PS 31-173 - Examining food web network structure variation as a function of productivity and fragmentation at regional and continental scales

Abstract: Food webs describe both the flow of energy moving through a system as well as the interactions among species by which energy is transferred. Energy for food webs begins with the primary productivity of the ecosystem, which varies depending on location and anthropogenic land-uses. Understanding the drivers of food web structure in response to disturbances can contribute to our understanding of biodiversity maintenance. In this study we investigate relationships between food web similarity, plant productivity, and landscape fragmentation for 170 African mammal communities, while simultaneously accounting for predicted defaunation and human pressure using partial Mantel tests. We hypothesize that if primary productivity is important for structuring food webs, then communities with similar levels of primary productivity will have more similar food web structures. If fragmentation is important for structuring food webs, then communities with similar levels of fragmentation will have more similar food webs. We first created a binary overarching metaweb for African mammal predator-prey interactions using publicly available data and literature on species interactions. We then extracted food webs for each African location from the metaweb based on publicly available community composition data and measured differences in network geometry among food webs with Gromov-Wasserstein distances. We gathered remotely sensed data for primary productivity (EVI) and calculated the main principal component from fragmentation metrics total core area, effective mesh size, patch density, edge density using land cover data and the R package “landscapemetrics”. We then calculated distance indices for these values across all sites. We performed cluster analysis in species similarity at sites, as well as biome type, to aggregate sites and reduce bias into 9 bio-graphical regions.

Across grasslands/savannas and deserts as well, food web similarity was positively and significantly correlated with similarity in primary productivity, respectively (r=0.216, p< 0.026; r=0.294, p=0.027). Within tropical rainforests, food web similarity was positively and significantly correlated with similarity in fragmentation (r=0.643, p=0.001). These results held when considering human pressure and defaunation as well. Understanding how habitat loss and other anthropogenic disturbances alter food webs can provide scientists with a toolkit for predicting the consequences of conservation decisions on a community.