Abstract
Abstract
Background
Arid regions serve as critical ecological barriers in Northern China, yet their carbon cycle stability is increasingly threatened by intensifying climate extremes. This study utilizes 1982–2022 extreme climate indices alongside vegetation Net Ecosystem Productivity (NEP) and Carbon Use Efficiency (CUE) to elucidate the spatiotemporal dynamics and driving mechanisms of the terrestrial carbon cycle. Integrated Theil-Sen trend analysis, Mann–Kendall tests, Geographical Detectors, and Structural Equation Modeling (SEM) to quantify the response of NEP and CUE to 16 extreme temperature and precipitation indices across an aridity gradient.
Results
(1) From 1982 to 2022, extreme temperatures in the arid regions of China underwent asymmetric warming dominated by nighttime temperature rises. The precipitation regime featured increased extreme heavy rainfall in the west and intensified aridification in the east, with a distinct trend of wet-dry polarization. (2) From 1982 to 2022, vegetation NEP in the arid regions of China assumed an overall distribution pattern of eastward increase and westward decrease; CUE demonstrated a staggered distribution pattern along the aridity gradient. (3) The interactive effects of extreme climate factors significantly outweighed single-factor impacts, with NEP demonstrating higher sensitivity than CUE. A dominant synergistic pattern between TNx (Minimum temperature extreme value) and extreme precipitation was identified. SEM results revealed that in semi-arid regions, the optimization of hydrothermal coupling maximized the "NPP-to-NEP" conversion efficiency by bypassing respiratory losses, whereas CUE stability was maintained by a biological offset effect between GPP and NPP.
Conclusions
Under climate extremes, semi-arid ecosystems exhibit a distinct dual response: NEP acts as a highly sensitive carbon ‘hotspot, ‘ while the biological offset effect maintains CUE stability. These findings support targeted carbon management and ecological restoration in fragile arid regions.