Missing in action: Reference sites

Abstract

White et al. (2023) measured land area with precision in coastal areas exposed to river diversions, but the absence of concurrent measurements in reference sites prohibits them from making conclusions, which they do, that river diversions have a beneficial or neutral consequences but not harmful effects. Their global conclusions about land gain are inappropriate for analyses of areas embedded in ecosystems when there are no reference areas included in the analysis. Their inferences about the temporal on-site changes are valid, but measurements in suitable reference sites are needed to separate out the environmental ‘noise’ common to both and to allow conclusions about putative causes. Environmental impacts are typically discerned within a Before-After-Control-Impact (BACI) analysis. White et al.'s study design is a’BA’, but not a BACI design. They are critical of one of two of Turner et al.'s (2019) photogrammetric methods, but neglect to realize that the different methods reveal different aspects of land loss and are used independently of each other within a BACI framework having multiple reference sites. They also neglect to account for 1) results from a BACI analysis of thirty-two sites from farther downstream, 2) a striking loss of land after a natural river diversion opened up in 1973, and 3) they omit important information placing sediment retention into context. The problem of omitting reference sites in evaluations of environmental impacts is commonly observed and deserves wider attention to improve the general ecological intelligence of responsible parties and lay communities. Treating environmental projects a priori as experiments with testable hypotheses with alternative outcomes and actionable exit strategies will reduce errors and bias in endeavors with many unexpected unknowns.