Fig. 4. Changes in spring phenology averaged for all species and stations for 2019–2099.

Spring phenological events were simulated using linear regression between chilling accumulations based on the 12 chilling models (from 1 November to the onset of spring events) and the heat requirement (accumulated temperature >0 °C from 1 January to the onset of spring events) for all records. The observed (thick gray line) and simulated trends across spring events of all species over 1980–2018 are shown. Red colors represent valid chilling models (C₁, C₂, C₄, C₅, and C₁₂), and blue colors represent invalid models. a An 11-year moving average of spring phenological events under Representative Concentration Pathways (RCP) 4.5. b An 11-year moving average of spring phenological events under RCP 8.5. c Advance in spring events by the 2090s compared to the 2010s under RCP 4.5. d Advance in spring events by the 2090s compared to the 2010s under RCP 8.5. Error bars in (c) and (d), mean values ± SD (n = 30 species).