Quantifying the distribution of species' abundance across the landscape is prerequisite for understanding biodiversity and its response to environmental change, as well as for making informed management and policy decisions. However, it is notoriously difficult to do, especially because available data are rarely from designed experiments. To cover appropriate geographic ranges, researchers are often forced to combine data from many different sources, such as from a diversity of contributors to GBIF. Even more problematic is the desire to include both presence-only observation data and presence-absence survey data. To overcome these difficulties, we have developed a point-process model that integrates the processes influencing the absolute rate of occurrence across the landscape with processes of acquiring observations of occurrence. The model enables inference of absolute abundance and probability of occurrence for any relevant spatial extent, and can relate those to a variety of covariates.
Results/Conclusions
Applications of our integrated species' abundance model to seven invasive species of New England reveal strengths of this approach. One is our inference of spatially explicit absolute abundance based upon integration of two types of data (presence-only and presence-absence) with very different statistical properties. This counters claims that such data cannot be integrated. Often ecologists are instead interested in the probability of occurrence rather than the absolute abundance. We also infer spatially explicit probability of occurrence for any relevant spatial extent, which can range from small plots located anywhere to large regions. While we are currently integrating the two main types of available data used for inferring species ranges, presence-only and presence-absence data, other types can be integrated as well when they are available. One obvious example is inclusion of population counts in addition to occurrence. Thus, our approach opens up new avenues for better inferring one of the most important components of biodiversity, species' abundance, and better leveraging the diversity of relevant information. As a result, we expect our approach will improve understanding of biodiversity and its determinants, and provide better information to guide decision-making.