A Novel Simulation Framework for Validation of Ecological Network Inference


Erik Kusch and Anna C. Vinton share joint-first authorship.

  1. Understanding how the differential magnitude and sign of ecological interactions vary across space is vital to assessing ecosystem resilience to biodiversity loss and predict community assemblies. This necessity for ecological network knowledge and their labour-intensive sampling requirements has spurred the creation of ecological network inference methodology. Recent research has identified inconsistencies in networks inferred using different approaches thus necessitating quantification of inference performance to facilitate choice of network inference approach.

  2. Here we develop a data simulation method to generate data products fit for network inference and subsequently quantify the validity of two well-established ecological interaction network inference methods – HMSC and COOCCUR. The simulation framework we present here can be parameterised using real-world information (e.g., biological interactions observed in-situ and bioclimatic niche preferences) thus representing network inference capabilities in real-world applications. Using this framework, it is thus possible to evaluate the performance of any ecological network inference approach.

  3. We find that network inference performance scores highlight concerningly low accuracy of inferred networks as compared to true association networks and suggest analysis procedures with which to explore network inference reliability with respect to bioclimatic niche preferences and association strength of association partner-species.

  4. With this study, we provide the groundwork with which to validate and compare ecological network inference methods, and ultimately vastly increase our ability to predict species biodiversity across space.

Erik Kusch
Erik Kusch
Senior Engineer & Research Infrastructure Manager

In my research, I focus on statistical approaches to understanding complex processes and patterns in biology using a variety of data banks.