Whole‐landscape modelling of compositional turnover in aquatic invertebrates informs conservation gap analysis: An example from south‐western Australia

Freshwater aquatic ecosystems are in decline due to intensifying land use, salinisation, water abstraction and climate change. Understanding compositional patterns in aquatic biota is a useful step towards better management of aquatic ecosystems. We used generalised dissimilarity modelling (GDM) to predict compositional turnover in riverine invertebrate fauna (primarily insects) as a function of environment. Conceptual understanding of major drivers of aquatic invertebrate species distribution helped decide which predictor variables to source and include. Five groups of environmental variables—waterscape, local habitat, climate, landscape and disturbance—were derived from either spatial layers or in situ (site) measurements. Predictive models and variance partitioning tests of variable groups demonstrated the importance of representing all conceptual drivers of ecological pattern and process. As expected, waterscape variables were independently the most important group, followed by local habitat and landscape variables; with complex interactions between groups. Climate variables independently contributed the least. To determine the information content for mapping patterns, we investigated the independent and combined contribution of site‐measured and spatial predictors. Even though predictive models developed using only site‐measured variables or only spatial variables explained around the same amount of deviance (DE), combined they increased explained model DE by 11.2%. Compositional dissimilarities between the 51 surveyed site pairs predicted by the model using only spatial variables were highly correlated (r² = .85) with dissimilarities predicted using site and spatial variables. These results support the use of the spatial model for conservation decisions. The spatial model was used to evaluate representativeness of both the conservation reserve network and biological monitoring locations. The location of aquatic monitoring sites was uneven, with comprehensive coverage south and coastward, and less representative of inland environments. Proportional protection of ecological environments (scaled by riverine invertebrate taxa) was found to vary between 20% and 30%, being higher in southern parts where more land has been allocated to reserves and less in northern and inland parts. This demonstrated local progress towards achieving the Convention on Biological Diversity's Aichi Target 11 for inland waters. These results provide a focus for improving the robustness of information used in decisions affecting the conservation of aquatic biodiversity, including places to target to fill gaps in the reserve network and additional aquatic monitoring locations (supporting Convention on Biological Diversity's Aichi Target 19). The GDM‐based approach to characterising ecological environments, provided a first quantitative foundation for comprehensively evaluating the conservation status of freshwater ecosystems in south‐western Australia. Potential future applications include assessing the ecological implications of land use and climate change.