The collapse and limited recovery of groundfish stocks have been a subject of concern for decades. Their failure to fully recover has been linked to various hypothetical causes including predation, overfishing, and changes in environmental conditions while this only tells part of the story. There remain uncertainties about how these stressors have interacted to alter groundfish stock populations, community abundance, and size structure. Our research uses a novel modelling methodology and hypothesis testing framework to evaluate and compare the individual and combined impact of these stressors on groundfish dynamics on the Eastern Scotian Shelf (Eastern Canada). Using the multispecies size spectrum model therMizer, a modification of Mizer, we reconstructed the past impact of environmental conditions, predation, fishing, and their combined impacts on groundfish stocks pre- and post-collapse. Using an in silico experimental framework of multiple stressor attribution and detection to compare to alternative models, we found that there was no single hypothesized stressor that best captured past change. Interacting effects of multiple stressors propagated from individual to higher levels of organization and differentially affected trends in population and community indicators. This work informs future strategies for testing fisheries management interventions effectively amid evolving stressors, advancing robust regional ecosystem modeling.