Towards a predictive model of species interaction beta diversity

Area of Science:

• Ecology
• Biodiversity Science
• Ecological Modeling

Background:

• Species interactions are crucial for community dynamics and ecosystem functions.
• Predicting changes in species interactions across space and time remains a significant challenge in ecology.
• Understanding interaction betadiversity is key to comprehending biodiversity patterns.

Purpose of the Study:

• To develop and validate a Bayesian approach for predicting species interactions.
• To improve forecasts of mutualistic interactions by incorporating occurrence and detectability probabilities.
• To explore seasonal variation in interaction betadiversity and its underlying drivers.

Main Methods:

• Utilized a multi-year hummingbird-plant time series data, split into training and testing sets.
• Developed a Bayesian model to disentangle probabilities of species co-occurrence, interaction, and detectability.
• Applied the model to assess seasonal differences in interaction betadiversity, hummingbird occurrence, and interaction frequency.

Main Results:

• Models incorporating detectability and occurrence significantly improved forecasts of mutualistic interactions.
• Despite seasonal differences in observed interactions, no significant changes in hummingbird occurrence or interaction frequency were detected.
• Low interaction detectability poses a challenge for inferring the causes of interaction betadiversity.

Conclusions:

• The proposed Bayesian framework enhances the prediction of species interactions.
• The study highlights the importance of considering detectability and occurrence in ecological forecasting.
• The model offers potential for integrating local interaction data with broader biogeographic and evolutionary contexts to understand biodiversity variation.