Rising, falling, stalling: Complex biodiversity trajectories in UK rivers

Freshwater ecosystems face intense anthropogenic pressures, with global indicators showing steep biodiversity declines in many regions. Yet in parts of Europe, including the UK, several recent analyses of long-term monitoring datasets have documented widespread improvements in macroinvertebrate communities following major water-quality investments. These gains are often interpreted as being nationally uniform, based on spatially aggregated data or simplified effect sizes, despite substantial site-level heterogeneity in temporal dynamics. Consequently, the prevalence, distribution, and causes of local biodiversity dynamics remain poorly resolved. We address this gap using macroinvertebrate time-series data from 803 river sites in Englands Environment Agency monitoring network. We applied an unsupervised, shape-explicit time-series clustering framework to classify sites into distinct shape-based clusters for three complementary metrics: pollution sensitivity of communities (Whalley, Hawkes, Paisley & Trigg (WHPT) Index), total abundance, and Simpson diversity index. We then characterized taxonomic and environmental differences among clusters using Linear Discriminant Analysis and non-metric multidimensional scaling. Finally, we quantified how temporal trajectories are structured across increasingly aggregated spatial scales (site, water body, catchment, typology, region) using a Bayesian variance-partitioning model, providing the first multi-metric, multi-scale assessment of macroinvertebrate trends in UK rivers. Overall, our analyses reveal substantial heterogeneity in ecological trajectories: although many sites showed improving trends, more than half (~53%) were in decline or lacked clear directional change. Sites with consistent increases or declines were strongly differentiated by longitudinal position, channel morphology, and water chemistry, indicating that network placement and local environmental context mediate river biodiversity changes. Variance partitioning showed that most explainable variation lies at the site and typology levels, demonstrating that broader regional or national summaries can obscure a mosaic of non-linear, site-level trajectories and persistent local declines. Collectively, these findings affirm that freshwater biodiversity change in UK rivers cannot be summarized by a national or even regional narratives of recovery, but represents a mosaic of asynchronous, site-specific trajectories. These results illustrate the need for a renewed investment in dense, long-term river monitoring networks capable of resolving biodiversity change across hierarchical spatial scales.