In RivFishTIME river fish (1964–2019), range position predicts abundance–temperature slopes between sites, but not between species at one site or abundance trends

Brown et al. (2024, PNAS) reported that river fish populations near the cold edge of the species range increase with warming more often than populations near the warm edge. Their model compared the same species at different sites, with species intercepts only. We tested this claim again with RivFishTIME: 89,993 native-range populations of 864 species in 10,636 time series (1964–2019, span of 10 years or more). We used native ranges from a basin-scale occurrence database and CRU TS temperature converted to water temperature. Range position is 0 at the warm edge and 1 at the cold edge. With species intercepts, the published response replicated: positive abundance–temperature slopes were 8.6 percentage points more frequent at the cold edge (95% CI 4.8 to 12.4, clustered by basin; logit 0.35 vs. 0.235 in the original). But range position did not predict the sign of the abundance trend over years (−0.9 points, CI −6.7 to 4.9). The association remained with detrended temperature (+6.8, CI 1.7 to 11.9), so it shows year-to-year covariation, not a trend. When we divided range position into the site mean and the deviation of each species from it, the between-site part was +13.7 (CI 9.2 to 18.2) and the within-site part was −6.4 (CI −14.4 to 1.6). The difference was −20.1 (CI −29.4 to −10.8). With site intercepts, species near their cold edge and species near their warm edge at the same site did not differ (−1.7, CI −6.0 to 2.6). A range-position effect for each population predicts the same sign in all these comparisons. Thus, in these data the pattern is better explained as a site-level relation: at cooler sites, catches of most species tend to be higher in warm years. The results were the same with thermal position, with 5 or more survey years, with or without the native filter, and without each of the four main countries. Limitations: basin-scale ranges and 0.5° air temperature are coarse; the within-site test has less leverage for Tmax and detrended temperature; the data come mostly from the USA, Great Britain, France and Sweden.

freshwater fish · RivFishTIME · range position · climate change fingerprint · population trends · replication · fixed effects · between-site confounding · native range · temperature sensitivity