https://orcid.org/0000-0001-8087-424X
Figures
Abstract
Land-use changes significantly influence carbon storage capacity by altering the structure, layout, and function of terrestrial ecosystems. Predicting the relationship between future land-use changes and carbon storage is essential for optimizing land-use patterns and making rational, ecology-based decisions. Using multi-period land-use data from Xinjiang, we analyzed the spatial pattern of carbon storage. Based on land-use change patterns in Xinjiang from 2000 to 2020, we coupled the Markov-Future Land Use Simulation (FLUS)-Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model to simulate and predict land-use spatial patterns in Xinjiang for 2035 under two scenarios: natural growth and ecological protection. Carbon storage and its spatiotemporal dynamic changes under these scenarios were evaluated, and the Geodetector was employed to analyze the spatial heterogeneity of carbon storage from a statistical perspective, revealing the influence of various driving factors. The results showed that: (1) From 2000–2020, grassland and unused land were the primary land-use types in Xinjiang, accounting for over 28.85% and 60.17% of the total area, respectively. By 2035, cropland, forest, water, and construction land areas are expected to increase, while grassland and unused land areas are projected to decrease. Under the ecological protection scenario, cropland, forest land, and grassland—major main contributors to carbon storage—will be effectively conserved to some extent. (2) From 2000 to 2020, Xinjiang’s carbon storage capacity exhibited an overall increasing trend, with a cumulative increase of 137.515×105 t and a growth rate of 1.58%. However, this capacity is projected to decline by 2035, with an estimated reduction of 168.344×105 t compared to that in 2020. Ecological protection is anticipated to mitigate this decline, increasing carbon storage by 13.227×105 t relative to the natural growth scenario. (3) Geodetector analysis indicated that land-use types had the greatest carbon storage explanatory power for carbon storage (q = 0.80), followed by soil types (q = 0.41), net primary productivity (q = 0.32), and geomorphology (q = 0.22). This highlights land-use types as the most critical environmental factor determining the spatial pattern of carbon storage. These findings provide scientific insights and recommendations for the sustainable development management and the enhancement of carbon storage functions.
Citation: Jin M, Duan X, Zhang Y, Xu Q (2025) Predicting the spatial pattern of land use change and carbon storage in Xinjiang: A Markov-FLUS-InVEST model approach. PLoS ONE 20(4): e0321929. https://doi.org/10.1371/journal.pone.0321929
Editor: Sudeshna Bhattacharjya,, ICAR-Indian Institute of Soil Science, India
Received: August 18, 2024; Accepted: March 13, 2025; Published: April 17, 2025
Copyright: © 2025 Jin et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data underlying this study are available from the Open Science Framework database at https://osf.io/nqyug/(DOI:10.17605/OSF.IO/NQYUG).
Funding: Mengting Jin was supported by the Science and Technology Innovation Foundation of the Command Center for Comprehensive Survey of Natural Resources (KC20230015) and the China Geological Survey Project (DD20220962). She played a role in data collection, methodology, and project administration. Quan Xu was supported by the China Geological Survey Project (DD20240740). He played a role in data collection and methodology.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Severe global climate change presents significant challenges to human survival and sustainable development [1,2]. Carbon storage in terrestrial ecosystems constisutes a critical component of global carbon storage, playing a key role in mitigating global climate change by absorbing and releasing greenhouse gases and effectively regulating regional climates [3]. Land-use and cover change (LUCC) is one of the primary driving factors influencing changes in carbon stocks. LUCC impacts vegetation and soil carbon storage in terrestrial ecosystems, thereby alteing regional carbon storage, ecosystem structure, and function, ultimately affecting carbon cycling processes [4,5]. In the context of China’s dual carbon strategy—aiming to peak carbon emissions by 2030 and achieve carbon neutrality by 2060—Xinjiang has emerged as a strategically significant region. Therefore, investigating the spatiotemporal patterns of land use and cover changes and their impacts on carbon storage is essential.
Current methods for assessing the impact of LUCC on carbon storage include field surveys and model simulations [6]. Unlike traditional field surveys, model simulations can evaluate carbon stock changes at various scales and provide spatially visualized evaluation results. Among the many available evaluation models (e.g., ARIES, GUMBO, MIMES, CITYgreen), the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model has been widely adopted due to its low data requirements, fast processing speed, and high assessment accuracy [7–9]. Developed by the Natural Capital Project in the United States, InVEST is a modeling system designed to evaluate the functional volume and economic value of ecosystem services, supporting ecosystem management and decision-making. He et al. [10] used a new model to assess the impacts of urban expansion on regional carbon storage by linking the Land Use Scenario Dynamics-urban and InVEST models. Similarly, Wang et al. [11] applied the InVEST model to estimate carbon storage in the Taihang Mountains from 2005 to 2020. They examined the main drivers of the spatial evolution of carbon storage and analyzed their driving mechanisms. These studies highlight that the InVEST model is one of the most widely used approaches for evaluating regional carbon storage.
Exploring the impact of LUCC on ecosystem carbon storage under different scenarios can provide quantified and spatially visualized distribution results. This information is valuable for decision-makers in formulating effective land-use policies and ecological protection plans, as well as optimizing the national spatial pattern [12–14]. However, most existing studies primarily focus on analyzing historical carbon stock changes resulting from land-use alterations, with relatively few studies addressing future land-use changes under multiple scenarios and their impact on carbon storage. Additionally, limited research investigates the spatial heterogeneity of carbon storage using spatial statistics to uncover the influence of diverse driving factors on the spatiotemporal evolution characteristics of carbon storage [15,16].
To address these gaps, we simulated future land-use patterns in Xinjiang under natural growth and ecological protection scenarios using the Future Land Use Simulation (FLUS) model [17], based on land-use data from 2000 to 2020. Subsequently, we employed the InVEST model to evaluate the spatiotemporal evolution of carbon storage in Xinjiang from 2000 to 2020 and predicted conditions for 2035 under different scenarios. Finally, we applied Geodetector to analyze the impact of various driving factors on the spatial distribution of carbon storage.
2. Study area and data
2.1. Overview of the study area
Xinjiang is located inland, between 73°40′E to 96°18′E longitude and 34°25′N to 48°10′N latitude. The region features complex terrain, including various landforms such as mountains and basins. It has a typical temperate continental climate characterized by abundant sunshine and significant temperature differences. The area is generally arid with relatively low annual precipitation—averaging approximately 150 mm. Precipitation varies significantly across the region, with northern Xinjiang receiving more rainfall than southern Xinjiang, while temperatures are higher in southern Xinjiang compared to the north [18]. The topography of Xinjiang affects carbon stocks in a number of ways, with vegetation cover and soil conditions in mountainous areas being conducive to carbon fixation and storage, while basins and unutilized areas have relatively low carbon stocks. Factors such as topographic relief, altitude and moisture conditions also influence carbon cycling and distribution to some extent [19].
Xinjiang is China’s largest provincial-level administrative region, covering 1.66 million km2, which accounts for one-sixth of China’s total land area. It has a border length exceeding 5,000 km, sharing boundaries with eight neighboring countries. Historically, Xinjiang was a critical hub of the ancient Silk Road, and today it plays an essential role in China’s “Belt and Road” policy. Its strategic geopolitical importance cannot be overstated. Consequently, studying the spatiotemporal evolution of carbon storage in Xinjiang holds significant practical value. A schematic of the study area is provided in Fig 1.
The basemap is obtained from the Standard Map Service System (http://bzdt.ch.mnr.gov.cn/).
2.2. Data
The data used in this study (Table 1) were obtained from the Resource and Environmental Science Data Platform.
The land use data is derived from remote sensing satellite imagery, including Landsat 8 OLI and GF-2, among others. This dataset is acquired through a high-resolution remote sensing technology system that integrates unmanned aerial vehicles and ground surveys, complemented by human-computer interactive interpretation methods grounded in geoscience knowledge. The classification and overall accuracy were assessed using a confusion matrix. The first category of land use achieved a comprehensive accuracy of over 93%, while the second category’s comprehensive accuracy exceeded 90%, satisfying the user’s requirement for a drawing accuracy at a scale of 1:100,000 [20,21]. Based on the “Classification Standard for Land Use Status” (GBT 21010–2017) and accompanying data descriptions, land-use types were categorized into six classes: cropland, forest land, grassland, water area, construction land, and unused land.
Based on previous research [18], the selected driving factors included the Digital Elevation Model (DEM), slope, aspect, rainfall, land surface temperature (LST), net primary productivity (NPP) of vegetation, luminous index, soil types, and geomorphology. To ensure consistency across datasets, the projection method was unified and transformed into the Albers_Conic_Equal_Area system. Additionally, the spatial resolution was resampled to 1 km using cubic convolution interpolation.
3. Method
The study involved three key steps: (1) a multi-scenario land use simulation using the FLUS model, (2) calculations of the spatiotemporal characteristics of carbon storage using the InVEST model, and (3) an analysis of the driving forces behind carbon stock using the Geodetector. The methods applied in each of these steps are detailed in the following subsections. The technological roadmap is illustrated in Fig 2.
3.1. Multi-scenario land-use change simulation based on the FLUS model
FLUS is a land-use prediction model based on cellular automata [22]. Using transformation rules and quantitative relationships between land-use types and driving factors, the model employs a roulette selection algorithm to simulate the spatial pattern of land use under different scenarios and years [23]. It also spatially allocates simulated land-use demand [24,25]. In this study, we used GeoSOS-FLUS V2.4 software to predict land-use changes in Xinjiang under different scenarios for the year 2035.
Based on existing land-use data, we applied the Markov model to simulate future land-use type areas in Xinjiang for 2035. Driving factors for land-use change included the DEM, slope, aspect, rainfall, temperature, land surface temperature (LST), net primary productivity (NPP) of vegetation, luminous index, soil types, and geomorphology. Notably, temperature refers to the thermal state of the atmosphere, measured in a sheltered environment, which indicates atmospheric warmth. In contrast, LST reflects the ground’s thermal condition, measured at the interface between the Earth’s surface and the air. LST is influenced by various factors such as terrain type, vegetation, and soil moisture.
We also incorporated neighborhood factors and transfer matrix parameters into the simulation. Neighborhood factors represent the difficulty of converting one land-use type to another, with values ranging from 0 to 1. A score closer to 1 indicates a strong expansion ability for that land-use type. Transfer matrices are represented by values of 0 and 1: a value of 1 indicates that a conversion between two land-use types is permissible, while 0 indicates that it is not. Detailed parameter settings are provided in Table 2.
Two land-use scenarios were simulated: natural growth and ecological protection. In the natural growth scenario, land-use types were allowed to evolve based on current conditions without deliberate modification, and the corresponding matrix value was set to 1. In the ecological protection scenario, enhanced protection measures were applied to ecological lands, such as forests, grasslands, and water bodies, while the expansion ability of other land-use types was restricted. For this scenario, the corresponding matrix value was set to 0.
Finally, we estimated the spatial distribution of various land-use types in Xinjiang for 2035 under both scenarios by modifying the input parameters of the FLUS model.
3.2. Carbon storage assessment based on the InVEST model
In this study, the carbon storage module of the InVEST model was used to assess the spatiotemporal changes in carbon storage in Xinjiang. This module calculates ecosystem carbon storage by integrating land-use and carbon density data [26]. It categorizes ecosystem carbon storage into four primary carbon pools: aboveground vegetation carbon (Cabove), belowground vegetation carbon (Cbelow), soil organic carbon (Csoil), and dead organic carbon (Cdead).
Aboveground vegetation carbon pool: Includes the carbon stored in all living vegetation above the ground. Belowground vegetation carbon pool: Refers to the carbon stored in the root systems of living plants. Soil organic carbon pool: Typically represents the organic carbon stored in mineral and organic soils. Dead organic carbon pool: Represents the carbon stored in litter and dead plants [27,28]. The formula for calculating carbon storage is as follows:
Where C represents the total carbon storage per unit area for each land-use type, Cabove represents the carbon density of aboveground biomass, Cbelow represents the carbon density of belowground biomass, Csoil represents the soil carbon density, Cdead represents the carbon density of dead biomass, and Ctotal represents the total carbon storage of all land cover types. Ak represents the area of each land cover type, Ck represents the carbon density of each unit, and n represents the number of land-use types. The carbon density data used in this study were obtained from previous relevant research [29], and the carbon density values for each land-use type are shown in Table 3.
3.3. Driving force analysis of spatiotemporal evolution of carbon storage based on Geodetector
The Geodetector is a software tool used to measure and attribute spatially stratified heterogeneity. This method does not make linear assumptions, and it has a simple form with clear physical meaning [30]. The basic idea is to divide the study area into several subregions. If the sum of the variances of the subregions is smaller than the total variance of the region, spatial heterogeneity exists. If the spatial distribution of two variables tends to be consistent, there is a statistical correlation between them [31,32]. The Geodetector includes four detectors:
- (1) Differentiation and factor detection
The spatial differentiation of the probe-dependent variable (Y) and the extent to which the probe driving factor (X) explains the spatial differentiation of Y, measured by the q value, is expressed as:
Where: h = 1,..., L represents the strata of variable Y or factor X; Nh and N are the number of units in layer h and the entire region, respectively; σh2 and σ2 are the variances of layer h and area Y values, respectively; and SSW and SST are the Within Sum of Squares and Total Sum of Squares, respectively.
- (2) Interaction detection
Recognizing the interaction between different driving factors involves assessing the combined effects of two factors (X1 and X2) on the dependent variable Y by determining whether their interaction increases or decreases the explanatory power of Y, or whether these factors have independent effects on Y. The relationships between these two factors can be classified into the following categories:
- (3) Risk detection
A t-test was used to determine whether there was a significant difference between the means of the attributes in two subregions. It was calculated as follows,
Where represents the average value of the attributes in subregion h, nh is the number of samples in subregion h, and Var is the variance.
- (4) Ecological detection
The F-test was used to assess whether there was a significant difference in the spatial distribution of attribute Y due to the influence of two factors, X1 and X2. This was achieved by comparing the effects on Y as follows,
where NX1 and NX2 represent the sample sizes of the two factors X1 and X2, respectively; SSWX1 and SSWX2 denote the sums of the within-group variances formed by X1 and X2, respectively; and L1 and L2 represent the numbers of strata for X1 and X2, respectively.
We used Geodetector Software (Beta) for ArcGlS Pro to investigate the impact of various driving factors on carbon storage in Xinjiang. Carbon storage was the dependent variable (Y), and the driving factors were the independent variables (X). Random sampling was used to avoid the influence of spatial autocorrelations and human factors. One thousand sample points were randomly generated within the study area, and the carbon storage and driving factor values for each point were extracted for analysis.
4. Results
4.1. Analysis of characteristics of land use change
Between 2000 and 2020, grassland and unused land were the dominant land-use types in Xinjiang, accounting for more than 28.85% and 60.17% of the total area, respectively (Fig 3 and Table 4). The most common land-use types were cropland, forestland, and water bodies, which accounted for 3.73%, 1.75%, and 2.07% of the total area, respectively. Construction land covered the smallest area, accounting for only 0.27% of the total area. Over the 20 years, cropland, grassland, and construction land showed increasing trends, with increases of 21,712 km2, 4,084 km2, and 3,371 km2, respectively. In contrast, areas of forestland, water bodies, and unused land showed decreasing trends, with decreases of 8,138 km2, 11,638 km2, and 8,023 km2, respectively. From the perspective of spatial distribution, the increases in cropland, grassland, and construction land are mainly concentrated in Tacheng, the Bortala Mongol Autonomous Prefecture, and the oasis areas in southern Xinjiang. The decreases in forest land and water bodies are primarily observed in the Korla and Ili regions.
Concerning land-use transfer, the increase in cropland and forestland was mainly due to the development of unused land, while the increase in construction land resulted primarily from the conversion of cropland, grassland, and unused land. Most forestland was converted to grassland, and areas of water bodies gradually decreased and became unused land. It is evident that, with the continuous development of the social economy between 2000 and 2020, Xinjiang focused on expanding farmland, conducting afforestation, and constructing infrastructure. Areas of unused land gradually decreased, accelerating urbanization and playing a positive role in Xinjiang’s economic development and environmental protection. However, due to the impacts of human activities and climate change, there have also been some negative effects on ecological lands such as forestland and water bodies.
In the 2035 natural growth scenario, cropland, forestland, water bodies, and construction land in Xinjiang are expected to increase by 5,629 km2, 6,020 km2, 11,850 km2, and 1,268 km2, respectively, with increases of 8.75%, 30.14%, 49.97%, and 19.50%. Areas of grassland and unused land are expected to decrease by 19,972 km2 and 6,353 km2, respectively, with decreases of 5.86% and 0.92%. From the perspective of land-use transfer, the increase in cropland, forestland, and construction land will mainly occur through the conversion of grassland. In contrast, the increased water body areas will mainly result from the conversion of grassland and unused land. Grassland areas are predicted to decrease significantly, primarily due to their conversion into unused land.
Under the ecological protection scenario, the change in the land-use pattern in Xinjiang was predicted to be similar to that under the natural growth scenario, but with different area changes. Compared to the natural growth scenario, cropland, forestland, and grassland would increase by 133 km2, 397 km2, and 180 km2, respectively, while water bodies, construction land, and unused land would decrease by 13 km2, 646 km2, and 51 km2, respectively. Under the ecological protection scenario, croplands, forestlands, and grasslands, which are the primary contributors to carbon storage, were effectively protected to a some extent.
4.2. Characteristics analysis of carbon storage changes under multiple scenarios
The spatial distribution of carbon storage in Xinjiang is shown in Fig 4. In terms of quantity, the total amount of carbon stored in Xinjiang every 5 years from 2000 to 2020 was 8689.373×105 t, 8692.023×105 t, 8697.929×105 t, 8700.356×105 t, and 8826.889×105 t, respectively, exhibiting an overall increasing trend. There was a cumulative increase of 137.515×105 t, with a growth rate of 1.58%. On average, the amount of carbon stored increased by 6.876×105 t per year. The most significant change in Xinjiang’s carbon storage occurred between 2015 and 2020, with an increase of 126.532×105 t, accounting for 92.01% of the cumulative increase. In terms of area, the area of carbon stock increase amounted to 188,928 km2 and the area of decrease amounted to 182,667 km2. From a spatial distribution perspective, regions demonstrating significant increases in carbon storage are predominantly concentrated in the mountainous areas of Xinjiang. In contrast, areas showing decreased carbon storage levels are primarily distributed along the periphery of both the Taklimakan Desert and the Gurbantunggut Desert in Xinjiang.
As shown in Fig 5, high carbon storage areas were mainly concentrated in cropland, forestland, and grassland areas due to the different land-use types with varying carbon sequestration capabilities. In contrast, the carbon densities of water bodies, construction land, and unused land were relatively low, resulting in a lower carbon storage content. Between 2000 and 2020, the increased cropland and grassland areas in Xinjiang resulted in a relative increase in carbon storage. The changes in carbon storage were mainly influenced by changes in land-use types.
Under the natural growth scenario, the overall trend of carbon storage in Xinjiang is expected to decrease, with a projected reduction of 168.344×105 t in 2035 compared to 2020. This decline in carbon storage is primarily attributed to the conversion of large areas of grassland into construction land and unused land, leading in a decrease in both soil carbon storage and the carbon storage of aboveground and belowground vegetation. As a result, the total amount of carbon stored in Xinjiang would decline. However, the ecological protection scenario would mitigate this reduction in carbon storage. By limiting the conversion of forestland and grassland to other land-use types, carbon storage could increase by 13.227×105 t compared to the natural growth scenario. The implementation of ecological protection policies could enhance the effectiveness of regional ecological conservation and facilitate sufficirnt carbon sequestration in Xinjiang.
4.3. Driving force analysis of carbon storage based on Geodetector
The results of the analysis based on the Geodetector are shown in Fig 6. Fig 6a presents the calculated results for all driver divergence probes. The divergence factor q ranges from 0 to 1, with larger values indicating stronger explanatory power of the driver regarding carbon stock, and vice versa. The results revealed that land-use types had the highest q value (0.80), followed by soil types (0.41), NPP (0.32), and geomorphology (0.22), while the q values for the other factors were relatively small. Furthermore, from the perspective of interaction detection, the q-value increased to 0.83 after the interactions between land-use types and both soil types and geomorphological factors. This further enhanced the explanatory power of carbon storage, suggesting that these factors jointly play a positive role.
Fig 6b presents the interaction types among all driving factors, enabling an assessment of whether the combined effect of any two factors enhances or reduces the explanatory power for carbon stocks. The results indicate that pairwise interactions between all factors can produce an enhancing effect. Specifically, land-use types serve as a two-factor enhancer when interacting with DEM, slope, rain, temperature, LST, NPP, soil types, and geomorphology, respectively. In contrast, land-use types act as a nonlinear enhancer when interacting with aspect and luminous index, respectively. All these interactions contribute to a further improvement in the explanatory power of carbon stocks.
Fig 6c presents the ecological detection results for all the driving factors, which indicate whether there was a significant difference in their impact on the spatial distribution of carbon storage. (“Y” indicates a significant difference, while “N” indicates no significant difference.) From the results, it can be seen that land-use types, LST, NPP, and soil types are all significantly different from other factors, indicating that these types of factors act independently on carbon stock distribution, respectively.
In conclusion, the analysis showed that, among these driving factors, land-use type was the most relevant and robust explanatory factor for change in carbon storage, and it was the primary environmental factor determining the spatial pattern of carbon storage.
5. Discussion
Rapid urbanization leads to large-scale encroachment of construction land on croplands, forests, and grasslands [33]. From 2000 to 2020, the overall trend of carbon storage in Xinjiang increased. However, under the two typical future scenarios, the carbon storage trend is expected to decrease. Employing the ecological protection scenario would mitigate this trend to some extent compared to the natural growth scenario because it comprehensively coordinates the functions of various land types and is more conducive to slowing the loss of carbon storage. Implementing such protectionist policies would help maintain a regional carbon balance and be of great significance for optimizing the spatial pattern of Xinjiang’s land.
The results showed that between 2000 and 2020, grassland was the dominant land-use type in Xinjiang, and its carbon storage capacity was higher than that of other ecosystems. Therefore, protecting grasslands is key to achieving the sustainable development of carbon storage functions in Xinjiang. The changes in carbon storage under different scenarios show that Xinjiang should prioritize protecting its grassland resources to prevent degradation. It is essential to focus on optimizing the land-use structure with a low-carbon orientation and to promote the transformation of low-coverage grassland and unused land into high-coverage grassland through policies that establish grazing bans and protection zones, restrict associated land development, and strengthen the legal management of grasslands. Additionally, it is crucial to improve the grassland monitoring network and consolidate the achievements of grassland ecological restoration.
Land-use change is a complex process influenced by multiple driving factors [34,35]. In this study, factors such as DEM, slope, aspect, precipitation, and temperature were selected as the driving factors for simulating future land-use patterns. These factors exhibited an excellent fitting effect on the various land-use types, and the simulated results were highly accurate (Kappa = 0.94), thus meetting the research requirements. Our results were supported by those of previous relevant studies[29].
In future research, policy-related factors should be added into the driving factor system to reveal the impact of land use policies on carbon storage. At the same time, more driving factors should be further screened and optimized to improve the accuracy of land-use simulation and driving force analysis. Moreover, to accurately estimate regional carbon storage changes and enhance the accuracy of model verification, efforts should be made to strengthen the acquisition of carbon density data, as well as to conduct localization calibration and field measurements of core indicators.
6. Conclusions
- (1) From 2000 to 2020, the land-use types in Xinjiang were dominated by grassland and unused land, which accounted for more than 28.85% and 60.17% of the total area, respectively. By 2035, cropland, forestland, water bodies, and construction land are expected to increase, while areas of grassland and unused land are expected to decrease. Under the ecological protection scenario, cropland, forestland, and grassland—key contributors to carbon storage—will receive adequate protection. Compared to the natural growth scenario, the areas of these land types are expected to increase by 133 km2, 397 km2, and 180 km2, respectively.
- (2) From 2000 to 2020, the overall trend of carbon storage in Xinjiang increased, with a cumulative rise of 137.515×105 t, or an increase of 1.58%, and an average annual increase of 6.876×105 t. By 2035, carbon storage is expected to decrease, with an estimated reduction of 168.344×105 t compared to 2020. Ecological protection would help mitigate this decrease in carbon storage: under the ecological protection scenario, carbon storage is expected to increase by 13.227×105 t compared to the natural growth scenario.
- (3) A Geodetector was used to explore the degree of influence each driving force on the spatial distribution of carbon storage. The results showed that land-use type had the highest explanatory power for carbon storage (q = 0.80), followed by soil types (0.41), NPP (0.32), and geomorphology (0.22). These findings indicate that land-use type is the primary environmental factor determining the spatial pattern of carbon storage.
References
- 1. McMonigal K, Larson S, Hu S, Kramer R. Historical changes in wind‐driven ocean circulation can accelerate global warming. Geophys Res Lett. 2023;50(4).
- 2. Liao W, Liu X, Xu X, Chen G, Liang X, Zhang H, et al. Projections of land use changes under the plant functional type classification in different SSP-RCP scenarios in China. Sci Bull (Beijing). 2020;65(22):1935–47. pmid:36738059
- 3. Xu C, Zhang Q, Yu Q, Wang J, Wang F, Qiu S, Ai M, Zhao J. Effects of land use/cover change on carbon storage between 2000 and 2040 in the Yellow River Basin, China. Ecol Indic. 2023;151:110345.
- 4. Wang Y, Liang D, Wang J, Zhang Y, Chen F, Ma X. An analysis of regional carbon stock response under land use structure change and multi-scenario prediction, a case study of Hefei, China. Ecol Indic. 2023;151:110293.
- 5. Wei B, Kasimu A, Reheman R, Zhang X, Zhao Y, Aizizi Y, et al. Spatiotemporal characteristics and prediction of carbon emissions/absorption from land use change in the urban agglomeration on the northern slope of the Tianshan Mountains. Ecol Indic. 2023;151:110329.
- 6. He Y, Ma J, Zhang C, Yang H. Spatio-temporal evolution and prediction of carbon storage in Guilin based on FLUS and InVEST models. Remote Sens. 2023;15(5):1445.
- 7. Zhao J, Shao Z, Xia C, Fang K, Chen R, Zhou J. Ecosystem services assessment based on land use simulation: a case study in the Heihe River Basin, China. Ecol Indic. 2022;143:109402.
- 8. Chen D, Jiang P, Li M. Assessing potential ecosystem service dynamics driven by urbanization in the Yangtze River Economic Belt, China. J Environ Manage. 2021;292:112734. pmid:33984640
- 9. Sun W, Liu X. Spatio-temporal evolution and multi-scenario prediction of ecosystem carbon storage in Chang-Zhu-Tan urban agglomeration based on the FLUS-InVEST model. Sustainability. 2024;16(16):7025.
- 10. He C, Zhang D, Huang Q, Zhao Y. Assessing the potential impacts of urban expansion on regional carbon storage by linking the LUSD-urban and InVEST models. Environ Modelling Softw. 2016;75:44–58.
- 11. Wang C, Luo J, Tang H. Analysis on the driving force of spatial and temporal differentiation of carbon storage in the Taihang mountains based on InVEST model. Ecol Environ Sci. 2023; 32, 215–225.
- 12. Xu H, Song Y, Tian Y. Simulation of land-use pattern evolution in hilly mountainous areas of North China: a case study in Jincheng. Land Use Policy. 2022;112:105826.
- 13. Mansour S, Al-Belushi M, Al-Awadhi T. Monitoring land use and land cover changes in the mountainous cities of Oman using GIS and CA-Markov modelling techniques. Land Use Policy. 2020;91:104414.
- 14. Cunha ER da, Santos CAG, Silva RM da, Bacani VM, Pott A. Future scenarios based on a CA-Markov land use and land cover simulation model for a tropical humid basin in the Cerrado/Atlantic forest ecotone of Brazil. Land Use Policy. 2021;101:105141.
- 15. Liang Y, Hu H, Crowther TW, Jörgensen RG, Liang C, Chen J, et al. Global decline in microbial-derived carbon stocks with climate warming and its future projections. Natl Sci Rev. 2024;11(11):nwae330. pmid:39440268
- 16. Zhu K, He J, Tian X, Hou P, Wu L, Guan D, et al. Analysis of evolving carbon stock trends and influencing factors in chongqing under future scenarios. Land. 2024;13(4):421.
- 17. Liu X, Liang X, Li X, Xu X, Ou J, Chen Y, et al. A future land use simulation model (FLUS) for simulating multiple land use scenarios by coupling human and natural effects. Landsc Urban Plan. 2017;168:94–116.
- 18. Xu Q, Guo P, Jin M, Qi J. Multi-scenario landscape ecological risk assessment based on Markov–FLUS composite model. Geomat Nat Hazards Risk. 2021;12(1):1449–66.
- 19. Fu K-X, Jia G-D, Yu X-X, Chen L-X. Analysis of temporal and spatial carbon stock changes and driving mechanism in Xinjiang Region by coupled PLUS-InVEST-Geodector model. Huan Jing Ke Xue. 2024;45(9):5416–30. pmid:39323159
- 20. Kuang W, Zhang S, Du G, Yan C, Wu S, Li R, et al. Monitoring periodically national land use changes and analyzing their spatiotemporal patterns in China during 2015–2020. J Geogr Sci. 2022;32(9):1705–23.
- 21. Liu J, Kuang W, Zhang Z, Xu X, Qin Y, Ning J, et al. Spatiotemporal characteristics, patterns, and causes of land-use changes in China since the late 1980s. J Geogr Sci. 2014;24(2):195–210.
- 22. Mamitimin Y, Simayi Z, Mamat A, Maimaiti B, Ma Y. FLUS based modeling of the urban LULC in Arid and Semi-Arid Region of Northwest China: a case study of Urumqi City. Sustainability. 2023;15(6):4912.
- 23. Zhu K, Cheng Y, Zang W, Zhou Q, El Archi Y, Mousazadeh H, et al. Multiscenario simulation of land-use change in Hubei Province, China based on the Markov-FLUS model. Land. 2023;12(4):744.
- 24. Chu M, Lu J, Sun D. Influence of urban agglomeration expansion on fragmentation of green space: a case study of Beijing-Tianjin-Hebei Urban agglomeration. Land. 2022;11(2):275.
- 25. Xiang H, Ma Y, Zhang R, Chen H, Yang Q. Spatio-temporal evolution and future simulation of agricultural land use in Xiangxi, Central China. Land. 2022;11(4):587.
- 26. Wei Y, Wang H, Xue M, Yin Y, Qian T, Yu F. Spatial and temporal evolution of land use and the response of Habitat Quality in Wusu, China. Int J Environ Res Public Health. 2022;20(1):361. pmid:36612683
- 27. Hu N, Xu D, Zou N, Fan S, Wang P, Li Y. Multi-scenario simulations of land use and habitat quality based on a PLUS-InVEST model: a case study of Baoding, China. Sustainability. 2022;15(1):557.
- 28. Valencia JB, Guryanov VV, Mesa-Diez J, Tapasco J, Gusarov AV. Assessing the effectiveness of the use of the InVEST annual water yield model for the rivers of Colombia: a case study of the Meta River Basin. Water. 2023;15(8):1617.
- 29. Wang Z, Li X, Mao Y, Li L, Wang X, Lin Q. Dynamic simulation of land use change and assessment of carbon storage based on climate change scenarios at the city level: a case study of Bortala, China. Ecol Indic. 2022;134:108499.
- 30. Shi S, Wang X, Hu Z, Zhao X, Zhang S, Hou M, et al. Geographic detector-based quantitative assessment enhances attribution analysis of climate and topography factors to vegetation variation for spatial heterogeneity and coupling. Glob Ecol Conserv. 2023;42:e02398.
- 31. Zhang X, Ma X, Wang X. Using the geographic detector model to identify factors controlling the bioavailability of Sr Isotopes in China. Front Earth Sci. 2023;10.
- 32. Liu Y, Wan C, Xu G, Chen L, Yang C. Exploring the relationship and influencing factors of cultivated land multifunction in China from the perspective of trade-off/synergy. Ecol Indic. 2023;149:110171.
- 33. Shi M, Wu H, Jiang P, Zheng K, Liu Z, Dong T, et al. Food-water-land-ecosystem nexus in typical Chinese dryland under different future scenarios. Sci Total Environ. 2023;880:163183. pmid:37030378
- 34. Kulaixi Z, Chen Y, Li Y, Wang C. Dynamic evolution and scenario simulation of ecosystem services under the impact of land-use change in an Arid Inland River Basin in Xinjiang, China. Remote Sens. 2023;15(9):2476.
- 35. Zhao X, Wang P, Gao S, Yasir M, Islam QU. Combining LSTM and PLUS models to predict future urban land use and land cover change: a case in Dongying City, China. Remote Sens. 2023;15(9):2370.