净生态系统生产力（Net Ecosystem Productivity, NEP）是生态系统碳吸收与呼吸消耗的差值，表征了生态系统的碳源/汇。准确了解NEP的变化趋势对于实现碳达峰和碳中和的“双碳”目标具有重要意义。在气候变化的背景下，干旱等极端气候事件频发，干旱风险持续增加，深入了解干旱背景下碳源/汇的时空演变及其驱动机制，对于更好地理解生态系统对气候的调节功能，以及碳平衡的动态过程具有重要意义。

本文基于FLUXNET 2015、ChinaFLUX、黑河流域站点的观测数据，以及全球降尺度的CHELAS等气候数据、叶面积指数（Leaf Area Index, LAI）等遥感产品，计算了77个站点（823站年）的标准化降水蒸散指数（Standardized Precipitation Evapotranspiration Index, SPEI）、冠层导度（Canopy Conductance, Gc）等指标，分析了不同生态系统的干旱环境与碳通量的时空变化特征，揭示了干旱累积时间、干旱强度、干旱发生时间对生态系统碳源/汇的影响，并进一步研究了碳源/汇变化的驱动机制。本研究的主要结论如下：

（1）通过77个站点（823站年）的数据分析发现，全球干旱趋势相对显著，夏半年是干旱的频发时段，主要表现为轻度和中度干旱，极端干旱则较少发生，站点间的干旱频率差异与其所处的气候区、纬度位置相关。对于碳通量而言，大部分生态系统都具有较好的碳汇能力（NEP>0），且各生态系统的植被总初级生产力（Gross Primary Productivity, GPP）和呼吸（Ecosystem Respiration, Re）具有相似的分布特征。其中，森林生态系统（DBF、EBF、ENF、MF）的平均固碳能力（GPP均值为125.84~190.79 gC·m-2·month-1）普遍高于其他植被类型（16.00~135.75 gC·m-2·month-1）。除EBF外，北半球生态系统的NEP月均值随时间变化呈现出倒“U”型，南半球则为“U”型。

（2）干旱对生态系统的影响具有累积效应，其中GPP和Re的干旱累积时间分别集中在3个月和4个月，相比之下，NEP的累积时间较为分散，累积时间为2个月的站点数量最多。随着干旱强度的增加，GPP和Re在3、6、12个月的累积时间下均有显著变化（p<0.05），主要表现为降低趋势。相比之下，NEP受短期（1、3个月尺度）累积干旱强度的影响并不显著（p>0.05），但随着中期（6个月尺度）和长期（12个月尺度）累积干旱强度的增加，NEP分别呈现出先增加后减少和显著降低的趋势（p<0.05）。此外，在海洋性气候区、大陆性湿润气候区和亚寒带气候区，春季累积干旱对NEP具有一定的促进作用，秋季干旱则普遍导致NEP的降低，其他气候区的植被在干旱条件下的NEP普遍低于正常情况。

（3）生态系统NEP在干旱环境下变化的直接原因是GPP的相对变化高于Re。在影响NEP变化的气候、地理、生物三类因素中，生物因素中的LAI在单一环境条件（正常或干旱）下均展现出最高的贡献率。当干旱发生时，下垫面类型（Land Cover, LC）和Gc对NEP的影响有所增加，这使得生物因素替代气候因素，成为NEP变化的主导因素。在干旱环境下，不同生态系统的生理结构和气候适应性不同，因此NEP的驱动机制也具有明显差异。对于大部分植被而言，在影响NEP变化的三种因素中，气候因素（以植被蒸腾对应的能量LEc和潜热通量LE为主）的贡献率最高，其次是生物因素（以LAI和Gc为主），地理因素的影响相对较小。

Net ecosystem productivity (NEP) is the difference between ecosystem carbon absorption and respiration consumption, representing the carbon source/sink of the ecosystem. Accurately understanding the changing trend of NEP is of great significance for achieving the "dual carbon" goals of carbon peaking and carbon neutrality. Meanwhile, against the background of climate change, frequent climate events such as droughts are occurring, with the risk of droughts continuing to escalate. A deep understanding of the spatiotemporal evolution of carbon sources/sinks and its driving mechanisms under drought conditions is of great significance for better understanding the regulatory function of ecosystems on climate, as well as the dynamic adjustment process of carbon balance.

This study used various datasets, including site observation data from FLUXNET 2015, ChinaFLUX, and the Heihe River Basin, as well as regional products like globally downscaled climate data (e.g., CHELAS) and remote sensing products such as leaf area index (LAI). Through the computation of standardized precipitation evapotranspiration index (SPEI) and canopy conductance (Gc) across 77 sites (823 site-years), the paper revealed the spatial and temporal characteristics of drought and carbon sources/sinks, the effects of cumulative drought duration, drought intensity, and drought timing on ecosystem carbon source/sink dynamics, and the driving mechanisms behind these changes. The main conclusions drawn from this study are as follows:

(1) The analysis of data from 77 sites (823 site-years) reveals a relatively significant global trend of aridity, with the summer half-year being a period of frequent aridity, mainly manifested as mild to moderate aridity, while extreme aridity occurs less frequently. The difference in drought frequency between sites is related to their climate zone and latitude. Regarding carbon flux, most ecosystems exhibit good carbon sink capacity (NEP>0), with similar distribution characteristics of ecosystem respiration (Re) and gross primary productivity (GPP) across ecosystems. Among them, forest ecosystems (DBF, EBF, ENF, MF) generally have higher average carbon sequestration capacities (with the mean GPP ranging from 125.84 to 190.79 gC·m-2·month-1) compared to other vegetation types (16.00~135.75 gC·m-2·month-1). Except for EBF, the monthly mean NEP of ecosystems in the Northern Hemisphere shows a inverted "U" shape over time, while in the Southern Hemisphere, it follows a "U" shape..

(2) The impact of drought on ecosystems has cumulative effect, with the cumulative time for GPP and Re concentrated at 3 months and 4 months respectively, while the cumulative time for NEP is more dispersed. In comparison, among the affected sites, the majority experience a cumulative time of 2 months. With increasing drought intensity, both GPP and Re show significant changes at the time scales of 3, 6, and 12 months (p < 0.05), mainly manifesting as a decreasing trend. In contrast, the influence of short-term droughts (e.g., 1- and 3-month drought) intensity on NEP is not significant (p > 0.05). However, as medium-term (6-month) and long-term (12-month) drought intensities increase, there is an observed trend of initial increase followed by decrease and significant decrease, respectively, in NEP (p < 0.05). Additionally, for marine climate, continental humid climate, and sub-glacial zone climate, cumulative drought in spring has a certain promoting effect on NEP, while autumn drought generally leads to a decrease in NEP. In other climate zones, NEP under drought conditions is generally lower than under normal conditions.

(3) The direct cause of changes in NEP during the drought is that the relative change in GPP is higher than that of Re. Among the climatic, geographic, and biological factors influencing NEP changes, the LAI within the biological factors showed the highest contribution under both normal and drought conditions. When drought occurs, the impact of land cover (LC) and Gc on NEP increases, causing biological factors to replace climatic factors as the dominant driver of NEP changes. In arid environments, for most vegetation types, among the three factors affecting NEP changes, climatic factors (primarily LEc and LE) have the highest contribution rate, followed by biological factors (primarily LAI and Gc), while the influence of geographical factors is relatively small.