Understanding Plant Water Stress and the Terrestrial Carbon Cycle in Tropical Ecosystems: the Roles of Plant Hydraulics, Phenology, and Competition

Abstract

Tropical terrestrial ecosystems play an important role in regulating land surface processes. However, the fates of tropical ecosystems are highly uncertain from predictions by terrestrial biosphere models (TBMs) partly because plant water stress functions in TBMs are often highly empirical and are inconsistent with the ecophysiology of plant water use. In this dissertation, I evaluate the importance of three key ecophysiological processes in plant water use: 1) variation in plant hydraulic traits that regulate water use, 2) canopy phenology that responds to seasonal water stress and 3) competition between species with differential plant water use strategies. I further propose frameworks to incorporate these processes into TBMs.

First, I investigated seasonally dry tropical forests (SDTFs) where coexisting plant groups employ contrasting strategies. I evaluated the role of both plant hydraulics strategy and canopy phenology in coping with drought stress by developing a novel hydraulic module in the ED2 model. Combining modelling and observations at SDTFs in Costa Rica, I showed that the incorporation of diversity in plant hydraulics and drought-driven phenology generated better spatio-temporal predictions on vegetation dynamics under seasonal rainfall climate. Second, I explored the canopy phenological patterns in Panamanian tropical moist forests with milder seasonal water stress. I tested the hypothesis that variations of leaf longevity, a major component of phenology, could be largely explained by the optimization of leaf lifetime carbon gain. Furthermore, the carbon optimality framework shows future potential to account for the timing of leaf exchanging and suggests that seasonal variations in atmospheric water demand may contribute to the canopy phenology in tropical moist forests. Third, I examined tropical savannas to investigate how tree-grass competition for soil water resources influences ecosystem response to changes in hydroclimatic regimes. To do this, I calibrated a biophysical tree-grass competition model coupled with a stochastic rainfall generator. I showed that tree-grass competition could explain the sensitivity of landscape tree cover to rainfall frequency in African savannas. Using multiple pan-tropical remote-sensing datasets, I further showed that this phenomenon is consistent across the global tropics. Altogether, these results help to enhance our understanding of tropical ecosystem dynamics under plant water stress.