Realized vs. potential dietary niche aggravates species loss to secondary extinctions

1.Biodiversity loss is occurring at an unprecedented rate, but its consequences on ecosystem robustness remain difficult to predict. Both theoretical and empirical work have suggested that species extinctions for taxa with different dietary niche in food webs (e.g., species generality) can produce unequal consequences. However, the diverse definitions of species generality impede us to obtain a comprehensive understanding of the impact of specialist/generalist (i.e., species with low/high generality) loss on ecosystem robustness. 2.Here, we consider five generality metrics (one binary generality, one effective link generality, and three weighted generality, respectively, by link strength, biomass and energy flux) at both species- and network-level. We use a multi-trophic food web model in combination with empirical data to explore the relationship between generality and species richness. We further perform secondary extinctions by sequentially removing species with either maximal or minimal generality under two generality ranking scenarios to examine food web robustness to species loss, where species generality is ranked based either on their realized or potential dietary niche. 3.Results show that at a network-level, all considered generality metrics increase with species richness, albeit at markedly different rates. Yet the relative richness of generalists to specialists stays roughly invariant, indicating that generalist-prone communities as a consequence of specialist loss might probably just a transitory state and may eventually return to a more specialized species assemblage. The increased network-level generality is due to the enhanced heterogeneity in species-level generality rather than the increased generalist richness. 4.Our sequential species elimination experiment demonstrates substantial variability in ecosystem robustness responses to different generality metrics. In both generality ranking scenarios, removing generalists (specialists) measured by biomass-weighted generality is the most (least) detrimental to food web robustness, suggesting that species with dominant biomass should be of high biodiversity concern. Further, robustness loss in response to generalist removal is larger in the realized niche ranking scenario than in the potential niche ranking scenario, implying that realized rather than potential dietary niche is more effective to identify species priority in conservation. 5.Our work demonstrates a consistent relationship between network-level generality and species richness, but the variability in metrics also indicates that different ecological processes are at play behind each generality metric. We recommend that studies investigating biodiversity conservation under various anthropogenic forcing should include biomass-weighted metrics, and pay more attention to species’ dietary niche realized in the current food webs instead of the dietary niche recorded in the historic food webs.