Simultaneously estimating food web connectance and structure with uncertainty.

Area of Science:

• Ecology
• Theoretical Ecology
• Network Analysis

Background:

• Food web models are crucial for understanding trophic interactions and inferring missing ecological links.
• The allometric diet breadth model (ADBM) uses foraging theory and allometric scaling of body size for predator-prey interactions.
• Previous ADBM parameterization had limitations: point estimates for parameters and unestimated food web connectance.

Purpose of the Study:

• To improve the allometric diet breadth model (ADBM) by addressing limitations in parameter estimation and connectance.
• To estimate parameter distributions rather than point estimates using approximate Bayesian computation.
• To allow food web connectance to emerge from model parameterization using the true skill statistic.

Main Methods:

• Employed approximate Bayesian computation (ABC) to parameterize the ADBM for 12 diverse food webs.
• Utilized the true skill statistic to measure model fit, incorporating both presence and absence of trophic links.
• Estimated distributions for ADBM parameters, allowing connectance to emerge from the fitting process.

Main Results:

• Estimated food web connectance was consistently higher than previously reported across the studied ecosystems.
• Significant variation in parameter distributions was observed in some food webs, leading to uncertainty in predicted food web structure.
• The refined ADBM suggests that observed food web data may be missing actual trophic links.

Conclusions:

• The enhanced ADBM, parameterized with approximate Bayesian computation, provides more robust estimates of food web structure and connectance.
• Acknowledging uncertainty in parameter estimates is vital for predicting food web responses to environmental changes.
• Future research should incorporate additional traits beyond body size into the ADBM to refine link predictions and reduce false positives.