Predicting predator-prey interactions in terrestrial endotherms using random forest

Abstract

Species interactions play a fundamental role in ecosystems. They affect where species can live, how their population sizes fluctuate through time, and how environmental perturbations cascade through communities. But few ecological communities have complete data describing such interactions, which is an obstacle to understanding how ecosystems function and respond to environmental perturbations. Because it is often impractical to collect empirical data for all potential interactions in a community, various methods have been developed to infer interactions. Random forest machine learning is emerging as one of the most accurate and frequently used methods for making interaction predictions, but its performance in inferring predator-prey interactions in terrestrial vertebrates remains untested. The sensitivity of random forest performance to variation in quality of training data is also unclear. We examined predator-prey interactions within and between two diverse, primarily terrestrial vertebrate classes: birds and mammals. Combining data from a global interaction dataset and a specific ecological community (Simpson Desert, Australia), we tested how well random forests predict predator-prey interactions for mammals and birds using species ecomorphological and phylogenetic traits. We also tested how variation in training data quality affected model performance by: removing records and switching interaction records to non-interactions (false non-interactions) in the entire training dataset, or restricting these changes to records involving focal prey, focal predators, or non-focal species (focal = species for which predictions are being made). We found that random forests could predict predator-prey interactions for birds and mammals using either ecomorphological or phylogenetic traits, and that these predictions were accurate even when there were no records in the training data for the focal predator or focal prey species. In contrast, false non-interactions for focal predators in the training data strongly degraded model performance. Our results demonstrate that random forests can be used to identify predator-prey interactions for bird and mammal species that have few or no trophic interaction records. Furthermore, our study provides a roadmap for predicting interactions using random forests which might help ecologists: (i) address knowledge gaps and explore network-related questions in data-poor situations and (ii) predict interactions for species invading new ecosystems.

Predicting predator–prey interactions in terrestrial endotherms using random forest

“…We also tested how the performance of these models was influenced by the type of species data used, the filtering of training data by removing predators with few interaction records in those data, and the quality of the training data. The two interaction datasets were: 1) a dataset from the 'Global Biotic Interactions database' (globalbioticinteractions.org; Poelen et al 2014) supplemented with interaction data from diet studies done in Australia (Llewelyn 2022), and 2) an ecosystem-specific dataset from the Simpson Desert in Australia that focuses on seven predators for which detailed dietary information is available (Llewelyn 2022). We focused on birds and mammals rather than on all tetrapods due to the availability of detailed trait (Wilman et al 2014) and phylogenetic (vertlife.org) information for birds and mammals (detailed trait databases for reptiles and amphibians are far from complete), thereby providing a wide range of traits for predicting interactions.…”

“…In both cases, we built these datasets by aggregating predator-prey records from diet studies (Supporting information). We then combined these additional interaction datasets with the GloBI data (adding 363 records) to create an enhanced global interaction dataset (Llewelyn 2022).…”

“…Data are available from the GitHub and Zenodo: https://doi. org/10.5281/zenodo.7037189 (Llewelyn et al 2023).…”

Navigating the integration of biotic interactions in biogeography

“…The accumulation of this collective knowledge now enables us to draw generalisations about interactions across diverse environments and taxa. When information is lacking for certain species, interactions can be inferred from known species to those that are functionally or phylogenetically similar (Caron et al, 2022; Llewelyn et al, 2023; Strydom et al, 2022). Through the integration and analysis of these diverse data sources, researchers can then enhance the understanding of interactions on a broader scale, bridging the gap caused by limited direct observations (Compson et al, 2018; Gravel et al, 2013; Maiorano et al, 2020).…”

“…Meta‐networks integrate information on species interactions from various sources, including observations (e.g. GLOBI), expert knowledge, literature reviews (Maiorano et al, 2020) and phylogenetic or trait inference (Caron et al, 2022; Llewelyn et al, 2023). Metanetwork data thus provide a comprehensive summary of all potential species interactions, without specifying the spatial and temporal variability in the strength and realisation of these interactions (Maiorano et al, 2020).…”