When Rain Meets Heat: Drivers of Peak Carbon Uptake in East African Drylands

Semi-arid landscapes dominate much of Kenya, yet their contribution to regional carbon cycling remains poorly constrained, particularly regarding how peak ecosystem photosynthetic capacity responds to highly variable wet-season rainfall. Here, we synthesize eddy covariance observations from four contrasting Kenyan dryland ecosystems, including natural savannas—a managed savanna grassland at Kapiti and a wooded savanna at Choke—and croplands—a smallholder system at Maktau and a commercial farm at Ausquest. We examine how rainfall, canopy development, and atmospheric demand jointly regulate maximum net ecosystem CO₂ uptake (NEEₘₐₓ) during the wet season, when most annual carbon assimilation occurs and interannual variability in precipitation pulses is pronounced.

Site-specific relationships between rainfall and NEEₘₐₓ were derived, and responses to temperature and vapour pressure deficit (T–VPD) were analysed under light-saturated conditions to disentangle water supply effects from atmospheric constraints on photosynthesis. Across all sites, rainfall primarily acted as a trigger for peak carbon uptake, with NEEₘₐₓ increasing rapidly following rainfall onset but saturating once sufficient soil moisture supported canopy development. In natural savanna ecosystems, increasing rainfall consistently led to higher maximum leaf area index (LAIₘₐₓ) and enhanced NEEₘₐₓ, while differences between grassland and wooded savanna reflected contrasts in vegetation structure and rooting depth. In contrast, croplands exhibited a muted rainfall–NEEₘₐₓ response, with peak uptake largely governed by cropping cycles, crop type, and management practices rather than total rainfall amounts.

Under high-light conditions, temperature and VPD imposed a common upper bound on NEEₘₐₓ across all ecosystems, defining a narrow envelope of maximum photosynthetic capacity. These results demonstrate that peak carbon uptake in East African drylands emerges from interacting controls of rainfall timing, canopy development, vegetation structure, and atmospheric demand and is modulated by management and land use. Our findings provide critical constraints for land–atmosphere coupling in understudied dryland regions and have important implications for modelling carbon cycle responses under increasing rainfall variability and land-use change.