Summary Excess loading of phosphorus (P) and nitrogen (N) triggers a shift in the trophic structure of shallow lakes from a clear‐water, macrophyte‐dominated state to an algal‐dominated turbid state. However, the role of N in the shift is debated, and experimental evidence is, with a few exceptions, based on short‐term studies (days to a few months). We studied the effect of N loading on macrophytes (dominated by Potamogeton lucens and Cabomba caroliniana), periphyton, filamentous algae and phytoplankton in mesocosms over 10 months (starting in October) in subtropical China (Wuhan). There were three N treatments: controls (CN) without nitrogen addition (mean TN = 1.9 mg L−1), low nitrogen (LN) addition (mean TN = 3.5 mg L−1) and high nitrogen (HN) addition (mean TN = 5.5 mg L−1). Total phosphorus (TP) concentration in the water column remained moderate (0.05–0.07 mg L−1) during the experiment in all treatments. Macrophyte abundance declined in the LN and HN treatments in the first 6 months, but not in controls, followed by a partial recovery in the LN treatments. They disappeared completely in the HN treatments the following summer. Periphyton (biofilm on plastic) and phytoplankton biomass remained unaffected during the first 6 months but increased over the summer by two or three times, compared with controls, in low and high nitrogen treatments, respectively. By contrast, the abundance of filamentous algae increased over winter but declined during the summer with no obvious relationship to the N treatments. There was no difference in the TN or nitrate concentrations or soluble protein, soluble sugar and Chl‐a content of P. lucens leaves and stems with increasing N load. Macrophyte populations are partially resilient to abrupt increases in N loading at moderate TP concentrations, but, after prolonged exposure, a complete collapse occurs. Our results further indicate that macrophyte loss is exacerbated by shading by filamentous algae during the winter, and by phytoplankton and periphyton in the summer, while there was no indication of direct N toxicity.

There is a pressing need to apply stability and resilience theory to environmental management to restore degraded ecosystems effectively and to mitigate the effects of impending environmental change. Lakes represent excellent model case studies in this respect and have been used widely to demonstrate theories of ecological stability and resilience that are needed to underpin preventative management approaches. However, we argue that this approach is not yet fully developed because the pursuit of empirical evidence to underpin such theoretically grounded management continues in the absence of an objective probability framework. This has blurred the lines between intuitive logic (based on the elementary principles of probability) and extensional logic (based on assumption and belief) in this field.