Recent climate changes have caused vegetation mortality in several ecosystems on the planet. Knowledge about how plants die during drought has advanced greatly in recent decades. However, due to the complexity of the physiological mechanisms involved in plant mortality, we have yet to be able to predict this process reliably. In this project, I propose a hypothesis based on evolutionary optimization theory to predict plant mortality during drought. According to this hypothesis, plants must dynamically adjust their physiological processes, such as photosynthesis, transpiration and carbon storage, to maximize their chances of survival during drought. This hypothesis can also explain how plant communities share water resources in arid environments. We will experimentally test this hypothesis with field observations from plant communities along a precipitation gradient.