A constant thermal-time model accounts for the change in leaf-out sensitivity to spring temperature at German and Swiss stations in 1951-2024; the 1999-2013 decrease did not continue in 2014-2023

Question: Fu et al. (2015) reported that spring leaf unfolding in European trees became less sensitive to preseason warming from 1980-1994 to 1999-2013. Is this a physiological change (less chilling), or the change that a constant thermal-time requirement gives in warmer springs, and did it continue? Data: 757,136 DWD leaf-out records (6 species, 6,472 German stations, 1951-2023) and 33,191 MeteoSwiss records (5 taxa, 175 stations, 1951-2024), with E-OBS v31.0e daily temperature. Method: within-station slopes of leaf-out day on the fixed 60-day preseason mean, by period, with two-way (station, year) cluster intervals and a year bootstrap. A forcing-only model with constant physiology (M1) and a chilling model (M2) were fitted on 1951-1979 only and applied to all years; excess change = observed minus M1 slope change. Plan written before outcome fits. Results: the German slope was -4.04 days per C in 1980-1994, -2.96 in 1999-2013 and -3.71 in 2014-2023. The Fu-period decrease (+1.08, 95% CI +0.14 to +2.02) was borderline in the year bootstrap (-0.04 to +1.98) and did not continue. M1 gave about half of it; the excess (+0.51, CI -0.51 to +1.52) and all excess changes relative to 1951-1979 (-0.43 to -0.13) included zero; 4 of 221 excess contrasts were above zero and 6 below, near chance. Swiss changes were small; M1 predicted a larger decrease. At the same spring temperature, warmer winters did not delay leaf-out in 11 of 11 samples. Relative to M1, German leaf-out in 2014-2023 was 3.8 days earlier (6 of 6 species), but a model with forcing from 1 January gave later-than-predicted dates (+2.3 days Germany, +5.8 days Switzerland), so the sign of this drift depends on model structure. Spring radiation increase explained about 11% of the German drift; an East-West test for sulfur dioxide decrease showed no consistent pattern. Limitations: simple models without photoperiod, no observer identity, shrinking network, E-OBS not homogenised (Swiss -0.12 to -0.19 C drift against Berkeley Earth), exploratory pollution checks.

spring phenology · leaf unfolding · temperature sensitivity · thermal time model · chilling · climate change · DWD phenology · MeteoSwiss phenology · E-OBS · Germany · Switzerland · Fagus sylvatica · Betula pendula · Quercus robur · Aesculus hippocastanum · observation bias · solar brightening · out-of-sample test · two-way clustering · replication