Spatial and temporal evolution of carbon stocks in Yulin City under changing environments

Abstract

Land use changes directly or indirectly affect the regional carbon balance. Investigating the spatial and temporal evolution of regional carbon stock and the contribution of land use driving factors is crucial for understanding the formation mechanisms of ecosystem carbon cycles and carbon budget balance.In this study, the researchers selected the model simulation method after comparing various carbon stock estimation methods. This paper uses the InVEST model to calculate the carbon stock of Yulin City from 2000 to 2020. It then applies the PLUS model to predict the land use of Yulin City under different scenarios in 2030 and forecasts the future carbon stock, providing a theoretical basis for the city’s future development planning. The main findings of this study are as follows: (1) The land use transfer in Yulin City, Guangxi, from 2000 to 2020, is mainly among forest land, arable land, and construction land. The net transfer out of cultivated land is 476.9982 km², and the net transfer in forest land and construction land is 245.5803 km² and 231.0048 km², respectively. (2) The study area’s carbon stock distribution closely follows the regional elevation pattern. The high-carbon stock areas concentrate in the mountainous and hilly regions of Yulin City at higher elevations. Medium-value areas lie in the study area’s relatively flat central and southern parts. In contrast, low-value areas are located in the reservoirs and rivers within Yulin City. (3) Compared with 2000, the carbon stock of the regional ecosystem in 2020 increased by 2.16 × 10⁶ t. Compared with 2020, the carbon stock of the regional ecosystem in 2030 increased under NDS and EPS by 0.3214 × 10⁶ t and 0.3286 × 10⁶ t, respectively, and decreased under CPS by 2.1524 × 10⁶ t. Overall, the carbon stock in 2030 increased by 0.3214 × 10⁶ t compared with 2020 and decreased by 2.1524 × 10⁶ t compared with 2020. Overall, the carbon stock in 2030 is expected to increase compared to 2020, but the growth rate is declining. The trend of increasing carbon stock in Yulin City is likely to continue decreasing in the future, and may even show a reduction.

Introduction

Global warming caused by greenhouse gas emissions such as carbon dioxide has resulted in problems such as urban flooding and extreme weather that jeopardizes human survival, and human beings are paying more and more attention to a series of carbon issues such as carbon emissions. Against this background, the goal of “carbon neutrality” has been proposed as a representative of sustainable development, i.e., the pursuit of net-zero carbon emissions and the decoupling of economic growth and resource consumption¹. China has even set the goal of peaking its carbon emissions by 2030 and reaching “carbon neutrality” by 2060². Studies show that human activity is closely linked to global climate activity. Soil is an essential component of terrestrial ecosystems and is a central vehicle for ecosystem material cycling, energy flow, and information transfer⁵. Land use/cover change has become the next most significant source of carbon emissions after energy consumption⁶. Land use and land cover change is the result of the interaction between human production, life, and natural environment community replacement, and land cover change affects the composition of ecosystem components, which will have a far-reaching impact on carbon sequestration rates and carbon stocks, thus changing the carbon source or sink patterns of terrestrial ecosystems⁷.

Currently, the methods for estimating terrestrial ecosystem carbon stocks mainly include field measurement, model analysis, and remote sensing inversion^7–10. The field survey method requires the collection of soil samples for physical and chemical tests from selected sample areas within the study area or calculating carbon stocks from vegetation and soil inventory data. This method is labor-intensive and time-consuming, unsuitable for large-scale environmental studies, and has some limitations in reflecting the spatial and temporal evolution of carbon stocks. The traditional mathematical and statistical methods ignore the spatial variability and complexity of the climate, which leads to regional environmental variability¹¹. Researchers commonly use remote sensing inversion methods to study aboveground and soil biomass in specific ecosystems, such as forests. They commonly use models like CASA, FORCCHN, LPJ-GUESS, and DNDC, which face challenges such as complex data acquisition and poor applicability. The model analysis method has become the primary means of quantifying carbon stocks because it can effectively assess and predict carbon stocks at different scales and achieve spatial visualization and expression of the assessment results. Therefore, many scholars have adopted modeling approaches for regional carbon stock assessment, among which CA-Markov, CLUE-S, and GeoSOS-FLUS, combined with InVEST models, have been widely used in the prediction of land use and carbon stock changes at different spatial and temporal scales because of their explanatory solid ability and superiority at spatial and temporal scales^12–14.

Many scholars have studied the spatial and temporal evolution of ecosystem carbon stocks and their influencing factors from various scales and perspectives. For example, they have shown that soil is a vast potential carbon reservoir, storing about 2500 Pg of carbon in the soil loop. Soil organic carbon (SOC) is a vital component of soil. It provides essential nutrients for plant growth and energy for soil microorganisms. SOC is also one of the most important indicators used to evaluate soil quality^16,17. The SOC stock accounts for 60% of the global soil carbon pool, which is about twice the amount of carbon in the atmosphere and three times that in vegetation^18,19.Raj Kumar²⁰ assessed and compared tree species’ biomass, carbon stock, and climate adaptation potential in degraded and afforested valleys in western India. The results indicated that afforesting degraded gully areas has a strong potential to accumulate biomass and carbon sinks in both vegetation and soil systems. This makes these areas a significant sink for sequestering atmospheric carbon and accelerating national efforts for mitigation and adaptation.Yue et al.²¹ simulated future land use data for Anhui Province under different SSP scenarios from 2030 to 2070, based on 2010 land use data. They calculated the spatial and temporal changes in carbon stock using the coupled PLUS-InVEST model. The study found that the spatial distribution of carbon stock in Anhui Province under the three SSP scenarios shows a general pattern of high carbon stock in the north and south, and low carbon stock in the center. The changes in carbon stock are closely related to the areas of cultivated land, forested land, and grassland.

Previous studies using the InVEST model to analyze carbon stocks have shown that, with the rapid growth of urbanization, the expansion of arable land into mountainous areas, and the increasing deforestation, the overall trend of carbon stocks and density is declining^22,23. However, in recent years, the government has placed more emphasis on the ecological environment. Relevant policies have been introduced to protect the environment, which has slowed the decline in carbon stock and even led to increases in some areas.To explore the impact of spatial planning policies on carbon stock, many experts and scholars have conducted simulations of land use under multiple development scenarios. Most studies show that the ecological protection scenario offers the greatest benefit for carbon stock enhancement, significantly outperforming other scenarios like cropland protection or economic development^24–26.Carbon stocks exhibit spatial variability, which researchers attribute to various factors. Many scholars have found that areas adjacent to high-carbon stock regions tend to have higher carbon values, while areas next to low-carbon stock regions generally have lower values²⁷. High-value regions are mainly found in mountainous, forested areas at higher elevations. However, when the elevation exceeds 4,000 m, unfavorable climatic conditions limit biodiversity growth, leading to a decrease in carbon stocks as elevation increases^28,29. Existing studies indicate that geographic factors, such as topography and slope, as well as climatic factors, such as average annual rainfall, are usually the main drivers of changes in carbon stocks³⁰. Economic development can also contribute to an increase in carbon stocks in some cases³¹.

Yulin City is a pilot zone for industrial integration and development in Guangxi’s “two gulfs” region. It serves as an important channel and key node city that links the east and west, connecting the Guangdong-Hong Kong-Macao Greater Bay Area and the Beibu Gulf Economic Zone. Located at the intersection of the southern China and the Great Southwest China economic circles, Yulin City plays a crucial role in bridging the economic development of these regions.This paper predicts and analyzes carbon stock changes in Yulin City by 2030 under different development scenarios. The analysis is based on land use change data from 2000 to 2020, combined with the PLUS-InVEST model. Kernel density and landscape analysis are used to study the spatial aggregation of carbon stock in Yulin City, further exploring how carbon stocks respond to land use changes under various future scenarios.The results show that while the carbon stock in Yulin City is increasing year by year, the growth rate is slowing down. This indicates that Yulin City should focus more on urban greening and land management in its future development. The findings of this study are significant for understanding the regional ecosystem carbon cycle and its formation mechanisms. They also provide valuable decision-making support for developing regional ecological resource protection policies for arable and forest land, reducing soil erosion, and rationally utilizing land resources. Additionally, the results offer theoretical support for short- and long-term ecological carbon sequestration goals and the formulation of regional energy conservation and emission reduction policies.

However, very few studies have considered the impact of spatial planning policies on land use and carbon stock changes. In addition, most of the previous prediction models have yet to explore the driving factors of land use in-depth and have yet to seriously examine the contribution of the driving factors to each land use type. This study, based on the Yulin City government’s future planning, set different development scenarios for the city and explored the contribution of driving factors to each land use type.

Materials and methods

Study area

Yulin City (E109° 32′–110° 53′, N21° 38′–23° 7′) in the Guangxi Zhuang Autonomous Region is located in southwestern China (Fig. 1), with a land area of 12,800 km².

Fig. 1.

Study area overview.

Yulin City lies in a low-latitude area, with the continent to the north and the tropical ocean to the south. It falls within the southern subtropical monsoon climate zone. Yulin City is located in a low-latitude area, with the continent to the north and the tropical ocean to the south. It falls within the southern subtropical monsoon climate zone. The city’s terrain is characterized by mountains, which slope from the center to the north and south. Rolling hills dominate the central region, connecting a section of the north-south watershed. The landforms in Yulin City belong to the southeast Guangxi hilly region, a part of the broader national landform classification.Yulin City is rich in both mineral and forest resources. The primary mineral resources include molybdenum, bismuth, tungsten, silver, mercury, lead, niobium, tantalum, and sulfurous iron ore, as well as non-metallic minerals like cement water chert and fluorite. In total, the region contains 44 types of minerals, accounting for 26.19% of Guangxi’s mineral resources.Additionally, Yulin City is home to two state-owned forest farms directly under the Autonomous Region: Bobai Forest Farm and Liuwan Forest Farm. The city also has one municipal forest farm and nine county-level forest farms.

Data sources

Meteorological data

In this study, we selected a dataset provided by the Tibet Multi-sphere Data Assimilation and Simulation Center of the Institute of Tibetan Plateau Studies, Chinese Academy of Sciences. This dataset includes temperature, precipitation, and solar radiation data with a spatial resolution of 0.1°. We used ArcGIS software to convert the data to the WGS 1984 UTM 49 N projected coordinate system and adjusted the spatial resolution to 1 km through a projective transformation. Researchers then used this data as one of the driving factors for the PLUS model.

Land use remote sensing data sets

The research team selected land use data from the Global 30 m Fine Ground Cover Dynamic Monitoring Product 1985–2020 (GLC_FCS30-1985_2020), released by Liu Liangyun’s team at the Institute of Space and Astronautical Information Innovation, Chinese Academy of Sciences (CAS). This study used 2000, 2005, 2010, 2015, and 2020 data. To meet the study’s requirements, the team reclassified and merged the secondary classifications of land use data into six primary categories: cropland, forest land, grassland, waterland, construction land, and wetlands.

Carbon density data

In this paper, to ensure the accuracy of the selected carbon density data, the carbon density of each class in Yulin City is summarized regarding the results obtained by the predecessors of carbon stock calculation in the Guangxi region^32–36, as shown in Table 1:

Table 1.

Carbon densities of each part of land use in Yulin.

Land use type	Aboveground carbon density Cabove	Subsurface carbon density Cbelow	Soil carbon density Csoil	Carbon density of dead organic matter Cdead	Total carbon density Ctotal
Cropland	13.65	2.62	47.40	1.00	64.67
Grassland	3.01	13.53	60.00	1.00	77.54
Forest land	58.30	14.58	96.00	3.50	172.38
Wetland	37.00	11.84	55.50	3.00	107.34
Construction land	11.45	0.93	31.40	0.00	43.78
Water land	2.80	2.40	0.00	0.00	5.20

Driver data

In this paper, the authors explored the driving factors affecting land use expansion in each category. Based on existing research results and the specific conditions of the study area^21,35,37, they selected 14 driving factors from socio-economic and natural characteristics, with the indicators shown in Table 2. The team obtained soil type, population density, and GDP data from the China Resource and Environment Science Data Center, which has a spatial resolution of 1 km. They also obtained DEM data from the Geospatial Data Cloud, which has a spatial resolution of 30 m.

Table 2.

Indicators of drivers of arable land change in Yulin.

Type	Data type	Descriptive	Unit
Socio-economic factors	Demographic GDP	Population density by grid GDP values for each raster	Persons/hectare Million yuan/kilometer
	Distance to county government, distance to the railroad, distance to highway Distance to primary roads, secondary roads, tertiary roads	Distance from the center of each grid to a location Distance from the center of each grid to a location	m m
Climate and Environmental Factors	Soil type	Soil type represented by each grid	–
	Distance from water	Distance from the center of each grid to the water system	m
	Average annual temperature	Average multi-year temperatures represented by each raster	degrees centigrade
	Annual precipitation	Multi-year average precipitation represented by each raster	millimetre
	DEM/slope	Elevation values/slope values for each grid	m/degree

Methods

Land use change dynamics

In this paper, the dynamic land use attitude reflects the change of land use types over a certain period. The attitude is a single, comprehensive, dynamic attitude³².

1. Individual land use dynamics, often expressed as K, represent the change in land-use types within a specific time frame in a study area. The greater the value of K, the larger the magnitude of change in the land use type during the study period, indicating a more active change. Its expression is:

   1

   where U_a and U_b represent the area of the pre and post-land use types, respectively, and T represents the time between the two periods.

2. Comprehensive land use dynamics, which analyzes a study area’s comprehensive land use dynamics, refers to the annual land use change rate in the study area. People often express its value as Lc:

   2

   where L_Ui−j is the absolute value of land use type I converted to non-I land use type, _Lui is the area of land use type in the previous period, and T represents the time between the two periods.

Land use transfer matrix

Many scholars use the land use transfer matrix to analyze the transfer between land use types³⁸, which is used in this study to analyze the transfer in and out between land use types. The land use transfer matrix is:

3

where S is the area; n is the number of land use types before and after the transfer; I and j are the land use types during the study period, respectively; and S_ij is the area of land type I in the early part of the study period converted to land type j in the late part of the study period.

Kernel density analysis

Kernel density analysis is a nonparametric method of estimating a probability density function that shows that the closer a thing is to a core element, the greater the value of density expansion acquired by the location. The larger the value, the higher the degree of aggregation of a particular place class. Considering the performance of the equipment, this paper first resamples the land category data with a resolution of 30 to 300 m raster data using ArcGIS10.6 software. The team processes the data into point data and performs kernel density analysis. The calculation is summarized as follows:

4

where f_n is the estimated kernel density of land classes in the study area; K(x) is the kernel function; h is the bandwidth; n is the number of land class point data in the broadband range; and x–x_i is the distance from the cultivated point x to the sample point x_i.

• (2)Fragmentation quantitative metrics.

The team used the landscape pattern index method to portray changes in land patterns and cropland fragmentation. In this paper, regarding previous research results³⁹, eight landscape pattern indices (Table 3) were selected to characterize the cropland patterns and fragmentation changes in Yulin City.

Table 3.

Selected landscape in dice and related denotes.

Landscape index	Designation	Unit	Range of values	Instruction
N.P.	Number of plaques	Pieces	[1,+∞]	Indicates the number of plaques
P.D.	Plaque density	Pieces/km2	[0,+∞]	Reflects degree of patch fragmentation
LSI	Landscape shape index	None	[1,+∞]	Reflects the complexity of the plaque shape
Division	Landscape segmentation index	%	(0,100]	Reflecting the degree of patch fragmentation
A.I.	Aggregation index (math.)	%	[0,100]	Reflects aggregation between patches
Clumpy	Clustering index (math.)	none	[-1,1]	Reflects the degree of aggregation between patches
Split	Separation index	%	(0,100]	Reflects the degree of separation between patches
Cohesion	Plaque cohesion index	%	(0,100]	Reflects overall aggregation of plaques

PLUS model

The PLUS model is based on meta-cellular automata. During the simulation, the LEAS and CARS modules collaborate to generate development probabilities and simulate land patch generation based on these probabilities⁴⁰. The PLUS model identifies the driving factors and their contributions to the expansion or contraction of specific land categories. Researchers used the “Extract Land Expansion” module of the PLUS model⁴¹ to simulate land use changes at the patch scale. They then analyzed the contribution of driving factors using the Random Forest classification algorithm.

In this study, researchers first tested the accuracy of the PLUS model. They simulated the spatial distribution of land use in Yulin City for 2020 using the CARS module of the PLUS model, which is a C.A. model based on multi-class stochastic patch seeding. The simulation was based on Yulin City’s 2010 land-use data and the land-expansion probability for each land-use type derived from the 14 selected drivers. To evaluate the accuracy, they compared the simulated 2020 land use with the actual 2020 land use data for Yulin City using Kappa coefficients^42,43. The results showed that the Kappa coefficient was 0.899107, with an overall accuracy of 0.94744 (Fig. 2). These results indicate that the PLUS model is suitable for simulating land use in Yulin City and is both feasible and reliable to a certain extent.

Fig. 2.

Actual land use in Yulin City in 2020 PLUS model simulation of land use in 2020.

The PLUS model primarily consists of the LEAS module and the CARS module. In this paper, researchers determine the development probability of each land use type in Yulin City by using land use data from 2010 to 2020, along with the operation of the LEAS module of the PLUS model. They then input the 2020 land use type data and the development probabilities for each land use type into the CARS module to make land use projections for 2030 under different development scenarios. First, the domain weights for each class are set. The domain weight W_i represents the conversion difficulty coefficient for different land use types, with values ranging from [0,1]. This coefficient can be determined using the following formula:

5

In the formula, S_i denotes the expansion area of a land category from 2010 to 2020, while S_max and S_min denote the locations of the largest and smallest expansion areas of land categories from 2010 to 2020, respectively. In this paper, researchers use Eq. (5) to determine the domain coefficients for each class in the 2030 NDS. They reference the domain weights for the remaining scenario classes based on relevant literature^37,44,42 and finalize the values after iterative debugging in the PLUS model. Next, they set the transfer matrix, which indicates the difficulty of mutual conversion between land classes. The values are generally 0 or 1, with zero indicating that converting between two land classes is complex and one suggesting that conversion is easy. Tables 4 and 5 show the domain weights and land use transfer matrix parameters used in this paper to simulate Yulin City’s land use under multiple scenarios in 2030.

Table 4.

PLUS model simulates the domain weights of each category under each development scenario in 2030.

Development scenario	Cropland	Grassland	Forest land	Wetland	Construction land	Waterland
NDS	0.7022	0.0009	1.0000	0.0000	0.5149	0.0226
CPS	1.0000	0.5000	0.5000	0.0000	0.0000	0.0000
EPS	0.3000	1.0000	1.0000	0.3000	0.3000	0.3000

Table 5.

PLUS model simulation of the transfer matrix by category for different development scenarios in 2030.

Development scenario	Soil type	Cropland	Grassland	Forestland	Wetland	Construction land	Waterland
NDS	Cropland	1	1	1	1	1	1
	Grassland	1	1	1	1	1	1
	Forestland	1	1	1	1	1	1
	Wetland	1	1	1	1	1	1
	Construction land	0	0	0	0	1	0
	Waterland	1	1	1	1	1	1
CPS	Cropland	1	0	0	0	0	0
	Grassland	1	1	1	1	1	1
	Forestland	1	1	1	1	1	1
	Wetland	1	1	1	1	1	1
	Construction land	1	1	1	1	1	1
	Waterland	1	1	1	1	1	1
EPS	Cropland	1	1	1	1	1	1
	Grassland	0	1	0	0	0	0
	Forestland	0	0	1	0	0	0
	Wetland	1	1	1	1	1	1
	Construction land	1	1	1	1	1	1
	Waterland	1	1	1	1	1	1

InVEST model

The carbon stock module of the InVEST model uses land use types as the units of measurement to assess carbon stocks⁴⁵. The carbon density data include four primary carbon pools for each land use type: aboveground biomass, belowground biomass, soil organic matter, and dead organic matter. By applying land use classification, researchers calculated the average carbon densities of these four carbon pools for different land categories to obtain the total carbon stock in the study area. In this study, the InVEST model was used to analyze the carbon stock in Yulin City for 2030 and to predict its spatial distribution based on the simulated land use. The specific calculation formula is as follows:

6

where C_total is the carbon stock of the region in t, A_i is the area of each land use type i in region x; C_{i, above}, C_{i, blow}, C_{i, soil}, C_{i, dead} are the carbon density of aboveground biomass, carbon density of belowground biomass, carbon density of soil organic matter, and carbon density of dead organic matter of land use type i in t/hm², respectively.

Multi-scene setting

Various factors affect future urban development and land use changes in Yulin City. Therefore, researchers must consider different environmental factors when predicting land use changes^46,47. Based on optimization policies and relevant literature from the “Yulin Statistical Yearbook,” this study, considering the actual situation of Yulin City, formulates conversion rules and conversion rates for the increase and decrease of land categories (Tables 4 and 5). Since 2004, Yulin City has strictly implemented measures for farmland protection. These policies primarily focus on limiting the conversion of agricultural land to non-agricultural uses, as well as strengthening farmland reclamation and restoration. Starting in 2013, with the national emphasis on ecological environment protection, Yulin City introduced a series of local ecological protection policies, with a focus on enhancing the protection of forests, wetlands, and other ecological environments. These policies include the designation of ecological function zones and the implementation of ecological compensation mechanisms.The study sets three scenarios^48,49: the natural development scenario, the cropland protection scenario, and the ecological protection scenario. The following describes each scenario:

1. Natural development scenarios (NDS): This scenario follows the law of land use structure development in Yulin City from 2000 to 2020 and sets the scenario to be brought into the model by the domain factors of each category in Yulin City from 2010 to 2020, and sets the conversion of construction land to other land use types in the land use transfer matrix as 0, and all values in the rest of the land use transfer matrix as one by default; and takes the watersheds as a limitation of the land use change Factor.

2. Cropland Protection Scenario (CPS): This scenario follows cropland protection policies, which restrict cropland conversion in the area to other land types.

3. Ecological Protection Scenario (EPS): This scenario ensures the sustainable development of ecologically functional land, such as woodland and grassland, and restricts the conversion of forest land and grassland in the region to other land types in the simulation.

Results

Characterization of land use status and change

From 2000 to 2020, the main land use types in Yulin City included forest land, cropland, and construction land, while grassland, waters, and wetlands made up a relatively small percentage (Fig. 3). In terms of spatial distribution, forest land was mostly found in high-altitude mountainous hills, away from urban areas. Construction land was primarily concentrated in towns and cities within counties and districts, with Yuzhou District having the largest coverage. Cropland was mainly located in suburban areas and flat terrain, typically surrounding construction land.

Fig. 3.

Land use distribution map of Yulin City from 2000 to 2020. (a) 2000; (b) 2005; (c) 2010; (d) 2015; (e) 2020.

In the past 20a, the area of cropland and water bodies decreased, while the area of grassland, forest land, wetland, and constructed land increased (Table 6). The cropland area has reduced the most, with a reduction of 476.9964 km². In contrast, forest land and construction land areas have increased significantly. Forest land has risen by 245.5911 km², and construction land has increased by 231.0048 km².

Table 6.

Changes in the area of various land in different periods.

Land use type	Area of change(km2)
	2000–2005a	2005–2010a	2010–2015a	2015–2020a	2000–2020a
Cropland	− 131.5062	− 150.5547	− 90.9954	− 103.9401	− 476.9964
Grassland	0.0684	0.1908	0.1089	0.1863	0.5544
Forestland	81.4473	98.3547	19.1466	46.6425	245.5911
Wetland	0.0279	0.0126	0.0279	0.0432	0.1116
Construction land	46.6011	54.5589	72.2484	57.5964	231.0048
WaterLand	3.3813	− 2.5641	− 0.5373	− 0.5337	− 0.2538

Within 2000–2020, taking five years as a period, it can be seen that cropland had the most significant decrease in 2005–2010, with a reduction of 150.5547 km²; forest land had the most significant increase in 2005–2010, with a rise in 98.3547 km²; construction land showed an increasing trend, with the most significant increase in 2010–2015, and the net transfer area of construction land was 72.2448 km². The net transfer area of land use is 72.2448 km². From 2000 to 2010, people had the highest attitude toward the integrated movement of land use (Fig. 4) and the lowest degree of land use integration from 2010 to 2015. The area of cropland and water bodies continued to decrease, with the most significant decrease in cropland mainly converted into forest land and construction land (Tables 7, 8).

Fig. 4.

Single and combined dynamics of land use change by type from 2000 to 2020.

Table 8.

Results of the Yulin City land use transfer matrix from 2000 to 2020.

Type of land use in 2000	Type of land use in 2000 (Unit: km2)						Outflow
	Cropland	Grassland	Forestland	Wetland	Construction land	Water land
Construction land	47.4102	0.0000	2.7594	0.0054	0.0000	0.3384	50.5134
Water land	13.7133	0.0207	10.5318	0.0216	1.1277	0.0000	25.4151
Inflow	626.4504	1.1970	828.8640	0.1665	281.5182	25.1622	–

Table 7.

Results of the Yulin City land use transfer matrix from 2000 to 2020.

Type of land use in 2000	Type of land use in 2000(Unit: km2)						Outflow
	Cropland	Grassland	Forestland	Wetland	Construction land	WaterLand
Cropland	0.0000	0.5985	815.3397	0.1224	264.8313	22.5567	1103.4486
Grassland	0.4032	0.0000	0.2331	0.0000	0.0009	0.0054	0.6426
Forestland	564.8985	0.5778	0.0000	0.0171	15.5520	2.2383	583.2837
Wetland	0.0252	0.0000	0.0000	0.0000	0.0063	0.0234	0.0549

By analyzing the kernel density distribution of cropland, forest land, and construction land in Yulin City for the years 2000, 2010, and 2020 (Fig. 5), researchers identified several trends. The area of very high kernel density for cropland gradually expanded. With urbanization, the kernel density of cropland in the central part of Yuzhou District steadily declined.The kernel density of forest land followed a “low in the middle, high in the surroundings” pattern. High-density areas were mostly found in mountainous and hilly regions at higher elevations. The kernel density of construction land showed a “high in the middle, low in the surroundings” pattern. Over time, urban areas in various counties and districts saw a gradual increase in kernel density.Cropland patches in Yulin City are becoming more dispersed (Fig. 6). However, their fragmentation shape remains simple. Spatial aggregation is increasing, and fragmentation is becoming more noticeable. Forest land area is growing, with patches becoming more dispersed and their shape stabilizing. Spatial aggregation remains relatively high. Construction land area is expanding. While patch fragmentation is becoming more complex, the degree of aggregation for construction land patches remains higher.

Fig. 5.

Spatial distribution of nuclear density levels of arable land, forest land, and construction land in Yulin during 2000–2020.

Fig. 6.

Fragmentation index of arable land, forest land, and construction land in Yulin during 2000–2020)

Land use drivers

Figure 7 shows the analysis of the contribution strength of the PLUS model to the expansion of six types of land in Yulin City, based on 14 driving factors. The results indicate that the main driving factors for cropland expansion are population density, slope, and distance from railroads. For forest land expansion, the primary driving factors are elevation, distance from a level 2 road, and slope. The main driving factors for construction land expansion, in order of importance, are distance from a level 2 road, elevation, and distance from a level 1 road. Each of these three factors contributes relatively equally. For grassland expansion, the primary driving factors are elevation, distance from water areas, and average annual precipitation, with each factor contributing similarly. Elevation is the most significant driving factor for the expansion of wetland and water areas, contributing far more than the other factors. This highlights the crucial role of elevation in expanding wetlands and water areas.

Fig. 7.

Drivers and Contribution of Site Expansion from 2000 to 2020. (a) Cropland; (b) Grassland; (c) Woodland; (d) Wetland; (e) Construction land; (f) Water areas.

Figure 8 shows the results of overlaying cropland, forested land, and construction land with their respective primary contributing drivers. The results reveal that cropland expansion is denser in the eastern part of Bobai County and the southern part of Luchuan County. In areas where population density is either too high or too low, cropland expansion tends to be more dispersed.The terrain in the eastern part of Bobai County and the southern part of Luchuan County is flat with a slight slope. This suggests that the distribution of cropland is influenced by elevation and slope. The reduction in cropland area is caused by the combined effects of multiple factors. Forest land expansion is limited in areas with higher elevation values. The most significant expansion of forest land occurs near lower elevation areas, particularly in the southern part of Fumian District, the south part of Beiliu City, Luchuan County, and the eastern part of Bobai County. Forest land expansion appears to be spreading gradually from mountainous hills to lower elevations.Construction land expansion is concentrated in the center of Yuzhou District and surrounding areas of Beiliu City and Fumian District. Other expansion areas are sporadically distributed in towns and cities across various counties. Yuzhou District has denser roads, and its urbanization and development levels are higher than those of other counties. As a result, construction land expansion is more prominent in Yuzhou District than in the surrounding areas.In conclusion, the increase in cropland area is the main driver of the expansion of both forest land and construction land.

Fig. 8.

Results of overlaying the areas of increased arable land, forest land, and construction land with their highest contribution factors in Yulin. (a) Cropland; (b) Forest land; (c) Construction land.

Characterization of Spatial and Temporal variations in carbon stocks

In this study, the InVEST model was used to analyze the temporal and spatial changes in Yulin from 2000 to 2020 and to predict the temporal and spatial changes in Yulin in 2030 under different scenarios. In terms of time, the carbon stocks in 2000, 2005, 2010, 2015, and 2020 were 159.83 × 10⁶ t, 160.59 × 10⁶ t, 161.55 × 10⁶ t, 161.61 × 10⁶ t, and 161.99 × 10⁶ t in that order, and the overall carbon stocks of the Yulin ecosystem increased year by year from 2000 to 2020, with a 2.16 × 10⁶ t during 20a, a cumulative increase of 1.35%. Overall, the carbon stock showed a slow increase, slowly increasing from large to small. The increase in carbon stock was smallest from 2010 to 2015, with only 0.059 × 10⁶ t of carbon stock added during those 5 years. After 2010, urbanization accelerated, but the carbon stock still maintained a steady rate of increase. This suggests that the region can reasonably control the speed of urbanization. The increase in carbon stock helped suppress the rise of CO₂ in the ecosystem, positively contributing to regional carbon neutralization.

Using ArcGIS software, researchers processed the carbon stocks in Yulin City for 2000, 2010, and 2020 with the natural breakpoint method and spatial overlay to obtain the spatial distribution and spatial changes of carbon stocks from 2000 to 2020 (Fig. 9). The spatial distribution of carbon stocks is grouped according to the carbon stocks per unit area as high-value area (5.20-43.22 t/hm²), medium-value area (43.22-106.82 t/hm²), and low-value area (106.82-172.38 t/hm²). In Yulin City, the land classes in the high-value area of carbon stock are mainly forest land, the middle-value area is mainly cropland, and the low-value area is mainly construction land. This paper uses eight landscape pattern indices to examine changes in land class patterns and the fragmentation of forest land, cropland, and construction land. The findings show that cropland patches in Yulin City are becoming more dispersed, although the fragmentation shape remains relatively simple. Spatial aggregation is increasing, and the fragmentation of cropland is becoming more pronounced.The area of forest land is expanding, with patches becoming more dispersed. The shape of the patches is gradually stabilizing, and spatial aggregation remains relatively high. The area of construction land is also increasing, and while the fragmentation shape is becoming more complex, the degree of patch aggregation remains higher.Carbon stock distribution in Yulin City shows spatial variability. From 2000 to 2020, the spatial pattern of carbon stocks remained largely similar, with high carbon stock areas scattered around the study area. This pattern aligns with regional elevation. High carbon stock values are concentrated in the higher elevation hills, while medium carbon stock values are found in the flatter areas in the middle and south of the study area. In the southern part of the study area, medium carbon stock values are near the Nanliujiang River, Jiuzhou River, and some reservoirs, while the northern part of the study area, particularly in Xingye County, also has medium carbon stock values.Low-carbon stock areas are found in reservoirs and rivers in Yulin City, such as Xiaojiang Reservoir and Huotou Reservoir in the southern part of the study area. In 2000, high and medium carbon stock areas were concentrated around the study area. By 2020, the high carbon stock areas were more densely distributed, with medium-value areas largely transforming into high-value areas. As urbanization progresses and human activities increase in the central part of the study area, low carbon stock areas gradually replace high carbon stock areas, while medium-value areas remain predominant.Overall, the distribution of carbon stocks in 2020 is more distinct than in 2000, with clearer boundaries between regions.

Fig. 9.

Classification of carbon storage levels and degree of carbon stock changes in Yulin. (a) 2000; (b) 2010; (c) 2020; (d) The extent of change for 2000–2020.

From Fig. 9d, researchers categorized the degree of spatial change in carbon stocks from 2000 to 2020 into five categories using the natural breakpoint method: significant increase, minor increase, basically unchanged, minor decrease, and considerable decrease. They observed that the carbon stock in the eastern part of Bobai County, the entire territory of Luchuan County, the southern part of Fumian District, and the southern part of Beiliu City intersperses significant increases and significant decreases. However, the overall area shows a dominance of substantial increase, with a considerable reduction occurring in Luchuan County and the junction with Bobai County, displaying an “S” distribution with elevation. Significantly increasing areas are also found in the northeast of Xingye County and the south of Bobai County. By comparing the land use distribution between 2000 and 2020, researchers found that cropland decreased in 2020, and much of it was converted into forest land with a higher carbon storage capacity, leading to a significant increase in carbon storage in Yulin City by 2020.

Characteristics of spatial and temporal land use changes in Yulin City under multi-scenario constraints

In this study, the PLUS model simulates the main land use types in Yulin City for 2030 under the NDP, CPS, and EPS scenarios (Fig. 10).

Fig. 10.

Distribution of land use in Yulin in 2030 under different scenarios.

Under the three development scenarios, the mainland types in Yulin City in 2030 are forest land and cropland. As shown in Table 9, the cropland, forest land, and construction land areas in the NDS are 4494.7431 km², 7556.5503 km², and 665.8092 km², respectively. Compared with the area of each category in 2020, the area of cropland, grassland, and wetland shrinks in this scenario, with cropland shrinking the most, at 178.5330 km², and the area of forested land and constructed land increasing compared with 2020, at 54.1422 km² and 124.4970 km², respectively. Compared with the 2010–2020 cropland, forested land, and constructed land amount of change, the decrease in the amount of cropland decreased, and the increase in the area of forest land and construction land decreased during the decade 2020–2030.

Table 9.

Area of each category in 2030 under the three scenarios and the amount of change compared to 2020.

Surface area /amount of change	Period	Cropland	Grassland	Forestland	Wetland	Construction land	Waterland
Surface area/km2	2020	4673.2761	2.5029	7502.4081	0.3843	541.3122	95.5827
	NDS	4494.7431	2.3733	7556.5503	0.2457	665.8092	95.7447
	CPS	5209.8561	0.3852	7248.1914	0.1368	262.5174	94.3794
	EPS	4494.9312	2.7972	7556.5503	0.2655	666.5427	94.3794
Surface area Amount of change/km2 (Compared to 2020)	2010–2020	− 194.9355	0.2952	65.7891	0.0711	129.8448	− 1.071
	NDS	− 178.5330	− 0.1296	54.1422	− 0.1386	124.4970	0.1620
	CPS	536.5800	− 2.1177	− 254.2167	− 0.2475	− 278.7948	− 1.2033
	EPS	− 178.3449	0.2943	54.1422	− 0.1188	125.2305	− 1.2033

In the CPS, except for the increase in the area of cropland, the area of the rest of the land use decreases, with the areas of cropland, forest land, and construction land being 5209.8561 km², 7248.1914 km², and 262.5174 km², respectively. Compared with the regions of each category in 2020, the scenario shows that the area of cropland increased by 536.5800 km², and the areas of forest land, construction land, grassland wetland, and water area decreased by 254.2167 km², 278.7948 km², 2.1177 km², 0.2475 km², and 1.2033 km² respectively.

In the EPS, the areas of cropland, forest land, and construction land are 4497.9312 km², 7556.5503 km², and 666.5427 km², respectively. Compared with the areas of each land category in 2020, the area of cropland in this scenario decreased by 178.3449 km², while on the contrary, the areas of forest land and construction land increased by 54.1422 km², 125.2305 km², respectively. The similar amount of change in the EPS and the three land categories of cropland, forest land, and building land in the NDS further suggests that the development of Yulin City in 2020–2030, according to the 2010–2020 pattern, is a development approach closer to the EPS.

In the NDS, spatial changes in various types of land occur based on inertia, with construction land increasing in a decentralized manner. In the CPS, cropland protection is prioritized, and the increase in cropland mainly results from the reduction of forest land and construction land. The expansion of cropland is concentrated, primarily occurring in forest land at lower elevations and in urban marginal construction land. In the EPS, forest land is mainly converted from cropland areas designated for return to the forest. Construction land also expands, mainly in economic development zones defined by policy, maximizing its contribution to economic growth. However, cropland continues to receive inadequate protection within this development context.

Distribution of future carbon stocks and characterization of their changes under multiple scenarios

Carbon stock estimates

We estimated carbon stocks for different development scenarios in 2030 using the InVEST model, and the results appear in Tables 10 and 11. In the NDS, the ecosystem carbon stock in Yulin City is 162.3130 × 10⁶ t. Compared with 2020, the carbon stock in this scenario increases by 0.3214 × 10⁶ t. Compared with the increase in carbon stock from 2010 to 2020, the decadal increase in carbon stock in this scenario decreases. In the CPS, the ecosystem carbon stock in Yulin was 159.8393 × 10⁶ t, and compared with the carbon stock in each category in 2020, the carbon stock under this scenario decreased by 2.1524 × 10⁶ t. In the EPS, the ecosystem carbon stock in Yulin was 162.3203 × 10⁶ t, and compared with the carbon stock in each category in 2020, it increased by 0.3286 × 10⁶ t. The results show that ecological protection measures can effectively promote carbon sequestration in Yulin.

Table 10.

 Carbon storage and changes by category in 2030 under different scenarios.

Carbon stock/Amount of change	Period	Cropland	Grassland	Forestland	Wetland	Construction land	Waterland
Carbon stock/106t	2020	30.2221	0.0194	129.3265	0.0041	2.3699	0.0497
	NDS	29.0675	0.0184	130.2598	0.0026	2.9149	0.0498
	CPS	33.6921	0.0030	124.9443	0.0015	1.1493	0.0491
	EPS	29.0687	0.0217	130.2598	0.0028	2.9181	0.0491

Table 11.

Carbon storage and changes by category in 2030 under different scenarios.

Carbon stock/Amount of change	Period	Cropland	Grassland	Forestland	Wetland	Construction land	Waterland
Carbon stock Amount of change/106t (Compared to 2020)	2010–2020	− 1.2606	0.0023	1.1341	0.0008	0.5685	−0.0006
	NDS	−1.1546	−0.0010	0.9333	−0.0015	0.5450	0.0001
	CPS	3.4701	−0.0164	−4.3822	−0.0027	−1.2206	−0.0006
	EPS	−1.1534	0.0023	0.9333	−0.0013	0.5483	−0.0006

The mechanisms by which land use type conversion responds to changes in carbon stocks vary across the three development scenarios (Table 10, 11). Under the NDS, the increase in carbon stock is mainly due to the conversion of cropland to forest land, which contributes 0.9333 × 10⁶ t. The conversion of cropland to construction land leads to the loss of 0.5450 × 106 t of carbon stock, which is generally increasing. Still, compared to the change in carbon stock from 2000 to 2020, the degree of change in carbon stock from 2020 to 2030 shows a significant decreasing trend in general. In the CPS, the conversion of forest land and construction land into cropland results in a loss of 4.3822 × 10⁶ t and 1.2206 × 10⁶ t, respectively, and the carbon stock of cropland increases by 3.4701 × 10⁶ t, leading to an overall decrease in carbon stock. In the EPS, the conversion of cropland to forest land and grassland increased forest cover and added 0.9356 × 10⁶ t of carbon stock. This increase surpassed the carbon stock loss caused by the cropland conversion to construction land by 0.6051 × 10⁶ t, resulting in an overall increase in carbon stock.

Projections of Spatial patterns of carbon stocks

In 2030, spatial variability in carbon stocks continues in Yulin City (Fig. 11). High-value carbon stock areas are found in the higher-elevation mountains and hills. Medium-value areas are mostly in the central part of the region, where urbanization is advancing quickly, and in the flat southern part. The southern portion of the study area, with rivers and some reservoirs, mainly contains low-value regions. Among the counties and districts, Bobai County has the highest carbon stock, while Yuzhou District has the lowest.

Fig. 11.

Classification of carbon storage levels and degree of carbon stock changes under different scenarios in 2030.

Under the 2030 NDS (Fig. 11d), the change in carbon stock from 2000 to 2030 generally shows a significant decreasing trend. In this scenario, cropland is converted to construction land and forest land, expanding the construction land area in the towns and cities of various counties and districts. Although forested land has increased, the expansion is smaller compared to the cropland converted to forest land between 2010 and 2020. The conversion of cropland with higher carbon storage capacity to construction land with lower carbon storage capacity is substantial. As a result, while the carbon stock has increased, the increase from the conversion of cropland to forest land is relatively small.

Under the CPS (Fig. 11e), the spatial carbon stock change from 2020 to 2030 is generally stable. Most areas of the city show sporadic regions with significant increases in carbon stock. The main increase occurs as construction land is converted into cropland, which leads to a considerable rise in carbon stock. The overall stability may be due to the reduction in forest land and grassland, both of which have higher carbon storage capacities. The decrease in carbon stock from the loss of these areas, along with the reduction in construction land, has offset the overall increase in carbon stock.

From Fig. 11f, under the EPS, the carbon stock change from 2020 to 2030 remains mostly stable, with a slight increase in areas near Yuzhou and Fumian districts, as well as Beiliu City and Yuzhou District. In this scenario, there is potential for expanding ecological lands, such as forest and grassland, which leads to an increase in carbon stocks. Developers still have room to expand construction land, while also promoting sustainable use of ecological lands. However, cropland has not been effectively protected. Among the three projection scenarios, the EPS offers the highest benefit for enhancing total carbon stocks.

Discussion

Impact of land use/cover change on carbon stocks in Yulin City

The primary driver of cropland expansion in Yulin is elevation, which has become more balanced since 2010. Most of the expansion occurs in high-elevation areas, such as northern Yulin, while cropland expansion in the central plains with lower elevation remains limited. This phenomenon can be attributed to the rapid urban development in the central region, where construction land has replaced cropland. As a result, cropland expansion has shifted to higher elevation areas. However, as developers utilize these higher elevation areas, they encounter greater challenges in converting forested land into cropland due to the increasing elevation. Consequently, the damage to carbon stocks from cropland expansion decreases as elevation rises.

Figure 7 shows that the main driving factors for expanding cropland are population density, slope, and distance from railroads. For expanding forest land, the key factors are elevation, distance from level 2 roads, and slope. The expansion of construction land is primarily driven by distance from level 2 roads and elevation, with a balanced contribution from the distance to level 1 roads. The main drivers of grassland expansion are elevation, distance from water, and average annual precipitation. The primary factor for expanding wetland and water areas is elevation, with its contribution being significantly higher than that of other factors. The high-value carbon stock areas are mainly composed of forest land, while the medium-value areas are mostly cropland, and the low-value areas consist primarily of construction land. The driving factors for high-value carbon stock areas include elevation, distance from level 2 roads, and slope. For medium-value carbon stock areas, population density, slope, and distance from railroads are the main factors. In low-value carbon stock areas, the driving factors, in order, are distance from level 2 roads, elevation, and distance from level 1 roads, with a relatively balanced contribution from all three factors.

Table 12 demonstrates the amount of carbon stock and carbon stock changes in each category from 2000 to 2020. The decrease in the area of cropland and waters resulted in a loss of about 3.08 × 10⁶ t of carbon stock, while the expansion of the area of grassland, forest land, wetland, and construction land increased the carbon stock by a total of 5.25 × 10⁶ t. The most significant amount of carbon sequestered was in forest land, and the carbon stock in forest land for the three periods of time were 125.09 × 10⁶ t, 128.19 × 10⁶ t, and 129.33 × 10⁶ t. In 2020, the carbon stock of forest land accounted for 79.84% of the total annual carbon stock, and the carbon stock of cropland accounted for 18.66% of the total yearly carbon stock. Although the carbon density of wetland and grassland is relatively high, the carbon stock of wetland and grassland has little influence on the total carbon stock because the area of wetland and grassland accounts for a relatively low percentage in the study area.

Table 12.

Carbon storage of each part of land use in Yulin unit: 10⁶t.

Certain year	Cropland	Grassland	Forestland	Wetland	Construction land	Waterland
2000	33.3068	0.0151	125.0930	0.0029	1.3585	0.0498
2010	31.4827	0.0171	128.1924	0.0034	1.8014	0.0503
2020	30.2221	0.0194	129.3265	0.0041	2.3699	0.0497
Net Variabie	-3.0847	0.0043	4.2335	0.0012	1.0113	-0.0001

In terms of land use change, forest land contributes the most to the total carbon stock in Yulin, followed by cropland. As a result, Yulin views both forest land and cropland as significant carbon sinks. From 2000 to 2020, land use change was the primary factor affecting the change in carbon stock. The transfer of land from cropland to forest land and construction land was the main contributor to the increase or decrease in carbon stock in Yulin.

Tables 13 and 14 provides a detailed breakdown of the increase or decrease in carbon stock following land use conversions from 2000 to 2020. The conversion of cropland to grassland, forest land, and wetland resulted in an increase of 0.0008 million tons, 0.8782 million tons, and 0.0005 million tons of carbon stock, respectively. On the other hand, converting cropland to construction land and water led to a reduction of 0.5532 million tons and 0.01341 million tons, respectively. Converting forest land to cropland, grassland, wetland, construction land, and water caused a decrease in carbon stock by 0.60845 million tons, 0.0055 million tons, 0.0001 million tons, 0.2000 million tons, and 0.0374 million tons, respectively. The conversion of construction land to cropland and forest land resulted in an increase of 0.0990 million tons and 0.0355 million tons in carbon stock. Other land use conversions had varying effects on carbon stock. Overall, the increase in carbon stock outweighed the decrease, resulting in an overall increase in carbon stock in Yulin City by 2020.

Table 14.

Changes in carbon storage under land class transfer in Yulin during 2000–2020.

Grassland–Waterland	0.54	0.0000	Waterland - Cropland	1371.33	0.0816
Forestland–Cropland	56,489.85	− 6.0845	Waterland–Grassland	2.07	0.0001
Forestland–Grassland	57.78	− 0.0055	Waterland–Forestland	1053.18	0.1761
Forestland–Wetland	1.71	− 0.0001	Waterland–Wetland	2.16	0.0002
Forestland–Construction land	1555.20	− 0.2000	Waterland–Construction land	112.77	0.0044

Under the NDS for 2030, the cropland area in Yulin decreases, while the areas of forest land and construction land expand. As a result, the carbon stock shows an increase compared to 2020 (Table 10, 11). In contrast, under the CPS, the cropland area significantly increases, but the forest land and construction land areas decrease. This leads to a decline in carbon stock compared to 2020. Under the EPS, the cropland area shrinks, while ecological lands such as forest land and grassland increase, and construction land expands slightly. In this scenario, carbon stock shows an increase compared to 2020, with the highest carbon stock improvement among the three scenarios. This paper compares the regional changes in carbon stock across different scenarios. The CPS significantly protects cropland but limits the sustainable development of ecological lands and urbanization, impacting both the environment and economic growth. Under the EPS, the conversion of ecological lands like forest land and grassland is restricted. This leads to an increase in carbon stock compared to 2020, showing that projects such as returning cropland to forest have a positive impact on carbon storage. However, cropland protection remains insufficient. To achieve sustainable development in the regional ecosystem, local governments must plan land use patterns and resources carefully and implement measures that ensure both food security and ecological development.

Table 13.

 Changes in carbon storage under land class transfer in Yulin during 2000–2020.

Classification transfer	Area/hectare	Change in carbon stocks/million t	Classification transfer	Area/hectare	Change in carbon stocks/million t
Cropland–Grassland	59.85	0.0008	Forestland– Waterland	223.83	− 0.0374
Cropland–Forestland	81533.97	8.7820	Wetland– Cropland	2.52	− 0.0001
Cropland–Wetland	12.24	0.0005	Wetland– Construction land	0.63	0.0000
Cropland–Construction land	26483.13	-0.5532	Wetland– Waterland	2.34	− 0.0002
Cropland–Waterland	2255.67	-0.1341	Construction land–Cropland	4741.02	0.0990
Grassland–Cropland	40.32	-0.0005	Construction land–Forestland	275.94	0.0355
Grassland– Forestland	23.31	0.0022	Construction land–Wetland	0.54	0.0000
Grassland–Construction land	0.09	0.0000	Construction land–Waterland	33.84	− 0.0013

Impact of policies on urban carbon stocks

Policies play a crucial role in promoting regional economic development and ecological restoration. However, socio-economic development policies have placed significant pressure on ecosystem protection. By implementing proactive environmental protection and restoration policies, it is possible to enhance ecosystem stability and integrity while increasing regional carbon stocks. For example, Wang et al.²⁷ found that active ecological policies influence carbon stock estimation in the Guangdong-Hong Kong-Macao Greater Bay Area. Promoting ecological policies can help protect forest and grassland areas, thus boosting regional carbon stocks⁵⁰, especially in economically developed urban agglomerations. As urbanization accelerates, future carbon stocks in this region are expected to decline. To mitigate this, appropriate policy guidance for land planning can limit the conversion of high-carbon-density areas to low-carbon-density ones, improving future carbon stocks. Liang et al.⁴⁹ observed that in the early stages of coordinated regional development in the Beijing-Tianjin-Hebei region, the relocation of high-energy-consuming and high-emission industries, along with expanded transportation networks, promoted economic development in neighboring areas. However, this also led to carbon emission transfer and imbalances in carbon metabolism. While the carbon metabolism systems in Beijing and Tianjin slightly improved, industrial land and other land uses in Hebei continued to compete intensively, resulting in a relatively low reciprocity index. Therefore, comprehensive land use planning, strict protection policies for arable land, and reasonable ecological compensation mechanisms are essential to ensure sustainable development in the Beijing-Tianjin-Hebei urban agglomeration.

Since 2004, Yulin City has strictly implemented measures for farmland protection. These policies primarily focus on limiting the conversion of agricultural land to non-agricultural uses, as well as strengthening farmland reclamation and restoration. Starting in 2013, with the national emphasis on ecological environment protection, Yulin City introduced a series of local ecological protection policies, focusing on enhancing the protection of forests, wetlands, and other ecological environments. These policies include the designation of ecological function zones and the implementation of ecological compensation mechanisms.

From 2000 to 2020 (Fig. 12), the carbon stock in each county of Yulin City increased, except in Yuzhou District. This trend is related to the “Yulin City Urban and Rural Planning Law” introduced in 2015, which emphasizes urban-rural integration, land intensification, and the standardized management of land development. Among these, the most significant increases were seen in Bobai County, Beiliu County, and Rongxian County. By 2020, the carbon stock in these three counties accounted for the top three proportions in Yulin City, which is closely linked to the series of local ecological protection policies introduced by the Yulin City government.

Fig. 12.

Distribution of carbon storage by counties in 2000 and 2020 (a 2000; b 2020).

Response relationship between land use and Spatial distribution of carbon stocks

Land use is closely linked to the spatial distribution of carbon stocks. In Yulin City, carbon stock changes are patchy and sporadic. Forest and grassland dominate Beiliu City, Rong County, and Xingye County, where high forest coverage and ecological integrity provide abundant carbon stock. As a result, carbon stocks in these areas remain highly aggregated. In Bobai County, Luchuan County, and Fumian District, cropland and forest land are dominant. High-altitude areas are mainly forested, preserving a high aggregation of carbon stocks. However, low-altitude regions are mostly cropland, with few ecological patches. This results in high fragmentation and limited space for dense carbon stocks, keeping carbon stocks in these areas less aggregated.

Assessing carbon stock causes with the invest model

Researchers typically assess carbon stocks using field surveys, remote sensing inversion, and model analysis. The field survey and remote sensing inversion methods are complex to implement in practice. Field sampling involves selecting sample areas, collecting soil samples for physical and chemical tests, or obtaining data through vegetation and soil inventory calculations. This method requires significant effort and time, and due to the large study area (the entire urban area of Yulin), sample collection is difficult. As a result, field sampling is not applicable to this study. Remote sensing inversion is often used to estimate aboveground and soil biomass in specific ecosystems. This study focuses on four carbon stock components: aboveground biomass, belowground biomass, soil organic matter, and dead organic matter. However, remote sensing inversion only captures two of these components. Compared to the carbon stock obtained through the InVEST model, the carbon stock estimated by remote sensing inversion is less accurate. Common remote sensing models, such as CASA, FORCCHN, LPJ-GUESS, and DNDC, have complicated data requirements and limited applicability. In contrast, the InVEST model requires less data and provides more accurate assessments. It also allows for the study of regional carbon stock changes at various scales. Therefore, this study uses the InVEST model for carbon stock estimation.

Limitations

This study provides insights into the regional carbon cycle and balance, offering theoretical references for the sustainable development of regional ecosystems. However, some limitations and uncertainties remain. (1) To minimize errors in carbon stock evaluation, this study uses carbon density pools measured in the same study area or similar geographic locations whenever possible. However, carbon density data from other scholars in the Guangxi region were used in this study, which may lead to inaccurate carbon stock estimates. The carbon stock calculations in this paper assume a single carbon density for each land use type. However, different carbon densities can exist within the same land use type, depending on cover changes. For instance, vegetation in the same land-use class can accumulate varying levels of carbon density, and changes in aboveground biomass (e.g., carbon sequestration from vegetation growth) can cause dynamic shifts in carbon density over time. To improve the accuracy of carbon stock estimation, future studies should conduct long-term monitoring of the carbon density of local land use types, minimizing uncertainty caused by changes in carbon density. Therefore, enhancing the accuracy of carbon stock calculations should be a key focus for future research. (2) The accuracy and timeliness of parameters often present challenges in these types of studies. Field sampling can help validate these parameters, reducing uncertainties and yielding more reliable results. However, conducting field surveys across large and medium-sized study areas is costly and difficult. Future research should prioritize improving the certainty of parameters. (3) The accuracy of land-use change assessments and carbon stock predictions can vary at different scales. Therefore, future studies should consider the scaling effects of spatial and temporal changes. Addressing this aspect is crucial for improving the accuracy and relevance of these studies. This study focused primarily on land use change and carbon stocks, without delving deeply into other ecosystem services or synergies. Future research should emphasize these additional aspects.

Conclusions

In this study, the ecosystem carbon stock in Yulin City from 2000 to 2020 was estimated using land use and carbon density data for each category. The InVEST model was applied to analyze the spatial and temporal characteristics of the carbon stock. Furthermore, predictions for land use and carbon stock changes in Yulin City in 2030 were made using the coupled InVEST-PLUS model. The main findings are as follows:

1. The cropland patches in Yulin are becoming more dispersed, but their fragmentation remains simple. The spatial aggregation of cropland is increasing, and its fragmentation is becoming more pronounced. The area of forested land is expanding, with patches becoming more dispersed. The shapes of these patches are stabilizing, and their spatial aggregation remains relatively high. The area of construction land is also increasing, and while the fragmentation of its patches is growing more complex, the aggregation of construction land patches remains higher.

2. The driving factors show that population, DEM, and slope are the main drivers affecting the change of carbon stock, and the spatial pattern of land use influences the increase or decrease of regional carbon stock. The regional ecosystem carbon stock increased by 2.16 × 10⁶ t from 2000 to 2020, with the smallest growth in carbon stock from 2010 to 2015. Researchers distribute high-value carbon stock areas in the mountains and hills with higher elevations in Yulin City. They mainly place medium-value areas in the central part of the region, where urbanization is progressing rapidly, and in the southern part, where the terrain is flat. Low-value areas mainly occupy the southern part of the study area, where rivers and some reservoirs exist. In terms of counties and districts, Bobai County has the highest carbon stock, and Yuzhou District has the lowest.

3. Carbon stock changes occur under different development scenarios due to land use transformations. In the EPS scenario, which limits the conversion of high-carbon-density forest land and grassland into other land uses, carbon stock increases compared to 2020. This suggests that policies such as returning cropland to forests positively impact carbon stock growth. While cropland protection remains insufficient, the EPS scenario achieves the highest carbon stock increase among the three. Currently, Yulin City is undergoing urbanization, leading to a decrease in cropland and an expansion of construction land, which could increase regional carbon emissions. Cities should leverage regional advantages based on resources and socio-economic conditions. Future development should prioritize protecting ecological resources within environmental protection zones. At the same time, Yushu District should manage land development intensity to improve construction land efficiency and economic productivity. As urbanization progresses, carbon stocks in Yulin’s urban areas may rise slowly or even decline. Thus, managing carbon stocks in different regions requires a targeted approach. Policymakers should create land-use guidelines to prevent the conversion of high-density forest and grassland areas into low-density construction zones, which would help boost Yulin’s future carbon stock. Additionally, the energy structure should be adjusted by promoting energy-efficient technologies, encouraging high-tech research and development, and shifting from carbon-based to renewable energy sources. Efforts to minimize carbon emissions through energy consumption reductions or industrial carbon capture will support the goal of “carbon neutrality.”

Author contributions

G.S. and Y.L. developed the method and validated the results. G.S., Y.L., R.H., and C.M., worked on data collection and analysis. G.S., Y.L., and R.H. wrote and edited the manuscript. All authors reviewed the manuscript.

Funding

The National Natural Science Foundation of China (Grant No. 52269002), the Guangxi Water Conservancy Science and Technology Promotion Project (Grant No. SK-2022-021), and the Guangxi Key Research and Development Program (Grant No. AB24010047) supported this work.

Data availability

All methodology codes and data supporting this study’s findings are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.