Warmer winters did not lower spring river invertebrate abundance at 1,503 English sites in 1991-2017; warm-summer richness gains were smaller at high-ammonia sites

We tested if a known warming effect on stream invertebrates transfers to a national river network. Durance and Ormerod (2007) found that spring macroinvertebrate abundance in Welsh upland streams fell by up to 21% for each 1 C of warming. We used the ChemPop data product (EIDC, Environment Agency samples) with 23,706 spring and 23,910 autumn samples from 1,503 English river sites in 1991-2017. Each sample has the air temperature, water quality and river flow of the previous 6 months. Models compared each site with itself across years and removed regional trends, sampling day and recording method. In spring, the log10 abundance slope was -0.001 per C (95% CI -0.014 to +0.013), or about -0.3% per C. All tests excluded the Llyn Brianne value of -0.102 per C: two periods, seven regions, upland sites, flow and water-quality controls, year fixed effects, region-year means and true counts only. Family richness and EPT family richness also had no clear slope. Higher 6-month high flows lowered spring abundance and richness, but did not change the temperature slope. In spring, high-ammonia sites did not respond differently. In autumn, warmer summers came with more families, but the main effect was uncertain with year fixed effects. The gain was smaller at high-ammonia sites: -0.46 core families per C (CI -0.75 to -0.17) and -0.25 EPT families per C (CI -0.36 to -0.13) with year fixed effects. This interaction held with extra site controls and in the later period, and a next-year placebo gave no interaction. Warm summers also lowered oxygen saturation more at these sites. Limits: air temperature is a proxy for water temperature, temperature anomalies are mostly national, and the autumn interaction is an association from a pre-specified check, not the primary test.

freshwater macroinvertebrates · climate warming · ammonia · multiple stressors · river flow · EPT · Environment Agency · England · replication · null result · observation bias