In streams, hydrology is a predominant driver of ecological structure and function. Providing adequate flows to support aquatic life, or environmental flows, is therefore a top management priority in stream systems.
Flow regime classification is a widely accepted approach for establishing environmental flow guidelines. However, it is surprisingly difficult to quantify relationships between hydrology and ecology (flow–ecology relationships) while describing how these relationships vary across classified flow regimes. Developing such relationships is complicated by several sources of spatial bias, such as autocorrelation due to spatial design, flow regime classification and other environmental or ecological sources of spatial bias.
We used mixed moving‐average spatial stream network models to develop flow–ecology relationships across classified flow regimes and to assess spatial patterns of these relationships. We compared relationships between fish traits and life‐history strategies with hydrologic metrics across flow regimes and assessed whether spatial autocorrelation influenced these relationships.
Trait–hydrology relationships varied between flow regimes and across all streams combined. Some relationships between traits and hydrologic metrics fit predictions based on life‐history theory, while others exhibited unexpected relationships with hydrology. Spatial factors described a large proportion of variability in fish traits and different patterns of spatial autocorrelation were observed in different flow regimes.
Synthesis and applications. Further work is needed to understand why flow–ecology relationships vary across classified flow regimes and why these relationships may not fit predictions based on life‐history theories. Managers determining environmental flow standards need to be aware that different hydrologic metrics are often important drivers of fish trait diversity in different flow regimes. Flow–ecology relationships may therefore be confounded by spatial structure inherent in flow regime classification and much existing biological data. Complex patterns of spatial bias should be considered when managing stream systems within an environmental flows framework.