Spatiotemporal Evolution, Regional Disparities, and Transition Dynamics of Carbon Effects in China’s Agricultural Land Use

Abstract

A precise understanding of the carbon dynamics of agricultural land use is essential for advancing China’s “dual carbon” goals and promoting sustainable rural development. Drawing on panel datasets for 31 Chinese provinces over the period 1997–2022, this study comprehensively analyzes the spatiotemporal evolution, regional disparities, and transition dynamics of agricultural carbon capture and emissions. Using a combination of the emission factor method, the Dagum Gini coefficient, kernel density estimation, and Markov chain models, the study finds that China’s total agricultural carbon capture has continued to increase, yet regional disparities are widening, with the central region leading and the northeastern region lagging. Meanwhile, agricultural carbon emissions exhibit a “strong west, weak east” spatial pattern and demonstrate a high degree of club convergence. Club convergence refers to the phenomenon where regions with similar initial levels converge to the same steady-state over the long run, while remaining persistently different from other regions. The net carbon effect exhibits a dual structure of carbon surplus zones and carbon deficit zones: 23 provinces act as carbon surplus zones, while 8 provinces are carbon deficit zones, primarily located in ecologically fragile or special-function regions. These findings highlight the spatial heterogeneity, path dependence, and policy sensitivity of carbon effects from agricultural land use. Accordingly, the study proposes differentiated policy recommendations, including region-specific carbon management strategies, the establishment of a unified agricultural carbon trading system, and the integration of technological and institutional innovations to achieve a balanced and low-carbon agricultural transformation.

1. Introduction

In recent years, the continuous rise in greenhouse gas (GHG) emissions has intensified global warming, making it one of the most pressing environmental challenges worldwide. The 2023 United Nations Environment Program report noted that in 2022, average global temperatures hit a record high, with more than 86 days exceeding pre-industrial levels by over 1.5 °C. Moreover, global GHG emissions increased by 1.2% from 2021 to 2022, reaching a record 57.4 gigatons of CO₂ equivalent (GtCO₂e) [1]. China, currently the world’s largest CO₂ emitter, has pledged to peak emissions before 2030 and achieve carbon neutrality by 2060, reflecting its commitment to low-carbon and sustainable development.

Achieving these ambitious climate targets demands not only technological innovation and industrial transformation but also a comprehensive understanding of the spatial and sectoral sources of carbon emissions. Among the various contributing sectors, land use and land use change stand out as critical determinants of carbon fluxes, particularly given China’s rapid urbanization and rural restructuring [2]. On the one hand, land plays dual roles in carbon emission and sequestration, with its utilization and transformation directly influencing terrestrial carbon storage, ecosystem processes, and atmospheric carbon levels. On the other hand, accelerating urban development has led to increasingly frequent and large-scale changes in land use, resulting in complex spatiotemporal patterns of carbon emissions [3]. These changes not only affect the carbon storage capacity of ecosystems but also reshape regional and global carbon fluxes. Consequently, analyzing the spatiotemporal dynamics and state transitions of land use-related carbon effects is essential for designing localized mitigation strategies and promoting climate-resilient land governance.

The impact of land use on regional carbon cycles can be divided into positive and negative aspects [4]. The net effect is usually measured by the balance between carbon sequestration and emissions. When sequestration exceeds emissions, regions form carbon surplus zones; the opposite leads to carbon deficit zones. Positive effects stem from practices such as afforestation, conservation tillage, and improved land management [5], whereas negative effects occur when land use driven by human activities adds to atmospheric CO₂ [6]. Agricultural land is central to these processes: farming activities release carbon through machinery, fertilizers, and energy inputs, but also enhance sinks via photosynthesis and soil carbon storage. The overall outcome depends on management practices and ecological conditions. In China, where cropland is extensive and farming intensive, agriculture has a strong influence on national emissions [7]. Therefore, a comprehensive understanding of the carbon effects of agricultural land use is essential for achieving China’s dual carbon goals and advancing sustainable rural development.

Existing studies show that land use strongly affects carbon processes, with land use change (LUC) exerting diverse impacts on soil carbon stocks under different conversion patterns [8,9,10]. Other scholars have focused on the spatial spillover impacts of land use-related carbon emissions in China, revealing strong spatial dependencies and significant interregional correlations [11,12]. However, few studies have simultaneously examined the spatial scales of both carbon storage and emissions in the context of land use. In addition, several studies have investigated the spatiotemporal evolution of land use carbon emissions, capturing the temporal trends and spatial heterogeneity of emissions at various geographic scales [13,14]. Research has also highlighted the potential of refining land use configurations alongside management practices for lowering greenhouse gas emissions [15]. Regarding land-based carbon sinks, prior studies have primarily concentrated on how land use change influences carbon sequestration potential, particularly through ecosystem restoration and vegetation cover dynamics [15,16,17]. Although prior research has deepened knowledge of land use-carbon linkages, it often centers on single spatial levels or emission categories and seldom provides an integrated perspective that combines regional heterogeneity, temporal trends, and dynamic shifts, particularly for agricultural land use. This gap highlights the necessity for a more systematic and regionally differentiated approach to assess the carbon effects of agricultural land use, with particular attention to temporal changes and transition states across regions. Such an approach is critical for informing evidence-based land management and advancing the realization of low-carbon agricultural development.

In recent years, the carbon effects of agricultural land use have garnered increasing academic attention. However, the analytical scope of existing studies remains relatively narrow, often focusing on discrete aspects of land use change without offering a holistic understanding of agricultural carbon dynamics. Few studies have simultaneously examined both carbon emissions and carbon sequestration when investigating the carbon effects of land use. The majority of research has explored how shifts in agricultural land use patterns, such as cropping structure adjustments, expansion or contraction of cultivated areas, and changes in input intensities affect carbon emissions and sequestration processes [18,19,20]. Beyond land use change, research has also explored socio-economic factors influencing agricultural carbon emissions, highlighting the roles of rural industrialization, urban-rural linkages, and farming intensification [21]. Others have explored the mediating role of land use transitions in the relationship between rural development and carbon emissions, shedding light on the indirect mechanisms underlying carbon output [22]. In parallel, an emerging strand of research has emphasized the potential of improved land management practices, including conservation tillage, precision fertilization, and crop rotation to mitigate agricultural emissions and enhance soil carbon sinks [23]. Despite these valuable contributions, few studies have adopted a systematic, spatiotemporally explicit approach to assess the carbon effects of agricultural land use. In particular, there is limited research that integrates carbon emission, regional disparities, and temporal trends into a unified analytical framework. Addressing this gap is critical for advancing low-carbon agricultural transformation, informing regionalized land use strategies, and supporting China’s broader pursuit of its “dual carbon” goals.

Despite significant advances in the study of land use carbon effects, existing research still faces key limitations. Most studies tend to focus separately on carbon emissions or carbon sequestration, lacking an integrated framework that captures their complex interactions within the carbon cycle. Additionally, research on the carbon effects of agricultural land use remains relatively limited in scope, with an overemphasis on emissions and insufficient attention to its potential as a carbon sink. This narrow perspective hampers a holistic understanding of agricultural land’s dual role in carbon dynamics. To address these gaps, this study aims to provide a comprehensive analysis of both carbon source and sink effects of agricultural land use, examine spatial heterogeneity across regions to support differentiated land management strategies, and explore the spatiotemporal evolution and state transitions of agricultural land use carbon effects to inform more effective, stage-specific mitigation policies. Compared to existing studies, this study constructs a systematic framework that simultaneously reveals the spatiotemporal pattern of “carbon sinks--carbon emissions-net effects” in agricultural land use. Moreover, it further disaggregates the sources of regional disparities and their dynamic evolution, and explores the dynamic transition mechanisms underlying carbon effects.

The subsequent sections are arranged as follows. Section 2 outlines the research design, including the construction of the evaluation framework for agricultural land use carbon effects, the methodologies employed, including the estimation of carbon sources and sinks, the Dagum Gini coefficient, kernel density estimation, and Markov chain analysis, as well as data sources. Section 3 presents the spatiotemporal distribution and evolving patterns of agricultural land use carbon sinks in China, highlighting regional disparities and transition trends. Section 4 focuses on the spatiotemporal dynamics of agricultural carbon emissions, further exploring their dynamic evolution across provinces and regions. Section 5 offers an in-depth discussion of the key findings in relation to existing literature. Lastly, Section 6 distills the key outcomes and offers tailored strategies to advance carbon reduction efforts and promote sustainable land management in China’s agricultural sector.

2. Materials and Methods

2.1. Methodology for Estimating the Carbon Effects of Agricultural Land Use

2.1.1. Methodology for Estimating Carbon Emissions from Agricultural Land Use

Drawing on the Intergovernmental Panel on Climate Change (IPCC)’s National Greenhouse Gas Inventories Guidelines [24] and relevant studies by scholars [25,26,27], this study identifies crop production and livestock farming as the primary sources of carbon emissions in China, consistent with the country’s dominant farming practices. Referring to the Provincial Greenhouse Gas Inventory Preparation Guidelines (Trial) issued by China’s National Development and Reform Commission in 2011, a coefficient-based estimation approach is applied to quantify carbon emissions from various categories of agricultural land use. This study calculates China’s total agricultural land use emissions by aggregating all carbon sources. The calculation is performed using the following equation.

$$E_{carbon}=E_1+E_2$$

(1)

where E_carbon denotes total agricultural land use emissions; E₁ denotes emissions from the planting sector; E₂ denotes emissions from animal husbandry.

Based on reference [25], this study categorizes the primary carbon sources in the planting industry process into six types, including chemical fertilizers, pesticides, agricultural plastic film, agricultural diesel, crop sowing, and agricultural irrigation. The emission factor method is applied to estimate the carbon emissions associated with each category. The specific formulas refer to reference [25].

In the animal husbandry sector, this study incorporates growth cycle adjustments and includes 11 types of livestock in the carbon emission accounting system: pigs, rabbits, poultry, cows, other cattle, horses, donkeys, mules, goats, sheep, and camels. These categories are representative of China’s livestock industry, covering different breeding types, and their breeding scale occupies an important position nationwide. To ensure accuracy, the annual average feeding quantity is adjusted based on the growth cycles of animals that are typically raised and slaughtered within a year. Specifically, the growth periods are set as 200 days for pigs, 105 days for rabbits, and 55 days for poultry. The adjusted average feeding quantity is then used in the emission estimation, as defined by Equations (2) and (3). Based on IPCC (2006), the emission factors corresponding to each animal type and greenhouse gas are detailed in

Table 1. Carbon Source Factors in Animal Farming Emissions.

Carbon Sources	Emission Factors of CH4		Emission Factors of N2O
	Gastrointestinal Fermentation	Fecal Fermentation
Pigs	1.000 kg per animal per year	3.500 kg per animal per year	0.530 kg per animal per year
Rabbits	0.254 kg per animal per year	0.080 kg per animal per year	0.020 kg per animal per year
Poultry	0 kg per animal per year	0.020 kg per animal per year	0.020 kg per animal per year
Cows	68.000 kg per animal per year	16.000 kg per animal per year	1.000 kg per animal per year
Other cattle	51.400 kg per animal per year	1.500 kg per animal per year	1.370 kg per animal per year
Horses	18.000 kg per animal per year	1.640 kg per animal per year	1.390 kg per animal per year
Donkeys	10.000 kg per animal per year	0.900 kg per animal per year	1.390 kg per animal per year
Mules	10.000 kg per animal per year	0.900 kg per animal per year	1.390 kg per animal per year
Goats	5.000 kg per animal per year	0.170 kg per animal per year	0.330 kg per animal per year
Sheep	5.000 kg per animal per year	0.150 kg per animal per year	0.330 kg per animal per year
Camels	46.000 kg per animal per year	1.920 kg per animal per year	1.390 kg per animal per year

[28].

$$E_2={\displaystyle \sum GW{P}_{C{H}_4}}\times M_i\times {\chi }_{1i}+{\displaystyle \sum GW{P}_{C{H}_4}}\times M_i\times {\chi }_{2i}+{\displaystyle \sum GW{P}_{N_{2}O}}\times M_i\times {\chi }_{3i}$$

(2)

$$M_i=\left\{\begin{array}{l}Days\_aliv{e}_i\times {\displaystyle \frac{A_i}{365}},Days\_aliv{e}_i<365\\ {\displaystyle \frac{C_{it}+C_{i(t-1)}}{2}},Days\_aliv{e}_i\ge 365\end{array}\right.$$

(3)

where E₂ denotes the total greenhouse gas emissions from the livestock sector; GWP_CH4 and GWP_N2O denote greenhouse efficiency index for the global warming potentials of CH₄ and N₂O, with values of 25 and 298; M_i represents annual average feeding quantity of class i animals in the animal husbandry; ${\chi }_{1i}$, ${\chi }_{2i}$ and ${\chi }_{3i}$ represent the emission factors of CH₄ from enteric fermentation, CH₄ from manure treatment, and N₂O from manure handling; E₂ is the overall agricultural emission; M_i represents the adjusted annual average feeding quantity of livestock class i in the animal husbandry sector; Days_alive_i represents the growth cycle (in days) of livestock class i animals; A_i is the yearly output of livestock class i animals; C_it denotes the end-of-year inventory for livestock class i.

2.1.2. Methodology for Estimating Carbon Sink from Agricultural Land Use

The study primarily estimates the carbon sequestration of agricultural land use by evaluating the capacity of crops on farmland to convert atmospheric CO₂ into organic carbon through photosynthesis. Based on reference [24] and relevant research by scholars [29,30,31], crops on agricultural land capture carbon from the atmosphere and synthesize it into carbohydrates via photosynthesis. The carbon absorption rates, moisture contents, economic coefficients, and root-to-shoot ratios of major crops selected for agricultural land in this study are based on the research by reference [31] (see

Table 2. Carbon absorption rate, moisture content, economic coefficient, and root-to-shoot ratio of major crops.

Crops	Carbon Absorption Rate	Moisture Content	Economic Coefficient	Root-to-Shoot Ratio
Corn	0.471	0.13	0.4	0.16
Rice plant	0.414	0.12	0.45	0.6
Wheat	0.485	0.12	0.4	0.4
Legume	0.45	0.13	0.35	0.13
Tubers	0.423	0.7	0.7	0.18
Oilseeds	0.45	0.09	0.25	0.04
Cotton	0.45	0.08	0.19	0.1
Tobacco	0.45	0.17	0.55	0.32
Fiber crops	0.45	0.13	0.1	0.4

). The corresponding carbon sequestration is estimated using the following Equation (4).

$$E_{sink}={{\displaystyle \sum C}}_i\times Y_i\times (1-W_i)\times (1+R_i)/H_i$$

(4)

where E_sink represents the total carbon sink of agricultural land use; C_i, Y_i, W_i, R_i and H_i denote the carbon absorption rate, economic yield, moisture content, root-to-shoot ratio, and economic coefficient of crop class i, respectively.

2.2. Dagum Gini Coefficient

The Dagum Gini coefficient, unlike conventional measures such as the Gini or Theil index, accounts for subgroup distributions and addresses the issue of overlapping samples, enabling the detection of subtle interregional differences. It decomposes total disparity into within-region, between-region, and transvariation components, offering clearer insights into the origins of inequality. These features make it well-suited for analyzing spatial variation in agricultural carbon emissions [32,33,34,35].

2.3. Kernel Density Estimation

As a non-parametric technique, kernel density estimation reveals regional distribution patterns across years without relying on a predefined model. The method uses continuous density curves to portray the distributional pattern of random variables, thus reflecting the location, shape and other information of the distribution of variables. To better reveal overall and regional dynamics, this study adopts Gaussian kernel density estimation. The specific formulas refer to references [35].

2.4. Markov Chain Analysis

The Markov chain approach, viewed as a probabilistic framework with discrete states and time steps, provides a means of simulating system evolution by classifying data into categories and evaluating the likelihood of state transitions [36]. In this paper, a Markov matrix is constructed to portray the feature of the dynamic evolution of distribution of agricultural land use carbon emissions. The specific formulas refer to references [35].

2.5. Data Sources

The research object of this study covers all 31 provinces in mainland China. Due to data limitations, regions such as the Hong Kong Special Administrative Region, Macao Special Administrative Region, and other territories are excluded from the analysis. The data used in this study are primarily sourced from authoritative publications, including the China Statistical Yearbook, China Environment Statistical Yearbook, China Science and Technology Statistical Yearbook, China City Statistical Yearbook, and China Population and Employment Statistical Yearbook, together with historical datasets from the National Bureau of Statistics of China. Missing provincial or annual data were supplemented using linear interpolation to ensure continuity. To better capture regional disparities, the country is categorized into four regions: Northeast, East, Central, and West.

3. Results

3.1. Spatial–Temporal Patterns and Evolutionary Dynamics of Carbon Capture in China’s Agricultural Land Use

The spatiotemporal characteristics of agricultural land use carbon capture across China and its four major regions are illustrated in Figure 1. From 1997 to 2022, the total carbon capture from agricultural land use in China exhibited an overall upward trend, increasing from 634.19 million tons to 888.36 million tons, with an overall growth of 28.62%. Specifically, throughout the observation period, the central region consistently recorded the highest carbon capture levels among the four major regions, while the northeastern region maintained the lowest levels. Spatially, this disparity not only reflects variations in natural conditions across regions but also highlights the combined influence of regional agricultural development strategies and production practices on the carbon capture capacity of agricultural land use.

Based on the estimated data of agricultural land use carbon capture across Chinese provinces, the carbon capture levels are categorized into four groups: low level [0.3451, 7.6911] million tons, lower-middle level (7.6911, 16.6355] million tons, upper-middle level (16.6355, 37.4502] million tons, and high level (37.4502, 98.782] million tons. The spatiotemporal characteristics of China’s total agricultural land use carbon capture in the years 1997, 2003, 2009, 2015, and 2022 are illustrated in Figure 2. From 1997 to 2022, the agricultural land use carbon capture in China exhibited a marked rise. As illustrated in the figure, the carbon capture levels gradually shifted from lower-middle levels toward upper-middle and high levels, with high-performing provinces increasing from 8 in 1997 to 11 in 2022. Among them, the Inner Mongolia Autonomous Region made a transition from a lower-middle level to a high level.

Table 3. Regional Disparities and Decomposition Results of Agricultural Land Use Carbon Capture in China.

Year	Overall	Inner-Regional	Inter-Regional	Intensity of Transvariation	Contributions
					Inner- Regional	Inter- Regional	Intensity of Transvariation
1997	0.432	0.104	0.196	0.131	24.130	45.478	30.391
1998	0.416	0.103	0.178	0.135	24.702	42.921	32.376
1999	0.432	0.105	0.190	0.137	24.407	43.935	31.658
2000	0.438	0.110	0.180	0.149	24.994	41.047	33.959
2001	0.436	0.106	0.193	0.137	24.364	44.294	31.341
2002	0.434	0.105	0.193	0.136	24.141	44.556	31.304
2003	0.425	0.105	0.176	0.144	24.761	41.379	33.860
2004	0.436	0.103	0.203	0.131	23.526	46.489	29.985
2005	0.441	0.105	0.197	0.139	23.766	44.671	31.563
2006	0.464	0.106	0.227	0.132	22.737	48.913	28.350
2007	0.462	0.107	0.221	0.135	23.067	47.760	29.173
2008	0.462	0.106	0.221	0.136	22.830	47.716	29.454
2009	0.463	0.107	0.215	0.141	23.083	46.455	30.462
2010	0.465	0.106	0.220	0.140	22.694	47.243	30.063
2011	0.468	0.105	0.229	0.133	22.507	49.046	28.447
2012	0.463	0.104	0.223	0.135	22.545	48.189	29.266
2013	0.466	0.105	0.226	0.136	22.543	48.374	29.084
2014	0.471	0.106	0.223	0.142	22.564	47.352	30.084
2015	0.473	0.107	0.226	0.140	22.519	47.786	29.695
2016	0.490	0.107	0.243	0.139	21.923	49.643	28.434
2017	0.494	0.108	0.246	0.140	21.926	49.750	28.324
2018	0.496	0.111	0.238	0.147	22.366	47.953	29.682
2019	0.498	0.111	0.242	0.144	22.325	48.674	29.001
2020	0.498	0.112	0.240	0.146	22.413	48.247	29.341
2021	0.495	0.111	0.240	0.144	22.366	48.548	29.086
2022	0.495	0.111	0.238	0.146	22.452	48.062	29.486
Average	0.462	0.107	0.216	0.139	23.140	46.711	30.149

presents China’s agricultural carbon sequestration disparities and contribution rates. From 1997 to 2022, the Gini coefficient for agricultural land use carbon capture exhibited an upward trend, increasing from 0.432 in 1997 to 0.495 in 2022, an increase of 14.5%. During the observation period, the average Gini coefficient was 0.462, indicating substantial disparities in carbon capture across different regions. Moreover, the data show that inter-regional differences account for most of the variation, contributing on average 46.71%, further underscoring the significance of regional heterogeneity in agricultural land use carbon capture.

Kernel density analysis was conducted on the agricultural land use carbon capture of 31 provinces and autonomous regions in China, as illustrated in Figure 3. The results exhibit a distribution characterized by both major and minor peaks, indicating a multimodal pattern. The major peaks are primarily concentrated in regions with lower levels of agricultural land use carbon capture, while the minor peaks tend to appear in regions with higher carbon capture values. It reflects pronounced spatial differences in the carbon capture capacity of agricultural land. Over time, the density curves become increasingly dispersed, indicating a widening regional disparity in carbon capture levels. This observation is consistent with the earlier findings derived from the Dagum Gini coefficient analysis, which also pointed to an expanding regional gap in carbon capture distribution. Additionally, the presence of a rightward tail in the density curves further underscores the persistent and significant regional disparities in agricultural land use carbon capture.

Figure 4 displays farmland carbon capture dynamics in China’s four regions (1997–2022). In the eastern region, the carbon capture exhibits a multimodal distribution, with major peaks concentrated at lower carbon capture levels and minor peaks at higher levels, reflecting considerable intraregional disparities. In the central region, the main peak is centered around higher carbon capture values, indicating a relatively high overall carbon capture level. Moreover, the distribution in the central region shows increasing dispersion over time, indicating widening regional differences. In the western region, the overall carbon capture curve gradually shifts to the right over time, suggesting that the concentration interval of carbon capture is moving toward higher values, implying an overall increase in carbon capture capacity. In the northeastern region, a bimodal distribution is evident, with the main peak gradually shifting rightward, signifying that carbon capture is increasingly concentrated at higher levels and that the regional average has risen markedly over time.

To further explore the dynamic evolution of agricultural land use carbon capture in China, this study conducts a Markov chain transition matrix analysis based on the calculated carbon capture levels.

Table 4. Traditional Markov Transition Probability Matrix of Agricultural Land Use Carbon capture in China.

t	t + 1
	Ⅰ	II	III	IV
Ⅰ	0.990	0.010	0	0
II	0.010	0.954	0.036	0
III	0	0.025	0.904	0.071
IV	0	0	0.053	0.947

presents the results of the traditional Markov chain transition matrix. As observed, the values along the diagonal are consistently higher than the off-diagonal elements, indicating a strong level of stability in China’s agricultural land use carbon capture. Specifically, the probability of remaining at a low carbon capture level is 0.990, with only a 0.01 probability of upward transition. Similarly, the probability of maintaining a high carbon capture level is 0.947, with a 0.053 probability of downward transition. These results suggest a pronounced club convergence characteristic in the evolution of agricultural land use carbon capture in China, where provinces with similar carbon capture levels tend to maintain their positions, showing limited likelihood of shifting significantly up or down the distribution.

3.2. Spatial–Temporal Patterns and Evolutionary Dynamics of Carbon Emissions in China’s Agricultural Land Use

Figure 5 presents China’s estimated agricultural carbon emissions. Nationally, emissions followed a generally rising yet fluctuating trajectory, climbing from 422.83 million tons in 1997 to 456.45 million tons in 2022, an overall increase of 7.4%. Over the study period, the national average of agricultural carbon emissions was 14.60 million tons, with 14 provinces exceeding this average, thereby falling into the high-emission category. Among the four major regions, the western region consistently recorded the highest carbon output from farming activities, while the northeastern region remained at the lowest level. This pattern reveals that overall agricultural carbon emissions in China are still relatively high, highlighting the urgent need to explore effective strategies to address this challenge and promote sustainable development.

Based on the estimated agricultural carbon emissions of each province in China, the emissions are categorized into four levels: low [0.372, 6.881] million tons; lower-middle (6.881, 13.138] million tons; upper-middle (13.138, 19.486] million tons; and high (19.486, 52.071] million tons. The spatiotemporal characteristics of China’s total agricultural carbon emissions in the years 1997, 2003, 2009, 2015, and 2022 are shown in Figure 6. China’s agricultural carbon emissions increased steadily from 1997 to 2022. The figure demonstrates a gradual transition from lower-middle emission areas to upper-middle and high emission areas. Specifically, the number of high-level regions increased from six in 1997 to eight in 2022, while the number of lower-middle-level regions grew slightly from nine to ten. The rise in agricultural carbon emissions is particularly evident in central and western China, driven by expanded production scales and shifts in farming practices. The moderate growth in the number of lower-middle-level areas suggests that regions with previously lower emissions are increasingly contributing to agricultural carbon outputs. The more significant increase in high-level areas indicates that carbon emission issues are more pronounced in these regions, which require focused attention.

The decomposition results of the Gini coefficient for China’s agricultural carbon emissions are presented in

Table 5. Regional Disparities and Decomposition Results of Agricultural Land Use Carbon Emissions in China.

Year	Overall	Inner-Regional	Inter-Regional	Intensity of Transvariation	Contributions
					Inner- Regional	Inter- Regional	Intensity of Transvariation
1997	0.345	0.093	0.117	0.135	26.889	33.838	39.273
1998	0.356	0.097	0.117	0.143	27.100	32.778	40.122
1999	0.368	0.100	0.115	0.153	27.279	31.216	41.505
2000	0.371	0.102	0.110	0.159	27.434	29.586	42.980
2001	0.368	0.101	0.104	0.163	27.510	28.153	44.337
2002	0.367	0.101	0.103	0.163	27.419	28.139	44.442
2003	0.363	0.100	0.095	0.168	27.480	26.139	46.381
2004	0.366	0.101	0.092	0.173	27.596	25.220	47.183
2005	0.370	0.102	0.095	0.173	27.628	25.590	46.782
2006	0.375	0.102	0.105	0.167	27.333	27.982	44.684
2007	0.371	0.102	0.106	0.163	27.491	28.666	43.843
2008	0.368	0.104	0.095	0.169	28.168	25.832	46.000
2009	0.366	0.104	0.092	0.170	28.356	25.171	46.473
2010	0.358	0.100	0.104	0.154	27.905	29.093	43.001
2011	0.357	0.099	0.111	0.147	27.668	31.019	41.314
2012	0.358	0.099	0.111	0.148	27.640	31.125	41.235
2013	0.357	0.098	0.115	0.145	27.502	32.060	40.437
2014	0.359	0.098	0.118	0.143	27.399	32.804	39.797
2015	0.363	0.099	0.120	0.144	27.294	33.150	39.555
2016	0.363	0.099	0.122	0.142	27.228	33.659	39.113
2017	0.360	0.100	0.114	0.147	27.680	31.571	40.749
2018	0.362	0.100	0.136	0.126	27.594	37.512	34.894
2019	0.362	0.098	0.149	0.114	27.226	41.141	31.633
2020	0.365	0.098	0.160	0.107	26.862	43.865	29.272
2021	0.361	0.096	0.160	0.105	26.560	44.394	29.046
2022	0.360	0.096	0.163	0.102	26.500	45.257	28.243
Average	0.363	0.100	0.117	0.147	27.413	32.114	40.473

. The overall Gini coefficient of agricultural carbon emissions in China shows a trend of rising first and then declining, which aligns with the broader context of low-carbon agriculture. Specifically, the overall Gini coefficient increased from 0.345 in 1997 to a peak of 0.375 in 2006, representing an increase of 8%. Thereafter, it began to decrease steadily, falling to 0.360 by 2022, a reduction of 4.17%. This trend indicates that regional disparities in agricultural carbon emissions initially widened and then narrowed over time. The overall Gini coefficient remained below 0.4, suggesting a narrowing disparity in China’s farmland carbon output across regions, though imbalances in distribution still persist. Furthermore, the primary source of regional disparities in agricultural carbon emissions was hypervariable density, with an average contribution rate of 40.47%. However, since 2017, inter-regional differences have increasingly become the main driver of agricultural carbon emission disparities in China, which is consistent with real-world observations.

Kernel density analysis was conducted for agricultural carbon emissions across 31 provinces in China, as illustrated in Figure 7. The graph shows a multi-peaked distribution, characterized by one prominent peak and several smaller ones. The primary peak is concentrated in regions with lower carbon emissions, while the smaller peaks are located in areas with higher emissions. This pattern indicates a clear imbalance in the distribution of agricultural carbon emissions across the country. Overall, the density curve demonstrates a right-then-left temporal shift, suggesting that China’s agricultural carbon emissions initially increased and then decreased. The peak height initially decreased before increasing again, while the peak width gradually widened, indicating that the distribution of provinces has shifted from a more dispersed pattern toward a more unified level of emissions. Additionally, the right-skewed tail reveals that regions with relatively high emissions have a pulling effect on the overall distribution. The contraction in the curve’s spread suggests a reduction in absolute differences across regions.

Figure 8 shows spatiotemporal trends in agricultural carbon emissions across China’s major regions between 1997 and 2022. All four regions display internal emission imbalances, with distinct patterns: the eastern region peaked initially before declining, as reflected in the distribution’s right-then-left shift. Over time, the peak height first declined and then rose, and the width gradually expanded, suggesting that the distribution of values became more concentrated. In the central region, the peak gradually moved leftward, reflecting a decrease in agricultural carbon emissions. Similarly, the peak height initially decreased before rebounding, and the peak width widened, indicating a convergence of emission levels. In the western region, the peak shifted rightward, indicating an increase in emissions. A multi-peaked pattern emerged, with one main peak and several smaller ones, suggesting an imbalance in agricultural carbon emissions. Over time, the peak height increased while the width narrowed, indicating a higher level of concentration and greater convergence. In the northeastern region, the peak of agricultural carbon emissions first shifts to the right and then to the left, indicating a trend of first increasing and then decreasing carbon emissions. The peak height first declined and then rose, reflecting a trend toward convergence in emission levels.

To further analyze the dynamic evolution of agricultural carbon emissions in China, a Markov transition matrix was constructed based on the estimated carbon emission data.

Table 6. Traditional Markov Transition Probability Matrix of Agricultural Land Use Carbon Emissions in China.

t	t + 1
	Ⅰ	II	III	IV
Ⅰ	0.9692	0.0308	0	0
II	0.0258	0.9021	0.0722	0
III	0	0.0622	0.8860	0.0518
IV	0	0	0.0415	0.9585

presents the results of the traditional Markov transition matrix. As shown in the table, the diagonal elements of the matrix have the highest values, indicating a strong stability in China’s agricultural carbon emissions. The probability of remaining at a low level is 0.9692, with an upward transition probability of 0.0308, while the probability of remaining at a high level is 0.9585, with a downward transition probability of 0.0415. These results suggest a strong club convergence. Furthermore, the difficulty of upward and downward transitions reflects the high level of stability in emissions, with limited likelihood of inter-regional transitions.

3.3. Carbon Effects from Agricultural Land Use in China

The measurement results of carbon effects from agricultural land use in the four major regions of China are shown in Figure 9. The eastern region consistently exhibits a carbon capture-dominated trend, with the absolute value gradually increasing, indicating a continuous strengthening of its carbon capture advantage. The central region has long been dominated by a strong carbon capture, with a rapid decline in the absolute value, possibly related to ecological protection in major grain-producing areas and the green transformation of agriculture. In the northeastern region, agricultural land use is primarily characterized by a carbon capture effect, although the absolute value reached its lowest point in 2000 and then showed an increasing trend, suggesting a renewed enhancement of carbon capture capacity and reflecting dynamic adjustments in regional agro-ecological systems. The western region was carbon emission-dominated before 2006, with an increasing trend in carbon effects, but shifted to a carbon capture-dominated pattern thereafter, reflecting the positive impact of policies such as Grain for Green. However, the carbon effects from agricultural land use in the western region have fluctuated around zero, indicating the need for further strengthening. In summary, the eastern, northeastern, and central regions have long functioned as carbon capture advantage zones, while the western region is undergoing a “carbon emission–carbon capture” transition, highlighting regional heterogeneity in the carbon effects of agricultural land use and production modes in China.

Based on the average carbon effects of agricultural land use across 31 provinces in China, regions can be classified into carbon surplus zones (carbon capture > carbon emission) and carbon deficit zones (carbon emission > carbon capture), as shown in

Table 7. Zoning of Carbon Effects in Agricultural Land Use in China.

Types	Provinces
Carbon surplus zone (carbon capture > carbon emission)	Tianjin, Hebei, Shanxi, Liaoning, Jilin, Heilongjiang, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Guangxi, Chongqing, Sichuan, Shannxi, Ningxia, Xinjiang
Carbon deficit zone (carbon emission > carbon capture)	Beijing, Neimenggu, Hainan, Guizhou, Yunnan, Xizang, Gansu, Qinghai

. The table indicates that 8 provinces fall into the carbon deficit category, while the remaining 23 provinces are classified as carbon surplus zones. The carbon deficit regions may be driven by high emissions and limited capture capacity. These areas are often located in ecologically fragile zones, such as Gansu and Qinghai, where low vegetation cover and land degradation result in prominent carbon emissions. Some deficit regions also coincide with special functional zones, such as Beijing, characterized by high urbanization, limited arable land, and concentrated carbon emissions. Overall, China exhibits a spatial pattern of “carbon surplus dominance with localized deficits.” However, it is essential to strengthen regional coordination and implement ecological compensation in carbon deficit areas. Through differentiated policy measures, the agricultural carbon effect nationwide can be enhanced by simultaneously reducing emissions and increasing captures, thereby promoting regional carbon balance and supporting the green transformation of agriculture and ecological security.

4. Discussion

This study provides a comprehensive assessment of the spatiotemporal characteristics and dynamic evolution of agricultural land use carbon effects in China. By examining carbon capture, emissions, and effects clearly, it offers new evidence on the complexity of agricultural carbon processes and highlights the urgent need for differentiated regional strategies.

4.1. Regional Differences and Dynamic Evolution Path of Agricultural Carbon Capture

Agricultural carbon capture in China has shown a steady upward trend over the past two decades, yet the pace of growth and its regional distribution remain markedly uneven. A distinct “central strong, northeast weak” spatial configuration has gradually emerged. This pattern reflects both the natural resource endowments of different regions and the varying degrees of adaptation in agricultural practices. The central region benefits from fertile plains and a high proportion of paddy rice cultivation, which has a relatively strong carbon sequestration capacity. Moreover, its relatively stable soil management practices have enabled it to maintain and even enhance sequestration potential. This finding is broadly consistent with Wen et al. [37], who highlighted the central region’s superior sequestration performance arising from rice-dominated systems and better irrigation infrastructure. In contrast, the Northeast, despite being endowed with some of the most fertile black soils in China, demonstrates comparatively weaker carbon capture growth. This result supports the conclusions of Liu et al. [38], who documented the rapid degradation of black soils due to decades of monoculture, excessive reclamation, and insufficient investment in soil conservation. Such degradation has reduced soil organic matter, undermining the region’s capacity to function as a long-term carbon sink. Furthermore, climatic conditions exacerbate this trend: rising temperatures in high-latitude areas accelerate soil respiration and carbon release, further weakening sequestration potential [39]. Taken together, the interplay of ecological degradation and climate stress helps explain why the Northeast lags behind despite its apparent natural advantages.

Meanwhile, results from the Dagum Gini coefficient decomposition indicate that inter-regional differences are the dominant source of variation in agricultural carbon capture, accounting for nearly half of the overall disparity. Kernel density estimation further reveals an increasingly dispersed distribution, with clear evidence of polarization. On one side, provinces with favorable agricultural structures, such as diverse crop rotations, conservation tillage, and integrated straw-returning practices are able to consolidate and expand their sequestration advantages. On the other side, provinces constrained by fragile ecosystems, soil erosion, or less adaptive farming systems continue to fall behind. This divergence signals the risk of a “two-speed” development trajectory in agricultural carbon capture, where advanced regions accumulate ecological benefits while vulnerable ones become increasingly marginalized.

Compared with previous literature, which has emphasized the natural determinants of soil carbon [37,38,39,40], the regional differences identified in this study suggest that agricultural management practices and institutional support may also play a critical role. The observed differences across various regions could be partly related to variations in practices such as crop diversification, conservation farming, and low-disturbance tillage, which can enhance the natural sequestration potential of soils. In addition, differences in policy environments such as the availability of ecological compensation schemes, targeted subsidies for green technologies, and the enforcement of soil protection regulations may help explain why some provinces have achieved stronger carbon capture outcomes than others. Therefore, future research should further disentangle and quantify the specific mechanisms through which management practices and policy frameworks shape regional carbon capture capacities.

From a governance perspective, strengthening agricultural carbon capture requires a two-pronged strategy. On the biophysical side, priority should be given to soil restoration projects, crop diversification, and the promotion of organic fertilizers to rebuild soil organic matter. On the institutional side, ecological compensation schemes should be expanded to incentivize farmers in fragile regions, while targeted subsidies and technical assistance can help ensure the adoption of sustainable practices. This is particularly urgent for the Northeast, where ongoing soil degradation threatens both food security and ecological stability. Aligning agricultural modernization with strict soil conservation policies, combined with the scaling-up of precision farming technologies and farmer training programs, will be essential to prevent further decline and to secure the region’s long-term contribution to national carbon neutrality goals.

4.2. Spatial Distribution and Stability of Agricultural Carbon Emissions

The spatial distribution of agricultural carbon emissions exhibits a markedly different pattern compared with carbon capture. A clear “strong west, weak east” trend is observed, with the western region consistently maintaining the highest emission intensity. This outcome is consistent with Zhu et al. [41], who identified the concentration of livestock production in western provinces as a key driver of agricultural emissions. Meanwhile, the observed westward shift in emissions also echoes findings by Li et al. [42], indicating that agricultural emissions in China have gradually relocated from traditional farming regions in the east to pastoral and mixed-farming areas in the west. This reflects broader structural adjustments in Chinese agriculture, as well as the increasing dominance of animal husbandry in western provinces.

Using the Dagum Gini coefficient, the results further demonstrate that intra-regional differences, especially within the western region, contribute significantly to overall inequality. For example, intensive livestock farming in the Sichuan Basin produces far higher emissions than extensive grazing systems in the surrounding plateau areas. This suggests that heterogeneity within regions is as important as differences between regions, and that provincial-level agricultural structures strongly influence overall emission outcomes. The role of technological access is particularly notable, provinces with more advanced manure treatment, feed efficiency improvements, and renewable energy adoption exhibit relatively lower emission intensities despite similar agricultural endowments. Consequently, dynamic analysis using the Markov chain provides additional insights. The results reveal that both high- and low-emission provinces exhibit strong stability, with limited transition between categories. This indicates that once a province develops a certain emission profile, it is difficult to reverse in the short term. Such inertia can be attributed to the path dependence of production systems, the limited diffusion of emission-reduction technologies across regions, and the relatively slow response of agricultural policies to emerging challenges. In other words, technological and institutional stickiness constrains regions’ ability to rapidly adjust their emission trajectories.

Compared with existing research, which has often emphasized input-based factors such as fertilizer use and energy consumption, this study highlights the significance of structural drivers such as livestock density, land use intensity, and the spatial reallocation of agricultural activities [32,41,42]. These findings enrich the understanding of agricultural emissions by underscoring that production modes and technological heterogeneity are equally important determinants alongside traditional input factors.

From a policy perspective, the evidence points to the need for more differentiated emission-reduction strategies rather than a uniform national approach. For western provinces dominated by animal husbandry, priority should be given to accelerating the adoption of advanced manure treatment systems, promoting clean energy substitution in rural areas, and piloting market-based mechanisms such as carbon trading for the livestock sector. In contrast, for eastern provinces with relatively low but stable emissions, the main task is to consolidate existing achievements and prevent rebounds. This can be achieved through stricter control of fertilizer application, continuous improvements in energy efficiency, and the promotion of precision agriculture to avoid waste. In addition, strengthening inter-regional cooperation and knowledge transfer, for example, disseminating emission-reduction technologies from advanced pilot provinces to lagging areas could help narrow the gap and enhance the overall efficiency of agricultural carbon mitigation in China.

4.3. Surplus–Deficit Dual Pattern of Agricultural Carbon Effects

The joint evaluation of agricultural carbon capture and emissions reveals a distinct surplus deficit dual structure across provinces, underscoring the imbalance between carbon sequestration and carbon release in China’s agricultural sector. A total of 23 provinces function as carbon surplus zones, where carbon capture exceeds emissions, while 8 provinces are in persistent deficit. The persistence of surplus provinces is closely associated with improvements in crop structure, adoption of conservation tillage, and promotion of straw returning and biogas utilization, all of which enhance sequestration capacity while reducing emissions [43]. These practices have been widely promoted in eastern and central regions, reinforcing their position as stable surplus zones. The clustering of surplus provinces also reflects the cumulative effects of policy support, relatively favorable ecological conditions, and higher levels of technology adoption.

By contrast, deficit provinces are mostly concentrated in ecologically fragile or highly urbanized regions. In provinces such as Gansu and Qinghai, the prevalence of desertification, limited vegetation coverage, and harsher climatic conditions restrict natural sequestration potential. At the same time, agricultural emissions remain relatively high because of resource-dependent production structures and limited diffusion of emission-reduction technologies. In megacities such as Beijing and Tianjin, the situation is shaped by severe land constraints, limited arable areas, and high production intensity, which together exacerbate their deficit status. These cases highlight how both ecological vulnerability and socio-economic development models can reinforce agricultural carbon deficits, making them harder to reverse without external intervention [29,32]. This surplus–deficit duality has important implications. On the one hand, surplus provinces represent valuable ecological assets that not only contribute to national carbon neutrality goals but also provide potential carbon offset opportunities for deficit regions [44]. Protecting these advantages requires establishing low-carbon agricultural demonstration zones, strengthening long-term ecological monitoring systems, and preventing ecological degradation from undermining existing gains. On the other hand, deficit provinces face the challenge of balancing food production with ecological sustainability [45]. Their deficit status, if left unaddressed, may widen further, thereby intensifying regional disparities and undermining the national strategy for coordinated and sustainable agricultural development.

From a policy perspective, addressing the dual structure calls for multi-level responses. In the short term, interregional coordination mechanisms are urgently needed. Ecological compensation and transfer payments can serve as effective instruments to channel financial and technical resources from surplus to deficit provinces, mitigating imbalances and supporting adaptive capacity. In the medium term, targeted interventions should focus on optimizing production structures such as promoting crop livestock integration, expanding conservation tillage, and scaling up the use of low-carbon agricultural technologies to gradually close the sequestration–emission gap. In the long run, institutional arrangements such as a national agricultural carbon accounting system, market-based trading of agricultural carbon credits, and regional cooperation platforms could create a more resilient framework for balancing agricultural carbon effects across the country. By integrating ecological protection with agricultural modernization, the surplus deficit duality can be progressively alleviated, fostering a more equitable and sustainable agricultural transition in line with China’s broader carbon neutrality ambitions.

5. Conclusions and Implications

Based on panel data from 31 provinces in China spanning 1997 to 2022, this study systematically examined the characteristics of agricultural land use carbon capture, agricultural carbon emissions, and the net carbon effect of land use. Overall, the study reveals significant spatiotemporal disparities in both carbon capture and carbon emissions across regions, with the overall carbon effect exhibiting a “carbon surplus with localized deficits” pattern. The key conclusions are as follows. Firstly, China’s agricultural land use carbon capture increased from 634.19 million tons in 1997 to 888.36 million tons in 2022, with widening regional disparities: the central region led while the northeast lagged, and Dagum Gini decomposition indicated that inter-regional variation was the main source of inequality (average 46.71%). From a dynamic perspective, kernel density estimation revealed multi-modal distribution, and Markov chain analysis suggested strong stability with club convergence. Then, agricultural carbon emissions rose from 422.83 to 456.45 million tons (7.4%), with the west consistently at the highest level; intra-regional differences, especially within the west, contributed most to overall variation (average 40.47%), and their evolution also showed stable club convergence. Lastly, the carbon effect displayed a surplus–deficit dual structure: 23 provinces formed surplus zones (mainly in eastern, central, and southwestern regions), while 8 provinces in ecologically fragile or special-function areas faced persistent deficits.

Drawing on the above findings, the following implications can be derived. First, regional coordination and ecological compensation mechanisms should be strengthened. Given that sequestration advantages are concentrated in the central region while emissions are more prominent in the west, a dynamic monitoring system should be established to identify deficit risk zones in real time. Ecological compensation and fiscal transfers should be provided to deficit provinces, while technical training on practices such as conservation tillage and crop rotation should be expanded to improve sequestration in lagging areas. Second, a unified national agricultural carbon emission trading system should be established. A differentiated quota allocation mechanism, cross-regional carbon transfer accounting, and the inclusion of agricultural CCER projects in the national carbon market can create incentives for emission reduction. By linking carbon responsibility to both production and consumption, and by enabling enterprises to purchase agricultural carbon credits, the system can encourage technological innovation and ensure balanced regional development. Finally, efforts should be made to optimize the carbon effects of agricultural land use. In surplus provinces, technologies such as straw returning, eco-friendly farming, and agroforestry should be scaled up to consolidate sequestration advantages. In deficit areas, ecological restoration measures including reforestation, grassland rehabilitation, and erosion control are critical to increasing carbon sinks. Special-function zones should strictly limit high-emission activities and promote sustainable practices such as water-saving irrigation and organic farming. Together, these measures can enhance regional synergy, narrow the surplus–deficit gap, and support the transition toward low-carbon sustainable agriculture in China.

6. Limitations and Reflection

This study has three main limitations. First, the estimation of land use carbon sequestration only accounts for carbon uptake by natural ecosystems through photosynthesis, which converts CO₂ into organic carbon. It does not consider the potential enhancement of carbon sequestration through agricultural soil carbon management practices. Similarly, agricultural emission estimates exclude rice methane and farmland nitrous oxide. Additionally, this study does not account for the downward trend in emission factors driven by policy interventions, which may result in an overestimation of actual carbon emissions. Moreover, the exclusion of technological advances that optimize these factors could further widen the discrepancy between estimated and actual emissions. Second, the long-term effects of climate warming on carbon capture and sources have not been quantified. Future research could integrate climate-agriculture coupled models to improve predictive accuracy. Moreover, this study does not quantify the dynamic impact of climatic factors on emission factors, which may weaken the accuracy of long-term trend analyses. On the one hand, rising temperatures accelerate the decomposition of soil organic matter, increasing the carbon-emission factor of soil respiration; on the other hand, warming can alter crop growth cycles, thereby indirectly affecting the carbon-uptake rate of crops used in carbon-capture estimates. Third, the study does not differentiate the carbon effects across specific land use types. Follow-up studies could leverage high-resolution remote sensing data to deepen spatial analysis. Furthermore, how to design effective regional coordinated emission reduction mechanisms such as inter-provincial carbon compensation schemes warrants further exploration.