Using joint species distribution models to quantify potential biotic interactions among lotic fish assemblages of the contiguous United States

Quantifying the influence of environmental factors on species distributions can elucidate their ecological requirements and potential response to anthropogenic disturbance. Single species distribution models are commonly used to quantify taxon-environment relationships but do not account for biotic interactions (e.g. competition, predation, mutualism). Joint species distribution models (JDSMs) are a multivariate extension that overcome this challenge by modeling all taxa in an assemblage simultaneously and explicitly accounting for associations between species. We compiled occurrence data for ~200 fish species at 3,865 sites surveyed by U.S. Environmental Protection Agency’s National Rivers and Streams Assessment. We then used JDSMs to quantify taxon-environment relationships and pairwise associations. These models accurately characterized fish assemblages (AUC > 0.7). Gradients of specific conductance, average maximum air temperature, discharge and substrate diameter were important variables for explaining species occurrences. In addition, we found several significant associations indicating that certain species co-occur more or less than expected from environmental conditions. Leveraging these associations, we used conditional prediction to assess the potential effects of removing non-native fishes. Interestingly, removing non-native fish could either increase or decrease native taxa richness, depending on the site and region. We demonstrate that data collected for monitoring purposes provide unparalleled opportunities to quantifying species’ relationships with key environmental factors and assessing potential biotic interactions across a large spatial extent. 

Impact/Purpose

Biological interactions (e.g. competition, predation, mutualism) can influence a species spatial distribution but are not included explicitly in single species distribution models. In this presentation, we leverage fish assemblage data collected by the National Rivers and Streams Assessment and joint species distribution models (JSDMs) to quantify taxon-environment relationships and potential biotic interactions between fishes. Beyond applications for biological assessment, these models can elucidate large scale patterns in fish biological diversity. We found that JSDMs accurately characterized fish assemblages throughout the United States and specific conductance, maximum air temperature, discharge and substrate diameter were important variables for explaining species occurrences. We also detected several significant associations between species, indicating that certain taxa co-occur more or less than expected from environmental conditions. Finally, using these associations and conditional prediction, we assessed the potential effects of removing non-native fishes. Interestingly, we found that by assuming that non-native fish were absent from all sites, native taxa richness could either increase or decrease, depending on the site and region. Future research is needed to decipher whether these associations are due to biological interactions between fishes or because the taxa have a similar response to an unknown environmental variable.      

Citation

Kopp, D., J. Stoddard, D. Peck, A. Herlihy, AND P. Kaufmann. Using joint species distribution models to quantify potential biotic interactions among lotic fish assemblages of the contiguous United States. Society for Freshwater Science, Philadelphia, PA, June 02 - 06, 2024.