Accurate estimations of phenophases in deciduous trees are important to understand forest ecosystems and their feedback on the climate. In particular, the timing of leaf senescence is of fundamental importance to trees’ nutrient stoichiometry and drought tolerance and therefore to trees’ vigor and fecundity. Nevertheless, there is no integrated view on the significance, and direction, of seasonal trends in leaf senescence, especially for years characterized by extreme weather events. Difficulties in the acquisition and analyses of hierarchical data can account for this. We collected four years of chlorophyll content index (CCI) measurements in thirty-eight individuals of four deciduous tree species (Betula pendula, Fagus sylvatica, Populus tremula and Quercus robur) in Belgium, Norway and Spain, and analyzed these data using generalized additive models for location, scale and shape (GAMLSS). As a result, (I) the phenological strategy and seasonal trend of leaf senescence in these tree species could be clarified for exceptionally dry and warm years, and (II) the daily average (air) temperature, global radiation, and vapor pressure deficit could be established as main drivers behind the variation in the timing of the senescence transition date. Our results show that the onset of the re-organization phase in the leaf senescence, which we approximated and defined as local minima in the second derivative of a CCI graph, was in all species mainly negatively affected by the average temperature, global radiation and vapor pressure deficit. All together the variables explained 89 to 98% of the variability in the leaf senescence timing. An additional finding is that the generalized beta type 2 and generalized gamma distributions are well suited to model the chlorophyll content index, while the senescence transition date can be modeled using the normal-exponential-student-t, generalized gamma and zero-inflated Box-Cox Cole and Green distributions for beech, oak and birch, and poplar, respectively.

准确估计落叶树的物候期对于了解森林生态系统及其对气候的反馈非常重要。特别是，叶片衰老的时间对树木的营养化学计量和耐旱性至关重要，因此对树木的活力和繁殖力至关重要。然而，对于叶子衰老的季节性趋势的重要性和方向，没有统一的观点，特别是在以极端天气事件为特征的年份。获取和分析分层数据的困难可以解释这一点。我们收集了 4 种落叶乔木（Betula pendula、Fagus sylvatica、Populus tremula和Quercus robur）的 38 个个体的四年叶绿素含量指数 (CCI) 测量值) 在比利时、挪威和西班牙，并使用位置、规模和形状的广义加法模型 (GAMLSS) 分析了这些数据。因此，（I）这些树种叶片衰老的物候策略和季节趋势可以在异常干燥和温暖的年份得到明确，并且（II）日平均（空气）温度、全球辐射和蒸汽压不足可以被确定为衰老过渡日期时间变化背后的主要驱动因素。我们的研究结果表明，叶片衰老重组阶段的开始，我们将其近似并定义为 CCI 图的二阶导数中的局部最小值，在所有物种中主要受到平均温度、全球辐射和蒸汽的负面影响压力不足。所有这些变量共同解释了叶片衰老时间变化的 89% 到 98%。另一个发现是广义 2 型和广义 gamma 分布非常适合模拟叶绿素含量指数，而衰老过渡日期可以使用正态指数学生-t，分别为山毛榉、橡木、桦木和杨树的广义 gamma 和零膨胀 Box-Cox Cole 和 Green 分布。