LB 30-294 - How much gut content data is required to predict trophic interactions?

Background/Question/Methods:

Presence-absence data of species interactions such as gut content data is used to parameterise food web models such as the allometric diet breadth model (ADBM), thereby predicting interactions. Collecting and analysing these data from the field is, however, an expensive and time-consuming task. Therefore, it is crucial to know how much gut content data is required to parameterise food web models with high accuracy and high precision. We apply to this problem a recently developed approach, namely Bayesian computation to parameterise the ADBM, and true skill statistics to measure the goodness of fit, and do so while varying the amount of gut content data used in the parameterisation.

Results/Conclusions:

We estimated the minimum amount of gut content data required for seven different food webs that resulted in 95% of the maximum true skill statistics achieved by the ADBM when all available gut content data was used. We found that incomplete gut content data can be used to parameterise the ADBM with the lowest amount of gut content data being 28% of the available gut content data. These results suggest that one need not collect such a large quantity of gut content data to predict the structure of a food web, thereby reducing sampling effort considerably, while having little effect on precision or accuracy of predictions.