Using networks to detect regime changes in aquatic communities across nutrient gradients

Abstract:

Networks capture links or interactions between organisms within ecological webs. When an environmental stress occurs, rapid changes in ecosystem state are expected in food webs with highly connected networks and functionally redundant species. These networks can dissipate local disturbances quickly and provide resistance to change at first until a threshold is reached, at which point, a critical transition occurs (nodes shift in synchrony). In contrast, in low connectivity (modular) heterogeneous networks, the response in ecosystem state to an environmental stressor is gradual. Given that these ecosystem-level shifts can be difficult to predict, hard to reverse and can have undesirable consequences, there is considerable interest in identifying what type of response (gradual vs. hysteresis) is most likely in nature. In this work, we thus aimed to test for the support for a bifurcated response in aquatic ecosystem across a landscape of human impact and track which of the above scenarios was most common. More specifically, using the US EPA National Lake Assessment water quality dataset (2007 sampling), we quantified differences in food-web structures across a spatial gradient of human impact (eutrophication). Preliminary results indicate that certain network properties vary nonlinearly with respect to nutrient enrichment.