Spatiotemporal dynamics of vegetation carbon sink and its driving mechanisms in the Huaihai Economic Zone, China

Summary

This study aims to advance urban low-carbon transformation and foster a sustainable, livable environment. By utilizing MOD17A3 NPP and meteorological data, the net ecosystem production (NEP) of the Huaihai Economic Zone from 2001 to 2023 was estimated. The research used trend analysis, centroid migration models, the Hurst index, spatial autocorrelation, partial correlation, and the optimal parameter geographic detector (OPGD) to investigate the spatiotemporal evolution and driving mechanisms of NEP. Results indicate an overall increasing trend in NEP, with 57.62% of the region showing growth, although some areas experienced declines. Most areas (51.89%) display weak anti-persistence. Gross domestic product (GDP) and population density significantly account for its spatial heterogeneity. Based on these findings, the study recommends prioritizing ecological restoration in areas characterized by high volatility and weak anti-persistence and optimizing carbon sink strategies in regions where economic growth and vegetation conditions present synergistic advantages.

Graphical abstract

Highlights

• •The annual average NEP in the Huaihai Economic Zone showed a fluctuating upward trend
• •The carbon sink capacity faces a potential risk of future degradation
• •Precipitation has a significantly greater impact on NEP than temperature
• •Among the driving factors, GDP has the strongest explanatory power for spatial variation

Earth sciences; Environmental monitoring; Environmental science

Introduction

Climate change represents a critical global challenge confronting human society. Since the onset of industrialization, rapid socioeconomic development combined with intensive fossil fuel use has triggered severe ecological degradation. The IPCC Sixth Assessment Report (2021) indicates that the global average temperature has risen by 1.1°C relative to pre-industrial levels, resulting in substantial impacts on terrestrial ecosystems.¹ As essential components of the global carbon cycle and primary substrates for biospheric energy-material transformations, terrestrial ecosystems underpin planetary life systems. Enhancing their carbon sequestration capacity is therefore a crucial strategy for mitigating atmospheric greenhouse gas concentrations and slowing global warming.² Consequently, the relationship between terrestrial ecosystem carbon cycling and climate change has become a central focus within international research frameworks such as the Global Carbon Project and the Global Change and Terrestrial Ecosystems Program. Vegetation, a core component of terrestrial ecosystems, sequesters atmospheric CO₂ through photosynthesis, effectively converting carbon sources into sinks and establishing essential carbon fixation mechanisms. Accurate quantification of their carbon sequestration potential, along with analyses of spatiotemporal patterns and identification of driving factors, is vital both for optimizing ecological conservation strategies and for advancing carbon neutrality goals.

With increasing demands for precise carbon accounting, academic focus has shifted toward comprehensive assessments of net ecosystem productivity (NEP).³ As a key parameter of carbon balance, NEP measures the net difference between photosynthetic carbon fixation and ecosystem respiration (including autotrophic and heterotrophic respiration, Rh) across spatiotemporal scales, directly reflecting ecosystem carbon storage efficiency. Its polarity carries clear ecological significance: positive values (NEP >0) indicate a carbon sink, whereas negative values (NEP <0) signify a carbon source.⁴ Particularly for evaluating long-term carbon sequestration potential, NEP integrates both vegetation-driven carbon fixation and carbon release processes such as soil respiration, allowing for a systematic characterization of ecosystems’ sustained ability to offset anthropogenic carbon emissions.

Extensive research on NEP has been conducted worldwide. Notable examples include Verduzco et al.,⁵ who applied coupled ecohydrological and soil carbon models to investigate soil-vegetation interactions in subtropical shrublands of northwestern Mexico under climate change conditions. Their results indicated that rising temperatures suppress NEP, whereas elevated atmospheric CO₂ concentrations can partially offset this effect by enhancing vegetation water-use efficiency. Using 20 years of eddy covariance observations from the Vielsalm site in Belgium, Aubinet et al.⁶ analyzed interannual NEP fluctuations in a mixed beech forest and identified the primary environmental drivers. Huang et al.⁷ employed multi-source remote sensing datasets to reconstruct the spatial and temporal dynamics of NEP across the ASEAN region, revealing a consistent decline in NEP from 2001 to 2020. In subtropical forests of Zhejiang Province, Zheng et al.⁸ utilized a terrestrial carbon cycle model, informed by remote sensing-derived forest classification data, to simulate NEP changes from 1985 to 2015 and to quantify NEP responses to temperature, precipitation, relative humidity, and solar radiation, providing valuable insights into the climate sensitivities of subtropical forest NEP. Zhang et al.⁹ estimated NEP across Central Asia by integrating the CASA model with empirical approaches and further evaluated NEP’s sensitivity to major climatic factors. Additionally, Cao et al.¹⁰ combined NEP simulation results with statistical trend, correlation, and clustering analyses to characterize the spatiotemporal variations and drivers of NEP in the Yellow River Basin from 2000 to 2020. Their study demonstrated steady NEP growth over the past two decades and highlighted the basin’s strong potential for future carbon sequestration.

Existing research on NEP has largely concentrated on regions that are highly sensitive to climate change or characterized by fragile ecological environments, including countries, river basins, arid zones, plateaus, and wetlands. In contrast, urban clusters with advanced economies and intensive human activities have received comparatively little attention. To address this knowledge gap, the present study focuses on major economic regions in China, specifically the Huaihai Economic Zone, to investigate NEP (Figure 1). While most prior studies have emphasized quantifying vegetation carbon sinks, few have conducted detailed analyses of their spatial and temporal distributions or explored the underlying drivers, often neglecting spatial heterogeneity. This study not only quantifies vegetation carbon sequestration but also examines its spatiotemporal dynamics and the mechanisms driving observed patterns. Methodologically, a comprehensive suite of analytical tools is employed, including the Theil-Sen median estimator, coefficient of variation (CV), Hurst exponent for resilience assessment, center-of-gravity migration modeling, and spatial autocorrelation analysis. Moreover, partial correlation analysis and the optimal parameter geographical detector (OPGD) are applied to investigate non-climatic drivers in depth. Although previous research has predominantly emphasized the influence of climatic factors on vegetation carbon sinks, the effects of socioeconomic, anthropogenic, topographic, and vegetation-related factors remain less explored. Accordingly, this study incorporates a broad range of driving factors, including human activities, socioeconomic indicators, terrain characteristics, and vegetation cover, to comprehensively assess their impact on NEP. The Huaihai Economic Zone, encompassing cities in Jiangsu, Anhui, Shandong, and Henan provinces, is a critical economic hub in eastern China and a nexus of multiple national strategies. Examining its carbon sink function holds both regional importance and broader implications for ecological conservation in other economically developed areas. Using MOD17A3HGF NPP data alongside meteorological records from 2001 to 2023, NEP was simulated through integrated NEP estimation and soil microbial respiration models (Table 1). Temporal and spatial patterns, fluctuations, trends, and future trajectories of NEP were analyzed using trend analysis, CV, Hurst exponent, centroid shift modeling, and spatial autocorrelation. Finally, the combined application of partial correlation analysis and OPGD enabled the assessment of the relative contributions of climatic, topographic, vegetative, socioeconomic, and anthropogenic factors, providing scientific support for ecological protection and carbon neutrality initiatives in developed region.

Figure 1.

Spatial distribution characteristics of the Huaihai Economic Zone, China

(A) Overview of China’s Location.

(B) Digital Elevation Model (DEM).

(C) Land cover classification.

Table 1.

Data types and sources

Data name	Resolution	Source	Data declaration
MODIS NPP	500 m	NASA (https://ladsweb.modaps.eosdis.nasa.gov/)	The MOD17A3HGF Version 006 product delivers global annual net primary productivity (NPP) estimates at L4 processing tier, synthesized through coupled light use efficiency and BIOME-BGC modeling frameworks.
Monthly Mean Temperature	1000 m	National Earth System Science Data Center (https://www.geodata.cn/main/)	This dataset combines global CRU climate data at 0.5° resolution with high-resolution climate records from WorldClim and applies the Delta downscaling method to enhance spatial detail over China.11
Monthly Mean Precipitation	1000 m
SPEI	1000 m	Global SPEI database (https://spei.csic.es/)	This data is a multi-scale drought index used for drought monitoring and climate change research. It takes into account the difference between precipitation and potential evapotranspiration, thereby providing a more comprehensive reflection of the water balance situation.
Sc_PDSI	1000 m	NOAA PSL (https://www.uea.ac.uk/groups-and-centres/climatic-research-unit/data)	This data is an improved type of drought index developed based on the traditional Palmer Drought Severity Index (PDSI). It overcomes the problem of poor applicability of the original PDSI in different climate regions. By introducing an adaptive climate region parameter calibration mechanism, this index has stronger spatial comparability and universality globally.
Soil Type	1000 m	Resources and Environmental Science Data platform (https://www.resdc.cn/)	The dataset was produced through the digitization of the 1995 National Soil Survey Office’s “1:1,000,000 Soil Map of China.”
Geomorphological Type	1000 m		Developed through a collaborative effort led by the Institute of Geographic Sciences and Natural Resources Research (CAS),12 this national geomorphic dataset integrates remote sensing interpretation, field surveys, and terrain analysis methodologies.
DEM	30 m	Geospatial Data Cloud (http://www.gscloud.cn/)	This data is sourced from the STRM data product provided by the Geospatial Data Cloud Platform.13 In ArcGIS 10.8 software, the DEM data of the Huaihai Economic Zone was obtained through clipping and stitching. Slope, aspect and curvature data were extracted in ArcGIS 10.8 software using relevant tools.
Vegetation Type	1000 m	Resources and Environmental Science Data platform (https://www.resdc.cn/)	This dataset classifies vegetation using a combination of field surveys, remote sensing image interpretation, and ecological modeling, following an internationally accepted vegetation classification standard.
NDVI	1000 m	NASA Earth Data (https://www.earthdata.nasa.gov/)	This data was generated through monthly synthesis, mosaic, and clipping based on MODIS MOD13A3 data.
Land Use Type	30 m	CLCD (https://www.ncdc.ac.cn/portal/)	This dataset is a 30-meter resolution annual land use time series product of China, constructed based on a large number of Landsat images on the Google Earth Engine platform.14 The overall classification accuracy of the data is generally above 85%.
Population Density	1000 m	World Pop (https://hub.worldpop.org/)	This data is a global high-resolution population spatial distribution data product generated through the fusion modeling of multiple sources of data such as remote sensing images, population censuses, statistical yearbooks, land cover, and nighttime lights.
GDP	–	Statistical Yearbooks Of Provinces and Cities	The dataset captures the aggregate economic output and developmental status of the research region, with primary sources being provincial and municipal statistical yearbooks.
Nighttime Light	500 m	National Earth System Science Data Center (https://www.geodata.cn/oldindex.html)	The “Class NPP-VIIRS” nighttime light remote sensing dataset overcomes the challenge of cross-sensor calibration, effectively addressing the inconsistency and discontinuity issues present in traditional nighttime light data. Compared with the conventional DMSP-OLS dataset, it provides a fourfold improvement in the representation of spatial details.

Results

Temporal characteristics of net ecosystem production

Analysis of the Huaihai Economic Zone’s annual mean NEP over the 2001–2023 period reveals an overall increasing trend accompanied by interannual fluctuations (Figure 2). The multi-year mean NEP reached 122.83 gC·m⁻²·a⁻¹, with an interannual change rate of 2.08 gC·m⁻²·a⁻¹. The NEP attained its minimum value of 58.19 gC·m⁻²·a⁻¹ in 2013 and peaked at 180.03 gC·m⁻²·a⁻¹ in 2021, consistently maintaining a carbon sink status (NEP >0). The temporal variation follows a “rise-fall-rise” trajectory. Between 2001 and 2007, the annual mean NEP exhibited a rising trend, increasing from 63.01 gC·m⁻²·a⁻¹ to 167.31 gC·m⁻²·a⁻¹. This was followed by a declining phase from 2007 to 2013, during which NEP decreased from 167.31 gC·m⁻²·a⁻¹ to 58.19 gC·m⁻²·a⁻¹. From 2013 to 2023, the annual mean NEP showed a resurgent increasing trend with pronounced interannual fluctuations, rising from 58.19 gC·m⁻²·a⁻¹ to 140.59 gC·m⁻²·a⁻¹.

Figure 2.

Annual NEP variation trends in the Huaihai Economic Zone from 2001 to 2023

As illustrated in Figure 3, municipal-level NEP trends were consistent with regional interannual variability throughout the study period. Zaozhuang City exhibited the most pronounced interannual NEP trend (3.89 gC·m⁻²·a⁻¹), whereas Suqian City displayed the smallest fluctuation magnitude (0.56 gC·m⁻²·a⁻¹). Lianyungang City consistently maintained higher NEP values than other municipalities across the temporal series.

Figure 3.

Temporal variation of NEP in various cities of the Huaihai Economic Zone

As shown in Figure 4, the proportional extent of carbon sink areas (NEP >0) within the Huaihai Economic Zone displayed an overall increasing tendency, accompanied by interannual fluctuations. The carbon sink area proportion peaked at 98.91% in 2007 but fell to a minimum of 78.10% in 2013. Between 2001 and 2023, the regional carbon cycle remained predominantly dominated by carbon sinks, reflecting a generally strengthening sequestration capacity.

Figure 4.

Percentage of carbon source and sink areas in the Huaihai Economic Zone between 2001 and 2023

Spatial characteristics of net ecosystem production

As shown in Figure 5A, from 2001 to 2023, the NEP of the Huaihai Economic Zone exhibited pronounced spatial heterogeneity, characterized by a patchy and dispersed distribution, with relatively low values in central areas and higher values in surrounding regions.¹⁵ Specifically, NEP gradually increased from central urban areas outward, forming a spatial gradient centered on major cities. Overall, the region displayed a distribution pattern of “high in the east and low in the west.” Across the study area, NEP values during 2001–2023 ranged from −227.933 to 414.086 gC·m⁻²·a⁻¹, with a multi-year average of approximately 122.83 gC·m⁻²·a⁻¹. High-value zones were primarily located in Lianyungang, Suqian, Xuzhou, Suzhou, and Linyi, with multi-year average NEP values of 205.07, 188.37, 141.30, 140.24, and 131.56 gC·m⁻²·a⁻¹, respectively. Medium-value zones were concentrated in Zaozhuang, Huaibei, and Shangqiu, with averages of 112.75, 111.44, and 102.93 gC·m⁻²·a⁻¹, respectively, while low-value zones were mainly distributed in Jining and Heze, with averages of 73.65 and 66.18 gC·m⁻²·a⁻¹, respectively.

Figure 5.

Spatial patterns of NEP in the Huaihai Economic Zone

(A) Mean NEP during 2001–2023.

(B) NEP distribution in 2001.

(C) NEP distribution in 2023.

(D) Carbon source and sink distribution for 2001–2023.

As illustrated in Figures 5B and 5C, the Huaihai Economic Zone experienced notable spatial shifts in NEP between 2001 and 2023. In 2001, the spatial distribution displayed a distinct “high in the east, low in the west” pattern, with elevated values concentrated in Lianyungang and Suqian and the lowest values in Heze and Shangqiu.¹⁶ By 2023, high-NEP areas had expanded substantially, whereas low-value regions diminished, becoming largely confined to northern Heze. Despite these changes, the overall spatial gradient remained consistent, with eastern cities such as Lianyungang and Suqian sustaining their roles as high-value cores, and western cities including Heze and Shangqiu remaining low-value zones. The general NEP level increased significantly, highlighting the positive impact of improved farmland management practices on regional carbon sink capacity. The maximum NEP in 2023 reached 530.47 gC·m⁻²·a⁻¹, representing a 42.8% increase relative to 2001, indicating a marked enhancement in carbon sequestration potential, with particularly strong growth observed in Linyi, Zaozhuang, Xuzhou, Huaibei, and Shangqiu.

As shown in Figure 5D, the spatial distribution of carbon sources and sinks from 2001 to 2023 was dominated by carbon sink regions, while carbon sources were scattered. Carbon sources accounted for only about 2% of the total area, whereas carbon sinks covered nearly 98%, confirming the region’s primary role as a carbon sink. Carbon-emission areas were mainly located in economically developed and densely populated zones with sparse vegetation, such as Linyi, Jining, Xuzhou, and Huaibei, suggesting that intensive human activities sustain carbon source conditions. Overall, over the past 23 years, the Huaihai Economic Zone has been predominantly characterized by carbon sink areas; however, the substantial difference between maximum and minimum carbon sink values, reaching 642.02 gC·m⁻²·a⁻¹, reflects the complexity and heterogeneity of the spatial distribution of NEP.

As illustrated in Figure 6, between 2001 and 2023, areas exhibiting an upward trend in NEP within the Huaihai Economic Zone accounted for 57.62% of the total area (Table 2). Among these, regions with highly significant and significant increases comprised 17.82% and 9.04%, respectively, and were primarily located in the northeast (Linyi), the east (Lianyungang), and the central area (Zaozhuang).⁴ Notably, Linyi, Lianyungang, and their surrounding regions formed large, continuous clusters of pronounced growth, closely associated with ecological restoration initiatives, improved farmland management practices, and favorable climatic variations, including increased precipitation and temperature conditions. Slight and non-significant increases were more dispersed yet covered a larger share, 5.71% and 25.04% of the study area, respectively, predominantly in southeastern cities such as Xuzhou, Suqian, Suzhou, and Huaibei. This indicates that although NEP improved across most of the region, the magnitude of change remained modest, likely due to limited ecological carrying capacity or the dominance of single land-use patterns. In contrast, areas of declining NEP accounted for only 6.85% of the total, including extremely significant decreases (0.80%), significant decreases (0.47%), slight decreases (0.33%), and non-significant decreases (5.25%). Despite their small proportion, these declining zones were often spatially clustered, particularly in parts of Xuzhou, Suqian, and Lianyungang, likely reflecting the impacts of population growth, urban expansion, industrial emissions, farmland degradation, or extreme weather events. Zones with stable NEP, where no statistically significant changes were detected, represented the largest share (35.53%), primarily distributed in the western (Shangqiu) and northwestern (Heze) parts of the region, reflecting a relatively balanced ecological state. Overall, the spatial extent of NEP increases (57.62%) far exceeded that of decreases, indicating that the Huaihai Economic Zone has generally experienced a continuous improvement of NEP over the past two decades. Nevertheless, localized declines underscore potential ecological vulnerabilities and highlight challenges for sustainable regional management.

Figure 6.

Temporal trend of vegetation NEP in the Huaihai Economic Zone

NC: no change; ESD: extremely significant decrease; SD: significant decrease; SSD: slightly significant decrease; NSD: non-significant decrease; NSI: non-significant increase; SSI: slightly significant increase; SI: significant increase; ESI: extremely significant increase.

Table 2.

Trend significance classification

β	Z	Trend
β > 0	2.58 < Z	Extremely significant increase
	1.96 < Z ≤ 2.58	Significant increase
	1.65 < Z ≤ 1.96	Slightly significant increase
	Z ≤ 1.65	Non-significant increase
β = 0	Z	No change
β < 0	Z ≤ 1.65	Non-significant decrease
	1.65 < Z ≤ 1.96	Slightly significant decrease
	1.96 < Z ≤ 2.58	Significant decrease
	2.58 < Z	Extremely significant decrease

As shown in Figure 7, NEP stability across the Huaihai Economic Zone spanned these five classes, with proportions quantified as follows: low fluctuation (35.53%), relatively low fluctuation (0.02%), moderate fluctuation (2.43%), relatively high fluctuation (7.70%), and high fluctuation (54.32%), indicating that low-fluctuation and high-fluctuation categories minated. Low-fluctuation zones were primarily concentrated in Shangqiu and Heze Cities, reflecting lower CV values in vegetation NEP and greater ecosystem stability. High-fluctuation areas were mainly observed in western Linyi, Zaozhuang, western Xuzhou, Suzhou, and Huaibei Cities, corresponding to elevated CV values and ecosystem instability, likely driven by significant environmental or anthropogenic disturbances. Moderate to relatively high fluctuations occurred in the localized areas of Linyi, Lianyungang, and Suqian, potentially influenced by environmental stressors or human activities. Overall, the NEP CV exhibits pronounced spatial heterogeneity, highlighting substantial disparities in ecosystem stability across subregions. Accelerated urbanization and associated anthropogenic pressures appear to destabilize vegetation carbon sequestration in ecoregions experiencing intensified human-nature interactions.

Figure 7.

Spatial distribution of the CV of NEP in the Huaihai Economic Zone

ArcGIS geostatistical tools generated standard deviational ellipses and centroid migration trajectories to analyze the spatiotemporal evolution of NEP spatial patterns in the Huaihai Economic Zone (2001–2023), as visualized in Figure 8.

Figure 8.

The standard deviation ellipses and centroid migration of NEP in the Huaihai Economic Zone from 2001 to 2023

The NEP centroid trajectory from 2001 to 2023 spanned 117°22′26.4″E−118°09′18.0″E and 34°23′47.8″N-34°43′17.8″N within the Huaihai Economic Zone.⁴ This southeastern displacement from the geometric centroid (117°17′05.9″E, 34°39′50.5″N) indicates greater spatiotemporal NEP variability in the eastern and southern sectors during the study period.

Analysis revealed that the NEP centroid primarily migrated across Xuzhou’s northern and Zaozhuang’s southern sectors, exhibiting complex spatial heterogeneity with a pronounced east-to-west directional shift.¹⁷ The average migration distance was 24.37 km, with peak interannual displacements occurring in 2002–2003 (45.55 km), 2008–2009 (45.16 km), 2012–2013 (45.79 km), 2013–2014 (62.06 km), and 2019–2020 (43.69 km). The centroid showed multidirectional azimuthal shifts: northwestward in 2002–2003, southeastward in 2008–2009, southeastward again in 2012–2013, southwestward in 2013–2014, and northwestward in 2019–2020, reflecting substantial disparities in vegetation carbon sequestration across opposing termini during these periods. Notably, the NEP centroid shifted rapidly northwestward from 2002 to 2003, likely due to a variety of factors. First, climate anomalies played a role. Around 2002, the Huaihai Economic Zone experienced significant climate fluctuations, particularly the uneven temporal and spatial distribution of summer precipitation. This may have led to a decline in vegetation productivity in eastern regions (such as Lianyungang and Suqian), while northern Xuzhou and southern Zaozhuang benefited from relatively favorable precipitation and temperature conditions, enhancing carbon sequestration capacity and shifting the centroid northwestward. Second, land use/land cover change contributed to the shift in the NEP centroid. After 2000, agricultural restructuring in western cities (such as Xuzhou and Zaozhuang) and the surrounding areas of the region underwent significant adjustments. In some areas, strengthened farmland management and artificial vegetation restoration increased regional carbon sequestration. In contrast, the eastern coastal region experienced urban expansion and farmland occupation, which limited NEP growth and further exacerbated the northwestward shift in the centroid. Third, human interference and policy effects also played a significant role. Around 2000, the government launched ecological projects such as “returning farmland to forest and grassland,” primarily in the hilly western regions and some mining subsidence areas. These measures significantly increased vegetation cover and regional carbon sequestration in the short term, while benefiting less in eastern China, accelerating the shift of the center of gravity to the northwest. During 2010–2011 and 2017–2018, centroid displacement magnitudes were 6.74 km and 7.34 km, respectively, with irregular directional changes, suggesting lower regional environmental disturbances. Overall, the NEP centroid in the Huaihai Economic Zone exhibited an east-to-west directional shift from 2001 to 2023.

Between 2001 and 2023, the standard deviational ellipse’s major axis showed phased expansion (120.98 → 152.07 km) followed by contraction (→147.38 km), reflecting spatial divergence and convergence along the NW-SE axis.¹⁸ The minor axis contracted steadily from 105.32 km in 2001 to 104.10 km in 2011 and 103.63 km in 2023, indicating progressive spatial clustering along the southwest-northeast direction. Regarding directional variation, the ellipse’s orientation increased from 89.34° in 2001 to 98.51° in 2016 before decreasing to 89.84° in 2023. These angular shifts denote an initially expanding and subsequently moderately contracting the spatial arrangement of NEP across the Huaihai Economic Zone over the 23-year period.

Analysis of future trends of net ecosystem production

Using the Hurst index method, this study evaluated potential future trajectories of NEP in the Huaihai Economic Zone for the period 2001–2023. As shown in Figure 9A, Hurst index values ranged from 0.160 to 0.903, with a mean of 0.434. Regions with H < 0.5 accounted for 51.89% of the total area, indicating an anti-persistent pattern, which suggests that the carbon sequestration capacity of ecosystems in these areas is likely to decline in the future.¹⁹ In contrast, regions with H > 0.5 represented 12.58% of the area, exhibiting persistent behavior and indicating potential for the sustained enhancement of NEP. Overall, anti-persistent zones substantially outnumbered persistent ones, highlighting a pronounced spatial gradient of declining carbon sink stability across the Huaihai Economic Zone and suggesting that future NEP dynamics in most areas may deviate from historical trends, tending toward ecological degradation.

Figure 9.

Spatial patterns of the Hurst exponent and its classification

(A) Distribution of the Hurst index.

(B) Classification of the Hurst exponent spatially.

To further interpret future NEP dynamics, the Hurst index was categorized into four classes: strong anti-persistence (0 < H ≤ 0.25), weak anti-persistence (0.25 < H ≤ 0.5), weak persistence (0.5 < H ≤ 0.75), and strong persistence (0.75 < H ≤ 1) (Figure 9B). Results show that strong anti-persistence areas comprise only 0.16%, while weak anti-persistence dominates at 51.73%, primarily concentrated in the southern and central parts of the Huaihai Economic Zone. Weak persistence covers 12.54%, and strong persistence occupies a mere 0.04%, mostly in the eastern and western regions. The dominance of weakly anti-persistent zones suggests that NEP trends in most areas will follow a weak anti-persistent trajectory, meaning that the direction of future NEP changes will likely differ from historical terns. The Huaihai Economic Zone lies at the intersection of China’s eastern coastal and central regions, where urbanization has accelerated.²⁰ Over the past two decades, cities such as Xuzhou, Shangqiu, Suzhou, and Lianyungang have experienced rapid expansion. Large areas of farmland and forest have been converted to construction land, reducing vegetation cover and depleting soil carbon pools, thereby lowering regional NEP. Simultaneously, agricultural production has intensified, with the increased use of fertilizers, pesticides, and water-intensive crops. While this has improved productivity per unit area, it has also adversely affected soil quality and ecosystem stability, amplifying NEP fluctuations. Conversely, national and local ecological initiatives, such as the Grain for Green Program, natural forest protection, and wetland restoration, have achieved positive outcomes in some areas, promoting vegetation recovery and enhancing carbon sequestration capacity, partially offsetting the negative impacts of urban expansion and agricultural intensification. The Hurst index results indicate that the extensive weak anti-persistence zones in the southern and central regions coincide with areas experiencing the most significant urban expansion and agricultural intensification, suggesting that NEP in these regions may continue to face pressure in the future. In contrast, weak and strong persistence zones in the eastern and western regions are closely linked to ecological project implementation and natural condition recovery, indicating the potential for stable NEP enhancement.

By performing a pixel-level Hurst index calculation for NEP in the Huaihai Economic Zone and integrating Sen’s slope trend analysis results that met the 95% confidence level (p < 0.05) (Figure 10A), the spatial pattern of future NEP changes in the region was derived (Figure 10B). The results indicate that areas expected to experience a declining NEP trend (48.44%) are substantially larger than those projected to show an increasing trend (16.03%). The proportions of different trend-persistence combinations, ranked from highest to lowest, are: increase-anti-persistence (46.74%), increase-persistence (10.88%), decrease-anti-persistence (5.15%), and decrease-persistence (1.70%). Spatially, persistent NEP growth is primarily observed in eastern Lianyungang and the southwestern parts of Shangqiu, Xuzhou, and Suzhou. Areas shifting from increasing to decreasing NEP are concentrated in the central, southern, and eastern portions of the study area. In contrast, zones with continuous NEP decline are mainly located in central Lianyungang and Suqian, while areas transitioning from decline to growth are clustered in southeastern Xuzhou and Suqian.

Figure 10.

Mann-Kendall significance trend assessment and future NEP trend analysis

(A) Mann-Kendall significance test.

(B) Future change trends of NEP in the Huaihai Economic Zone.

Overall, the Huaihai Economic Zone faces a potential risk of NEP decline in the future, highlighting the need for targeted ecological management measures and differentiated policies tailored to regions with varying Hurst types. In areas exhibiting persistent enhancement, existing ecological protection policies should be further reinforced, and ecological compensation mechanisms established to maintain stability and carbon sink capacity. For anti-persistent regions, priority should be given to designating them as key ecological restoration zones and implementing precise management interventions. In both weak and strong anti-persistent areas, it is essential to conduct thorough analyses of the driving factors behind vegetation degradation, such as land-use changes, climate variability, and human disturbances, and to reverse degradation trends through measures including the conversion of farmland to forests or grasslands, mountain closure for afforestation, and ecological resettlement. Additionally, an NEP dynamic monitoring and early-warning system should be developed, integrating remote sensing data, meteorological observations, and field surveys, to enable the focused monitoring of anti-persistent areas and timely adjustments to management strategies.

Analysis of spatial aggregation characteristics in net ecosystem production

Global spatial autocorrelation analysis

Analysis of NEP spatial clustering characteristics in the Huaihai Economic Zone (2001–2023) incorporated Moran’s I estimates, with associated Z-scores and significance testing outcomes presented in Table 3. From 2001 to 2023, Moran’s I value for NEP was consistently positive, with normalized statistics (Z-scores) exceeding the critical threshold of 1.96. Throughout the study period, Z-scores predominantly surpassed 5.0, and both metrics met significance criteria at the 99.9% confidence level, confirming a significant positive spatial autocorrelation in regional NEP patterns.²¹

Table 3.

Global Moran’s I index of NEP in the Huaihai economic zone from 2001 to 2023

Year	NEP		Year	NEP		Year	NEP
	Moran’s I	Z-value		Moran’s I	Z-value		Moran’s I	Z-value
2001	0.346∗∗∗	7.1929	2009	0.318∗∗∗	6.6133	2017	0.291∗∗∗	6.0199
2002	0.371∗∗∗	7.6785	2010	0.294∗∗∗	6.1051	2018	0.346∗∗∗	7.1437
2003	0.300∗∗∗	6.2465	2011	0.303∗∗∗	6.2912	2019	0.337∗∗∗	6.9730
2004	0.280∗∗∗	5.8180	2012	0.317∗∗∗	6.5767	2020	0.339∗∗∗	6.9988
2005	0.310∗∗∗	6.4320	2013	0.293∗∗∗	6.0560	2021	0.329∗∗∗	6.7960
2006	0.342∗∗∗	7.0862	2014	0.346∗∗∗	7.1727	2022	0.311∗∗∗	6.4322
2007	0.334∗∗∗	6.9267	2015	0.322∗∗∗	6.6710	2023	0.244∗∗∗	5.0812
2008	0.322∗∗∗	6.6806	2016	0.318∗∗∗	6.5944

Note: ∗∗∗ indicates that the clustering level is significant at the 1% level.

Moran’s I for vegetation NEP in the Huaihai Economic Zone during 2001–2023 exhibited a five-phase evolution: Phase I (2001–2005) saw Moran’s I decrease from 0.346 to 0.310, reflecting a geospatial shift toward dispersion; Phase II (2005–2009) experienced oscillatory growth to 0.318, marking a transition toward clustering; Phase III (2009–2013) declined to 0.293 through fluctuations, reverting to dispersion; Phase IV (2013–2020) surged sharply to 0.339, reestablishing clustering; and Phase V (2020–2023) decreased to 0.244. Overall, Moran’s I values remained relatively high during 2001–2014, indicating stronger spatial autocorrelation during this period. A decline in Moran’s I began in 2015, accompanied by weakened spatial dependence, likely attributable to environmental policy interventions, land-use changes, or other factors affecting NEP spatial distribution patterns (Table 3).

Local spatial autocorrelation analysis

While global spatial autocorrelation analysis revealed overall spatial dependence of NEP in the Huaihai Economic Zone, this method quantifies only the average spatial association across the entire region and cannot detect localized clustering patterns. To address this limitation, the study conducted a local spatial autocorrelation analysis of NEP data for 2001, 2011, 2021, and 2023 to generate LISA cluster maps. This approach enables a detailed examination of spatial associations among local units and reveals intra-regional spatial aggregation and differentiation patterns. As shown in Figure 11, the agglomeration patterns of NEP exhibit significant spatiotemporal heterogeneity across these four years. Overall, the spatial distribution of NEP is characterized by five cluster types: statistically non-significant areas, high-high (HH) clusters, high-low (HL) clusters, low-high (LH) clusters, and low-low (LL) clusters. HH and LL clusters emerged as the primary manifestations of local spatial agglomeration, demonstrating dynamic shifts in their spatial distribution. Specifically, HH clusters display relatively concentrated and contiguous spatial patterns, whereas HL, LH, and LL clusters are more fragmented.²²

Figure 11.

LISA clustering of NEP Values in the Huaihai Economic Zone

(A) LISA clusters of NEP values in 2001.

(B) LISA clusters of NEP values in 2011.

(C) LISA clusters of NEP values in 2021.

(D) LISA clusters of NEP values in 2023.

Quantitatively, cluster distribution throughout the study period consistently followed the hierarchy: HH > HL > LH > LL. In 2001, HH clusters predominantly occupied Lianyungang City (eastern Huaihai Economic Zone), Linyi City (northeastern), and the Suqian, Xuzhou, and Suzhou Cities (southeastern), regions characterized by higher forest and grassland coverage and enhanced carbon sequestration capacity. LL clusters, minimally distributed, were concentrated in built-up and coastal areas, reflecting diminished carbon sequestration capacity due to prolonged anthropogenic disturbances. Spatiotemporal analysis revealed dynamic shifts in HH cluster distribution: from 2001 to 2011, HH clusters expanded in Xuzhou, Huaibei, and Suzhou Cities while contracting in Linyi City, resulting in net spatial expansion; from 2011 to 2022, HH clusters expanded across Linyi, Xuzhou, Huaibei, Zaozhuang, Suzhou, and Shangqiu Cities; and from 2022 to 2023, HH clusters contracted in Xuzhou, Suzhou, Huaibei, and Shangqiu. Cumulatively, the 2001–2023 period exhibited an east-to-west directional expansion of HH clusters.

Partial correlation analysis

This study employs partial correlation analysis to examine the relationship between NEP and two key climatic factors, temperature and precipitation, in the Huaihai Economic Zone. Analyses were performed at a spatial resolution of 500 m, with pixel-wise partial correlation coefficients (R) computed between NEP and each climatic factor for the period 2001–2023. The statistical significance of these coefficients was subsequently assessed. Based on both the magnitude of the partial R and their corresponding significance levels, the relationships were classified into five categories: highly significant positive correlation (R > 0, p < 0.01), significant positive correlation (R > 0, p < 0.05), highly significant negative correlation (R < 0, p < 0.01), significant negative correlation (R < 0, p < 0.05), and non-significant correlation (p > 0.05).²³

Across the Huaihai Economic Zone, the partial R linking annual mean temperature with NEP ranged from −0.732 to 0.672, with a mean of 0.129 (Figure 12A) (Figure 12B). Positive correlations were observed across 76.46% of the total study area. Within these positively correlated regions, areas exhibiting highly significant and significant positive correlations accounted for 0.18% and 3.94%, respectively. These regions were primarily concentrated in the northern Huaihai Economic Zone, encompassing cities such as Linyi, Zaozhuang, Jining, Shangqiu, and Heze. This pattern can be attributed to the concentration of forest land and grassland in this area, which feature high vegetation coverage and heightened sensitivity to temperature changes. In particular, Linyi, Zaozhuang, and Jining formed a high-correlation cluster, where NEP values increased with rising temperatures, indicating that thermal climate factors significantly influence NEP variation in this region. Conversely, areas exhibiting negative correlations with temperature accounted for 23.54% of the Huaihai Economic Zone. Highly significant and significant negative correlations represented 0.45% and 0.04% of the area, respectively, and were primarily located in the southern Huaihai Economic Zone, notably in Suqian, Suzhou, and Huaibei. In these regions, NEP decreased as temperatures rose, likely due to increased evaporation leading to reduced soil moisture. Since plant growth depends on adequate water availability, decreased soil moisture negatively impacted vegetation growth, resulting in lower NEP. Areas showing no significant correlation accounted for as much as 95.39% of the region. This pattern is closely associated with the regional land use structure, which is dominated by cultivated land and construction land, with cultivated land occupying the largest share. In agricultural ecosystems, productivity is influenced not only by temperature but also by factors such as precipitation, soil fertility, crop varieties, and management practices, limiting the direct effect of temperature on NEP and resulting in statistically insignificant correlations in most agricultural areas. Similarly, construction land typically has low NEP and limited vegetation coverage, leading to a weak carbon sink function that is minimally affected by temperature changes. The highly centralized land use pattern of the Huaihai Economic Zone, dominated by farmland and construction land, along with the strong anthropogenic regulation of agricultural production, further weakens the direct impact of temperature on NEP. Overall, the spatial distribution indicates that regions with significant correlations are limited, while areas with non-significant correlations dominate, highlighting the importance of considering land use patterns and human activities when analyzing regional NEP responses to climatic factors.²⁴

Figure 12.

Partial correlation coefficients and significance tests between NEP and temperature in the Huaihai Economic Zone

(A) Partial correlation coefficients for temperature in the Huaihai Economic Zone.

(B) Significance test for temperature.

Within the Huaihai Economic Zone, the partial R between annual precipitation and NEP ranged from −0.758 to 0.841, with an average of 0.362 (Figure 13A) (Figure 13B). Overall, NEP exhibited a strong correlation with precipitation, underscoring the critical role of precipitation as a climatic driving factor. Areas showing positive correlations accounted for 97.06% of the study region, with highly significant positive correlations and significant positive correlations representing 11.74% and 25.84%, respectively. These positively correlated regions were primarily located in Shangqiu and Heze in the west, as well as Jining and Zaozhuang in the north. These areas are likely water-limited ecosystems or regions with rain-fed agriculture, making them highly sensitive to precipitation changes. A dense high-correlation zone is formed at the junction of Shangqiu, Linyi, Zaozhuang, and Jining in the west, predominantly occupied by farmland and forest, where increased precipitation promotes vegetation growth and enhances carbon sequestration capacity. Regions exhibiting a negative correlation accounted for 2.94% of the study area. Among them, areas with highly significant and significant negative correlations represented 0.06% and 0.03%, respectively, and were mainly concentrated in the urban areas of Lianyungang and Suqian in the southeast, as well as in coastal regions.²⁵ These negative correlations may arise from unique environmental conditions, such as low-lying areas where excessive rainfall leads to waterlogged soils, regions with limited sunlight, submerged vegetation, or coastal wetlands. In such environments, excessive precipitation can inhibit NEP. Meanwhile, 62.32% of the study area showed no significant correlation, suggesting that in these regions, precipitation is not the primary limiting factor for NEP. Ecosystem productivity fluctuations may instead be influenced more by temperature, radiation, nutrient availability, or human management practices. Overall, these results highlight the spatial variability of precipitation’s impact on NEP, providing a scientific basis for implementing differentiated ecological management and water resource allocation strategies. Specifically, areas with significantly positive correlations should prioritize water supply and drought mitigation, whereas negatively correlated areas require optimized drainage and land use planning to mitigate potential carbon losses associated with excessive precipitation.

Figure 13.

Partial correlation coefficients and significance tests between NEP and precipitation in the Huaihai Economic Zone

(A) Partial correlation coefficients for precipitation in the Huaihai Economic Zone.

(B) Significance test for precipitation.

Analysis of the impact of land use change on net ecosystem production

Changes in land use reflect the interplay between land-use intensity, human activities, and natural environmental factors. Such changes can directly alter ecosystem structure and function, thereby influencing NEP. To assess the impact of land-use dynamics on NEP in the Huaihai Economic Zone, a land-use transition matrix for the period 2001–2023 was constructed (Table 4). The results indicate that the region has experienced substantial shifts in land-use patterns over the past 23 years. Farmland underwent the largest total conversion, totaling 13,773.25 km², primarily transitioning to construction land (11,885.25 km²), water bodies (995.5 km²), and forest land (522.5 km²). This suggests that the primary driver of farmland loss is the expansion of construction land, followed by the growth of water bodies and forested areas. Forest land conversion reached 522 km², mainly to farmland (371.25 km²), indicating that part of the forest area had been reclaimed for agricultural purposes. Grassland conversion totaled 917.75 km², largely to farmland (567 km²) and forest land (236.5 km²), highlighting the dual role of grassland in supporting agricultural expansion and enhancing vegetation cover. Construction land conversion amounted to 5,713.25 km², predominantly to farmland (5,429 km²), which may be linked to urban land adjustments and policies encouraging the reclamation of construction land for agriculture. Overall, land-use changes in the region during the study period were characterized by a substantial transformation of farmland into construction land and water bodies, partial conversion of forest land and grassland to farmland, and some reversion of construction land back to farmland. These trends reflect the combined influence of regional economic development, urbanization, and the strategic optimization and adjustment of land-use structures.

Table 4.

Land use change transition matrix of the Huaihai Economic Zone, 2001–2023 (km²)

2001	2023						Total
	Cropland	Forest	Grassland	Water	Unutilized	construction
Cropland	62392.50	522.50	369.75	995.50	0.25	11885.25	76165.75
Forest	371.25	760.25	76.50	9.75	0.00	64.5	1282.25
Grassland	567.00	236.5	330.5	5.75	0.50	108	1248.25
Water	661.50	7.25	1.00	1988	0.25	411.50	3069.5
Unutilized	2.00	0.00	0.00	3.50	0.25	1.75	7.50
construction	5429	40.00	17.75	226.50	0.00	7826.75	13540
Total	69423.25	1566.5	795.5	3229	1.25	20297.75	95313.25

Changes in NEP were quantified using a land use transition matrix, with separate calculations for NEP inflow and outflow, thereby constructing the NEP transition matrix for the Huaihai Economic Zone (Table 5). The NEP transfer matrix for this region from 2001 to 2023 indicates that land-use changes have exerted a significant impact on the regional carbon balance. The conversion of farmland to construction land resulted in the largest reduction in NEP, amounting to 788,847.20 t C a⁻¹, primarily because construction land typically lacks substantial vegetative photosynthetic capacity, leading to a sharp decline in NEP. Similarly, the transformation of farmland into water bodies caused a NEP loss of 66,084.29 t C a⁻¹, largely due to differences in carbon sequestration capacity between cropland vegetation and aquatic ecosystems. NEP losses from forest land were mainly associated with its conversion to farmland (24,758.10 t C a⁻¹) and grassland (5,100.84 t C a⁻¹), reflecting the superior carbon sink function of forest vegetation, which diminishes considerably when converted to other land types. Grassland conversion to farmland and construction land led to NEP reductions of 17,272.02 t C a⁻¹ and 3,289.68 t C a⁻¹, respectively, highlighting the role of grasslands in maintaining carbon sinks. In contrast, the reversion of construction land to farmland contributed to an NEP increase of 312,116.21 t C a⁻¹, as reclamation restored vegetative productivity and significantly enhanced carbon sequestration capacity. Overall, NEP dynamics in the Huaihai Economic Zone were driven by both the negative impacts of farmland-to-construction land conversion and the positive effects of construction land-to-farmland reversion. While urban expansion has caused substantial carbon sink losses, land reclamation and adjustments to land-use structure have partially restored regional NEP. These findings underscore the importance of future land-use planning that balances economic development with ecosystem carbon sink functions to mitigate carbon losses arising from unsustainable land conversions.

Table 5.

NEP transition matrix of the Huaihai Economic Zone, 2001–2023 (t C a⁻¹)

2001	2023
	Cropland	Forest	Grassland	Water	Unutilized	construction
Cropland	0	34683.95	24543.89	66084.29	16.60	788847.20
Forest	24758.10	0	5100.84	650.13	0.00	4299.66
Grassland	17272.02	7204.09	0	175.09	15.23	3289.68
Water	20314.07	222.62	30.71	0	7.68	12635.37
Unutilized	86.80	0.00	0.00	151.90	0	75.97
construction	312116.21	2299.60	1020.48	13021.82	0.00	0

Analysis of driving mechanisms for net ecosystem production variations

Discretization of continuous factors

Vegetation carbon sequestration is influenced by multiple factors. In this study, 16 evaluative parameters were analyzed using four classification schemes for continuous variables (Table 6): equal interval, natural breaks, quantile, and geometric interval techniques. Interval classifications ranged from 3 to 10 classes. Q-values were calculated for all configurations, and parameter sets exhibiting the highest q-values were identified as optimal.²⁶ Specifically, for the evaluation factors: X5 (GDP) was classified into 9 classes using quantile classification; X6 (Nighttime Light) into 6 classes via natural breaks; X7 (Population Density) into 9 classes using quantile classification; X8 (Normalized Difference Vegetation Index) into 10 classes via equal interval classification; X9 (Annual Mean Temperature) into 10 classes via natural breaks; X10 (Standardized Precipitation Evapotranspiration Index) into 10 classes via geometric interval classification; X11 (Annual Precipitation) into 8 classes via equal interval classification; X12 (Self-Calibrated Palmer Drought Severity Index) into 8 classes via natural breaks; and X13 (Aspect), X14 (Curvature), X15 (Slope), and X16 (Elevation) into 10 classes using quantile classification.²⁷

Table 6.

Factor selection for geodetector

Type	Symbol	Factor
Human Activity Factor	X1	Land Use Type
Vegetation Factor	X2	Vegetation Type
Surface Factor	X3	Soil Type
Surface Factor	X4	Geomorphological Type
Socioeconomic Factor	X5	GDP
Socioeconomic Factor	X6	Nighttime Light
Human Activity Factor	X7	Population Density
Vegetation Factor	X8	NDVI
Climate Factor	X9	Annual Mean Temperature
Climate Factor	X10	SPEI
Climate Factor	X11	Annual Mean Precipitation
Climate Factor	X12	Sc_PDSI
Surface Factor	X13	Aspect
Surface Factor	X14	Curvature
Surface Factor	X15	Slope
Surface Factor	X16	Elevation

Factor detector analysis

The optimized geographical detector’s factor detector was employed to quantify the influence of sixteen drivers, including human, vegetation, topographic, socioeconomic, and climatic factors, on the spatial variation of NEP in the Huaihai Economic Zone, with their effects assessed using q-values.

Analysis with the factor detector revealed considerable differences in the explanatory power of these variables for NEP heterogeneity across the region (Figure 14; Table 7). The average explanatory strength, as indicated by q-values, was ranked as follows: X5 (GDP) > X7 (population density) > X16 (elevation) > X11 (mean annual precipitation) > X10 (SPEI) > X8 (NDVI) > X6 (nighttime light) > X2 (vegetation type) > X3 (soil type) > X9 (mean annual temperature) > X15 (slope) > X12 (Sc_PDSI) > X1 (land use type) > X14 (curvature) > X4 (geomorphological type) > X13 (aspect).

Figure 14.

Q values from single-factor detection

Table 7.

Single factor detection results

Factor	q	p	q Sorting
X1	0.035∗	0.000	13
X2	0.094∗	0.000	8
X3	0.084∗	0.000	9
X4	0.025∗	0.000	15
X5	0.443∗	0.000	1
X6	0.134∗	0.000	7
X7	0.418∗	0.000	2
X8	0.229∗	0.000	6
X9	0.084∗	0.000	10
X10	0.262∗	0.000	5
X11	0.306∗	0.000	4
X12	0.044∗	0.000	12
X13	0.001	0.580	16
X14	0.031∗	0.000	14
X15	0.059∗	0.000	11
X16	0.334∗	0.000	3

Note: ∗ indicates significance at the 0.05 level (p-value <0.05).

Specifically, the explanatory power of GDP (q = 0.443) and population density (q = 0.418) exceeded 40%,¹⁵ establishing them as the dominant factors driving NEP spatial differentiation. GDP, with the highest explanatory power, indicates that regional economic development, including industrialization, urbanization, and the intensity of agricultural activities, primarily shapes the NEP spatial pattern, as areas of high economic activity generally experience greater ecosystem pressure and pronounced land cover changes. Closely following GDP, population density exhibits a strong correlation, reflecting the direct and indirect impacts of concentrated anthropogenic activities, such as residential, industrial, transportation, and resource utilization, on ecosystem structure and function. This interaction significantly shapes the spatial configuration of NEP. Other factors, including X16 (elevation, q = 0.334), X11 (mean annual precipitation, q = 0.306), X10 (SPEI, q = 0.262), and X8 (NDVI, q = 0.229), also demonstrated substantial explanatory power (>20%). As a key natural background factor, topographic elevation gradients are typically associated with systematic variations in climate (temperature, precipitation), vegetation type, soil properties, and land use, thereby affecting ecosystem productivity and carbon sequestration. In the semi-humid monsoon climate of the Huaihai Economic Zone, mean annual precipitation represents a critical climatic constraint on vegetation growth and ecosystem productivity, with its spatial variation strongly influencing the NEP pattern. The SPEI captures regional moisture and aridity conditions, whose spatial differences directly affect carbon sink function by regulating vegetation physiological processes and soil water availability, with explanatory power second only to mean annual precipitation. The remaining factors, vegetation type, soil type, mean annual temperature, slope, Sc_PDSI, land use type, curvature, and geomorphological type, had an explanatory power below 10%, indicating relatively limited influence. Notably, the aspect factor (X13) did not pass the 95% significance test, demonstrating no statistically significant effect on NEP within the Huaihai Economic Zone.

This study demonstrates that the spatial distribution of human activities, as reflected by GDP and population density, is the primary driver of spatial heterogeneity in NEP across the Huaihai Economic Zone.²⁸ Land-use and land-cover changes, such as urbanization, industrialization, and agricultural intensification, alongside associated fossil fuel combustion from industry, transportation, and residential sources, directly shape the spatial patterns of regional carbon sources and sinks.²⁹ Nonetheless, key natural background factors, topography (represented by elevation) and climate (represented by precipitation and the SPEI), retain significant explanatory power for NEP spatial variation. Topography establishes the underlying ecological framework, while precipitation serves as a critical constraint on vegetation productivity. Although the single-factor detector confirmed the strong independent explanatory power of GDP and population density (high q-values), moderate q-values (e.g., q < 0.35 for elevation and precipitation) suggest that the spatial pattern of NEP arises from the combined influence of multiple factors.¹⁶

Interaction detector analysis

Geodetector interaction analysis reveals that multi-driver synergies significantly govern the spatial configuration of NEP in the Huaihai Economic Zone (Figure 15). Key manifestations include:

Figure 15.

Interaction factor detection analysis results

The q-values of all dual-factor combinations exceeded those of any single factor alone. All interactions displayed either bivariate or nonlinear enhancement, confirming that regional NEP heterogeneity arises not from individual drivers but from the synergistic effects of multiple factors, including land use type (X1), vegetation type (X2), soil type (X3), geomorphological type (X4), GDP (X5), nighttime light (X6), population density (X7), NDVI (X8), mean annual temperature (X9), SPEI (X10), mean annual precipitation (X11), Sc_PDSI (X12), aspect (X13), curvature (X14), slope (X15), and elevation (X16).

Socioeconomic factor interactions were particularly prominent. The strongest interaction occurred between GDP (X5) and NDVI (X8) (q = 0.716), demonstrating nonlinear enhancement. This suggests that during early stages of economic growth, GDP expansion may initially coincide with vegetation degradation (negative effect); however, once GDP surpasses a critical threshold, such as during advanced industrialization, investments in green technologies and ecological restoration can promote NDVI improvement (positive synergy). This indicates that the Huaihai Economic Zone may be in a transitional phase of its economic-ecological relationship, highlighting the need for policies that balance short-term developmental pressures with long-term ecological resilience. The interaction between population density (X7) and NDVI (X8) was similarly strong (q = 0.684), showing that in certain urban cores, such as Xuzhou City, initiatives such as three-dimensional greening and park-city development can enhance vegetation coverage despite high population density, supporting the ecological viability of compact urban development. Furthermore, interactions involving core socioeconomic factors with other elements, for example, GDP (X5) with nighttime light (X6) (q = 0.611) and nighttime light (X6) with population density (X7) (q = 0.579), collectively demonstrated explanatory power exceeding 50%, underscoring the profound influence of human activity on regional NEP patterns.³⁰

Synergistic effects between vegetation and climate were found to be highly significant. Interactions between the vegetation factor (NDVI) and key climatic variables, including SPEI and mean annual precipitation, exhibited strong explanatory power, with q-values of 0.553 and 0.594, respectively.³¹ Annual precipitation (X11) proved particularly influential, as its interactions with multiple other drivers consistently yielded high q-values exceeding 0.30. These results indicate that under the relatively variable moisture conditions of the Huaihai Economic Zone, a region characterized by a semi-humid monsoon climate, climatic factors, especially precipitation and derived moisture/aridity indices, synergistically interact with vegetation and environmental variables to regulate ecosystem productivity and carbon sequestration capacity.

Findings from the interaction detector further substantiate the critical role of multi-factor synergy in driving NEP spatial differentiation across the Huaihai Economic Zone. While single-factor detection highlighted the dominance of GDP and population density, interaction analysis revealed that their influence is substantially amplified through complex nonlinear synergies with vegetation, climate, and other factors. A comprehensive assessment integrating both factor and interaction detection identifies GDP, population density, vegetation status (NDVI), and key climatic variables (particularly precipitation and moisture/aridity conditions) as the central determinants shaping NEP spatial patterns. This emphasizes that understanding NEP drivers requires prioritizing the interactive synergistic effects among factors, rather than considering the independent impact of individual elements alone.³²

Suggestion

Establish an ecological spatial zoning system

Based on the analysis of the spatial patterns and temporal trends of NEP in the Huaihai Economic Zone, a differentiated ecological management strategy of “zoning and classified control” should be adopted. The first step toward efficient ecological management is to establish a scientifically grounded ecological spatial zoning system. Remote sensing inversion and estimation models can be employed to obtain annual average NEP values and trends, enabling the ecological zoning of the Huaihai Economic Zone. The region can be categorized into four major functional zones: carbon sink stability zone, carbon sink enhancement zone, carbon source control zone, and degradation restoration zone. The carbon sink stability zone is primarily concentrated in areas with high forest coverage, dense vegetation, extensive wetlands, and consistently high NEP values. These areas should be strictly protected as ecological red lines, prohibiting any development. The carbon sink enhancement zone is mainly located in regions with dense farmland and medium NEP levels showing positive trends, where ecosystem structure optimization can be promoted through active policy guidance. The carbon source control zone corresponds to areas experiencing rapid urban expansion and industrial activities, where NEP exhibits negative growth or high volatility; in these areas, the strict regulation and promotion of green industries should be implemented. Finally, the degradation restoration zone encompasses regions such as coal mining subsidence areas, heavily polluted water bodies, and zones subjected to excessive development, which should be prioritized in targeted ecological remediation plans.

Guide the transformation and structural optimization of high-carbon industries

The Huaihai Economic Zone is a key economic region in eastern China, characterized by an industrial structure dominated by traditional high-carbon sectors such as energy, chemicals, building materials, and metallurgy, which impose a substantial environmental burden. In some industrial-intensive areas, NEP exhibits declining or unstable trends. Enhancing NEP and promoting the coordinated development of ecological security and economic growth require optimizing the industrial structure, guiding high-carbon industries toward green, low-carbon transitions, and advancing regional sustainable development.³³ First, local governments should implement differentiated industrial guidance policies, including total energy consumption controls and technical access thresholds for high-energy-consuming and high-pollution industries, while raising environmental protection and emission reduction standards. Second, leveraging smart manufacturing and the industrial internet, efforts should focus on promoting energy-saving technological transformations in sectors such as metallurgy, building materials, and coal power. Key low-carbon technologies, including carbon capture and utilization, waste heat recovery, and efficient combustion, should be deployed, while outdated production capacities are gradually phased out. Finally, in ecologically sensitive and agriculturally favorable areas, priority should be given to developing green integration models, such as ecological agriculture, organic farming, and rural tourism, to achieve industrial transformation that enhances ecological value.

Strengthen the restoration of ecological functional areas

The Huaihai Economic Zone spans the border areas of four provinces, Jiangsu, Shandong, Henan, and Anhui, featuring complex terrain and unevenly distributed water resources. Ecosystem services play a crucial role in regulating regional environmental processes. To enhance NEP and ensure the integrity of ecological security functions, it is essential to strengthen the systematic restoration of key ecological function zones through ecosystem engineering. First, by integrating remote sensing monitoring with dynamic NEP assessments, areas exhibiting significant ecosystem degradation, biodiversity loss, and weakened soil and water conservation capacities should be identified. Ecological protection red lines, key ecological restoration zones, and ecologically fragile areas must then be delineated. Second, for wetland systems, efforts should prioritize restoring water connectivity, expanding water source conservation capacity, and reinforcing wetland vegetation reconstruction. In farmland and forest-grassland ecosystems, strategies such as converting farmland to forest or wetland, treating slope farmland, and implementing ecological agriculture techniques should be adopted to enhance soil carbon sinks and crop NPP, thereby stabilizing material and energy cycles within agricultural ecosystems. Finally, in urban-rural interfaces, along transportation corridors, and near industrial park boundaries, green corridors, ecological barriers, and rainwater storage systems should be strategically implemented to integrate ecological restoration with urban functions, enhancing the spatial connectivity of NEP across the region.

Promote the enhancement of carbon sinks in agricultural ecosystems

The Huaihai Economic Zone is a major grain-producing region in China, with extensive farmland ecosystems, making it a critical area for enhancing NEP and carbon sequestration potential. Due to intensive farming and significant human disturbances, NEP exhibits substantial inter-annual fluctuations. Improving the NPP of agricultural systems is essential for advancing the region’s carbon neutrality goals.³⁴ It is recommended to implement an “Agricultural Carbon Sink Enhancement Project,” comprising four key measures: 1. Systematic construction of farmland shelterbelt systems: Increase biomass in agricultural fields through measures such as restoring windbreaks, establishing green strips along ditches, and greening farmland edges. 2. Promotion of conservation tillage practices: Apply agricultural techniques that minimize carbon release, including no-till and reduced tillage methods, as well as straw return, to help maintain soil carbon levels. 3. Development of carbon sink agriculture tailored to local conditions: Encourage carbon sequestration-based practices such as biogas power generation, water management in rice paddies, and crop rotation with fallow periods. 4. Establishment of a carbon sink quantification mechanism for agriculture: Utilize remote sensing data and ground-based monitoring models to develop unit-area-based carbon sink estimation formulas, while also exploring pilot projects for agricultural carbon sink trading mechanisms.

Build and improve the technological and policy support system

The Huaihai Economic Zone encompasses diverse ecological types and a complex industrial structure. To achieve the sustained enhancement of NEP, it is essential to establish a comprehensive support system led by technological innovation and reinforced by policy mechanisms. First, a dynamic monitoring system for NEP should be implemented, combining remote sensing technologies, in situ observations, and simulation models. This system should integrate high-resolution remote sensing imagery, multi-temporal vegetation indices (e.g., NDVI and GPP), climate variables, and soil information to generate accurate spatiotemporal maps of NEP. Second, a “Huaihai NEP Intelligent Platform” should be developed, incorporating modules for remote sensing monitoring, data analysis, ecological early warning, and policy simulation, to create a visual and interactive ecological management tool. The platform should enable real-time analysis of NEP trends in key ecological function zones, assessment of responses to human activity disturbances, and provide scientific decision-making support for government authorities. Finally, NEP enhancement should be integrated into local governments’ ecological civilization assessment systems, promoting a shift from “area-based” to “quality-based” ecosystem management. Additionally, an ecological compensation and fiscal transfer payment mechanism should be established to provide policy support and performance incentives for projects that enhance NEP, including forest carbon sequestration, wetland protection, and restoration of degraded farmland.

Discussion

Summary and implications

Over the past 23 years, the average annual NEP in the Huaihai Economic Zone exhibited a fluctuating upward trend, with a mean value of 122.83 g C m⁻²·a⁻¹ and an inter-annual variation rate of 2.08 g C m⁻²·a⁻¹. Spatially, NEP displayed an “east-high, west-low” characteristic. Areas with increasing NEP accounted for 57.62%, while areas with decreasing NEP (6.85%) were primarily concentrated in urban built-up zones and coastal regions, indicating the potential weakening effect of urbanization and human activities along the coast on ecosystem carbon sink capacity. Stability analysis revealed a high proportion of highly volatile zones (potentially vulnerable areas) that required special attention. Between 2001 and 2023, the majority of NEP in the Huaihai Economic Zone (51.73%) exhibited weak anti-persistent characteristics, indicating a potential overall degradation trend in the future. It is recommended to adopt differentiated ecological management strategies based on Hurst index classifications, with particular emphasis on strengthening ecological restoration and dynamic monitoring.

From 2001 to 2023, the centroid of NEP in the Huaihai Economic Zone generally migrated westward. Its spatial distribution exhibited both dispersal and concentration along the southeast-northwest axis. During 2001–2023, NEP in the Huaihai Economic Zone exhibited significant positive spatial autocorrelation. Moran’s I values fluctuated over time, maintaining high autocorrelation levels from 2001 to 2014. However, spatial autocorrelation weakened thereafter, likely due to environmental policies and land use changes. Throughout this period, NEP displayed significant spatial clustering, particularly in the expansion of high-high clustering areas. Notably, the high-high clustering regions in the eastern and western parts of the zone showed significant dynamic changes over time.

Among the climate factors, precipitation has a significantly greater impact on NEP than temperature, underscoring the central role of water resource management in regional carbon neutrality efforts. Changes in precipitation should be incorporated into the core indicators of carbon management to optimize hydrological regulation in watersheds and agricultural lands. Between 2001 and 2023, land use changes in the Huaihai Economic Zone were considerable, particularly the conversion of farmland into construction land, which led to notable NEP losses. Conversely, land reclamation and the conversion of construction land back into farmland contributed to the restoration of carbon sink functions. These observations suggest that future land use planning should carefully balance economic development with the preservation and enhancement of ecological carbon sink functions. OPGD analysis indicates that GDP exerts the strongest explanatory influence on the spatial variability of NEP, exhibiting the highest q-value, and demonstrates the most pronounced interaction with NDVI. Collectively, GDP, population density, vegetation type, and climatic factors constitute the primary drivers of NEP’s spatial heterogeneity.

Based on the above research findings, the following five recommendations are proposed for the sustainable enhancement of NEP in the Huaihai Economic Zone: 1. Establish an ecological spatial zoning system to optimize land-use planning and resource allocation. 2. Promote the transformation and structural optimization of high-carbon industries to reduce emissions and enhance sustainability. 3. Strengthen the restoration of ecological function areas to maintain ecosystem integrity and resilience. 4. Enhance carbon sequestration in agricultural ecosystems through sustainable management practices. 5. Develop and improve technological and policy support systems to facilitate the effective implementation of ecological and carbon management strategies.

Validation and discussion of research findings

This study reveals a fluctuating but overall increasing trend in annual NEP across the Huaihai Economic Zone during 2001–2023, indicating a dominant regional carbon sink function. This pattern is consistent with the findings of Chen et al.,³² who, using meteorological data and AVHRR satellite observations within CASA and GSMSR models, reported a similar trajectory of increasing NEP variability and enhanced carbon sequestration across China from 1982 to 2020.

Hurst exponent analysis of the Huaihai Economic Zone’s NEP time series (2001–2023) produced values ranging from 0.160 to 0.903, with a mean of 0.434. These results indicate weak anti-persistence in future NEP dynamics, suggesting that trends in most parts of the region are likely to diverge from historical patterns. This finding aligns closely with Han et al.’s Hurst analysis of NEP in the Yangtze River Delta coastal zone, which reported values between 0.06 and 0.99 (mean: 0.46), also indicating weak anti-persistence, potential divergence from historical trends, and the risk of future vegetation NEP degradation.³⁵ Given the analogous coastal settings of both regions in eastern China, the Yangtze Delta results provide valuable insights. Consequently, the Huaihai Economic Zone’s future vegetation cover and carbon sink capacity may face potential declines. To address this challenge, strategic regional coordination and innovation are recommended, emphasizing green development and ecological conservation strategies to position the Huaihai Economic Zone as a model for sustainable and ecologically resilient development.

This analysis demonstrates significant associations between temperature, precipitation, and NEP in the Huaihai Economic Zone. Precipitation shows a stronger correlation (mean partial R: 0.362) than temperature, underscoring its predominant climatic influence in the region. These findings are consistent with Zhang et al.’s nationwide analysis of terrestrial NEP spatiotemporal dynamics and climate responses from 1981 to 2018, which indicated that although both temperature and precipitation affected NEP variations over the past four decades, precipitation played a dominant role across broader geographical areas. Collectively, these results highlight the critical importance of precipitation for NEP.³⁶

Using OPGD analysis, sixteen potential drivers were screened to evaluate their interaction effects on NEP in the Huaihai Economic Zone. The results identified GDP and population density as the primary factors influencing regional NEP variability, with the interaction between GDP and the NDVI exhibiting the strongest effect. While Yang et al.³⁷ also applied OPGD analysis to their study of Huaibei City, their analysis considered only six drivers. In contrast, the present study incorporates a substantially broader set of drivers, providing a more comprehensive assessment.

Limitations of the study

Based on MOD17A3 data (500 m resolution), NEP estimation provides clear advantages for large-scale assessments of terrestrial ecosystem carbon cycles. However, in highly fragmented urban and peri-urban areas, the coarse resolution of this dataset may result in significant underestimation of spatial heterogeneity. The main contributing factors are as follows: (1) Mixed pixel effect - a 500 m pixel often contains multiple surface types, including buildings, roads, grasslands, and forests. At this resolution, capturing fine-scale urban surface structures is challenging. Since the model assumes uniform coverage within each pixel, the contribution of small, highly productive vegetation patches is diluted. (2) Nonlinear scale effect - ecological processes frequently respond nonlinearly to environmental drivers, and direct averaging at coarse resolution can introduce systematic bias. (3) Parameter adaptability limitations - urban surfaces differ from natural ecosystems in thermal properties and vegetation physiological states. Generic model parameters (e.g., light use efficiency, respiration ratios) may not accurately represent urban conditions, potentially increasing bias. Collectively, these factors can lead to the underestimation of NEP levels and carbon source-sink patterns in urbanized regions, affecting the accuracy of local carbon flux totals and spatiotemporal change analyses. Future studies could incorporate high-resolution remote sensing data (such as Sentinel-2, PlanetScope, and Landsat) to implement mixed-pixel decomposition, scale conversion, or machine learning-based downscaling corrections, thereby reducing coarse-resolution biases and improving the accuracy and reliability of NEP assessments in urban and peri-urban areas.

Resource availability

Lead contact

• •Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Yu Zhang (yuzhang@jsnu.edu.cn).

Materials availability

• •This study did not generate new unique reagents.

Data and code availability

• •[Adjective or all] data reported in this article will be shared by the lead contact upon request.
• •This article does not report the original code.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

This study was supported by the National Natural Science Foundation of China (Grant No. 42101256). The comments and suggestions of the editor and the anonymous reviewers are gratefully acknowledged.

Author contributions

Conceptualization: C.M.; methodology: C.M. and Y.Z.; data curation: C.M., Y.Z., and Y.W.; writing – original draft: C.M.; supervision: Y.Z.; validation: Y.Z. and Y.W.; writing – review and editing: Y.Z.; investigation: Y.W. and K.Z.; visualization: K.Z.; funding: Y.Z. All authors have read and approved the final article.

Declaration of interests

There are no conflicts to declare.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
MODIS NPP	NASA	https://ladsweb.modaps.eosdis.nasa.gov/
Monthly Mean Temperature	National Earth System Science Data Center	https://www.geodata.cn/main/
Monthly Mean Precipitation	National Earth System Science Data Center	https://www.geodata.cn/main/
SPEI	Global SPEI Database	https://spei.csic.es/
Sc_PDSI	NOAA PSL	https://www.uea.ac.uk/groups-and-centres/climatic-research-unit/data
Soil Type	Resources and Environmental Science Data platform	https://www.resdc.cn/
Geomorphological Type	Resources and Environmental Science Data platform	https://www.resdc.cn/
DEM	Geospatial Data Cloud	http://www.gscloud.cn/
Vegetation Type	Resources and Environmental Science Data platform	https://www.resdc.cn/
NDVI	NASA Earth Data	https://www.earthdata.nasa.gov/
Land Use Type	CLCD	https://www.ncdc.ac.cn/portal/
Population Density	World Pop	https://hub.worldpop.org/
GDP	Statistical Yearbooks Of Provinces and Cities	N/A
Nighttime Light	National Earth System Science Data Center	https://www.geodata.cn/oldindex.html
Software and algorithms
ArcGIS10.8	Esri	https://www.esri.com/
MATLABR2024a	MathWorks	https://www.mathworks.com/
Origin2024	OriginLab	https://www.originlab.com/
NEP Estimation Model	Cao et al.38	https://doi.org/10.1080/15481603.2023.2194597.
Sen's Trend Analysis + Mann-Kendall Test	Pang et al.39	https://doi.org/10.15244/pjoes/182893.
Hurst Exponent	Zhang et al.40	https://doi.org/10.3390/rs15030619.
Moran’s Index	Ai et al.41	https://doi.org/10.1016/j.ecolind.2024.111555.
Partial Correlation Analysis	Wei et al.42	https://doi.org/10.1016/j.ecolind.2022.108834.
OPGD Model	Luo et al.43	http://www.geodetector.org/
Coefficient of Variation (CV)	Dong et al.44	https://doi.org/10.1016/j.jenvman.2024.121158.
Centroid Migration	Xu et al.4	https://doi.org/10.3390/f15091484.

Method details

Research framework

This study investigated NEP in the Huaihai Economic Zone using a comprehensive analytical framework that combined Theil-Sen median trend analysis, Mann-Kendall significance tests, CV, Hurst index, centroid migration, standard deviation ellipse, spatial autocorrelation, partial correlation analysis, and the OPGD. NEP for 2001-2023 was first estimated by integrating an NEP modeling approach with a soil microbial respiration model.⁴⁵ Temporal dynamics were assessed through trend analysis, CV, and the Hurst index, enabling quantification of interannual variability, evaluation of time-series persistence, and characterization of overall evolutionary and fluctuation patterns, thus providing a foundation for risk assessment. Spatial patterns were examined via center-of-gravity migration and spatial autocorrelation analyses; the former captured shifts in NEP spatial distributions over time, while the latter identified spatial dependence and clustering. Localized hotspots and broad-scale trends were further detected using LISA clustering, establishing a robust spatial diagnostic framework. To identify the key drivers of NEP changes,⁴⁶ partial correlation analysis evaluated the effects of climatic factors such as temperature and precipitation, while the OPGD method quantified the relative contributions and interactions of natural, social, and economic factors. These integrated analyses provide a scientific basis for policy recommendations aimed at promoting sustainable vegetation carbon sequestration in the Huaihai Economic Zone.

Study area

The Huaihai Economic Zone, strategically located at the Jiangsu-Shandong-Henan-Anhui junction in eastern coastal China, spans approximately 96,000 km² within a warm-temperate to subtropical ecotone characterized by pronounced land-sea interactions and seasonal variability (Figure 1). This key region functions as a national economic hub, linking China’s coastal belt (east), the Central Plains economic zone (west), the Yangtze River Delta megalopolis (south), and the Bohai Rim economic circle (north), thereby holding critical geopolitical importance in domestic dual-circulation strategies and trans-Eurasian trade corridors. Dominated by fertile alluvial plains that constitute 80% of its terrain, interspersed with localized low hills, the zone sustains over 10 million hectares of cultivated land (66.7 million mu), contributing roughly 10% of China’s total grain output while serving as a major base for energy production, mineral resources, and advanced manufacturing. The Huaihai Economic Zone also hosts diverse and relatively intact ecosystems, including forests, wetlands, and farmland, interconnected by 22 major ecological corridors totaling 1,557.06 kilometers, which are crucial for species migration and biodiversity conservation. As the convergence point of three national strategies, Coordinated Development of Beijing-Tianjin-Hebei, Yangtze River Delta Integration, and Yellow River Basin Ecological Conservation, the region’s ecosystem stability is essential for local sustainability and serves as a critical safeguard for transregional ecological security.

Data collection

This study integrates multi-source datasets, including remote sensing imagery, administrative boundary vectors, meteorological observations, land surface parameters, vegetation indices, anthropogenic activity indicators, and socioeconomic metrics. Specifically, it incorporates the following data:

The MOD17A3 NPP dataset, produced by NASA using MODIS remote sensing observations combined with ecological modeling, provides a quantitative measure of global vegetation carbon sequestration capacity and ecosystem photosynthetic productivity. It has been widely used in regional carbon cycle studies and offers a robust basis for calculating NEP. Mean monthly temperature and precipitation were selected as key climatic indicators because they directly influence the spatial distribution and seasonal dynamics of net primary productivity (NPP). The standardized precipitation-evapotranspiration index (SPEI), which integrates precipitation, temperature, and evapotranspiration, quantitatively reflects regional water balance dynamics; meanwhile, the self-calibrating palmer drought severity index (Sc_PDSI) incorporates soil moisture and historical climate context, enabling a more precise assessment of drought and wetness effects on ecosystem water availability and productivity. Regarding natural conditions, soil type regulates nutrient availability and water supply for vegetation growth; landform type affects hydrological processes, soil properties, and local climate, thereby shaping carbon cycling patterns; and vegetation type directly determines the efficiency of carbon uptake, storage, and release. DEM data provide terrain attributes such as elevation, slope, and curvature, which reveal how topography modulates the carbon cycle. The normalized difference vegetation index (NDVI) is employed to monitor vegetation cover and condition, serving as a dynamic indicator of carbon absorption processes.⁴⁷ Human activities also exert a significant influence: land-use changes, such as urban expansion and agricultural development, disrupt ecosystem functioning and substantially reduce regional carbon sequestration capacity. Population density reflects the intensity of anthropogenic pressures and is often associated with agricultural expansion, urbanization, and industrialization, indirectly altering land use, degrading habitats, and reducing biodiversity. Regional gross domestic product (GDP) indicates the scale of economic development, with associated energy consumption, industrial growth, and urbanization processes exerting direct or indirect effects on ecosystems. Nighttime light data further quantify urbanization and industrialization, capturing patterns of energy use and infrastructure expansion. In summary, this study integrates multi-source datasets to analyze the spatiotemporal evolution of vegetation carbon sequestration and its driving mechanisms in the Huaihai Economic Zone of China, considering climate, hydrology, topography, vegetation, and anthropogenic influences (Table 1).

Research methodology

NEP estimation model

Quantifying NEP allows for a systematic characterization of regional differences in vegetation carbon sink capacity, providing a robust scientific foundation for localized carbon sequestration assessments. This study’s modeling framework uses NEP as its central metric, defined by the ecosystem carbon balance equation: NEP = NPP - Rh. The estimation of Rh employs a climate-driven algorithm that dynamically simulates soil carbon emissions through multivariate regression equations incorporating air temperature (T) and precipitation (R).³⁸ This model has been applied successfully in numerous regions, yielding reliable results, with computational protocols expressed as:

$$\begin{array}{c}NEP=NPP-R_h\end{array}$$ (Equation 1)

$$\begin{array}{c}{R}_h=0.22\times \left[\mathit{exp}\left(0.0913T\right)+\mathit{ln}\left(0.3145R+1\right)\right]\times 30\times 46.5\%\end{array}$$ (Equation 2)

where R_h represents the carbon emission flux from soil microbially mediated respiration (g C·m^-2); T corresponds to the mean air temperature (°C); R denotes the mean precipitation (mm); 30 is the time scale conversion coefficient; and 46.5% is the coefficient representing soil organic matter carbon content.

Sen's trend analysis + Mann-Kendall test

The combination of Theil-Sen median trend analysis and the Mann-Kendall significance test provides a robust framework for evaluating long-term trends and their statistical significance in time series data,⁴⁸ particularly when the data are nonlinear, heteroscedastic, or deviate from normality. By integrating these two methods, the long-term direction and variability of NEP in the Huaihai Economic Zone can be accurately assessed, facilitating the identification of trend changes driven by climatic variability and human activities. In this study, the Theil-Sen median approach was used to estimate the slope of temporal variations in NEP, offering a quantitative measure of its long-term dynamics,³⁹ as mathematically expressed in Equation 3:

$$\begin{array}{c}\beta =Median\left(\frac{x_j-x_i}{j-i}\right),\forall j>i\end{array}$$ (Equation 3)

where Median denotes the median operator, calculated over all pairwise slopes between NEP values (xᵢ, xⱼ) at time points i and j. A positive slope β (β > 0) indicates an increasing NEP trend, whereas β < 0 signifies a decreasing trend.

The Mann-Kendall test is a nonparametric statistical method widely used in time series analysis to assess the significance of trends within a dataset. The temporal dynamics of NEP in the Huaihai Economic Zone were analyzed using the Mann-Kendall trend test, with corresponding Z-statistics calculated to evaluate statistical significance (Table 2). By applying a two-tailed test with reference to standard normal distribution thresholds (|Z| > 1.65, 1.96, 2.58), statistically significant trends were identified at the 90%, 95%, and 99% confidence levels, respectively.⁴⁹ Its computational procedures are formalized in Equations 4, 5, 6, and 7:

$$\begin{array}{c}\text{sgn}\left(x_j-x_i\right)=\{\begin{array}{c}+1,x_j-x_i>0\\ 0,x_j-x_i=0\\ -1,x_j-x_i<0\end{array}\end{array}$$ (Equation 4)

$$\begin{array}{c}\mathrm{S}=\sum _{i=1}^{n-1}\sum _{j=i+1}^{i=n}\text{sgn}\left(x_j-x_i\right)\end{array}$$ (Equation 5)

$$\begin{array}{c}{\mathrm{V}}_{\text{ar}}\left(\mathrm{S}\right)=\frac{n\left(n-1\right)\left(2n+5\right)}{18}\end{array}$$ (Equation 6)

$$\begin{array}{c}\mathrm{Z}=\{\begin{array}{c}\frac{S-1}{\sqrt{{\mathrm{V}}_{\text{ar}}\left(\mathrm{S}\right)}},\mathrm{S}>0\\ 0,\mathrm{S}=0\\ \frac{S+1}{\sqrt{{\mathrm{V}}_{\text{ar}}\left(S\right)}},\mathrm{S}<0\end{array}\end{array}$$ (Equation 7)

where sgn() denotes the signum function; S represents the test statistic; V_ar() is the variance function; Z indicates the standardized test statistic; n is the number of data points in the series, with S approximating a normal distribution when n > 10; and the indices i and j correspond to discrete time points, with x_i and x_j representing the NEP values observed at those respective times.

Coefficient of variation (CV)

The coefficient of variation (CV) provides an effective means to quantify the relative variability of net ecosystem productivity (NEP) across different temporal and spatial scales, thereby revealing the stability and fluctuations of regional ecosystem productivity. By calculating the CV of NEP, it is possible to quantitatively assess its relative stability and uncertainty, as well as to identify the sensitivity of different ecosystems within the region to external disturbances.⁴⁴ In this study, a pixel-level CV analysis was conducted to quantify the temporal stability of NEP across the Huaihai Economic Zone during 2001-2023. The calculation formula is given in Equation 8:

$$\begin{array}{c}CV=\frac{1}{\overline{x}}\sqrt{\frac{1}{n-1}\sum _{i=1}^n{\left(x_i-\overline{x}\right)}^2}\end{array}$$ (Equation 8)

where CV denotes the coefficient of variation of NEP; x_i represents the NEP value of each pixel in year i; and $\overline{x}$ is the multi-year mean NEP of each pixel. Based on established thresholds, NEP stability was classified into five categories: CV values below 0.05 indicate low fluctuation; 0.05-0.10 denote relatively low fluctuation; 0.10-0.15 represent moderate fluctuation; 0.15-0.20 suggest relatively high fluctuation; and values exceeding 0.20 reflect high fluctuation.

Hurst exponent

The Hurst exponent (H) is a statistical measure used to evaluate the long-term dependence or persistence within a time series. Originally introduced by British hydrologist Harold Edwin Hurst in his study of the long-term fluctuations of the Nile River’s water levels, it has since been widely applied across various fields, including finance, geophysics, climate studies, and network traffic analysis. The Hurst exponent quantifies whether a time series tends to reinforce trends (persistence), revert to the mean (anti-persistence), or behave randomly (no memory). For NEP in the Huaihai Economic Zone, the Hurst exponent helps analyze the relationship between long-term trends and seasonal fluctuations, particularly under external disturbances such as climate change and land use change. When 0.5 < H < 1, NEP exhibits persistence with long-term memory, indicating that future changes are likely to follow past trends; the closer H is to 1, the stronger the persistence. When H = 0.5, the NEP time series behaves independently across scales, resembling a purely random process without long-term correlation. When 0 < H < 0.5, the NEP time series shows anti-persistence, with fluctuations stronger than pure randomness, meaning future changes are likely to oppose past trends; the closer H is to 0, the stronger the anti-persistence.⁴⁰ It is mathematically defined as:

$$\begin{array}{c}{⟨\xi ⟩}_{\tau }=\frac{1}{\tau }\sum _{t=1}^{\tau }\xi \left(t\right)\tau =1,2,\cdots \end{array}$$ (Equation 9)

Cumulative deviation (X (t, τ)):

$$\begin{array}{c}\mathrm{{\rm X}}\left(t,\tau \right)=\sum _{u=1}^t\left(\xi \left(u\right)-{⟨\xi ⟩}_{\tau }\right)1\le t\le \tau \end{array}$$ (Equation 10)

Range (R):

$$\begin{array}{c}R\left(\tau \right)=\underset{1\le t\le \tau }{\max }X\left(t,\tau \right)-\underset{1\le t\le \tau }{\min }\left(t,\tau \right)\tau =1,2,\cdots \end{array}$$ (Equation 11)

Standard deviation (S):

$$\begin{array}{c}S\left(\tau \right)={\left[\frac{1}{\tau }\sum _{t=1}^{\tau }{\left(\xi \left(t\right)-{⟨\xi ⟩}_{\tau }\right)}^2\right]}^{\frac{1}{2}}\tau =1,2,\cdots \end{array}$$ (Equation 12)

S-τ functional relationship:

$$\begin{array}{c}\frac{R\left(\tau \right)}{S\left(\tau \right)}=c·{\tau }^H\end{array}$$ (Equation 13)

Implementation of least squares regression yields the H, which gives:

$$\begin{array}{c}\log \left(\frac{R}{S}\right)\tau =\lg \left(\tau \right)+H\lg \left(\tau \right)\end{array}$$ (Equation 14)

where H denotes the Hurst exponent; R(τ) represents the range at scale τ; S(τ) signifies the standard deviation at scale τ; c is a scaling constant; log(R/S)τ serves as the independent variable, while log(τ) constitutes the dependent variable.

Moran’s index

The Moran index is an effective tool for analyzing the spatial autocorrelation of NEP and for revealing spatial distribution patterns and similarities within a region. In the Huaihai Economic Zone, NEP, serving as a representative indicator of ecosystem carbon absorption, may be influenced by multiple factors, including climatic conditions, soil types, and land use patterns, which often exhibit significant spatial clustering. The Moran index helps identify the spatial clustering effect of NEP. A statistically significant positive value (I > 0) indicates clustered spatial patterns, with the magnitude of I reflecting the strength of spatial association. Conversely, a significant negative value (I < 0) denotes dispersed spatial distributions, with the absolute value of I corresponding to the intensity of spatial heterogeneity. If I approaches zero or lacks statistical significance, the spatial pattern is random, showing no detectable autocorrelation.⁴¹ The computational formula is provided below:

Global Moran’s I:

$$\begin{array}{c}I=\frac{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_i-\overline{x}\right)}{\frac{1}{n}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2{\sum }_{j=1,j\ne i}^n{w}_{ij}\left(x_i-\overline{x}\right)}\end{array}$$ (Equation 15)

Local Moran’s I:

$$\begin{array}{c}{I}_i=\frac{x_i-\overline{x}}{\frac{1}{n}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}\sum _{j=1,j\ne i}^n{w}_{ij}\left(x_i-\overline{x}\right)\end{array}$$ (Equation 16)

where n denotes the sample size; xᵢ represents the NEP value observed at the i-th geographic location; $\overline{x}$ is the arithmetic mean of NEP; and a binary adjacency matrix w is constructed such that wᵢⱼ = 1 if spatial units i and j share topological contiguity, and wᵢⱼ = 0 otherwise.

Partial correlation analysis

Partial correlation analysis is a powerful tool for identifying independent relationships among multiple variables.⁴⁸ By controlling for the influence of other factors, it allows precise assessment of the impact of a specific variable on NEP. This approach eliminates confounding effects, enabling accurate identification of the true relationship between a particular factor, such as temperature or precipitation, and NEP. Partial correlation analysis thus provides a more detailed analytical framework, facilitating a deeper understanding of how various factors in the Huaihai Economic Zone influence NEP under different conditions. This, in turn, offers a more precise scientific basis for ecological management, resource allocation, and the development of climate change adaptation strategies.⁴² It is mathematically defined as:

$$\begin{array}{c}{R}_{wx,yz}=\frac{R_{wx,y}-R_{wz,y}{R}_{xz,y}}{\sqrt{\left(1-R_{wz,y}^2\right)\left(1-R_{xz,y}^2\right)}}\end{array}$$ (Equation 17)

where R_{wx, v}, R_{wz, v}, and R_{xz, v} represents first-order partial R, quantifying the relationships between w and x, w and z, and x and z, respectively, while adjusting for the effect of variable v.

OPGD model

The OPGD is a powerful tool for quantifying the influence of multiple geographical variables on NEP spatial patterns and identifying the key driving factors. By detecting spatial correlations among different variables, it enables the quantification of the explanatory power of individual factors or factor combinations for NEP variations, making it particularly suitable for studying complex ecosystems. This method provides a scientific basis for assessing ecosystem service functions in the Huaihai Economic Zone. In particular, it offers strong data support for regional ecological protection, land use planning, and climate adaptation decision-making. The application of OPGD assists policymakers in accurately identifying the key factors influencing NEP, optimizing resource management, and formulating more precise strategies for ecological protection and sustainable development.⁴³ The OPGD framework integrates five principal classification algorithms, Natural Breaks, Equal Interval, Standard Deviation, Geometric Interval, and Quantile, within its discretization module, while adhering to the “Information Density Equilibrium Principle” for stratification tiers. Empirical simulations indicate that maintaining 3-10 classification tiers optimally balances the risk of overfitting caused by excessive granularity against the need to preserve critical information, with the underlying principles outlined as follows:

$$\begin{array}{c}q=1-\frac{\sum _{h=1}^L{N}_h{\sigma }_h^2}{N_{{\sigma }^2}}=1-\frac{SSW}{SST}\end{array}$$ (Equation 18)

$$\begin{array}{c}SSW=\sum _{h=1}^L{N}_h{\sigma }_h^2,SST=N_{{\sigma }^2}\end{array}$$ (Equation 19)

where the Geodetector-derived q-statistic (0 < q < 1) quantifies the strength of causal influence, with values approaching 1 indicating maximum explanatory power. Stratum h represents the classification index for explanatory factors, where N_h and N denote the sample counts within stratum h and across the entire region, respectively. Variance components are defined as follows: σ² represents the total regional variance of Y, while σ_h denotes the variance within stratum h. Correspondingly, SSW signifies the sum of within-strata variances, and SST represents the total regional variance.⁴³

Data preprocessing

MODIS NPP data, monthly mean temperature data, and monthly mean precipitation data were collected and preprocessed using ArcGIS 10.8, including mask extraction, projection unification, and spatial resampling. Bilinear interpolation was applied to resample all datasets to a spatial resolution of 500 m, thereby obtaining NPP, temperature, and precipitation datasets with consistent scales for 2001–2023. In addition, other auxiliary datasets such as land use, SPEI, Sc_PDSI, soil type, geomorphology, vegetation type, NDVI, DEM, population density, GDP, and nighttime light data were processed in the same way to meet the requirements of subsequent analyses.

Soil respiration and NEP estimation

Based on monthly mean temperature and precipitation data, a soil microbial respiration model was constructed and applied to estimate the monthly soil microbial respiration of the study area during 2001–2023, and annual respiration values were derived by accumulation. Combined with MODIS NPP data, the NEP estimation model was then employed to generate annual NEP sequences for the Huaihai Economic Zone from 2001 to 2023.

Spatiotemporal analysis

ArcGIS 10.8 and Origin 2024 were used to conduct statistical and visualization analyses of NEP data, producing annual variation maps, carbon source/sink percentage maps, county-level annual variation trend maps, mean annual NEP spatial distribution maps, and spatial distribution maps for typical years (2001 and 2023), thereby revealing the overall spatiotemporal patterns of NEP. Furthermore, Sen’s trend analysis, the Mann-Kendall test, Hurst index analysis, and coefficient of variation analysis were performed using MATLAB R2024a, yielding NEP spatial trend maps, Hurst index distribution and classification maps, future trend prediction maps, and coefficient of variation maps, which provided insights into the temporal evolution and persistence of NEP. Additionally, spatial autocorrelation analysis in ArcGIS 10.8 was applied to calculate Moran’s I and produce both global Moran’s I tables and NEP clustering maps, revealing spatial aggregation characteristics. The mean center and directional distribution (standard deviational ellipse) methods were further employed to analyze the migration trajectory of NEP gravity centers and the spatial orientation of NEP evolution.

Driving mechanism analysis

MATLAB R2024a was used to calculate partial correlation coefficients between NEP and climatic factors (temperature and precipitation), generating spatial distribution maps of partial correlations to explore their impact and regional heterogeneity. Land use data from 2001–2023 were used to construct a land use transition matrix, which was then combined with an NEP change matrix to quantitatively assess the contribution of land use changes to NEP dynamics. Finally, multiple datasets, including temperature, precipitation, SPEI, Sc_PDSI, soil type, geomorphology, vegetation type, NDVI, land use, DEM, population density, GDP, and nighttime lights, were integrated. The OPGD model was applied to quantitatively evaluate the explanatory power of different factors, thereby identifying the dominant drivers of NEP spatiotemporal variations and their interactions.

Comprehensive results and policy recommendations

Based on the above spatiotemporal analyses and driving mechanism investigations, this study systematically discusses the spatiotemporal patterns, future trends, and influencing factors of NEP in the Huaihai Economic Zone from 2001 to 2023. Finally, targeted policy recommendations are proposed to enhance regional NEP sustainability and improve carbon sink capacity.

Quantification and statistical analysis

Quantitative analysis was conducted using ArcGIS 10.8, Origin 2024, MATLAB R2024a and Excel. The results were reflected in Figures 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, and 15 and Tables 3, 4, 5, 6, and 7.

Published: November 6, 2025