Global sensitivity analysis of a complex marine ecosystem model: Advancing the understanding of ecosystem functioning

Huang Y.; Halouani G.; Luján C.; Lasram F.B.R.; Girardin R. · Ecological Modelling (2026) · DOI: 10.1016/j.ecolmodel.2026.111725

Resumen

Ecosystem models are characterized by high dimensionality and strong non-linear dynamics, making it essential to assess their sensitivity to parameter uncertainty to improve the robustness of model interpretation and their use in decision-support contexts. In this study, we conducted a global sensitivity analysis of the OSMOSE ecosystem model, which exhibits substantial variability due to numerous parameters and complex interactions among processes.The analysis was based on the Morris screening method, selected for its computational efficiency and its ability to detect global effects, non-linearities, and parameter interactions. A broad set of parameters related to growth, mortality, predation, resource accessibility, and fishing was included. Their effects were evaluated using five ecological indicators (biomass, catch, large-fish index in catch, mean length, and mean trophic level) at both the community and species levels.Strong non-linear responses and interactions among parameters were consistently observed across processes and indicators, confirming the necessity of global methods for analyzing complex ecosystem models. This analysis enables the early identification of influential parameters, such as those related to growth and sexual maturity. Beyond identifying key sources of uncertainty, the results also provide insights into ecosystem functioning by distinguishing species that play a structuring role at the community level (e.g., herring) from those that primarily insert local effects (e.g., lesser spotted dogfish).Overall, global sensitivity analysis emerges not only as a methodological tool but also as a means of ecological interpretation, enhancing the credibility and transparency of ecosystem models used for environmental impact assessment.