Process-based phenological models are currently used for assessing the effects of climatic warming on the timing of spring phenological events, such as leafout and flowering, in trees. However, the biological realism of the models may be undermined by the practices of often formulating the models solely on the basis of observational records of the phenological event rather than addressing the physiological processes actually modelled. Here we introduce a framework for developing process-based phenological models on the basis of experiments explicitly designed for this purpose and apply the framework to developing process-based models for leafout in four subtropical tree species. Our method is based on a hypothetico-deductive approach, where the air temperature responses of the simulated processes are inferred from their implications for the occurrence and timing of leafout in the experimental conditions. That approach has only rarely been taken with boreal or temperate trees, and to the best of our knowledge, never before with subtropical trees. Big differences were found between the four species in the air temperature responses modelled, and these differences implied major differences in the dormancy dynamics predicted for the four species. Together with the results of a recent modelling study based on observational data, our results highlight the importance of experimental studies for the development of biologically realistic process-based models of spring phenology in trees. The framework developed in the present study can be applied to developing such models for all tree species that show the phenomena of rest (endodormancy) and chilling requirement, no matter whether the trees are boreal, temperate, or subtropical.

基于过程的物候模型目前用于评估气候变暖对春季物候事件（例如树木的落叶和开花）时间的影响。然而，模型的生物学真实性可能会因经常仅根据物候事件的观察记录而不是解决实际建模的生理过程来制定模型的做法而受到破坏。在这里，我们在为此目的明确设计的实验的基础上引入了一个用于开发基于过程的物候模型的框架，并将该框架应用于开发基于过程的四种亚热带树种的叶出模型。我们的方法基于假设演绎法，其中模拟过程的空气温度响应是从它们对实验条件下叶子脱落发生和时间的影响推断出来的。北方或温带树木很少采用这种方法，据我们所知，亚热带树木以前从未采用过。在模拟的空气温度响应中发现四种物种之间存在很大差异，这些差异意味着对四种物种预测的休眠动态存在重大差异。连同最近基于观察数据的建模研究的结果，我们的结果强调了实验研究对于开发基于生物现实过程的树木春季物候学模型的重要性。