Marine ecosystems are influenced by multiple stressors in both linear and non-linear ways. Using generalized additive models (GAMs) fitted to outputs from a multi-ecosystem, multi-model simulation experiment, we investigated 14 major ecological indicators across ten marine ecosystems about their responses to fishing pressure under: (i) three different fishing strategies (focusing on low-, high-, or all-trophic-level taxa); and (ii) four different scenarios of directional or random primary productivity change, a proxy for environmental change. From this work, we draw four major conclusions: (i) responses of indicators to fishing mortality in shapes, directions, and thresholds depend on the fishing strategies considered; (ii) most of the indicators demonstrate decreasing trends with increasing fishing mortality, with a few exceptions depending on the type of fishing strategy; (iii) most of the indicators respond to fishing mortality in a linear way, particularly for community and biomass-based indicators; and (iv) occurrence of threshold for non-linear-mixed type (i.e. non-linear with inflection points) is not prevalent within the fishing mortality rates explored. The conclusions drawn from the present study provide a knowledge base in indicators’ dynamics under different fishing and primary productivity levels, thereby facilitating the application of ecosystem-based fisheries management worldwide.