An Evaluation of Spatial and Temporal Heterogeneities in the Carbon and Water Cycles of Savanna Ecosystems

Abstract

Carbon cycling in dryland ecosystems is complicated by three interrelated factors that pose challenges in characterizing current dynamics and predicting future change: 1) large pulses of ecosystem respiration occur when dry soils are rewetted, 2) a patchy distribution of vegetation leads to spatial heterogeneity in carbon stocks and fluxes, and 3) a large fraction of carbon is stored in belowground pools making it inherently more difficult to study. In this dissertation, I seek to address these issues through the development and application of landscape-scale ecosystem models, as well as field experiments and field observations from the Kalahari savannas of Southern Africa. In Chapter 2, I use new measurement techniques to characterize spatial and temporal variability. I compare the spatial pattern of soil carbon, woody plant canopies and root systems, which I mapped using ground penetrating radar, and use continuous in situ measurements of soil CO2 concentrations during experimental wetting treatments to determine the relationship between soil moisture and soil respiration at a fine temporal scale. Woody plant roots are the primary determinant of the spatial distribution of soil carbon, and soil respiration responds to fluctuations in soil moisture in a way that most large-scale models are unable to account for. To account for these factors I develop in Chapter 3 a steady-state, semi-analytical model of soil carbon stocks that uses a probabilistic description of vegetation structure and a multiplicative noise approximation of decomposition dynamics. The model results are sensitive to the parameters describing the spatial extent of woody plant root systems. I present the results of the excavation and mapping of complete root systems in Chapter 4 that better characterize rooting structure for modeling applications. I observed a high degree of species-level diversity that is not accounted for in the previously presented modeling framework. I conclude in Chapter 5 by developing a stochastic above- and belowground model of the abundance and distribution of biomass that incorporates these new results. It provides a framework for the future development of models of dryland carbon, water, and energy balance.