Understanding the relative importance of multiple stressors is valuable to prioritize conservation and restoration measures. Yet, the effects of multiple stressors on ecosystem functioning remain largely unknown in many freshwaters. Here, we provided a methodology combining ecosystem modelling with linear regression to disentangle the effects of multiple stressors on matter flow, an important ecosystem function. Treating a shallow lake as the model ecosystem, we simulated matter flow dynamics during 1950s-2010s with different combinations of stressors by Ecopath with Ecosim (EwE) modelling, and determined the relative importance of each stressor by generalized linear mixed models. We found that matter flow of the lake food web was highly dynamic, attributing to effects of multiple anthropogenic stressors. Biological invasion played the strongest role in driving the matter flow dynamics, followed by eutrophication, while biomanipulation (i.e. phytoplankton control by planktivorous fish stocking) was of little importance. Eutrophication had a stronger role on primary producers, pelagic food chain and top predators, while biological invasion on consumers in the middle of food chains. The former was more important in driving the quantity of matter flow, while the latter on trophic transfer efficiencies. Scenario forecasting showed that reducing nutrients contents could largely shape the matter flow pattern, while biomanipulation had little effect. Our findings provided new insights into understanding the mechanistic links between anthropogenic stressors and ecosystem functioning by combining ecosystem modelling with linear regression.