Distinguishing the main climatic drivers of terrestrial vegetation carbon dynamics in pan-Arctic ecosystems

Pan-Arctic terrestrial ecosystems have experienced widespread greening over the past few decades due to climate warming. However, the dynamic variations and climatic causes of such greening in the pan-Arctic are still unclear due to limitations in high-latitude ground-based measurements. The study first evaluated the applicability of three satellite-based vegetation indices (VIs) for pan-Arctic carbon dynamics in previous decades compared with multisource observations. Three VIs presented significant increasing trends up to 0.0005–0.0017 yr −1 for pan-Arctic ecosystems in the past 23 years, and near-infrared reflectance of vegetation (NIRv) could better capture pan-Arctic terrestrial carbon dynamic variations (e.g., increasing trends) than other VIs among most vegetation types, such as grasslands. The separate contributions of climatic factors to satellite-based NIRv variability were further quantified using partial correlation and ridge regression analysis over pan-Arctic ecosystems from 2001 to 2023. Carbon dioxide (CO 2 ) and air temperature (AT) presented positive partial correlations with the satellite-based NIRv over entire pan-Arctic ecosystems, due to fertilization and warming effects. However, the vapor pressure deficit (VPD) showed positive (negative) partial correlations with NIRv variations in high (low) pan-Arctic ecosystems because of diverse stomatal responses to air dryness. Therefore, increases in VPD contributed 11.2 % of the pan-Arctic NIRv variability, although positive (negative) effects were detected for high (low) pan-Arctic ecosystems. These findings highlight the advantages of the NIRv for representing terrestrial carbon dynamics in fragile ecosystems (e.g., the pan-Arctic) and reveal the diverse effects of the regional climate (e.g., air dryness) on pan-Arctic ecosystems under climate warming. • Three VIs (NDVI, NIRv and kNDVI) showed significant increasing trends across pan-Arctic ecosystems. • NIRv can better capture observed pan-Arctic greening trends than other vegetation indices. • VPD contributed NIRv variability with positive (negative) effects in high (low) pan-Arctic ecosystems.