Detection of fishing pressure using ecological network indicators derived from ecosystem models

Marine ecosystems are exposed to multiple stressors, mainly fisheries that, whenever mismanaged, may cause irreversible damages to whole food webs. Ecosystem models have been applied to forecast fisheries impact on fish stocks and marine food webs. These impacts have been studied through the use of multiple indicators that help to understand ecosystem responses to stressors. This study focused on a category of ecological indicators derived from the network theory to quantify energy flows inside the food web. These indicators were computed using two ecosystem models applied to the Eastern English Channel (i.e. Atlantis and OSMOSE). This work aimed at investigating how several ecological network indicators respond to different levels of fishing pressure and evaluating their robustness to model structure and fishing strategies. We applied a gradient of fishing mortality using two ecosystem models and carried out ecological network analysis to obtain network-derived indicators. The results revealed that the indicators response is highly driven by the food web structure, although the model assumptions buffered some results. The indicators computed from OSMOSE outputs were more sensitive to changes in fishing pressure than those from Atlantis. However, once the food web from Atlantis was simplified to mimic the structure of OSMOSE model, the indicators of the modified Atlantis became more sensitive to the intensity of fishing pressure. The indicators related to amount of energy flow and to the organization of the flows in the food web were sensitive to the increase of fishing mortality for all fishing strategies. These indicators suggested that increasing fishing mortality jeopardizes the amount of energy mobilized by the food webs and simplifies the ecological interactions, which has implications for the resilience of marine ecosystems. The study shed light on the trophic networks structure and functioning of the ecosystems whenever exposed to distur-bances. Furthermore, these indicators might be adequate for whole ecosystem assessments of health and contribute to ecosystem management.