Gross primary production variability associated with meteorology, physiology, leaf area, and water supply in contrasting woodland and grassland semiarid riparian ecosystems

Understanding ecosystem-atmosphere carbon exchanges in dryland environments has been more challenging than in mesic environments, likely due to more pronounced nonlinear responses of ecosystem processes to environmental variation. To better understand diurnal to interannual variation in gross primary productivity (GPP) variability, we coupled continuous eddy-covariance derived whole ecosystem gas exchange measurements with an ecophysiologic model based on fundamental principles of diffusion, mass balance, reaction kinetics, and biochemical regulation of photosynthesis. We evaluated the coupled data-model system to describe and understand the dynamics of 3 years of growing season GPP from a riparian grassland and woodland in southern Arizona. The data-model fusion procedure skillfully reproduced the majority of daily variation GPP throughout three growing seasons. While meteorology was similar between sites, the woodland site had consistently higher GPP rates and lower variability at daily and interannual timescales relative to the grassland site. We examined the causes of this variation using a new state factor model analysis that partitioned GPP variation into four factors: meteorology, physiology, leaf area, and water supply. The largest proportion of GPP variation was associated with physiological differences. The woodland showed a greater sensitivity than the grassland to water supply, while the grassland showed a greater sensitivity to leaf area. These differences are consistent with hypotheses of woody species using resistance mechanisms, stomatal regulation, and grassland species using resilience mechanisms, leaf area regulation, in avoiding water stress and have implications for future GPP sensitivity to climate variability following wood-grass transitions.