Ecosystem Carbon Storage in Southwest China’s Ecological Security Barrier Zone: Spatiotemporal Dynamics and Multi-Scenario Analysis

Abstract

Land use/cover change (LUCC) strongly regulates ecosystem carbon storage and provides a critical entry point for carbon-oriented territorial spatial governance. However, balancing carbon sequestration, food security, urban expansion, and ecological protection remains challenging in Southwest China’s Ecological Security Barrier Zone (ESBZ). In this study, we coupled the Patch-generating Land Use Simulation (PLUS) model with the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) carbon module to reconstruct LUCC and carbon-storage dynamics during 1999–2024 and to project land-use patterns and carbon storage in 2049 under four scenarios: Natural Development (NDS), Urban Development (UDS), Cultivated land Protection (CPS), and Ecological Protection (EPS). Unlike most existing PLUS–InVEST studies focused on cities, watersheds, or single provinces, this study targets a national ecological security barrier and integrates land-use evolution, carbon-storage responses, scenario trade-offs, and zoning-oriented governance into one analytical framework. It therefore provides spatially explicit evidence not only for carbon-oriented land management but also for interprovincial ecological compensation and coordinated ecological security governance in ecologically fragile regions. The 2024 land system was dominated by forest land (56.40%), cultivated land (25.47%), and grassland (16.09%). From 1999 to 2024, forest land expanded by 1.966 × 10⁴ km², whereas cultivated land and grassland decreased by 9.738 × 10³ km² and 1.874 × 10⁴ km², respectively; 92.65% of construction-land expansion originated from cultivated land conversion. Correspondingly, total carbon storage followed a “fluctuation–decline–recovery” trajectory, decreasing from 3.833 × 10¹⁰ t in 1999 to 3.820 × 10¹⁰ t in 2014, before rebounding to 3.831 × 10¹⁰ t in 2024. Pronounced provincial heterogeneity was observed: Sichuan and Yunnan jointly contributed about 76% of regional carbon storage, while Chongqing and Guizhou remained relatively low. By 2049, EPS produced the highest carbon storage (3.854 × 10¹⁰ t), whereas CPS, UDS, and NDS all led to lower values than in 2024. These contrasts indicate that the four scenarios do not represent a simple ranking of “better” or “worse”, but rather different trade-offs among carbon sinks, cultivated land protection, urban development, and regional equity. Overall, the results support province-differentiated, zoning-based land governance and highlight the need to coordinate ecological protection, cultivated-land conservation, urban growth control, and interprovincial ecological compensation to enhance carbon sequestration and safeguard ecological security in the ESBZ.

1. Introduction

Global warming has transcended its status as a discrete environmental concern to become a systemic crisis profoundly threatening human welfare and development trajectories [1,2]. Anthropogenic CO₂ emissions—driven primarily by fossil fuel combustion—are propelling the climate system toward extreme thermal states, destabilizing the global carbon budget and amplifying ecosystem risks [3,4]. This destabilization not only alters ecosystem structure and function but also undermines the delivery of critical services, including carbon sequestration, water retention, and climate regulation [5,6]. In response, the Paris Agreement established binding targets for temperature control and net-zero emissions, defining the scientific and policy boundaries essential for mitigation [1,2]. Within this framework, terrestrial ecosystems serve as vital “natural buffers” through vegetation and soil carbon sequestration; these services constitute a cornerstone for both climate change mitigation and the synergistic enhancement of ecosystem functions [7,8]. Concurrently, rapid land-system transformation has emerged as a primary driver of global change, significantly restructuring habitats, carbon pools, and productivity [9,10,11]. In China, a dual dynamic is evident: while national-scale vegetation “greening” and land management have substantially bolstered carbon sinks [12], rapid urbanization has precipitated the expansion of construction land and the reorganization of ecological spaces, thereby reshaping regional source-sink dynamics [13,14]. Consequently, optimizing the balance between emission reduction and sink enhancement at the territorial scale—amidst the parallel progression of ecological restoration and urbanization—remains a critical imperative for understanding coupled human–natural systems and informing spatial governance [8,15,16,17,18]

Extensive assessments and simulations of ecosystem carbon storage have spanned global, regional, watershed, and urban scales, systematically elucidating the impacts of land use change on carbon dynamics [4,9]. Traditional methodologies—underpinned by forest inventories, plot surveys, and empirical coefficient frameworks—have been instrumental in establishing carbon storage baselines. However, these approaches are often constrained by coarse spatial resolution and limited scenario comparability, hindering their ability to capture fine-scale spatiotemporal responses and variations driven by LUCC [15,19]. With the advancement of ecosystem service modeling, the InVEST model has emerged as a pivotal tool for assessment. By assigning carbon density parameters to specific land-use types, InVEST characterizes spatial carbon patterns and their evolution under LUCC at the grid scale, enabling trade-off analyses alongside biodiversity, production, and human well-being [8,20]. Consequently, InVEST-based frameworks have been widely deployed to quantify the potential shocks of urban expansion on regional carbon pools and to inform decision-making regarding multi-scenario spatial governance and ecological compensation [14,15].

To further this understanding, researchers have coupled InVEST with land-use change models such as CLUE-S, CA-Markov, and PLUS. This integration has established a critical paradigm for scenario analysis, enabling the prediction of land-use evolution and corresponding carbon storage responses under diverse policy or developmental trajectories [21,22,23]. However, traditional models often struggle to faithfully reproduce the spatial details of land-use evolution, exhibiting limitations in simulating patch growth processes, landscape connectivity, and the complex interactions of multiple driving factors [21,22]. Addressing these deficiencies, the PLUS model incorporates a specific patch-generating expansion strategy and data-driven factor mining. This approach superiorly characterizes the spatial competition and expansion of land-use patches, demonstrating enhanced accuracy and interpretability in multi-scenario simulations [24]. Recently, the coupled PLUS–InVEST framework has gained traction across urban agglomerations, watersheds, and resource-dependent regions. In this synergy, PLUS improves the spatial reliability of future land-use pattern simulations, while InVEST provides a unified platform for carbon storage assessment. Together, they allow for the coherent tracking of the coupled evolution of “land patterns and carbon storage” within a consistent scenario system [25,26,27,28,29]. Over the past two years, land-use prediction research has rapidly advanced from traditional land-cover transfer modeling to a new paradigm of coupled simulations integrating climate change, socio-economic scenarios, and ecosystem services. At the national scale, Guo et al. [30] developed an integrated LUH2–PLUS–InVEST framework for the synergistic projection of land-use patterns and carbon storage in China under SSP-RCP scenarios. This shift indicates an evolution from singular spatial pattern simulations to joint assessments encompassing structural forecasting, spatial allocation, and ecological impacts. Concurrently, the application of the PLUS model coupled with SSP-RCP scenarios to coal-resource-based cities by Li et al. [31] highlights land-use projection as a critical tool for balancing urban expansion, cropland conservation, and ecological security. At regional and applied scales, recent studies have enhanced the capacity to capture multi-scale dynamics, complex terrains, and policy constraints. For instance, Wu et al. [32] utilized the PLUS–InVEST framework to simulate the spatiotemporal evolution of land use and carbon storage in Dalian across various scenarios. Expanding to the urban agglomeration scale, Hao Wu et al. [33] introduced CMIP6 climate scenarios and a multi-scale optimization perspective, driving the field from basic scenario comparisons to comprehensive optimization analyses under climate constraints. Furthermore, Wu and Wang [16,32] confirmed the PLUS–InVEST framework’s robust applicability for assessing land-use changes and carbon storage responses in complex karst mountainous cities. Ultimately, research by Zhang et al. [25] in Hohhot substantiates that this coupled framework possesses strong replicability and robust explanatory power for evaluating policy scenarios.

Building upon this theoretical and methodological foundation, this study focuses on the ESBZ—a region of strategic significance for national ecological security—to investigate the spatiotemporal responses of ecosystem carbon storage to land-use change and multi-scenario development. Encompassing Sichuan, Chongqing, Yunnan, and Guizhou, this region serves as a critical headwater conservation zone for major basins like the Yangtze and Pearl Rivers and represents a global biodiversity hotspot. It is simultaneously characterized by ecological fragility and a high concentration of major ecological engineering projects. Long-term ecological restoration initiatives have significantly influenced regional carbon pools and ecosystem services, particularly regarding the remediation of degraded sloping cultivated land and vegetation recovery [34,35,36]. Concurrently, intensive development and rapid urbanization within this complex terrain continue to encroach upon limited high-quality ecological spaces, exacerbating the conflict between ecological conservation, economic development, and carbon sequestration [13,14]. To address these challenges, we applied a coupled PLUS–InVEST framework to systematically delineate the spatiotemporal evolution of ecosystem carbon storage within the ESBZ since 1999. Furthermore, we simulated the land-use patterns and corresponding carbon storage responses for 2049 under four distinct scenarios: Natural Development, Urban Development, Ecological Protection, and Cultivated land Protection. The spatial optimization strategy proposed herein is rooted in China’s territorial spatial planning system, embodying a governance logic that synergistically manages ecological conservation redlines, permanent basic cultivated land, and urban development boundaries. By focusing on the ESBZ, a comprehensive national-level region characterized by complex terrain, ecological fragility, and stringent policy constraints, this study broadens the application scope of the existing PLUS–InVEST framework for regional-scale carbon storage assessments. Our findings provide spatially explicit scientific evidence to inform territorial spatial planning, the layout of ecological restoration projects, and low-carbon development pathways. Moreover, this study offers a transferable methodological paradigm for achieving high-quality regional development under ecological security constraints.

2. Materials and Methods

2.1. Study Area

Located in the hinterland of Southwest China (approx. 21–34° N, 97–110° E), the ESBZ encompasses Sichuan, Yunnan, Guizhou, and Chongqing (Figure 1), covering a total area of approximately 1.13 × 10⁶ km². This region functions as a vital ecological barrier for the upper Yangtze River and a critical zone for water conservation and biodiversity protection [37,38]. The region exhibits highly complex topography, situated within the transitional belt from the Qinghai–Tibet Plateau to the lower-elevation eastern plains. This transition creates a distinct gradient pattern ranging from “alpine gorges to plateau hills and basin plains.” Specifically, the western and southwestern margins are dominated by the eastern edge of the Qinghai–Tibet Plateau and the Hengduan Mountains, featuring significant elevational relief. The central region is defined by the Sichuan Basin, while the southeast is characterized by the extensive Yunnan–Guizhou Plateau and Karst hills [37,38,39]. Climatologically, the area is dominated by monsoon-controlled subtropical and plateau-mountain climates. Orographic uplift and elevational disparities drive pronounced vertical differentiation and regional variation. While precipitation is generally abundant, it exhibits strong seasonal and spatial heterogeneity [38,40]. Driven by the synergy of natural gradients and ecological engineering, the region hosts diverse ecosystem types. Predominant vegetation includes evergreen broad-leaved forests, coniferous forests, mixed coniferous–broad-leaved forests, and alpine shrub–meadows. Notably, recent years have witnessed a significant trend of vegetative “greening” and ecological quality recovery across much of the landscape [38,39,40].

Figure 1. Study area overview.

2.2. Data Sources and Processing

2.2.1. Land Use Data

Land-use data for the ESBZ (Yunnan, Sichuan, Guizhou, and Chongqing) spanning the period 1999–2024 were derived from the dataset constructed by Yang and Huang (2023) [41] (https://zenodo.org/record/8176941, accessed on 10 May 2025), with a spatial resolution of 300 m × 300 m. This dataset underwent rigorous standardization in accordance with the Current Land Use Classification (GB/T 21010-2017 [42]). To precisely characterize the dynamic evolution of land use across the region, we harmonized diverse land-use types into six primary categories: cultivated land, forest land, grassland, water area, construction land, unused land. While this classification scheme may simplify specific categories—such as wetlands, which possess distinct carbon cycling functions—their limited spatial distribution within the study area and their incorporation into the “Water” category render their impact on the overall carbon storage assessment negligible. Moreover, this classification framework has been widely adopted in related fields, ensuring procedural standardization and the cross-regional comparability of results. The primary objective of this standardized scheme is to streamline subsequent analyses while maintaining consistency across all spatiotemporal dimensions.

2.2.2. Driver Data

To simulate land-use dynamics within the study area, we selected 14 driving factors, grounding our choice in regional land-use characteristics and established literature (Figure 2, Table 1). The socio-economic category comprises six variables: Gross Domestic Product (GDP), population density, and distance to railway, road, county-level settlements, and river. The climate and environmental category encompasses eight variables: Digital Elevation Model (DEM), slope, slope direction, soil organic matter content, soil pH, soil sand content, mean annual temperature, and annual precipitation. Climatic data (e.g., precipitation and temperature) were generated via Anusplin interpolation, while topographic attributes (slope and slope direction) were derived directly from the DEM. Socio-economic metrics—specifically distance to infrastructure and settlements—were computed using the Euclidean Distance tool in ArcGIS 10.8, based on data sourced from OpenStreetMap (OSM) and the National Bureau of Statistics. The 14 driving factors utilized in this study were systematically selected based on empirical precedents, regional geographical characteristics, and data availability. Given that the LEAS module of the PLUS model relies on a random forest algorithm, which is inherently robust against multicollinearity, a separate multicollinearity diagnostic was omitted as a preliminary screening step.

Figure 2. Main driving factors of land use change in ESBZ. (a) Temperature, (b) Precipitation, (c) DEM, (d) Slope, (e) Slope direction, (f) GDP, (g) Population, (h) Distance to county-level settlements, (i) Distance to railway, (j) Distance to river, (k) Distance to road, (l) Soil sand content, (m) Soil pH, (n) Soil organic matter content.

Table 1. The driving factors of simulated land-use change.

Data Type	Data Name	Year	Data Accuracy	Source
Land use data	Land use	1999–2024	300 m	https://zenodo.org/record/8176941 (accessed on 10 May 2025)
Natural factor	Precipitation	1999–2024	300 m	Resource and Environment Science and Data Center (https://www.resdc.cn/)
	Temperature	1999–2024	300 m
	DEM	2024	300 m	SRTM
	Slope direction	2024	300 m	From DEM
	Slope	2024	300 m
	Soil pH	2024	300 m	HWSD (https://lpdaac.usgs.gov/products/srtmgllv003/, accessed on 10 May 2025)
	Soil sand content	2024	300 m
	Soil organic matter content	2024	300 m
Social factor	Population	1999–2024	300 m	Resource and Environment Science and Data Center (https://www.resdc.cn/)
	GDP	1999–2024	300 m
	Distance to road	1999–2024	300 m	OpenStreetMap (https://openmaptiles.org/languages/zh/, acessed on 10 May 2025)
	Distance to railway	1999–2024	300 m
	Distance to river	1999–2024	300 m
	Distance to county-level settlements	1999–2024	300 m

2.3. Research Methods

2.3.1. PLUS Model

The PLUS model has emerged as a widely adopted patch-generating Cellular Automata (CA) framework for land use and land cover scenario simulation. By integrating the Land Expansion Analysis Strategy (LEAS) to extract conversion rules and a CA model based on Multi-type Random Patch Seeds (CARS) to simulate patch dynamics, PLUS excels in reproducing landscape patterns and enhancing simulation accuracy [24]. In standard applications, researchers typically employ Markov Chains—based on historical transition matrices—to project future demands for various land categories. Subsequently, the LEAS–CARS module executes spatial allocation under specific driving factors and scenario constraints, thereby supporting multi-scenario landscape deduction and policy assessment [43,44]. Specifically, LEAS identifies expansion units by overlaying land-use data from two distinct periods to extract transition statistics. By coupling this with non-parametric learning algorithms such as Random Forest, it characterizes the non-linear relationships between driving factors and land expansion, estimating growth probabilities and revealing critical conversion mechanisms [24]. Random Forest offers robust capabilities for evaluating variable importance within high-dimensional, collinear, and non-linear contexts [45,46]. Consequently, it has been successfully applied to identify core drivers of urban land-use change and quantify their relative contributions [47]. Building on this foundation, PLUS automatically generates patch-level simulation outputs under defined constraints and spatial parameters. This capability has led to its extensive application in multi-regional and multi-scale scenario simulations and ecological effect assessments [27,48].

2.3.2. Accuracy Verification

Following parameter calibration using historical datasets from 1999 and 2024, the PLUS model was employed to simulate the 2024 land-use scenario. Comparative validation against actual 2024 observations yielded a Kappa coefficient of 0.85, the FoM value was 0.20, and an overall accuracy of 0.91. These robust metrics substantiate the model’s high reliability and applicability for projecting future LUCC dynamics within the study area. The detailed methodological workflow is presented in Figure 3.

Figure 3. Flow chart of correlation analysis.

2.3.3. Setting of Multi-Scenario Transition Matrix

To systematically characterize the potential evolutionary pathways of land-use transition in the ESBZ through 2049, we constructed four representative scenarios grounded in distinct developmental orientations. These scenarios elucidate the potential divergence in land-use patterns under varying trajectories and policy preferences.

NDS: Grounded in the evolutionary inertia of land-use dynamics from 1999 to 2024, this scenario assumes an absence of intensified future interventions, thereby sustaining the historical continuity of land-use transition probabilities. It serves as a baseline scenario to identify the natural evolutionary trends of regional land use and carbon storage.

UDS: Driven by the prevailing context of sustained urbanization, infrastructure expansion, and escalating demands for construction land, this scenario assumes a prioritization of future development. Specifically, it applies a moderate increase to the probabilities of various land types transitioning into construction land, while simultaneously reducing the probability of construction land reverting to other classifications. The primary objective is to delineate the risks associated with the compression of ecological space and subsequent alterations in carbon storage under high-intensity development. Here, transition probabilities from cultivated land, forest land, grassland, water area, and unused land to construction land were upregulated by 30%, 25%, 25%, 10%, and 20%, respectively. Concurrently, conversions from construction land to cultivated land, forest, and grassland were down regulated by 40% to emphasize the net expansion of built-up areas.

CPS: Guided by policy directives encompassing food security, cultivated land protection redlines, and permanent basic cultivated land management, this scenario posits a future emphasis on the stabilization and replenishment of cultivated land. Specifically, it restricts the conversion of cultivated land to non-agricultural and non-grain purposes while moderately elevating the probability of other land categories transitioning into cultivated land. The objective is to elucidate the trade-offs between cultivated land protection and carbon storage dynamics. Probabilities of converting forest land, grassland, water area, construction land, and unused land to cultivated land were increased by 15%, 15%, 10%, 20%, and 25%, respectively. Furthermore, outflows from cultivated land to construction land, forest land, and grassland were reduced by 30% to mitigate the risks of conversion to non-agricultural or non-grain production purposes.

EPS: Contextualized within the frameworks of ecological civilization, the “Dual Carbon” goals, ecological conservation redlines, and ecological restoration initiatives, this scenario prioritizes the safeguarding of ecological space. Specifically, it facilitates the conversion of cultivated land and unused land into forest land and grassland, while strictly constraining the degradation of ecological land into construction land. The objective is to evaluate the potential for regional carbon sink enhancement and to identify spatial optimization pathways under an ecology-first paradigm. In this scenario, transition probabilities from cultivated land and unused land to forest and grassland were both enhanced by 30%. Conversely, conversions from grassland and water area to construction land were reduced by 40%, while forest-to-construction and forest-to-cultivated land transitions were lowered by 40% and 20%, respectively. Additionally, construction land-to-cultivated land conversion was increased by 10%, collectively reinforcing the expansion of ecological spaces and constraining construction intensity in ecologically sensitive areas.

Importantly, the four scenarios established in this study—NDS, UDS, CPS, and EPS—function as policy-driven exploratory scenarios. They are primarily designed to compare the potential responses of land-use patterns and carbon storage under divergent development pathways, rather than to serve as deterministic predictions of the land-use state in 2049. Future land-use dynamics remain subject to multiple uncertainties, including policy shifts, economic growth rates, urbanization trajectories, and climate change. Consequently, this study emphasizes identifying the relative discrepancies, directional shifts, and spatial heterogeneity across different scenarios, thereby providing a reference framework for regional land-use optimization and carbon management.

In the PLUS model, the neighborhood factor weights according to the actual situation are shown in the table below (Table 2). The cost matrix indicates whether the conversion of land use types is allowed. We set the corresponding value to 1 when the conversion is allowed, while setting it to 0 means that the conversion is forbidden [49,50]. The cost matrix of the scenarios is set out in Table 3. In addition, The patch generation threshold was set to 0.5, and the patch seed generation threshold was set to 0.0001. A 3 × 3 neighborhood was used in the CA simulation. The iterative allocation process continued until the projected land-use demand was satisfied.

Table 2. Neighborhood factor weight.

Land Use Type	Cultivated Land	Forest Land	Grassland	Water Area	Construction Land	Unused Land
Neighborhood Factor Weight	0.2	0.25	0.15	0.4	1	0.2

Table 3. Conversion cost matrix of land use types under different scenarios.

	NDS						UDS						CPS						EPS
Land Use Type *	1	2	3	4	5	6	1	2	3	4	5	6	1	2	3	4	5	6	1	2	3	4	5	6
CL	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	0	1	1	1	0	0	0
FL	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	0
GL	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1
WA	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1
CSL	0	0	0	1	1	0	0	0	0	1	1	0	0	0	0	1	1	0	0	0	0	1	1	0
UL	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1

* 1, 2, 3, 4, 5, 6 represent CL, FL, GL, WA, CSL, UL respectively.

2.3.4. InVEST Model

By assessing ecosystem service dynamics under diverse land-use scenarios, the InVEST model establishes a rigorous scientific foundation for decision-makers to balance anthropogenic benefits against their environmental costs [51]. Specifically, the model’s Carbon Storage and Sequestration module quantifies the carbon sequestration potential of terrestrial ecosystems by synthesizing data from four key carbon pools: aboveground biomass, belowground biomass, soil organic carbon, and dead organic matter. Within this framework, carbon density is defined as the quantity of carbon stored per unit area, and the total regional carbon storage is derived using the following equation:

$$C_j = C_{j-\mathit{above}} + C_{j-\mathit{below}} + C_{j-\mathit{soil}} + C_{j-\mathit{dead}}$$

(1)

$$C_{\mathit{total}}={\sum }_{j=1}^n{C}_j \times S_j$$

(2)

In this equation, C_j denotes the carbon density of land use type j. The terms C_j-above, C_j-below, C_j-soil, and C_j-dead correspond to the carbon densities of aboveground biomass, belowground biomass, soil carbon, and dead organic carbon, respectively. Furthermore, S_j signifies the area of land use type j, while C_total represents the total carbon storage.

Carbon density is substantially influenced by climatic conditions, particularly temperature and precipitation [52]. Consequently, we corrected the carbon densities for aboveground biomass, belowground biomass, and soil within the study area by applying the formulas proposed by Tang et al. (2018) [53], as expressed below:

$$C_{\mathit{ST}}= -3.4 \times t + 157.7$$

(3)

$$C_{\mathit{BT}}=-0.4 \times t+43.0$$

(4)

$$C_{\mathit{BP}}=0.03 \times P +14.4$$

(5)

$$C_{\mathit{SP}}=0.07 \times P+79.1$$

(6)

In the equations, C_ST and C_BT denote the biomass carbon density and soil carbon density, respectively, derived from mean annual temperature (t·hm⁻²). Similarly, C_SP represents the biomass and soil carbon densities derived from annual precipitation (t·hm⁻²). The variable P signifies the annual precipitation (mm), while t indicates the mean annual temperature (°C).

$$K_{\mathit{SP}}={\displaystyle \frac{C_{\mathit{SP}1}}{C_{\mathit{SP}2}}};K_{\mathit{ST}}={\displaystyle \frac{C_{\mathit{ST}1}}{C_{\mathit{ST}2}}}$$

(7)

$$K_{\mathit{BP}}={\displaystyle \frac{C_{\mathit{BP}1}}{C_{\mathit{BP}2}}};K_{\mathit{BT}}={\displaystyle \frac{C_{\mathit{BT}1}}{C_{\mathit{BT}2}}}$$

(8)

$$\mathit{KS}=K_{\mathit{SP}} \times K_{\mathit{ST}}$$

(9)

$$\mathit{KB}=K_{\mathit{BP}} \times K_{\mathit{BT}}$$

(10)

In the equations, K_SP and K_BP denote the precipitation-based correction coefficients for soil and biomass carbon densities, respectively. C_SP1 and C_SP2 represent soil carbon densities derived from annual precipitation at the ESBZ and national scales, while C_BP1 and C_BP2 refer to the corresponding biomass carbon densities. Similarly, K_ST and K_BT signify the temperature-based correction coefficients for soil and biomass carbon densities. C_ST1 and C_ST2 correspond to soil carbon densities derived from mean annual temperature at the regional and national scales, whereas C_BT1 and C_BT2 denote the respective biomass carbon densities. Furthermore, K_S and K_B represent the overall correction coefficients for soil and biomass carbon densities. Dead organic carbon density was parameterized following Wan et al. (2024) [54]; the final adjusted carbon density values are listed in Table 4.

Table 4. Carbon density of each land use type in ESBZ (t·hm⁻²).

Land Use Type	Density of ABC	Density of BBC	Density of SC	Density of DOC
CL	47.09	120.64	126.59	0.42
FL	63.38	173.26	143.99	1.49
GL	51.72	129.31	116.78	0.42
WA	0.00	0.00	0.00	0.42
CSL	0.00	0.00	91.09	0.00
UL	17.94	0.00	82.91	0.00

3. Results and Analysis

3.1. Land Use Change Patterns in the ESBZ from 1999 to 2024

3.1.1. Changes in Land Use Area

From 1999 to 2024, the land-use composition in the ESBZ was dominated by cultivated land, forest land, and grassland, which collectively accounted for approximately 97.95% of the total regional area (Figure 4, Table 5). Specifically, forest land constituted the largest share (56.38%), followed by cultivated land (25.49%) and grassland (16.07%). The remaining categories occupied minor proportions, with water area, construction land, and unused land comprising approximately 0.63%, 0.84%, and 0.56%, respectively. In terms of areal dynamics, the three dominant land-use types exhibited the most pronounced shifts. Forest land experienced the most significant expansion, with a net increase of 1.966 × 10⁴ km², whereas grassland and cultivated land underwent net reductions of 1.874 × 10⁴ km² and 9.738 × 10³ km², respectively, corresponding to single dynamic degrees of 0.32%, −0.94%, and −0.33% for the 1999–2024 period. Cultivated land exhibited a fluctuating downward trend throughout the study period, with phasic dynamic degrees of 0.05%, −0.11%, 0.13%, −0.33%, and −0.07%. Conversely, forest land displayed a trajectory of fluctuating growth, with phasic dynamic degrees recording −0.01%, 0.09%, −0.04%, 0.17%, and 0.11%. Grassland, however, underwent continuous contraction, indicated by sequential phasic dynamic degrees of −0.12%, −0.20%, −0.23%, −0.15%, and −0.27%. Among other categories, water area followed a trajectory of “initial expansion followed by retraction,” with phasic dynamic degrees of 0.41%, 1.16%, 0.79%, −0.02%, and −0.86%. Notably, construction land exhibited the most rapid relative expansion, accumulating a net increase of 6.442 × 10³ km² by 2024. This corresponds to a single dynamic degree of 20.63%, underscoring the potent driving effect of urbanization on reshaping the landscape pattern. Unused land also demonstrated a marginal increase, expanding by 1.492 × 10³ km² with a dynamic degree of 3.04%.

Figure 4. Distribution of land use types: (a) 1999, (b) 2004, (c) 2009, (d) 2014, (e) 2019, (f) 2024.

Table 5. Land use/cover area changes and single dynamic degree.

	Area (km−2)						Single Land Use Dynamics (%)
Land Use Type	1999	2004	2009	2014	2019	2024	1999– 2004	2004– 2009	2009– 2014	2014– 2019	2019– 2024	1999– 2024
CL	297,100	298,596	295,458	299,152	289,373	287,363	0.05	−0.11	0.13	−0.33	−0.07	−0.33
FL	615,929	615,332	620,640	618,177	628,715	635,588	−0.01	0.09	−0.04	0.17	0.11	0.32
GL	199,963	197,635	193,636	189,115	186,337	181,220	−0.12	−0.20	−0.23	−0.15	−0.27	−0.94
WA	6237	6492	7243	7815	7796	7124	0.41	1.16	0.79	−0.02	−0.86	1.42
CSL	3123	3879	5126	6883	8591	9565	2.42	3.21	3.43	2.48	1.13	20.63
UL	4902	5322	5153	6114	6443	6395	0.86	−0.32	1.86	0.54	−0.07	3.04

Analyzed at the regional scale, the land-use structure within Sichuan Province underwent substantial reconfiguration between 1999 and 2024. Cultivated land area contracted from 1.206 × 10⁵ km² to 1.101 × 10⁵ km², a net reduction of 1.055 × 10⁴ km² (−8.74%), reducing its areal share from 24.83% to 22.66%. This category exhibited continuous shrinkage, with the most precipitous decline occurring during the 2014–2019 period (−4.880 × 10³ km²). Conversely, forest land expanded from 1.896 × 10⁵ km² to 2.031 × 10⁵ km², achieving a net gain of 1.345 × 10⁴ km² and elevating its proportion from 39.04% to 41.80%. The largest increment occurred between 2014 and 2019 (+4.942 × 10³ km²), indicating a distinct trend of ecological space expansion. Grassland generally displayed a consistent downward trend, decreasing from 1.664 × 10⁵ km² to 1.589 × 10⁵ km² (a net loss of 7.493 × 10³ km²), causing its share to drop from 34.25% to 32.70%. Construction land demonstrated the most dramatic expansion, surging from 1.783 × 10³ km² to 4.845 × 10³ km². This cumulative increase of 3.063 × 10³ km² (+171.77%) raised its share from 0.37% to 1.00%. Water area showed a slight overall increase but was characterized by significant volatility, following a pattern of “initial growth followed by decline.” Finally, unused land increased from 4.412 × 10³ km² to 5.832 × 10³ km², representing a net growth of 1.420 × 10³ km² (+32.19%).

Land use patterns in Yunnan Province evolved through a distinct configuration characterized by the stable dominance of forest land, fluctuating growth in cultivated land, continuous contraction of grassland, and marked expansion of construction land. Specifically, forest land consistently maintained absolute predominance; its area increased from 2.704 × 10⁵ km² to 2.726 × 10⁵ km², a net gain of 2.186 × 10³ km², raising its proportion slightly from 70.60% to 71.17%. This trajectory exhibited phasic fluctuations within an overall trend of stable growth. Cultivated land expanded from 8.027 × 10⁴ km² to 8.523 × 10⁴ km², a net increase of 4.963 × 10³ km² (+6.18%), yet displayed significant volatility: a decline of 1.497 × 10³ km² (2004–2009) was followed by a recovery of 3.169 × 10³ km² (2009–2014) and further growth of 2.465 × 10³ km² (2019–2024). In contrast, grassland receded from 2.929 × 10⁴ km² to 2.080 × 10⁴ km², a net reduction of 8.493 × 10³ km² (−28.99%), causing its proportion to drop from 7.65% to 5.43%. Decreases occurred across all intervals, with the sharpest decline (−2.489 × 10³ km²) observed between 2019 and 2024, indicating a persistent and intensifying contraction. Concurrently, the water area generally increased by 3.229 × 10² km², despite a downturn in 2019–2024. Notably, construction land underwent rapid expansion, growing from 4.941 × 10² km² to 1.477 × 10³ km²—a net gain of 9.829 × 10² km² (+198.93%)—with its proportion rising from 0.13% to 0.39% amid continuous growth across all phases, while unused land showed a slight overall increase, keeping a largely stable proportion.

In Guizhou Province, cultivated land expanded slightly from 5.723 × 10⁴ km² to 5.749 × 10⁴ km², with its proportion rising marginally from 32.52% to 32.67%. However, this trajectory followed an “increase-then-decrease” pattern: the area grew continuously between 1999 and 2014, peaking at 6.177 × 10⁴ km² in 2014, before declining significantly from 2014 to 2024. Forest land exhibited overall expansion, increasing from 1.138 × 10⁵ km² to 1.148 × 10⁵ km² (a net gain of 0.86%). Its evolution was characterized by “initial contraction followed by recovery”: the area contracted to 1.101 × 10⁵ km² during 1999–2009 but increased rapidly after 2014, with a notable accumulation of 4.543 × 10³ km² between 2014 and 2024. Grassland experienced the most pronounced contraction, shrinking from 4.179 × 10³ km² to 1.595 × 10³ km²—a net reduction of 61.82%. Its share of the total area fell from 2.37% to 0.91%, reflecting a consistent downward trend. The water area generally increased, albeit with distinct fluctuations, growing from 4.025 × 10² km² to 6.377 × 10² km² (a net increase of 58.43%). It reached a phase peak of 6.654 × 10² km² in 2019 before receding slightly. Construction land demonstrated remarkable expansion intensity, surging from 3.691 × 10² km² to 1.473 × 10³ km². This represents a cumulative increase of 1.103 × 10³ km² (+299.00%), lifting its proportion from 0.21% to 0.84%. Growth persisted across all stages, with the largest increment occurring between 2014 and 2019. Although unused land showed an overall increase, its proportion remained negligible.

Land use evolution in Chongqing Municipality was characterized by a distinct pattern of steady forest expansion, continuous compression of cultivated land, substantial grassland shrinkage, and rapid growth of construction land. In 1999, cultivated land (3.890 × 10⁴ km², 47.20%) and forest land (4.204 × 10⁴ km², 51.01%) constituted the dominant categories. Together, they accounted for over 98% of the total area, leaving negligible shares for other land types. By 2024, cultivated land had declined to 3.425 × 10⁴ km²—a relative decrease of 11.97%—dropping its share to 41.55%. Conversely, forest land expanded to 4.529 × 10⁴ km², a net gain of 3.245 × 10³ km², raising its proportion to 54.94%. Grassland contracted by approximately 68.97%, with its share shrinking from 0.24% to a mere 0.08%, indicating a persistent retreating trend throughout the study period. The water area showed a slow, fluctuating increase, rising from 7.969 × 10² km² to 1.037 × 10³ km² (a net rise of 30.13%) and lifting its proportion from 0.97% to 1.26%. Construction land recorded the most dramatic expansion, surging from 4.820 × 10² km² in 1999 to 1.792 × 10³ km² in 2024. This net increase of 1.310 × 10³ km² (+271.82%) elevated its proportion from 0.58% to 2.17%, highlighting the profound impact of urbanization and infrastructure development in reshaping the land use landscape. Unused land remained negligible in both area and proportion, exerting limited influence on the overall configuration.

3.1.2. Land Use Transition Analysis

Between 1999 and 2024, the total land use transfer area in the ESBZ amounted to 1.413 × 10⁵ km², representing 12.5% of the total regional area (Figure 5, Table 6). Cultivated land experienced an outflow of 6.289 × 10⁴ km², exhibiting a pronounced directional bias: 78.65% converted to forest land, 9.07% to grassland, and 10.12% to construction land, with negligible transfers to other categories. The outflow from forest land totaled 4.623 × 10⁴ km² and was highly concentrated, with 92.42% reverting to cultivated land and 7.02% shifting to grassland; other transfers were insignificant. Grassland outflow reached 2.915 × 10⁴ km², primarily flowing into forest land (55.90%; 1.629 × 10⁴ km²) and cultivated land (33.46%; 9.754 × 10³ km²), while 8.02% transitioned to unused land. Construction land saw a minor outflow of 4.278 × 10² km², 87.12% of which converted to the water area. The outflow from the water area was 1.577 × 10³ km², distributed mainly between cultivated land (39.31%) and grassland (36.24%). Finally, unused land outflow was 1.032 × 10³ km², with 85.70% converting to grassland; shares for other destinations remained below 10%. From the inflow perspective, forest land received 6.589 × 10⁴ km², with 75.07% originating from cultivated land. This proportion aligns closely with the core objectives of the “Grain for Green” program, underscoring the policy as the dominant driver of forest expansion in the ESBZ. Simultaneously, construction land received an inflow of 6.870 × 10³ km², 92.65% of which was derived from cultivated land, indicating that the occupation of cultivated land for urbanization is the primary source of construction land expansion.

Figure 5. Chart of land use transfer matrix in each period: (a) 1999–2024, (b) 1999–2004, (c) 2004–2009, (d) 2009–2014, (e) 2014–2019, (f) 2019–2024.

Table 6. Land use/cover transfer matrix in ESBZ (km²).

2004
1999	CL	FL	GL	WA	CSL	UL	Total
CL	276,265	15,002	4759	320	754	0	20,835
FL	19,317	595,644	963	1	5	0	20,286
GL	2902	4659	191,030	320	62	990	8933
WA	111	28	446	5595	38	19	642
CSL	0	0	0	110	3013	0	110
UL	1	0	436	146	6	0	589
Total	22,331	19,689	6604	896	866	1008	51,395
2009
2004	CL	FL	GL	WA	CSL	UL	Total
CL	273,799	19,985	3027	530	1256	0	24,797
FL	17,663	596,639	1013	2	15	0	18,693
GL	3888	3974	188,832	307	111	522	8802
WA	108	42	143	6127	34	39	365
CSL	1	0	0	169	3709	0	170
UL	0	0	621	108	1	4591	730
Total	21,660	24,001	4803	1116	1416	562	51,395
2014
2009	CL	FL	GL	WA	CSL	UL	Total
CL	271,451	18,175	3408	676	1748	0	24,007
FL	23,214	596,388	1012	5	22	0	24,253
GL	4269	3579	184,070	341	89	1287	9565
WA	216	33	225	6595	62	112	647
CSL	1	0	0	163	4961	0	164
UL	0	2	400	34	1	4715	438
Total	27,700	21,789	5044	1219	1922	1399	51,395
2019
2014	CL	FL	GL	WA	CSL	UL	Total
CL	270,236	23,656	3509	264	1484	2	28,916
FL	15,182	601,835	1127	1	31	0	16,342
GL	3696	3206	180,662	369	220	962	8453
WA	257	16	306	7076	37	123	739
CSL	1	0	0	64	6818	0	66
UL	0	3	733	22	1	5356	758
Total	19,138	26,880	5676	721	1773	1087	51,395
2024
2019	CL	FL	GL	WA	CSL	UL	Total
CL	263,521	21,836	2925	238	853	0	25,853
FL	18,006	609,737	949	2	22	0	18,978
GL	5196	3962	175,837	254	135	954	10,501
WA	639	49	439	6547	23	99	1249
CSL	0	0	0	59	8531	0	59
UL	1	3	1071	24	1	5342	1100
Total	23,842	25,850	5384	577	1034	1053	51,395
2024
1999	CL	FL	GL	WA	CSL	UL	Total
CL	234,209	49,463	5705	1337	6365	21	62,891
FL	42,723	569,702	3245	98	160	2	46,227
GL	9754	16,293	170,814	526	238	2338	29,149
WA	620	125	571	4660	97	164	1577
CSL	54	0	0	373	2696	0	428
UL	3	5	885	130	9	3870	1032
Total	53,154	65,886	10,407	2464	6870	2525	51,395

The period from 1999 to 2004 marked the foundational phase, during which cultivated land served as the core supply source with an outflow of 2.084 × 10⁴ km². Specifically, the transfer of 1.500 × 10⁴ km² to forest land highlighted the initial implementation of the “Grain for Green” program, while the conversion of 7.540 × 10² km² to construction land signaled the onset of urbanization encroachment (Figure 6a). The 2004–2009 interval functioned as the trend intensification phase, reinforcing the trajectory established in the previous period. Cultivated land outflow rose to 2.450 × 10⁴ km², with transfers to forest land expanding further (1.998 × 10⁴ km²; Figure 6b), reflecting intensified efforts in the “Grain for Green” initiative. Concurrently, the inflow to construction land increased to 1.416 × 10³ km², indicating a significant acceleration in urbanization. In contrast, the inflow to unused land dropped from 1.008 × 10³ km² to 5.618 × 10² km², suggesting an initial contraction of its stock. The years 2009–2014 represented a supplementary adjustment phase. While the outflow trends of cultivated land to forest and construction land persisted (Figure 6c), cultivated land inflow peaked at 2.770 × 10⁴ km², sustained primarily by conversions from forest land. Although construction land inflow continued to grow, the deceleration in its growth rate signaled a subtle reconfiguration of resource allocation. The period 2014–2019 emerged as a phase of prominent contradictions. Cultivated land outflow reached a temporal maximum of 2.892 × 10⁴ km², with transfers to forest and construction land remaining the dominant directions (Figure 6d). Forest land inflow hit a new high of 2.688 × 10⁴ km², demonstrating a reinforced commitment to ecological restoration. The outflow from unused land doubled compared to the previous period, with 96.71% converting to grassland, thereby continuing an eco-friendly development pattern. Meanwhile, a slight decline in construction land inflow reflected an initial adjustment in the pace of urbanization. Finally, 2019–2024 served as the rhythm optimization phase, where core trends persisted with modulated intensity. Both the outflow (2.585 × 10⁴ km²) and inflow (2.384 × 10⁴ km²) of cultivated land receded from their peaks (Figure 6e), indicating a bidirectional convergence between protection and loss. Forest land inflow remained substantial, sustaining the momentum of ecological restoration. Crucially, the marked decline in construction land inflow signified a staged transition in urbanization from “acceleration” to “quality-oriented development.”

Figure 6. Spatial patterns of LUCC in each period.(a) 1999–2004, (b) 2004–2009, (c) 2009–2014, (d) 2014–2019, (e) 2019–2024, (f) 1999–2024.

3.2. Projection of Future Land Use Under Diverse Scenarios

Building upon historical land use data (1999–2024) for the ESBZ, this study utilized the PLUS model to project spatial land use patterns for the year 2049 across four distinct scenarios (Figure 7). The specific dynamics of change occurring between 2024 and 2049 are detailed in Table 7.

Figure 7. Distribution of land use types under the four scenarios in 2040: (a) NDS, (b) UDS, (c) CPS, (d) EPS.

Table 7. Projected areas and corresponding percentages of land use change under four scenarios over the next 25 years (to 2049).

Land Use Type	NDS		UDS		CPS		EPS
	Area Change (km2)	Percentage Change (%)	Area Change (km2)	Percentage Change (%)	Area Change (km2)	Percentage Change (%)	Area Change (km2)	Percentage Change (%)
CL	−5798	−2.02	−7943	−2.77	23,375	8.14	−33,159	−11.55
FL	13,838	2.18	13,372	2.06	−10,076	−1.58	40,177	6.32
GL	−14,632	−8.07	−14,738	−8.13	−17,653	−9.73	−12,210	−6.73
WA	1345	18.80	1418	19.83	1345	18.81	1241	17.35
CSL	4368	45.56	7013	73.15	2149	22.42	3475	36.25
UL	880	13.83	877	13.78	859	13.50	477	7.49

With 2024 serving as the baseline, the total land use area in the ESBZ was approximately 1.127 × 10⁶ km². The composition was dominated by forest land (6.358 × 10⁵ km²; 56.40%), cultivated land (2.871 × 10⁵ km²; 25.47%), and grassland (1.813 × 10⁵ km²; 16.09%), while the water area, construction land, and unused land collectively accounted for less than 2.1%. Across the four scenarios projected for 2049, the total regional area remained conserved, with changes manifesting primarily as inter-category conversions. Under the NDS (Figure 8a), cultivated land and grassland declined by 2.02% and 8.07%, respectively. Conversely, forest land expanded by 1.384 × 10⁴ km² (+2.18%), and construction land surged to 1.395 × 10⁴ km² (+45.56%), reaching a proportion of 1.24%. Provincially, the extremes of change were concentrated in Sichuan Province, which recorded the largest net increase in forest land (8.868 × 10³ km²) and the steepest decline in grassland (9.702 × 10³ km²). In this scenario, forest expansion and grassland contraction emerged as the core transformations, reflecting the combined influence of natural succession and mild anthropogenic activity.

Figure 8. Spatial patterns of LUCC under the four scenarios in 2049.(a) NDS, (b) UDS, (c) CPS, (d) EPS.

Under the UDS (Figure 8b), construction expansion was further amplified. Construction land reached 1.660 × 10⁴ km²—a net increase of 7.013 × 10³ km² (+73.15%)—accounting for 1.47% of the total area. Concurrently, cultivated land and grassland contracted to 2.791 × 10⁵ km² (−2.77%) and 1.666 × 10⁵ km² (−8.13%), respectively, while forest land retained a modest growth of 2.10%. Most notably, Chongqing Municipality exhibited the most dramatic shift, with construction land escalating from 1.792 × 10³ km² to 3.366 × 10³ km² (+87.81%). In Sichuan Province, the reduction in grassland (8.243 × 10³ km²) and the expansion of forest land (7.676 × 10³ km²) represented the most significant regional decline and accumulation, respectively. These dynamics reflect the encroachment of land demand—driven by the Chengdu-Chongqing urban agglomeration and transportation corridors—upon cultivated land and grassland. However, the sustained growth of forest land suggests that ecological constraints remain effective.

The CPS (Figure 8c) is characterized by a distinct strategy of “quantity preservation and proportional increase.” Under this scenario, cultivated land expanded to 3.104 × 10⁵ km² (+8.14%), capturing 27.54% of the total area. In contrast, forest land and grassland declined to 6.257 × 10⁵ km² (−1.58%) and 1.637 × 10⁵ km² (−9.73%), respectively. Although construction land continued to grow, its expansion was moderated, reaching 1.174 × 10⁴ km² (+22.42%). At the provincial level, Yunnan recorded the most substantial gains in cultivated land (1.539 × 10⁴ km²) and the steepest reductions in grassland (9.317 × 10³ km²). Notably, construction land in Yunnan and Guizhou provinces decreased by 4.198 × 10² km² and 6.272 × 10² km², respectively. These dynamics indicate that, constrained by mechanisms such as “permanent basic cultivated land” protection, the “requisition-compensation balance,” and land consolidation, cultivated land expansion is driven primarily by conversions from grassland and forest land, accompanied by stricter controls on construction land.

Under the EPS (Figure 8d), the dominant trajectory shifted toward “forest expansion and cultivated land compression.” Forest land ascended to 6.759 × 10⁵ km², a net gain of 4.018 × 10⁴ km² (+6.32%), comprising 59.96% of the total area. Conversely, cultivated land and grassland declined to 2.539 × 10⁵ km² (−11.55%) and 1.691 × 10⁵ km² (−6.73%), respectively, while construction land increased to 1.306 × 10⁴ km² (+36.25%). Regionally, Sichuan Province recorded the largest variations across the entire domain, with a net forest gain of 2.107 × 10⁴ km² and a cultivated land loss of 1.494 × 10⁴ km². These results suggest that under assumptions prioritizing “ecological red lines” and the “Grain for Green” program, cultivated land reverts more readily to forest. However, due to rigid demand from population and industrial agglomerations, construction land continued to expand—albeit at a rate significantly lower than in the urban development scenario—even under strict ecological constraints.

3.3. Dynamic Characteristics of Carbon Storage in ESBZ

3.3.1. Temporal Variability of Carbon Storage

Between 1999 and 2024, regional carbon storage in the ESBZ exhibited a multi-stage evolutionary trajectory characterized by “fluctuating adjustment, phasic decline, and gradual recovery” (Table 8). Regionally, carbon storage initiated at 3.833 × 10¹⁰ t in 1999, dipping slightly to 3.830 × 10¹⁰ t in 2004 (a reduction of 3.588 × 10⁷ t) and maintaining relative stability through 2009. A marked decline occurred in 2014, reaching 3.820 × 10¹⁰ t (a decrease of 9.338 × 10⁷ t relative to 2009). Subsequently, storage rebounded to 3.825 × 10¹⁰ t in 2019 (+4.924 × 10⁷ t vs. 2014) and further to 3.831 × 10¹⁰ t in 2024, nearly recovering to the initial 1999 baseline. At the provincial level, carbon storage dynamics exhibited distinct regional heterogeneity. Chongqing Municipality followed a fluctuating downward trend from 1999 to 2014, declining from 2.764 × 10⁹ t to 2.747 × 10⁹ t. This was followed by a gradual recovery, reaching 2.758 × 10⁹ t in 2024—an increase of 1.142 × 10⁷ t over 2014. Yunnan Province maintained a stable range around 1.357 × 10¹⁰ t tons. With a fluctuation amplitude of only 1.414 × 10⁷ t between 1999 and 2024, it emerged as the most stable zone for carbon storage within the ESBZ. Sichuan Province experienced a decline from 1.582 × 10¹⁰ t to 1.578 × 10¹⁰ t (−4.140 × 10⁷ t) during 1999–2014. However, it demonstrated a continuous recovery from 2014 to 2024, reaching 1.585 × 10¹⁰ t (+6.327 × 10⁷ t vs. 2014), distinguishing it as the region with the most significant recovery magnitude. Guizhou Province displayed a fluctuating decline from 6.163 × 10⁹ t to 6.114 × 10⁹ t (1999–2014), followed by a gradual recovery to 6.141 × 10⁹ t in 2024, essentially restoring carbon storage to 2004 levels.

Table 8. Provincial Carbon Storage in 1999, 2004, 2009, 2014, 2019 and 2024 (10⁹ t).

Region	1999	2004	2009	2014	2019	2024
Chongqing Municipality	2.764	2.768	2.766	2.747	2.754	2.758
Yunnan province	13.581	13.570	13.589	13.560	13.567	13.567
Sichuan province	15.824	15.819	15.822	15.784	15.810	15.847
Guizhou province	6.163	6.141	6.122	6.114	6.123	6.141

Under the NDS (Table 9), the projected carbon storage for the ESBZ in 2049 is 3.828 × 10¹⁰ t—a decrease of approximately 2.978 × 10⁷ t relative to the 2024 actual value. This figure occupies an intermediate position among the four scenarios, reflecting the baseline trajectory of carbon storage evolution in the absence of additional anthropogenic intervention. In the CPS, carbon storage is projected to decline to 3.812 × 10¹⁰ t, a reduction of roughly 1.942 × 10⁸ t compared to the 2024 observation (3.831 × 10¹⁰ t). This significant loss is primarily attributed to the target-driven expansion of cultivated land, which encroaches upon carbon-dense forest and grassland ecosystems. This displacement directly diminishes the scale of regional carbon carriers, triggering substantial carbon losses. Furthermore, physical soil disturbance during reclamation accelerates the decomposition of soil organic matter, thereby exacerbating carbon depletion. Conversely, under the EPS, carbon storage ascends to 3.854 × 10¹⁰ t, an increase of approximately 2.303 × 10⁸ t over the 2024 baseline. Representing the highest level among the scenarios, this gain underscores the substantive positive regulatory effect of ecological protection measures on regional carbon sequestration. Finally, under the UDS, carbon storage is projected at 3.822 × 10¹⁰ t, a decrease of about 8.986 × 10⁷ t relative to 2024. The core driver of this decline is the rapid sprawl of built-up areas, which displaces surrounding forest and grassland. This replacement directly reduces the proportion of carbon sink lands, while land hardening and native vegetation removal during development further weaken local carbon fixation capacities.

Table 9. Provincial Carbon Storage under Four Scenarios in 2049 (10⁹ t).

Region	NDS	UDS	CPS	EPS
Chongqing Municipality	2.747	2.745	2.754	2.735
Yunnan province	13.585	13.547	13.643	13.584
Sichuan province	15.797	15.689	15.948	15.745
Guizhou province	6.155	6.138	6.199	6.160

Fundamentally, the NDS, UDS, CPS, and EPS do not constitute a simplistic hierarchy of merit; rather, they represent four distinct land-use trajectories driven by divergent developmental orientations: baseline continuity, prioritized urban expansion, prioritized food security, and prioritized ecological conservation, respectively. Among these, the NDS reflects a baseline trend of a marginal decline in regional carbon storage in the absence of additional interventions. The UDS illustrates the sustained encroachment of construction-land expansion upon forest land, grassland, and cultivated land spaces; although its associated carbon loss is lower than that of the CPS, it potentially exacerbates the appropriation of cultivated land, spatial sprawl, and urban–rural developmental imbalances. While the CPS is conducive to stabilizing the quantity of cultivated land and food production space, it achieves cultivated land replenishment at the expense of forest land and grassland, thereby precipitating the most substantial decline in regional carbon storage. Conversely, the EPS achieves the highest carbon storage; however, the concurrent reduction in cultivated land implies that, in the absence of refined spatial zoning and supporting compensation mechanisms, it may exert potential adverse impacts on food security, the incomes of certain agricultural stakeholders, and balanced urban–rural development. Consequently, the comparative analysis of these four scenarios aims not to identify a singular “optimal” pathway, but to elucidate the multi-objective trade-off boundaries among carbon sink enhancement, cultivated land protection, urban development, and regional equity within the ESBZ.

3.3.2. Spatial Distribution Patterns of Carbon Storage in ESBZ

Carbon storage in the ESBZ exhibits a distinct spatial gradient, characterized by “high values in Sichuan and Yunnan, and low values in Chongqing and Guizhou” (Figure 9). Yunnan and Sichuan provinces constitute the high-value zones, relying on vast ecological lands—primarily forests and grasslands—as core carbon carriers. Between 1999 and 2024, the average annual carbon storage stabilized at approximately 1.357 × 10¹⁰ t in Yunnan and 1.582 × 10¹⁰ t in Sichuan. Together, these two provinces underpin roughly 76% of the total regional carbon stock. Conversely, Chongqing Municipality and Guizhou Province represent low-value zones. Constrained by topography and land use structures, these areas feature relatively high proportions of cultivated and construction land alongside limited ecological land. During the same period, average annual carbon storage was approximately 2.759 × 10⁹ t for Chongqing and 6.134 × 10⁹ t for Guizhou. In terms of regional contribution, Yunnan and Sichuan serve as the foundational pillars of the ESBZ’s carbon storage, driven by their extensive ecological land cover. In contrast, the Chongqing-Guizhou region, restricted by terrain and land composition, consistently occupies the lower range of the carbon storage spectrum.

Figure 9. Spatial distribution of carbon storage in ESBZ: (a) 1999, (b) 2004, (c) 2009, (d) 2014, (e) 2019, (f) 2024.

Under the NDS (Figure 10a), provincial carbon storage remained in a state of minor fluctuation and overall stability, showing no significant upward or downward trends. Carbon storage in Sichuan was 1.580 × 10¹⁰ t (a decrease of 4.974 × 10⁷ t vs. 2024), while Yunnan (1.358 × 10¹⁰ t), Guizhou (6.155 × 10⁹ t), and Chongqing (2.747 × 10⁹ t) all maintained fluctuation amplitudes within 1.775 × 10⁷ t relative to 2024. This pattern reflects the natural equilibrium of the regional ecosystem carbon cycle in the absence of additional anthropogenic intervention. Here, changes are regulated primarily by natural processes—such as vegetation growth and soil carbon dynamics—with fluctuations remaining within the ecosystem’s self-regulatory capacity. Under the UDS (Figure 10b), a divergent pattern emerged: “decline in Chongqing-Sichuan, stability in Yunnan-Guizhou.” Chongqing recorded the region’s largest relative decline, falling to 2.735 × 10⁹ t (−2.303 × 10⁷ t vs. 2024), while Sichuan dropped to 1.574 × 10¹⁰ t (−1.023 × 10⁸ t). In contrast, Yunnan (1.358 × 10¹⁰ t) and Guizhou (6.160 × 10⁹ t) remained stable, with fluctuations below 2.000 × 10⁷ t. This disparity stems from the regional concentration of urban expansion. The more pronounced urban sprawl in Chongqing and Sichuan exerted a stronger substitution effect on surrounding ecological lands, whereas the lower intensity of urban development in Yunnan and Guizhou resulted in minimal disturbance to carbon storage. In the CPS (Figure 10c), all provinces exhibited varying degrees of carbon deficits, with no regions showing growth. Sichuan experienced the most pronounced decline, dropping to 1.569 × 10¹⁰ t—a reduction of 1.580 × 10⁸ t from 2024. Guizhou followed with 6.138 × 10⁹ t (a decrease of 3.608 × 10⁷ t), while Chongqing and Yunnan declined to 2.745 × 10⁹ t (−1.272 × 10⁷ t) and 1.355 × 10¹⁰ t (−1.987 × 10⁷ t), respectively. This universal deficit is attributed to the systemic encroachment of cultivated land expansion upon carbon sinks like forests and grasslands. As key zones for cultivated land protection, Sichuan and Guizhou suffered more extensive reductions in ecological land, leading to disproportionately high carbon losses. Under the EPS (Figure 10d), Sichuan’s carbon storage climbed to 1.595 × 10¹⁰ t, an increase of 1.012 × 10⁸ t, marking the largest gain in the region. Guizhou reached 6.199 × 10⁹ t, adding 5.762 × 10⁷ t. Yunnan and Chongqing rose to 1.364 × 10¹⁰ t (+7.600 × 10⁷ t) and 2.754 × 10⁹ t (+4.462 × 10⁶ t), respectively. Notably, Sichuan and Chongqing—regions that suffered significant losses under the cultivated land protection and urban development scenarios—demonstrated the most robust carbon sink recovery under ecological protection. This region-wide growth substantiates the carbon enhancement efficacy of ecological restoration projects. Furthermore, as a core carbon storage zone, the implementation of protection measures in Sichuan proved critical to the aggregate gains of the entire ESBZ. Crucially, the simulation outcomes of the four scenarios reflect the relative response trends of regional land use and carbon storage dynamics under varying combinations of policy constraints. Consequently, their interpretation centers on comparing divergent future development trajectories within the ESBZ and their implications for spatial optimization, rather than serving as deterministic forecasts of actual future conditions.

Figure 10. Spatial distribution of carbon storage in four scenarios 2049: (a) NDS, (b) UDS, (c) CPS, (d) EPS.

4. Discussion

4.1. Analysis of Driving Forces of Land Use Change in the ESBZ

Situated within the critical ecological security barrier zone of the upper Yangtze and Pearl River basins, the ESBZ is characterized by complex topography and pronounced vertical zonality, which engender a typical “alpine–canyon–basin” gradient differentiation in land use. The natural landscape establishes a spatial “basemap” for diverse land categories through topographic accessibility and ecological suitability, significantly modulating the direction and intensity of transitions among cultivated land, forest land, and grassland [55]. Concurrently, in the ecologically fragile Karst mountainous areas, the superposition of karst geomorphology and hydrothermal conditions renders land use hypersensitive to natural constraints. Furthermore, the evolution of rocky desertification—driven by the interplay of land use and population pressure—reinforces the “locking effect” of this “natural substrate” on ecological land patterns [56].

Superimposed on this natural substrate, the formation of regional growth poles and urban agglomerations since the Reform and Opening-up has intensified the demand for land for urban expansion and infrastructure development, driven by the spatial agglomeration of population and industries. On a national scale, the expansion of construction land is predominantly governed by socioeconomic factors, exhibiting distinct spatial patterns of agglomeration and spillover [55]. Regarding the mechanism of conversion from cultivated land to construction land, multi-scale factors collectively shape the spatial selectivity of land occupation. On one hand, cultivated land loss induced by urban expansion demonstrates pronounced peri-urban characteristics, with cultivated land in urban fringes being highly susceptible to rapid conversion [57]. On the other hand, hierarchical socioeconomic and policy variables jointly elucidate the variations in construction land expansion and cultivated land conversion within hotspots, reflecting the constraining and amplifying effects of the “macro-institutions—local growth—land use behavior” chain on land transitions [58]. Furthermore, policy-oriented spatial carriers, such as development zones, can significantly alter the intensity and efficiency of urban land expansion in the short term. This reinforces the capacity of population-industry agglomerations to absorb construction land, driving land use spillover along transportation corridors and basin-valley belts.

Analyzed by land category (Figure 11), the dynamics of cultivated land are characterized by the coexistence of “demand-driven intensification” and “encroachment by urban expansion.” Driven by population agglomeration and market demand, intensive cultivation is more readily maintained or reinforced in basins, river valleys, and gentle terraces. Conversely, the superposition of mountainous terrain and labor outmigration—interacting with household scale-management decisions and off-farm employment opportunities—significantly elevates the probability of abandonment on marginal cultivated lands. Consequently, the stability of cultivated land becomes “locked” into regions characterized by low-to-mid elevations, high mechanization potential, and superior accessibility [59]. The evolution of forest land and grassland more distinctly embodies a pattern of “joint shaping by policy and nature.” Ecological engineering initiatives such as the “Grain for Green” program have generated significant carbon sink gains and land cover transformations on a national scale; however, their regional manifestations depend on the coupled patterns of topographic-climatic suitability and the intensity of anthropogenic activity [60]. In western China, such policies also induce identifiable trade-offs between grain supply and land system structure, thereby influencing the long-term configurational relationship between cultivated land and forest land [61]. In the Southwest Karst region, ecological engineering contributes significantly to vegetation dynamics, exhibiting a differentiated dominance relationship with climatic drivers across various geomorphological units. This implies that the expansion and quality improvement of forest and grass ecological lands predominantly occur within naturally suitable zones, modulated by the intensity of project implementation and management [62].

Figure 11. Contribution of driving factors: (a) Cultivated land, (b) Forest land, (c) Grassland, (d) Water area, (e) Construction land, (f) Unused land.

The dynamics of the water area and unused land can be interpreted as the combined outcome of anthropogenic engineering and ecological governance constrained by natural topography. Topography dictates the “skeleton” of river valley systems and reservoir siting, whereas demands for water supply, flood control, and irrigation in population centers often alter local surface water configurations via hydraulic projects, thereby enhancing the artificial attributes of the water area. It is imperative to note that while water areas have been integrated into the carbon storage assessment framework, their influence on the regional carbon cycle extends far beyond direct carbon storage. Hydrological cycling, soil moisture dynamics, and groundwater conditions can, by modulating vegetation growth, soil organic matter decomposition, and carbon migration processes, indirectly shape the regional carbon source-sink patterns [63,64,65,66,67]. In Karst mountainous regions, escalated land use intensity and population pressure can trigger a phasic shift in the dominant drivers of rocky desertification, consequently altering the potential and direction of transitions between unused land (e.g., bare land, exposed karst) and forest/grassland [56]. From the perspective of ecological governance, comprehensive control of rocky desertification—while enhancing regional ecological functions—may induce new ecological trade-offs (e.g., regarding habitat and biodiversity). This implies that converting unused land to ecological land necessitates zonal optimization balancing “restoration intensity–ecological risk–spatial suitability” [63]. In summary, land use change in this region represents a process of selective reshaping by demographic and economic activities within the constraints of the natural landscape. Natural factors delineate the spatial “basemap” and convertible boundaries of land categories, whereas urbanization, engineering development, and ecological projects drive the redistribution of land types within suitable zones, thereby profoundly influencing the spatiotemporal patterns and regulatory space of regional ecosystem carbon storage [55,64,65,66,67].

4.2. Impacts of LUCC Dynamics on Carbon Storage Changes

An integrated analysis of carbon storage data and land use transfer processes reveals a significant quantitative correlation between land use type conversion and total regional carbon storage in the ESBZ. This correlation serves as the core driver underlying carbon storage fluctuations during 1999–2024 and scenario differentiation in 2049. Between 1999 and 2024, the impact of land use conversion on carbon storage was characterized by a pattern of “loss dominance with localized gains.” Specifically, the average regional carbon density stood at 340.06 t·hm⁻² in 1999 (Table 10). However, transitions from high-carbon-density to low-carbon-density land types—such as cultivated land to construction land, grassland to unused land, and forest land to cultivated land/grassland—directly depressed the regional average. Consequently, the density dropped to 339.74 t·hm⁻² in 2004, remained suppressed after slight fluctuations in 2009, and further declined to 338.90 t·hm⁻² in 2014. This continuous decrease in average carbon density corresponded to the fluctuating decline in carbon storage observed in regions such as Chongqing and Guizhou, ultimately resulting in a net reduction of 1.284 × 10⁸ t in the ESBZ’s total carbon storage from 1999 to 2014. Post-2014, the gradual recovery of ecological land drove a rebound in average carbon density, which rose to 339.35 t·hm⁻² in 2019 and reached 339.88 t·hm⁻² in 2024. As the density approximated the initial 1999 baseline, the total regional carbon storage concomitantly approached its initial levels.

Table 10. Carbon density per unit area in ESBZ (t·hm⁻²).

Year	1999	2004	2009	2014	2019	2024
ESBZ	340.0639	339.7353	339.7421	338.9050	339.3532	339.8776
Chongqing Municipality	335.2780	335.7984	335.5743	333.2301	334.0662	334.6155
Yunnan province	354.5960	354.2948	354.7833	354.0438	354.2313	354.2274
Sichuan province	325.7274	325.6023	325.6588	324.8753	325.4203	326.1775
Guizhou province	350.2230	348.9417	347.8770	347.4412	347.9347	348.9827

Scaling down to the provincial level, Table 10 reveals a persistent and stable spatial heterogeneity in the average carbon density across Yunnan, Guizhou, Chongqing, and Sichuan. This demonstrates that the carbon storage pattern within the ESBZ is governed not merely by fluctuations in total area but is profoundly shaped by the structural composition of land use and the stability of ecological spaces within each province. Overall, Yunnan exhibits a consistently high average carbon density with minimal fluctuation, indicating a stable ecological spatial pattern dominated by forest land; consequently, high-carbon-density ecological land plays a commanding role in sustaining the region’s carbon storage. In contrast, the average carbon density in Guizhou demonstrates heightened sensitivity, reflecting the amplified response of ecologically fragile karst regions to land-use disturbances and ecological degradation. Chongqing manifests a comparatively low average carbon density with a fluctuating downward trajectory, signifying the sustained encroachment of urban expansion and construction land development upon high-carbon-density ecological spaces. Although Sichuan boasts a substantial total carbon storage capacity, it lacks a dominant average carbon density. This indicates that its high carbon storage relies heavily on the vast expanse of its ecological land area, rather than an inherent advantage in per-unit-area carbon density. Evidently, the genesis of these inter-provincial disparities in carbon density is fundamentally attributable to variations in both the compositional structure and the stability of forest land, grassland, cultivated land, and construction land. This underscores that interpretations of regional carbon storage spatial patterns must transcend mere total volumetric changes; it is imperative to integrate shifts in average carbon density to accurately identify the divergent mechanisms underpinning carbon sink maintenance across different regions.

Under the diverse development scenarios projected for 2049, the divergent orientations of land use conversion directly dictated the evolutionary trends of carbon storage. In the UDS, the increased conversion of cultivated land to construction land in Chongqing Municipality led to a reduction in total regional carbon storage of 2.303 × 10⁷ t compared to 2024. In the CPS, the expansion of cultivated land in Sichuan Province resulted in a decrease of 1.580 × 10⁸ t in total regional carbon storage relative to 2024. Conversely, under the EPS, the orientation of land use conversion was reversed. The restoration and expansion of high-carbon-density lands emerged as core characteristics, driving enhancements in carbon sequestration capacity in regions such as Sichuan and Yunnan. This ultimately achieved a net regional increase of 2.303 × 10⁸ t over 2024 levels. In summary, the dynamic evolution of carbon storage in the ESBZ is intrinsically the result of a dynamic trade-off between high-carbon and low-carbon land types during land use conversion. The expansion of low-carbon-density land serves as the core driver of carbon loss, whereas the restoration and expansion of high-carbon-density land constitute the critical pathway for carbon gain. This pattern provides a clear direction for enhancing regional carbon sequestration functions through land use regulation.

4.3. Study Limitations

By coupling the PLUS and InVEST models, this study integrates dynamic simulations of land use change with ecosystem carbon storage assessments in the ESBZ, effectively leveraging the strengths of both approaches. The PLUS model specializes in the spatially explicit simulation of land use dynamics, thereby overcoming the limitations of traditional Markov chain models, which often suffer from low spatial resolution and inadequate consideration of ecological mechanisms [24,68]. Simultaneously, the InVEST model provides precise carbon storage calculations based on ecosystem service assessment mechanisms, addressing the inability of conventional static analysis methods to simulate future scenario variations [14,69]. The coupling of these models integrates the drivers of land pattern evolution with ecological process mechanisms. This synthesis renders carbon storage assessment both dynamic and ecologically rational, enabling more accurate projections of regional carbon storage under diverse future scenarios [14,32,70]. Notwithstanding the demonstrated advantages of the combined PLUS–InVEST framework, several limitations warrant attention. First, the carbon density parameters used for estimation were derived primarily from literature rather than field surveys within the study area. This reliance may introduce regional representativeness bias, as it fails to capture variations in carbon density driven by local heterogeneity in climate, soil, and topography, potentially affecting the accuracy of the assessment [69,71,72]. Second, the relatively coarse spatial resolution of the land use data may obscure fine-scale ecological processes and detailed land changes. Coarse-grained rasters risk masking subtle patch-level variations, leading to potential deviations in the simulated spatial patterns [73,74]. Third, the model does not explicitly incorporate specific ecological engineering measures, such as afforestation and the “Grain for Green” program. Consequently, it is difficult to quantify the individual contributions of different projects to carbon storage changes. Although ecological protection and restoration initiatives are widely implemented in the region, their carbon sink effects were not independently evaluated, leaving the attribution of carbon storage dynamics incompletely resolved [32,70]. Notably, the ESBZ is characterized by pronounced topographical relief, diverse climate types, and a highly heterogeneous land use pattern. This complex heterogeneity poses severe challenges to simulation accuracy; the current framework may struggle to fully capture the integrated effects of multi-scale and multi-factor interactions, potentially introducing additional uncertainty by underestimating local process influences [70]. Furthermore, inherent uncertainties persist within the future scenario simulations presented in this study. These uncertainties primarily stem from variations in the intensity of policy implementation, fluctuations in the rates of economic growth and urbanization, and the potential impacts of future climate change on land-use demands and ecosystem carbon storage. Although this study incorporates both natural environmental and socio-economic drivers and calibrates carbon density parameters to enhance regional adaptability, it has yet to explicitly couple future climate scenarios with shifting policy intensities. Consequently, the omission of these variables could compromise the accuracy of the carbon storage simulations and the overall reliability of the scenario projections [58,63]. Moreover, the comparative analysis of the four future scenarios predominantly focuses on land-use patterns and their corresponding carbon storage responses, without integrating broader socio-economic metrics such as agricultural yields, household incomes, and the urban–rural development divide. Therefore, the NDS, UDS, CPS, and EPS are most appropriately utilized as reference scenarios to delineate land-use trade-offs under varying policy orientations; they cannot serve as direct substitutes for comprehensive policy evaluations targeted at food security, livelihood welfare, and regional equity.

Based on the aforementioned limitations, future research can be improved in the following aspects. First, high-resolution remote sensing, UAV monitoring, and ground plot surveys should be integrated to acquire more accurate regional carbon density data. Optimizing carbon pool parameters through these multi-source data will enhance the precision of carbon storage assessments [71,72,74,75]. Second, adopting land use data with higher spatial resolution and introducing intelligent interpretation technologies, such as deep learning, can improve land use simulation accuracy and refine the characterization of patch-scale changes [76,77]. Third, establishing an ecological engineering database to incorporate different project types into model scenarios is essential. Supplemented by methods such as factor analysis or Geographically Weighted Regression (GWR), this approach would enable the quantitative evaluation of the impacts of various engineering measures on carbon storage dynamics [32,78]. Fourth, future research should endeavor to construct a multi-objective synergistic evaluation framework that integrates carbon storage, food security, income welfare, and spatial equity, thereby enhancing the policy explanatory power of the scenario simulations. Such efforts will help identify the carbon sink benefits of diverse ecological interventions, providing a scientific basis for differentiated ecological protection strategies. In conclusion, refining methods and data to address these limitations will enhance simulation accuracy and reveal the spatiotemporal dynamics of carbon storage in the ESBZ more comprehensively. Ultimately, this will provide robust scientific support for regional carbon neutrality goals and ecological security strategies.

4.4. Policy Implications

Compared with existing research on ecological barrier zones, the land-use and carbon storage responses within the ESBZ demonstrate more pronounced features of composite governance. Previous studies indicate that within the Sichuan-Yunnan ecological barrier, multi-scenario simulations primarily emphasize the efficacy of land-use optimization in enhancing ecosystem services [79]. In the karst regions of southern China, ecological conservation scenarios prove beneficial in mitigating carbon storage depletion; however, inherent regional vulnerabilities perpetuate stark trade-offs between construction land expansion and ecological restoration [80]. Similarly, in the Qilian Mountains ecological barrier, an ecology-first orientation is more conducive to sustaining regional carbon storage and its associated ecological values [81], extending further into discussions on ecological compensation standards and conservation strategies [82]. Consistent with these findings, our study corroborates that the ecological protection scenario is optimal for bolstering regional carbon sink functions. However, unlike research predominantly focused on individual mountain ranges, single national parks, or isolated ecologically fragile zones, the ESBZ spans multiple provincial administrative units—Yunnan, Guizhou, Sichuan, and Chongqing. Consequently, it is subject to a confluence of multiplex constraints, encompassing cultivated land protection, urban expansion, karst vulnerability, and cross-provincial governance. Therefore, its policy imperative extends beyond mere ecological conservation; it centrally hinges upon differentiated provincial regulation predicated on ecological conservation redlines, cross-provincial ecological compensation and carbon revenue sharing, and hierarchical constraints on land-use intensity.

Between 1999 and 2024, forest land within the ESBZ experienced a net increase of 1.966 × 10⁴ km², whereas cultivated land and grassland saw net decreases of 9.738 × 10³ km² and 1.874 × 10⁴ km², respectively. Concurrently, 92.65% of construction land expansion originated from the conversion of cultivated land. This indicates that regional land-use dynamics are not an isolated ecological process, but rather the cumulative result of interactions among ecological conservation, food security, urban expansion, and regional development. The four future scenarios further elucidate these multi-objective trade-offs. The NDS represents a baseline pathway characterized by a gradual decline in carbon storage under the continuation of current developmental inertia. The UDS demonstrates that prioritizing urban expansion, while meeting construction demands, is likely to continuously compress cultivated land and ecological spaces, thereby exacerbating the risks of regional developmental imbalance. The CPS illustrates that a singular emphasis on stabilizing the quantity of cultivated land, though beneficial for safeguarding food production space, may undermine the region’s carbon sink capacity by encroaching upon forest land and grassland. Conversely, the EPS highlights the significant efficacy of an ecology-first orientation in enhancing carbon storage; however, without the support of optimized cultivated land layout, agricultural compensation, and livelihood transitions, it may pose potential implications for food security, household incomes, and regional equity. Therefore, land-use optimization in the ESBZ should not regard any single scenario as a universally applicable solution. Instead, utilizing these four scenarios as references, it should seek spatial synergy and balanced governance among carbon sink enhancement, cultivated land protection, urban development, and social equity.

Building upon this foundation, spatial optimization must emphasize differentiated regulation. As regions with high concentrations of population, industry, and construction activities, Sichuan and Chongqing should focus on curbing urban sprawl along the Chengdu-Chongqing corridor. They should prioritize urban renewal, unlocking the potential of existing land stock, and promoting intensive land use to substitute new construction land expansion, thereby alleviating the continuous appropriation of cultivated land and the pressure on surrounding forest land and grassland carbon pools. Relying on its high proportion of forest land, Yunnan should consolidate its existing ecological baseline. While strengthening ecological conservation redlines and biodiversity protection, it should focus on halting the continuous shrinkage of grassland, synergistically advancing carbon sink enhancement and ecological security maintenance. In light of the fragile ecological baseline and high degradation risk in its karst regions, Guizhou should intensify rocky desertification control and degraded land restoration. Priority should be given to restoring forest land and grassland carbon pools to enhance regional ecological resilience, while concurrently addressing the developmental constraints and livelihood needs of vulnerable mountainous areas. Integrating the ecological functional patterns and ecological conservation redlines of the ESBZ, Sichuan should prominently feature the redline units located in the Western Sichuan Plateau and the crucial water conservation areas in the upper reaches of the Yangtze River. Yunnan should prioritize the alpine gorges of Northwest Yunnan, critical biodiversity hotspots, and the peripheral areas of key national parks. Guizhou should focus on rocky-desertification-sensitive areas and key soil and water conservation zones, while Chongqing should highlight the ecological barrier zones in its northeast and southeast, alongside the “Four Mountains” of its main urban area.

Considering that Sichuan and Yunnan collectively contribute approximately 76% of the region’s carbon storage, while Chongqing and Guizhou face intense urban expansion pressures and ecological restoration mandates respectively, a pronounced spatial mismatch of “carbon sink supply—developmental constraints—conservation costs” exists within the region. Consequently, future carbon-oriented governance in the ESBZ should not be confined merely to the provincial spatial optimization level but must further establish cross-provincial ecological compensation and carbon revenue-sharing mechanisms. Specifically, within the existing framework of ecological protection compensation systems, metrics such as carbon storage scale, carbon sink increments, ecological conservation costs, developmental opportunity costs, and the paying capacity of beneficiary regions could be incorporated as bases for cross-provincial compensation accounting. This would provide sustained compensatory support to regions bearing the primary functions of ecological conservation and carbon sink supply. Concurrently, it is crucial to explore the integration of natural forest protection, the return of cultivated land to forest land and grassland, rocky desertification control, and key ecological restoration projects with carbon sink value realization mechanisms. This approach would empower ecologically advantaged regions to translate ecological product values and carbon revenues into sustainable development capacities, thereby mitigating the structural imbalance where “high-carbon-sink regions bear heavy conservation burdens but receive disproportionately low direct benefits.” Such institutional arrangements already possess a pragmatic policy foundation at the national level, exemplified by the “Regulations on Ecological Protection Compensation” implemented in 2024 and the “Guiding Opinions on Further Improving the Horizontal Ecological Protection Compensation Mechanism” issued in 2025. These provide a clear institutional window and policy basis for the ESBZ—particularly for Sichuan, Yunnan, and Chongqing, which are integral to the ecological security pattern of the upper Yangtze River—to advance cross-provincial ecological compensation, carbon revenue sharing, and regional synergistic governance.

Regarding the control of land-use intensity, a system of zoned and hierarchical rigid constraints must be established. For areas within ecological conservation redlines, new industrial, mining, and large-scale real estate developments should be strictly restricted, permitting only ecological restoration, essential infrastructure, and livelihood-security projects. For ecological buffer and synergistic management zones on the periphery of the redlines, emphasis should be placed on restricting the disordered conversion of forest land, grassland, and wetlands into construction land and low-efficiency orchards, thereby preventing peripheral development from continuously encroaching upon core ecological spaces. For areas within urban development boundaries, priority must be given to urban renewal, the redevelopment of inefficient land, and tapping the potential of existing construction land. Particularly in urban agglomerations such as the Chengdu-Chongqing corridor, Central Guizhou, and Central Yunnan, metrics including the growth rate of new construction land, the conversion rate of cultivated land, the net loss rate of forest land and grassland, and the carbon loss per unit of new construction land should be incorporated into local annual performance evaluations to establish a binding mechanism. Only through this combined governance framework of “redline control—compensation incentives—intensity constraints” can the overall carbon sink capacity of the ESBZ be more effectively elevated, simultaneously strengthening its comprehensive supportive role for national ecological security and coordinated regional development. It must be noted that this study has not yet conducted further calculations of cross-provincial compensation standards or simulations of carbon revenue allocation ratios. Consequently, the aforementioned discussion primarily serves as a mechanistic deduction based on the heterogeneity of provincial carbon storage results; future quantitative research remains necessary, integrating fiscal transfer payments, ecological engineering costs, and carbon market prices.

5. Conclusions

Based on the coupled PLUS–InVEST framework, this study systematically analyzed land-use dynamics and their impacts on ecosystem carbon storage within the ESBZ from 1999 to 2024. The results indicate that the overall land-use evolution in the study area was characterized by the concurrent recovery of forest land and the expansion of construction land, with the latter primarily appropriating cultivated land. Despite significant land-use transitions, regional carbon storage remained relatively stable overall, exhibiting a phased pattern of initial decline followed by subsequent recovery. Evidently, carbon storage dynamics within the ESBZ are not the result of the simple augmentation or reduction of a single land category; rather, they represent a comprehensive response to land-system restructuring driven by the synergistic interactions of ecological conservation, cultivated land requisition-compensation, and urban expansion.

Future scenario simulations reveal pronounced disparities in carbon storage responses across the four scenarios by 2049. The NDS, reflecting a baseline pathway devoid of additional interventions, projects a marginal decline in carbon storage. Under the UDS, the encroachment of construction land expansion upon ecological spaces precipitates a further reduction in regional carbon storage. Although the CPS is conducive to stabilizing the quantity of cultivated land, this is primarily achieved through the appropriation of forest land and grassland, thereby incurring the most substantial carbon loss. Conversely, guided by ecological conservation and restoration, the EPS attains the highest carbon storage. Overall, the NDS, UDS, CPS, and EPS do not constitute a simplistic hierarchy of merit; instead, they represent four distinct development trajectories prioritizing baseline continuity, urban expansion, food security, and ecological conservation, respectively. Their comparison elucidates the multi-objective trade-offs inherent in the regional land system among carbon sink enhancement, cultivated land protection, and spatial development.

In terms of spatial heterogeneity, Sichuan and Yunnan serve as the core supporting areas for carbon storage within the study region, whereas levels in Chongqing and Guizhou are comparatively lower. Consequently, future land-use governance must adhere to principles of spatial optimization and differentiated regulation: the Chengdu-Chongqing corridor should strictly curb urban sprawl, promoting urban renewal and intensive land use; Yunnan should consolidate its forest land-dominated spatial pattern and halt the continuous shrinkage of grassland; and Guizhou should intensify ecological restoration and degraded land management within its fragile karst regions. Simultaneously, it is imperative to establish cross-provincial ecological compensation and carbon revenue-sharing mechanisms. Integrating ecological engineering performance evaluations and carbon monitoring and accounting with territorial spatial use controls will ultimately elevate the overall carbon sink capacity and the level of synergistic regional governance within the ESBZ.