Metacommunity structure can be summarised by fitting joint species distribution models and partitioning the variance explained into environmental, spatial and codistribution components. Here we identify how these components respond through time with directed environmental change and propose this as an indicator of sustained directional pressure. Through simulations, we identify how declines in the codistribution component can diagnose ecological breakdown, while rises in environmental and spatial components may indicate losses in peripheral areas and dispersal limitations. We test the method in two well-studied systems. Butterflies are known to be strongly responding to climate change, and we show that over 21 years the codistribution component declines for butterfly communities in southern England. By contrast, birds in the same region are under less climate pressure and, despite high occupancy turnover, show minimal change in metacommunity structure. The approach has high potential to summarise and compare the impact of external drivers on whole communities.

The landscape scale response of ecological communities to environmental drivers can be challenging to efficiently summarise and differentiate from expected background turnover through time. Metacommunity structure can be encapsulated by fitting joint species distribution models (JSDMs) and partitioning the variance explained into environmental, spatial and species‐codistribution components. Here we identify how these components respond through time with directed environmental change and propose these changes in metacommunity structure as an indicator of sustained directional pressure. Through simulations, we identify how declines in the variation explained by species codistribution could diagnose ecological disintegration, while increases in the explanatory power of environmental and spatial predictors may indicate losses in peripheral areas and dispersal limitations. We then test these results in two well‐studied systems. Butterflies are known to be strongly responding to climate change, and we show that over 21 years the codistribution component declines for butterfly communities in southern England. By contrast, birds in the same region are thought to be responding less strongly to climatic pressure and, despite high occupancy turnover, do not show clear changes in metacommunity structure as measured by this approach. Our results suggest that these approaches could have a high potential to summarise and compare the impacts of external drivers on whole communities.