Quantitative Flow-Ecology Relationships Using Distributed Lag Nonlinear Models: Large Floods in the Murray River Could Have Delayed Effects on Aquatic Macroinvertebrates Lasting More Than Three Decades

We aimed to quantify the delayed effects of flows on aquatic ecosystems using 33 years of monitoring data on macroinvertebrates and water quality and 51 years of hydrological data, spanning 2,300 km of the Murray River, Australia. By incorporating the delayed effects of hydrological indices, along with physicochemical variables, into generalized additive models to form distributed lag nonlinear models for macroinvertebrate richness and abundance, we found that the effects of floods on macroinvertebrates could last up to 32 years, depending on their magnitude. The models indicated that a large flood can cause an initial depression in macroinvertebrate abundance and richness, followed by a sustained increase persisting for over 25 years before returning to preflood levels. Twelve hydrological indices representing the magnitude of flow events were evaluated, with the q90/med (the 6-monthly flow exceeded 10% of time, divided by the long-term median) performing (slightly) better than other indices. Competing hypotheses for the mechanisms underlying the apparent long-term effect of floods were considered, with the most plausible explanation being the flood-mediated influx of allochthonous organic matter, especially coarse and large woody debris, that might drive the persistent change in the aquatic community. It is generally believed that macroinvertebrate communities recover quickly after floods, but our findings suggest that we may need to reconceptualize the effects that floods can have on aquatic macroinvertebrates and the communities they support, especially in the context of flow restoration and climate change.