Perhaps the greatest challenge for water managers around the world is maintaining reliable water supplies and healthy river ecosystems.  There can be direct trade-offs between these goals because surface and groundwater extraction and regulation of river flows can lead to dramatic declines in native fish assemblages. This dichotomy has led to the burgeoning field of environmental flows where managers – using tactics such as managed releases from dams or extraction limits – allocate river flows for the benefit of the river itself, not just for human use.

Understanding important flow-ecology relationships can help tailor environmental flows to maximize ecosystem benefits, and a big focus in the Olden lab has been to explore relationships between historical flow regimes and freshwater fishes (for examples, see earlier lab research on the importance of native and non-native origins and life history strategies in relation to river flows). These studies are part of a rich literature dedicated to understanding relationships between aquatic species and variation in river flow.  Typically, these relationships are derived from a statistical model with a set of hydrologic metrics used as predictor variables; these might include metrics such as mean daily flow, or CV of daily flow to predict species abundances (such as those used in this 2011 study on fish responses to temporal variation in flow).

Fish - such as this mountain whitefish - respond to actual river flows rather than metrics which describe flows

An analytical challenge in developing flow-ecology relationships is that fish (such as this mountain whitefish) experience actual river flows, not summary metrics like daily mean or CV. Photo credit: J. Monroe, Freshwaters Illustrated

With literally hundreds of different flow metrics that could be used in flow-ecology models, however, it is a genuine challenge to select a subset from a bewildering array of possible choices. Another challenge is defining environmental flow recommendations based on different studies which may have used very different flow metrics. Beyond these practical challenges is the fundamental issue that flow metrics summarize the hydrograph, or some aspect of it, rather than depicting the flow itself, so there is limited information about the mechanistic link between the response variable and the actual flow. For example, fish in the river don’t experience the mean daily flow, they experience the actual flow. It is also possible for two very different hydrographs to produce similar values of a flow metric.

These challenges motivated us to investigate new methods to identify key flow-ecology relationships. The field of functional data analysis (FDA), which is a growing area of statistical research, provides an interesting way forward for river ecologists and managers alike wrestling with questions of flow ecology.  A key tool in FDA is the functional linear model, where the predictor variable in the model is itself a mathematical function rather than scalar variables. In a new research paper, we use functional linear models to capture the entire hydrograph as a predictor variable to identify relationships between single observations of fish abundances and river flow over the course of a year. This solves the problems of selecting a subset of flow metrics and quantifies a direct link between fish in the stream and the river flow they experienced.

This is an exciting development in flow ecology research, providing an avenue to define a range of possible hydrographs with little or no ambiguity and predict the responses of the fish fauna to each.

– Ben Stewart-Koster