A 1 January heat-sum frost measure overpredicted frost after observed leaf-out and fruit-tree flowering after warm winters at German stations, 1959-2026

Zohner et al. (2020) measured late-spring frost risk with climate data only: the heat sum above 0 C from 1 January to the last spring frost. This measure is a valid risk index only if plants need the same heat sum every year. We tested this assumption with 1.07 million DWD onset records (leaf unfolding of five trees, flowering of four fruit trees) from 6,079 German phenology stations, each paired with daily minimum and mean temperature from the nearest DWD climate station (median 11 km). Observed exposure is a frost after the observed onset day to 30 June. Proxy exposure places onset on the day when the 1 January heat sum reaches the 1959-2017 station mean. For each extra 50 degree-days in January-February, proxy exposure (below 0 C) increased by 12.4 percentage points for leaf-out and 11.9 for flowering, but observed exposure increased by only 2.0 (95% CI -1.3 to 5.3) and 4.6 (0.5 to 8.7). The observed-minus-proxy difference was negative with a CI below zero in all 23 replication subsets (9 taxa, 4 regions, 3 periods including 2018-2026), also at -2 C. Winter heat raised the heat sum at onset by 35-49 degree-days per 50 and advanced onset by only 2-5 days. A heat sum above 5 C from 1 March removed most of the bias for leaf-out. Observed exposure at -2 C had no clear 1959-2017 trend (+0.7 pp per decade, CI -1.9 to 3.3, leaf-out; +0.3, CI -2.9 to 3.5, flowering), and a 1959-2017 trend model predicted 2018-2026 leaf-out exposure (28.3% observed, 28.0% predicted), while the proxy gave 42.8%. Limitations: 2 m screen temperature at a nearby station, no damage data, unknown observer changes, Germany only, and the full-period trend gap is a late-1980s step rather than a steady trend. An increase in the climate-only measure is not sufficient evidence of more frost on leaves or flowers.

late spring frost · phenology · leaf-out · flowering · growing degree days · climate change · Germany · DWD · false spring · chilling · frost risk · observation bias