Adult male birds of 105 MAPS species were 0.2% lighter per °C of same-summer heat with year controls (1989–2018); future-summer heat and wing length showed no effect, and the mass trend was uncertain

Question: Is the temporal link between breeding-season temperature and adult male bird mass in MAPS banding data (Youngflesh et al. 2022) a year-to-year response or an artefact of shared trends? Data: 253,488 captures of 105 North American landbird species at 1,124 stations, 1989-2018 (CC0 table of the original study), joined to May-July mean daily maximum temperature from Daymet V4 (1 km) and, as a check, CRU TS 4.10. Methods: fixed-effects regressions within species x station, with year or species x year fixed effects, a split of temperature into station trend and anomaly, temperature 1-2 summers before and after capture (future summers are a placebo), wing chord as a structural control, cell-mean and Huber fits, a 200-fold year permutation, and two-way clustered errors. Results: In hotter summers, birds were lighter: -0.195% mass per °C (95% CI -0.289 to -0.102; -0.026 SD/°C) with year fixed effects, which is about two times the published -0.012 SD/°C. Detrended anomalies gave the same slope. Temperature 1 and 2 summers after capture had no effect (-0.002 and +0.003 SD/°C). No permuted series gave a slope as large as the observed one. National mass year effects followed the temperature anomaly (r = -0.65; first differences -0.80); the lightest years were the hot years 2006, 2007 and 2012. Wing chord did not respond to same-summer heat (-0.006 SD/°C, CI -0.013 to +0.001). The pooled mass trend (-0.16% per decade) agrees with the published decline, but it is clear only if each bird is independent; with station, year or two-way clustering, p = 0.08-0.24. With the same-summer temperature term, the trend fell by 43%. The heat slope times local warming predicts -0.06% per decade. CRU temperature gave the same results. Conclusion: in these data the temperature-mass link is a reversible same-summer mass response, not a cumulative structural shrinkage, and the 30-year mass decline is small and uncertain. Limitations: the table has no capture dates, ages, bander or individual identity; capture-timing effects in hot years can contribute; wing chord is a weak structural measure; we did not refit the original Bayesian model.

body size · Bergmann's rule · climate change · bird banding · MAPS · body mass · temperature-size rule · placebo test · fixed effects · Daymet · re-analysis · North America