Co-occurrence is not evidence of ecological interactions

Area of Science:

• Ecology
• Community Ecology
• Ecological Modeling

Background:

• Co-occurrence (presence-absence) data have been widely utilized to infer community assembly processes.
• Numerous modeling approaches, including null model analysis, co-occurrence networks, and joint species distribution models, have been developed based on this data.
• While ecological interactions are theorized to influence co-occurrence, the extent to which this signal is detectable in observational data remains debated.

Purpose of the Study:

• To critically assess whether co-occurrence data serve as a reliable proxy for ecological interactions.
• To evaluate the limitations of using spatial associations in presence-absence data to infer ecological processes.
• To propose alternative interpretations and methods for extracting meaningful ecological information from co-occurrence data.

Main Methods:

• A theoretical assessment using probability, sampling, food web, and coexistence theories.
• Critical analysis of existing modeling approaches applied to co-occurrence data.
• Review of empirical evidence regarding the relationship between species co-occurrence and ecological interactions.

Main Results:

• Significant spatial associations (or lack thereof) between species are demonstrated to be a poor proxy for ecological interactions.
• The influence of factors beyond direct ecological interactions, such as dispersal limitation and habitat filtering, can create misleading co-occurrence patterns.
• Current modeling approaches may overemphasize the role of interactions when interpreting co-occurrence data.

Conclusions:

• Co-occurrence data alone are insufficient to reliably infer ecological interactions and community assembly.
• A re-evaluation of how co-occurrence data is interpreted is necessary, moving beyond simplistic assumptions of interaction signals.
• Future research should focus on integrating co-occurrence data with other ecological information and developing more nuanced analytical frameworks.