Causal Inference Reveals the Dominant Role of Interannual Variability of Carbon Sinks in Complicated Environmental-Terrestrial Ecosystems

Chaoya Dang,Zhenfeng Shao,Peng Fu,Qingwei Zhuang,Xiaodi Xu,Jiaxin Qian

DOI: https://doi.org/10.1016/j.rse.2024.114300

IF: 13.5

2024-01-01

Remote Sensing of Environment

Abstract:Climate factors (CFs) are key variables shaping the interannual variability (IAV) of terrestrial ecosystem carbon sinks. However, the dominant CFs influencing the IAV of terrestrial carbon sinks remains debated, as CFs are coupled via land-atmosphere interactions. Here, the dominant factors influencing the IAV of global terrestrial net ecosystem production (NEP) were quantified using the convergent cross-mapping (CCM) technique. This analysis was conducted with distinct global terrestrial NEP datasets deriving from process-based ecosystem models, machine learning techniques, and eddy covariance flux towers. Results revealed that the spatial patterns of IAV of global terrestrial NEP were dominated by water availability (WA) and temperature (Ts). Ts mainly controlled the IAV of terrestrial NEP in mid to high-latitude regions of the Northern Hemisphere (NH), while WA exerted dominance over low and mid-latitude regions in both the NH and the Southern Hemisphere. Moreover, the energy limitation and water limitation explained the spatial pattern of Ts and WA dominant on NEP. Further analysis found that WA and Ts also played a dominant role in gross primary productivity (GPP) and terrestrial ecosystem respiration (TER), proving that WA and Ts were the dominant factors affecting NEP. In addition, we found a weakening trend in causal linkages of CFs to NEP in the temporal domain. This study used causal analysis to reveal the spatial patterns of water and heat dominating the NEP, providing support for improved assessment and prediction of terrestrial carbon sinks under climate change.