Stock et al., Exploring multiple stressor effects

Stock, A., C.C. Murray, E.J. Gregr, J. Steenbeek, E. Woodburn, F. Micheli, V. Christensen and K.M.A. Chan (2023). “Exploring multiple stressor effects with Ecopath, Ecosim, and Ecospace: Research designs, modeling techniques, and future directions.” Science of The Total Environment 869: 161719. Doi: 10.1016/j.scitotenv.2023.161719

Understanding the cumulative effects of multiple stressors is a research priority in environmental science. Ecological models are a key component of tackling this challenge because they can simulate interactions between the components of an ecosystem. Here, we ask, how has the popular modeling platform Ecopath with Ecosim (EwE) been used to model human impacts related to climate change, land and sea use, pollution, and invasive species? We conducted a literature review encompassing 166 studies covering stressors other than fishing mostly in aquatic ecosystems. The most modeled stressors were physical climate change (60 studies), species introductions (22), habitat loss (21), and eutrophication (20), using a range of modeling techniques. Despite this comprehensive coverage, we identified four gaps that must be filled to harness the potential of EwE for studying multiple stressor effects. First, only 12% of studies investigated three or more stressors, with most studies focusing on single stressors. Furthermore, many studies modeled only one of many pathways through which each stressor is known to affect ecosystems. Second, various methods have been applied to define environmental response functions representing the effects of single stressors on species groups. These functions can have a large effect on the simulated ecological changes, but best practices for deriving them are yet to emerge. Third, human dimensions of environmental change – except for fisheries – were rarely considered. Fourth, only 3% of studies used statistical research designs that allow attribution of simulated ecosystem changes to stressors’ direct effects and interactions, such as factorial (computational) experiments. None made full use of the statistical possibilities that arise when simulations can be repeated many times with controlled changes to the inputs. We argue that all four gaps are feasibly filled by integrating ecological modeling with advances in other subfields of environmental science and in computational statistics.