Functional traits predict species co‐occurrence patterns in a North American Odonata metacommunity

Abstract

Abstract The probability of occurrence of a given species in a target locality and assemblage is conditioned not only by environmental/climatic variables but also by the presence of other species (i.e., species co‐occurrence). This framework, already complex in nature, becomes even more complicated if one considers the functional traits of species that, in turn, might influence the structure of metacommunities in various ways. Depending on the ecological and environmental setting, functional similarity (i.e., convergence in morphological and ecological traits) between species might either reduce their co‐occurrence due to high niche overlap driving negative interactions or promote it if the similar traits are associated with local habitat suitability. Similarly, functional divergence might either promote species co‐occurrence by limiting negative interactions through niche separation or reduce it through trait mediated environmental filtering. Therefore, discriminating between these alternative scenarios—predicting whether two species will tend to co‐occur or not based on their traits—is extremely challenging. Here, we develop a novel protocol to tackle the challenge, and we demonstrate its effectiveness by showing that ecological species traits can predict species co‐occurrence in a large dataset of North American Odonata. To this end, we first used the Hierarchical Modeling of Species Communities framework to quantify the pairwise species co‐occurrence after controlling for environmental and climatic factors. Then, we used machine learning to generate models which proved capable of predict accurately the observed co‐occurrence patterns from species functional traits. Our approach offers a generalizable analytical framework with the potential to clarify long‐standing ecological questions.

Keywords

• assembly rules
• co‐occurrence
• HMSC
• machine learning
• Odonata
• species interactions