understand the reasons behind recent biodiversity loss.
DataAnalysis: Contains code for modeling, analysis, and results for both the simulation and case studies
DataFormat: Contains raw data, code to format raw data for analysis, and formatted data
PostAnalysis: Contains code to create tables & figures
SupportingInformation: Contains supporting information and code to generate supporting information
PublishedPDF: PDF of published paper
See the following subdirectories for data and metadata: DataFromatting/RawData
See the following subdirectories for code and metadata: DataAnalysis, DataFormatting, PostAnalysis, SupportingInformation