Two temperature records give opposite estimates of mean water temperature change at Bonneville sockeye passage, 1970–1993 to 2002–2025

We tested how temperature source choice affects a thermal index at adult sockeye passage at Bonneville Dam. We used daily counts in June–August and equal year weights. Complete windows gave 23 years in 1970–1993 and 24 in 2002–2025. A record that uses Bonneville temperature first gave a mean change of −0.46 °C (nominal 95% interval −1.25 to +0.38). A record that uses The Dalles first gave +0.49 °C (−0.22 to +1.24). The paired difference was 0.95 °C (0.57 to 1.47). Thus the source records do not establish the direction of mean temperature change at passage. An exact split gave similar passage-calendar components: −0.50 and −0.51 °C. With the later temperature calendar fixed, the later passage calendar reduced mean temperature by 0.55–0.58 °C. It also reduced overlap with water at or above 20 °C by 6.15–7.55 percentage points. These are descriptive replacements under independent year pairing. The negative passage component reproduces an earlier SCI result for different periods. The source comparison adds a continuous-temperature audit. A raw Bonneville-only check supports the direction of the passage component in a smaller, later sample. The package preserves fixed source files, code, all estimates, uncertainty draws, figures and independent checks. Mixed stocks, blank count cells and changes in temperature sources limit interpretation. The intervals do not include systematic gauge error. This index does not measure fish body temperature, heat dose, survival or adaptation.

sockeye salmon · Bonneville Dam · migration timing · water temperature · source sensitivity · reproducibility · thermal overlap · environmental monitoring