Warmer May–June did not advance the onset of autumn canopy greenness decline at 110 North American PhenoCam deciduous forest sites in 2008–2025, against a reported 1.9 days per °C

Question: does warming before the summer solstice make autumn senescence start earlier, and does warming after it make senescence last longer, as Zohner et al. (2023) reported (−1.9 and +2.6 days/°C)? Data: public PhenoCam archive canopy greenness (gcc_90, 3-day files, processed 2026-10-23) for deciduous broadleaf ROIs in North America, 2008–2025, with Daymet V4 R1 daily temperature at each camera. Method: a plan written before model fitting; one ROI per site; within-site regressions with site fixed effects, and also year effects or site trends; 95% intervals from 2,000 cluster-bootstrap replicates over years and over sites (the wider interval is reported). Outcomes: the PhenoCam standard 50% and 10% falling dates, and our onset and midpoint dates measured from the midsummer greenness level, so that the early-summer Gcc peak does not set the threshold. Results: warmer 1 May–21 June did not make onset earlier: +0.09 days/°C (−1.01 to +1.20; 869 site-years, 110 sites). In 23 model and threshold variants, no lower limit was below −1.77, so the reported −1.9 is excluded. The effect of warmer 22 June–15 September on duration was +0.23 (−1.63 to +2.93), which neither confirms nor rejects +2.6. The standard midpoint had the solstice-switch signs (−0.41 and +1.22, both intervals include zero). With year effects, the values were −1.34 (−2.42 to −0.41) and +2.00 (+0.15 to +3.59), but with site trends they were near zero. In monthly models, warm Julys made onset earlier (−1.20, −2.05 to −0.36) and all autumn dates earlier, while warm August–October made the midpoint and end later (+1.6 to +2.3). Thus the sign change occurs in late summer, not at the solstice. A later spring green-up came with a later autumn (+0.17 to +0.24 days per day). A second camera used as an instrument did not remove this, but its intervals were wide. Limits: provisional archive data, short series (median 7 years per site), 18 year clusters, gridded temperature, mostly eastern US sites, observational design.

autumn phenology · leaf senescence · summer solstice · PhenoCam · green chromatic coordinate · deciduous forest · Daymet · carry-over effect · temperature sensitivity · fixed effects · cluster bootstrap · North America