Rivers and streams are exemplars of inherently variable ecosystems. Their flows fluctuate on timescales from seconds to decades, and they experience intensities of flow events ranging from supraseasonal droughts to massive, infrequent floods. The prevailing view is that they are resilient to these variations in the sense that their flora and fauna recover or ‘bounce back’ after these disturbances.
Unfortunately, what constitutes ‘recovery’ in such systems remains vague. Classical theory that presupposes the existence of an ‘equilibrium’ condition seems inapplicable. Oddly, alternative concepts such as strange attractors, stochastic boundedness, robustness and anti-fragility have been little used by river ecologists, probably because the empirical demands of these alternatives are high.
Instead, most empirical studies have resorted to informal or short-term baselines to provide benchmarks for measuring recovery. It is uncommon for studies to include more than one or two sampling times before a flood or drought, and the few long-term studies often reveal multiple flow events that would prevent the populations of many species returning to an ‘equilibrium density’ or the community composition prior to the disturbance. Conceptually we have developed frameworks or models that focus on maintaining key habitat and flow features to ensure the persistence of species.
Nevertheless, a key problem facing water managers is when to declare that a river’s ecology has recovered from a disturbance. While posing considerable conceptual and empirical challenges, this problem has profound and urgent practical importance because of intensifying demands for water abstraction. I argue that we need to re-engage with debates about the most appropriate concepts surrounding recovery and resilience so that we get clearer prescriptions about the types of data we collect and the sampling regimes required to properly monitor our actions and responses.