Assessing the use of RIVPACS-derived invertebrate taxonomic predictions for river management

The River Invertebrate Prediction and Classification System (RIVPACS) is used widely in freshwater management to set targets for macroinvertebrate ecological health based on the expected scores of metrics such as WHPT or LIFE in the absence of anthropogenic stressors. An underutilised capability of RIVPACS-type models is the capability to predict expected macroinvertebrate community composition, which could function as a novel management metric for river health. We present a novel Monte-Carlo simulation approach that generates simulated expected communities for Englands rivers based on RIVPACS predictions. This allows for assessments of macroinvertebrate health using similarity calculations between observed and expected communities. We assess 10-year trends in similarity between 2010 and 2019 at 4172 sites in England, and contrast these trends with WHPT ASPT O/E trends in the same period. Similarity scores include both Chi-Squared and Hellinger methods, to prioritise rare and common species, respectively. We find that whilst most sites (63.3%) showed improvement in WHPT ASPT O/E in this period, most sites showed declines in similarity for Chi-Squared and Hellinger O/E (51.1% and 58.8%, respectively). We identified three case study regions showing contrasting trends and illustrate how the new RIVPACS-derived similarity calculations can track meaningful shifts in composition associated with water quality and multiple stressors including invasive species. RIVPACS-derived similarity calculations potentially provide a sensitive and practical management metric to assess ecosystem health, although further work is required to understand the composition of communities in changing environments with clear changes in stressor regimes.