Death by dehydration: How savanna trees die in drought

植物如何应对干旱正在成为一个关键的不确定性，限制了我们预测生态系统如何应对气候变化的能力。最近的研究质疑的既定理论，假设水力故障，碳饥饿和生物相互作用的组合可以预测干旱死亡率（麦克道尔模型）。我们的目的是测试一个替代假设：在干旱期间，脱水，而不是水运输故障诱导生理性能的降低，这可能导致死亡。为了实现这一目标，我们建议监测个别稀树草原树木和操纵他们的供水。我们将在稀树草原生态系统中进行实验，因为稀树草原树木的特征（高水平的重新发芽和高非结构性碳储量）似乎违反了麦克道尔模型的前提。我们将监测54个不同的物种（6种），大小和竞争力的邻居稀树草原树木。对于每棵树，监测变量将包括液流、茎含水量、叶水势、叶面积和茎直径。草和土壤蒸散也将在每棵树的邻居监测。我们将使用一种新的状态空间模型来解释三个生长季节中单个树木的动态。在过程层面上，我们将加深理解，通过使用状态空间建模来推断难以观察的过程，如从根表面吸收水分，从测量的树液流，茎含水量和土壤含水量的观察。由于状态空间方法依赖于观测数据系列的过程中的变化，我们通过使用BACI（前后控制影响）实验设计对个别树木进行实验性干旱来创建额外的变化。拟议研究的主要目标是检验我们的工作假设（即在干旱期间导致死亡的是脱水，而不是水运输失败）。我们将测试这一假设，检查是否茎生长，茎死亡率和整个植物死亡率在我们的时间序列中观察到的最好的解释由工厂的模拟储水水库或工厂的模拟水运输赤字。该研究将评估是否需要通过正式考虑脱水死亡假设来扩展McDowell模型。

How plants respond to drought is emerging as a critical uncertainty that limits our ability to forecast how ecosystems will respond to climate change. Recent studies question the established theory which posits that a combination of hydraulic failure, carbon starvation and biotic interactions can predict drought mortality (the McDowell model). We aim to test an alternative hypothesis: during drought dehydration rather than water transport failure induces reductions in physiological performance which can lead to mortality. To achieve this, we propose to monitor individual savanna trees and manipulate their water supply. We will perform our experiment in a savanna ecosystem because the traits of savanna trees (high levels of resprouting and high non-structural carbon reserves) appear to violate the premises of the McDowell model. We will monitor 54 savanna trees that differ in species (6 species), size and competitive neighbourhood. For each individual tree, monitoring variables will include sap flow, stem water content, leaf water potential, leaf area and stem diameter. Grass and soil evapotranspiration will also be monitored in each trees' neighbourhood. We will use a novel state-space model to interpret the dynamics of individual trees over three growing seasons. On a process level we will deepen understanding by using state space modelling to infer difficult to observe processes, such as the uptake of water from the root surface, from measured observations of sap flow, stem water content and soil water content. Because the state space approach relies on variation in processes underlying the observational data series, we create additional variation by experimentally imposing drought on individual trees using a BACI (before after control impact) experimental design. The proposed research's primary objective is to test our working hypothesis (i.e. dehydration rather than water transport failure leads to mortality during droughts). We will test this hypothesis by examining whether stem growth, stem mortality and whole plant mortality observed in our time series are best explained by the plant's modelled water storage reservoir or by the plant's modelled water transport deficit. The study will evaluate whether the McDowell model needs to be expanded by formally considering the death by dehydration hypothesis.