Abstract:

[Objective] Net ecosystem productivity (NEP), defined as the difference between vegetation net primary productivity and soil heterotrophic respiration, directly reflects net carbon exchange between terrestrial ecosystems and the atmosphere and better characterizes regional carbon budgets than net primary productivity. Chongqing, which combines extensive mountainous terrain with a large metropolitan area, is representative of mountainous cities where fragmented topography and intensive human activities produce complex spatial differentiation in NEP. However, how biophysical conditions and human activities interact to drive carbon source-sink differentiation remains unclear, and few studies have characterized its spatiotemporal evolution from a broad-scale geographical perspective. This study aims to reveal the spatiotemporal evolution of NEP in Chongqing and clarify the mechanisms driving carbon source-sink patterns in mountainous cities from the perspective of natural-anthropogenic interactions. [Methods] Using multisource remote-sensing and meteorological data from 2003 to 2023, we estimated NEP in Chongqing at a 250 m resolution using the CASA-GSMSR model, which couples the CASA light-use-efficiency model with the GSMSR geostatistical soil respiration model. The estimates were evaluated against three NEP products (ZENG, FLUXCOM, and BESS) in terms of interannual variability and magnitude. Pixel-scale NEP trends and their significance were assessed using Theil-Sen and Mann-Kendall analyses. NEP was classified into six classes: high and low carbon-source classes, a carbon-balance class, and low, moderate, and high carbon-sink classes. Standard deviational ellipse (SDE) analysis and gravity-center migration modeling were used to quantify the spatial extent, distribution orientation, and migration trajectory of each class. Geodetector was applied to 14 natural and anthropogenic factors to identify the dominant factors, optimal ranges associated with high NEP, and the types and strengths of pairwise interactions. [Results] The estimated NEP exhibited interannual variations broadly consistent with those of the three reference products, with the strongest correlation observed for BESS (r = 0.84). Over the common period (2003‒2015), the multiyear mean NEP estimated in this study was 188.05 g·m⁻²·a⁻¹, comparable in magnitude to ZENG (166.68 g·m⁻²·a⁻¹) and FLUXCOM (165.56 g·m⁻²·a⁻¹). These results indicate that our estimates are consistent in magnitude with flux-based machine-learning upscaling products while retaining the interannual variability captured by process-based mechanistic models. Moreover, the 250 m resolution provides an advantage over existing coarse-resolution products for characterizing carbon source-sink heterogeneity in complex terrain. From 2003 to 2023, NEP in Chongqing increased with fluctuations at an average rate of 7.03 g·m⁻²·a⁻¹. The trend can be divided into two distinct stages. During the first stage (2003‒2015), NEP increased rapidly at a rate of 10.07 g·m⁻²·a⁻¹, whereas during the second stage (2015‒2023), it remained relatively stable but declined slightly at a rate of 4.54 g·m⁻²·a⁻¹. At the pixel scale, 90.42% of the study area showed an increasing trend, with significant increases occurring across 55.16% of the study area, mainly in northeastern districts and counties such as Kaizhou and Yunyang. Decreasing trends occurred across 9.58% of the study area, whereas significant decreases covered only 2.52% and were strongly concentrated in the central urban districts and the built-up areas of other districts and counties. Widespread increases in NEP offset carbon-sink losses in locally urbanized areas, driving an overall strengthening of the regional carbon-sink function. Spatially, the multiyear mean NEP from 2003 to 2023 was 213.49 g·m⁻²·a⁻¹, with carbon-sink and carbon-source areas accounting for 79.35% and 20.65% of the total area, respectively. Core carbon-sink areas were located in the Daba Mountains of northeastern Chongqing and the Wuling Mountains of southeastern Chongqing, whereas carbon-source areas were concentrated in the central urban districts, western Chongqing, and areas along the Yangtze River. Chongqing's carbon source-sink pattern evolved toward “carbon-source retreat and carbon-sink expansion.” The SDEs of the high and low carbon-source classes and the carbon-balance class all contracted and remained concentrated in western Chongqing. The contraction was greatest for the high carbon-source class, whose SDE area decreased by 11665.80 km² and whose gravity center migrated 54.27 km toward the central urban area. Conversely, the SDEs of the low, moderate, and high carbon-sink classes all expanded, and their gravity centers migrated relatively long distances, all exceeding 63 km. The gravity centers of the moderate and high carbon-sink classes shifted toward the Wuling Mountains in the south, whereas that of the low carbon-sink class shifted westward toward areas of gentler terrain. The factor detector identified vegetation type, air temperature, and elevation, represented by the digital elevation model (DEM), as the dominant factors (q > 0.7). The risk detector indicated that high NEP values occurred in areas with relatively low mean annual temperatures (5.12‒12.10 ℃), abundant annual precipitation (1620‒1790 mm), and high annual total solar radiation (3910‒4050 MJ·m⁻²), particularly in mid- to high-elevation mountainous areas (1380‒2610 m) with slopes of 11.50‒31.80°. Areas characterized by evergreen broadleaf forests, karst landforms, and yellow-cinnamon soils also exhibited relatively high NEP. All pairwise interactions had higher q values than the corresponding individual factors and exhibited bivariate or nonlinear enhancement, indicating synergistic controls on the spatial differentiation of NEP. Anthropogenic factors had limited independent explanatory power (q = 0.13‒0.44), but their interactions with DEM increased q to 0.72‒0.75, indicating pronounced spatial coupling among topography, human activities, and NEP. [Conclusion] The 250 m NEP estimates were consistent in magnitude with flux-based machine-learning upscaling products and in interannual variability with process-based mechanistic products, supporting their overall reliability. From 2003 to 2023, Chongqing’s ecosystems acted predominantly as a carbon sink, with an overall strengthening of the carbon-sink function and a clear pattern of “carbon-source retreat and carbon-sink expansion.” The spatial differentiation of NEP was governed primarily by natural factors, including vegetation type, air temperature, and topography. Anthropogenic factors had limited independent explanatory power, but their explanatory power increased substantially when they interacted with DEM, reflecting close spatial coupling between natural conditions and human activities. In particular, mountainous terrain constrains population concentration and land development and reduces anthropogenic disturbance, thereby allowing high-elevation mountainous areas to retain low population densities and forest-dominated land-use patterns. These conditions help preserve ecosystems with high NEP and play a critical role in maintaining the regional carbon-sink function. These findings provide scientific support for ecosystem carbon accounting, the formulation of differentiated ecological compensation policies, and the low-carbon optimization of territorial spatial patterns in mountainous cities of southwestern China.

Key words: net ecosystem productivity, carbon source/sink, Geodetector, spatiotemporal evolution, mountainous city

摘要：

【目的】揭示重庆市净生态系统生产力（NEP）的时空演变特征及驱动因素，为该区差异化生态补偿和国土空间低碳优化提供依据。【方法】基于遥感和气象数据，采用CASA-GSMSR耦合模型估算250 m NEP，并与3种NEP产品对比评估；运用Theil-Sen趋势分析、Mann-Kendall检验、标准差椭圆、重心迁移模型及地理探测器等方法，分析NEP时空动态、演变轨迹及驱动因素。【结果】估算的NEP年际变化与3种产品一致，量级与ZENG、FLUXCOM相当。2003－2023年NEP总体呈波动上升趋势，2015年前后由快速增长转为稳中略降。重庆市多年平均NEP为213.49 g·m⁻²·a⁻¹，碳源和碳汇面积占比分别为20.65%和79.35%，碳汇核心区位于渝东北、渝东南高海拔山区，碳源区集中在渝西、中心城区及长江沿岸。空间格局呈“源退汇进”演变，碳源区向中心城区收缩集聚，碳汇区持续巩固并向外围扩张。植被类型、气温和DEM是主导NEP空间分异的自然因子，而人口密度等人文因子独立解释力有限，但与DEM交互后解释力明显提升。【结论】重庆市以碳汇为主，碳汇功能总体增强。NEP空间分异由自然因子主导，人文因子与DEM的交互作用揭示了山地地形对人类活动的约束是维持区域碳汇功能的关键。

关键词: 净生态系统生产力, 碳源/汇, 地理探测器, 时空演变, 山地城市