Co-Occurrence Patterns Do Not Predict Mutualistic Interactions Between Plant and Butterfly Species.

Area of Science:

• Ecology
• Community Ecology
• Biotic Interactions

Background:

• Biotic interactions shape community structure and dynamics.
• Directly measuring interactions is resource-intensive, especially with many species.
• Inferring interactions from co-occurrence patterns is a common but potentially limited approach.

Purpose of the Study:

• To evaluate the accuracy of co-occurrence methods for detecting mutualistic interactions in terrestrial ecosystems.
• To compare the performance of three co-occurrence analysis methods against empirical data.
• To assess the impact of using flower abundance versus plant abundance data on detection accuracy.

Main Methods:

• Compiled literature-based interaction data for butterfly-plant mutualisms in Germany.
• Collected observational abundance and presence/absence data.
• Applied three co-occurrence analysis methods (Relative Interaction Intensity, probabilistic, Spearman's rank correlation) and compared results using sensitivity and specificity analyses.

Main Results:

• All co-occurrence methods showed low sensitivity in detecting actual mutualistic interactions.
• The Relative Interaction Intensity index with pairwise null models performed best among the tested methods.
• Empirical data revealed numerous interactions missed by co-occurrence methods; flower abundance data improved specificity but not sensitivity.

Conclusions:

• Current co-occurrence methods have significant limitations in accurately capturing species interactions, particularly mutualisms.
• Ecological dynamics are complex, and predicted co-occurrence does not always reflect actual interaction strength.
• Further development of methods is needed to reliably infer species interactions from observational data.