PS 86-250
Predicting food web structure and robustness
Anthropomorphic disturbances are causing, what many refer to, as the "sixth mass extinction." To manage such large scale species decline, conservation plans needs to incorporate food web dynamics as the loss of a single species can cause pervasive cascading secondary species loss. However, there is still considerable debate about how food web structural properties trend with species richness, connectance, regionalised productivity, and their global localities. There is also great uncertainty about what structural properties determine a food web's robustness. Historically it was proposed that food web stability increases with diversity, this was later challenged by mathematical models that suggested the opposite was true. At present, studies still show disagreement regarding the determinants of food web stability. We used 179 well resolved binary food webs to examine how 30 metrics of food web structure trend with food web size, location, and regionalised evapotranspiration. We also examined how food web robustness to random species loss, and the sequential loss of species with the highest connectivity, generality, vulnerability, and mean path length to a basal species, can be predicted using bivariate trends, multi-metric linear models, and artificial neural networks from the metrics calculated.
Results/Conclusions
We found that the total number of links in a web increases following a quadratic trend, whilst connectance decreased following a hyperbolic trend with increasing species richness. Larger food webs had greater mean links per species, with a greater proportion of the links being skewed towards a small proportion of species. We also found that robustness to random species loss showed a strong, significant, increase with species richness. The best multi-metric linear models and artificial neural network models were also strong predictors of robustness to random species loss. Longer mean food chains destabilised the food webs, whilst increased nestedness and increased entropy of the generality distribution stabilised the food webs.