Article révisé par les pairs
Résumé : Constraints of temperature on spring plant phenology are closely related to plant growth, vegetation dynamics, and ecosystem carbon cycle. However, the effects of temperature on leaf onset, especially for winter chilling, are still not well understood. Using long-term, widespread in situ phenology observations collected over China for multiple plant species, this study analyzes the quantitative response of leaf onset to temperature, and compares empirical findings with existing theories and modeling approaches, as implemented in 18 phenology algorithms. Results show that the growing degree days (GDD) required for leaf onset vary distinctly among plant species and geographical locations as well as at organizational levels (species and community), pointing to diverse adaptation strategies. Chilling durations (CHD) needed for releasing bud dormancy decline monotonously from cold to warm areas with very limited interspecies variations. Results also reveal that winter chilling is a crucial component of phenology models, and its effect is better captured with an index that accounts for the inhomogeneous effectiveness of low temperature to chilling rate than with the conventional CHD index. The impact of spring warming on leaf onset is nonlinear, better represented by a logistical function of temperature than by the linear function currently implemented in biosphere models. The optimized base temperatures for thermal accumulation and the optimal chilling temperatures are species-dependent and average at 6.9 and 0.2°C, respectively. Overall, plants’ chilling requirement is not a constant, and more chilling generally results in less requirement of thermal accumulation for leaf onset. Our results clearly demonstrate multiple deficiencies of the parameters (e.g., base temperature) and algorithms (e.g., method for calculating GDD) in conventional phenology models to represent leaf onset. Therefore, this study not only advances our mechanistic and quantitative understanding of temperature controls on leaf onset but also provides critical information for improving existing phenology models.