Introduction

Cultivated land serves as the foundation of China’s agricultural production and food security, providing the spatial basis for grain cultivation1. As a key indicator of green transformation in agriculture, the ecological efficiency of cultivated land use is essential for achieving high-quality and sustainable agricultural development2,3. This efficiency reflects the overall capability and sustainability of national agricultural production. The ecological efficiency of cultivated land represents a complex, multidimensional system mainly shaped by five key dimensions2,3,4:

  1. (1)

    Strategic orientation: Centered on ensuring food security, protecting cultivated land reserves, and promoting green development through optimized resource allocation and reduced agricultural carbon emissions5,6;

  2. (2)

    Spatial variability: Manifested through regional disparities in natural endowments (climate, soil quality, water availability) that create distinct geographical patterns of efficiency7;

  3. (3)

    Indirect drivers: Influenced by socioeconomic factors and technological advancements, particularly precision agriculture innovations that optimize production methods8;

  4. (4)

    Policy regulation: Guided by government interventions including incentive mechanisms and support for green technology adoption9,10,11;

  5. (5)

    Dynamic balance: Requiring careful equilibrium between economic returns and ecological preservation for sustainable development12,13.

Despite advancements in irrigation and fertilization technologies aimed at boosting food output, China faces significant challenges4. While grain stocks have steadily increased, leading to lower market prices, the economic benefits of grain crops are diminishing in comparison to cash crops. This discrepancy not only dampens farmers’ enthusiasm but also threatens the long-term sustainability of the food industry14. Additionally, shifts in agricultural production structures have exacerbated “non-agricultural” and “non-grain” land use trends15, highlighting growing imbalances in cultivated land utilization16.

These challenges reveal a critical policy gap in prioritizing ecological efficiency. Many regions continue pursuing yield maximization through intensive chemical inputs, degrading soil health and ecosystems17. This paradox, where short-term production gains undermine long-term sustainability, makes improving ecological efficiency particularly urgent in China’s major grain-producing regions18,19.

China has implemented various policies to address these issues20, including the 2016 Central Rural Work Conference, which emphasized the need for agricultural supply-side reforms focused on increasing land output, resource efficiency, and labor productivity while ensuring food security. Additionally, the 2023 No. 1 Central Document proposed the construction of high-standard farmland to improve land quality and enhance resilience against natural disasters. However, the implementation of these policies has faced challenges, with lax enforcement of land protection measures leading to resource degradation21 and hindering improvements in ecological efficiency22.

In addition, China’s demographic and consumption trends further amplify the urgency of improving ecological efficiency23. Population growth, coupled with a shift toward more resource-intensive foods, has placed increased demand on agricultural land24. As a result, future food security requires not only higher agricultural output but also a reduction in the ecological damage caused by land use practices25. Balancing these dual objectives including ensuring food security while minimizing environmental harm is central to sustainable agricultural development in China.

Despite the importance of ecological efficiency, existing research shows several limitations. Prior studies often use traditional Data Envelopment Analysis (DEA) methods, which struggle to handle zero or negative data and do not fully incorporate slack variables or spatial-temporal variations. Moreover, there is a lack of comprehensive, spatially explicit assessments of ecological efficiency that consider multi-dimensional inputs, desirable outputs, and unintended outputs such as carbon emissions across China’s major grain-producing regions. This limits understanding of regional heterogeneity and temporal dynamics essential for informed policy making.

To fill these gaps, this study employs a super-efficiency EBM model that integrates both radial and non-radial efficiency measures, overcoming the inflexibility of conventional DEA and providing more precise efficiency rankings, especially for regions with input redundancy or excess pollution outputs. The model is applied to 180 prefecture-level cities in 13 provinces and autonomous regions, incorporating a robust index system that includes land, labor, water, mechanization, fertilizers, pesticides, agricultural film, energy, agricultural output, grain production, and carbon emissions as an unintended output.

Furthermore, this study investigates spatial distribution patterns and long-term trends of ecological efficiency in key river basins including the Songhua, Yellow, and Yangtze, highlighting spatial agglomeration, polarization, and convergence-divergence dynamics. These insights provide a foundation for regionally tailored policies that promote sustainable land use and enhance ecological efficiency while maintaining food security.

The conceptual model of this research is presented in Fig. 1.

Fig. 1
figure 1

Conceptual model of this research.

Full size image

Literature review

Impacts of agroecosystems, ecosystem services, and land use

Agroecosystems are fundamental for human survival, providing essential goods such as food, fiber, and fuel26. They rely heavily on ecosystem services, including nutrient cycling, water regulation, and carbon sequestration, which sustain agricultural productivity and environmental health27,28. However, these ecosystem services are frequently undervalued or omitted in traditional market transactions, leading to a disconnect between agricultural output and the ecological processes that support it29. A pressing global challenge is balancing the supply of these services with increasing human demands to ensure economic viability and social sustainability30.

Irrational land use practices such as excessive conversion, fragmentation, or degradation, disrupt vegetation cover and species habitats, posing significant threats to biodiversity and ecosystem integrity31. Increased anthropogenic pressure on land ecosystems has accelerated biodiversity loss, undermining vital services like soil protection and natural self-regulation, which in turn drives land degradation and reduced ecosystem functionality32.

Agricultural intensification, while boosting food production, has exacerbated environmental problems, including soil erosion, agrochemical overuse, and water pollution, thereby degrading landscape quality and ecosystem services33,34. Excessive fertilizer use notably depletes soil health, emphasizing the urgent need to reconcile food production goals with ecological conservation17,35,36.

A critical research gap exists in the lack of integrated approaches that transcend mere quantification and spatial distribution of cultivated land. There is a need to better understand the processes driving land-use change, spatial disparities, and the complex interactions within the “cultivated land–socioeconomic–ecological environment” system37,38. This study addresses this gap by proposing a comprehensive framework to analyze these multifaceted interactions, advancing sustainable agricultural development.

Carbon emission and ecological efficiency of agricultural land

Agriculture is a significant contributor to global greenhouse gas (GHG) emissions, primarily due to carbon emissions from land-use changes and farming practices39. Recent estimates indicate that agriculture and land use activities contribute roughly 70% of CO2 emissions within the global food system40,41. These emissions impair the ecological efficiency of cultivated land by reducing the effectiveness of food production and intensifying environmental degradation.

Extensive research has explored mitigation strategies for agricultural carbon emissions, including substituting organic for chemical fertilizers, improving land management, and promoting biofuel use42,43. Practices such as enhanced farmland management, degraded land restoration, and soil organic matter enrichment have proven effective in reducing emissions and enhancing carbon sequestration44,45.

However, the expansion of biofuel production raises sustainability concerns, particularly when it involves land conversion that may increase net emissions unless biofuel crops are cultivated on existing farmland46,47. This highlights a gap in understanding the net environmental impacts of biofuel policies. Our study addresses this by exploring alternative strategies focused on land restoration and soil management to sustainably increase productivity while mitigating emissions.

The need for resource use and sustainable agricultural practices

The inefficient use of agricultural resources is a major contributor to environmental degradation, particularly through the irrational application of nitrogen fertilizers and agricultural plastics42,48. For example, global nitrogen use efficiency for cereal crops has declined from approximately 80% in the 1960s to around 25% by 2000, due largely to disproportionate increases in nitrogen application relative to yield gains38,49. In addition to nitrogen fertilizers, the use of agricultural plastic film, while providing short-term economic benefits, contributes to the accumulation of microplastics in the soil50,51.

Prior studies indicate that sustainable agricultural practices are effective in mitigating the negative impacts of resource overuse, with organic farming serving as a notable example14,29,52,53. Organic farming employs various strategies to minimize ecological impacts, sustain soil and ecosystem health, and reduce reliance on non-renewable resources54. This research gap highlights the importance of improving resource use efficiency and adopting sustainable practices to ensure long-term agricultural sustainability. This study contributes by evaluating how sustainable agricultural practices can reduce resource inefficiency and environmental degradation, offering practical insights for enhancing ecological sustainability in land use.

Measurement and drivers of land use eco-efficiency

To enhance the ecological efficiency of cultivated land, it is essential to establish a comprehensive measurement system that integrates technological and environmental factors, particularly, an accurate measurement framework is crucial for assessing the interplay between economic, social, and ecological benefits55.

Researchers have employed various methodological frameworks, such as Data Envelopment Analysis (DEA) and Stochastic Frontier Analysis (SFA), to assess economic, environmental, and ecological efficiency in agricultural production across different regions56. Many of these studies incorporate indicators of cropland performance alongside ecosystem service metrics to identify optimal eco-efficiency outcomes57. In addition, some scholars have explored the ecological aspects of cropland use by examining practices that mitigate soil degradation and promote soil restoration. These findings suggest that effective cultivation strategies can simultaneously support ecological balance and economic growth58,59.

The drivers of eco-efficiency are complex and multifactorial. Rational use of chemical inputs plays a critical role; reducing the overuse of fertilizers and pesticides has been shown to substantially improve ecological outcomes. Moreover, improved management of agricultural resources and the adoption of ecologically sustainable practices contribute to enhanced eco-efficiency, food security, and lower carbon emissions2,39,60. However, the gap remains in integrating these diverse factors into a unified framework that assesses both the drivers and outcomes of eco-efficiency in land use. This study addresses this gap by employing a comprehensive index system that incorporates both input factors (e.g., labor, water, and fertilizers) and unintended outputs (e.g., carbon emissions) to assess land use sustainability comprehensively.

This study addresses this gap by implementing a comprehensive index system that incorporates a broad range of input factors such as labor, water, fertilizers, pesticides, machinery, and energy, as well as unintended outputs, particularly agricultural carbon emissions. This integrated approach enables a more accurate and holistic assessment of cultivated land use sustainability.

In conclusion, this study specifically seeks to address the research gaps below:

  1. (1)

    Spatial and temporal trends in ecological efficiency: By developing a comprehensive approach to analyzing the spatiotemporal dynamics of ecological efficiency across China’s major agricultural regions.

  2. (2)

    Policy and regional characteristics: By investigating the influence of policy interventions on ecological efficiency, with particular attention to regional differences in natural endowments such as climate and water availability.

  3. (3)

    Integrated measurement framework: By introducing a unified and data-driven index that captures both resource inputs and unintended environmental outputs (e.g., carbon emissions) for assessing land use sustainability.

By addressing these gaps, this study contributes significantly to the academic literature on sustainable land use. It provides practical insights for improving the ecological efficiency of cultivated land in China and supports the development of policies aimed at promoting sustainable agricultural practices.

Materials and methods

Data source and preprocessing

This study evaluates the ecological efficiency of cultivated land use in China’s 13 major grain-producing provinces (grouped into the Yangtze, Yellow, and Songhua River basins), which contributed 77.9% of national grain output in 2023. The analysis covers 180 prefecture-level cities from 2011 to 202061.

Data source includes:

  1. 1)

    Land-use data: Derived from the CLCD dataset (Fig. 2), with 2020 land classifications validated against remote sensing imagery.

  2. 2)

    Economic/social data: Collected from provincial statistical yearbooks, bulletins, and the China Rural Statistical Yearbook. Missing values were imputed using linear interpolation or moving averages (for trends).

  3. 3)

    Carbon emissions: Calculated using the IPCC tier-1 method with region-specific emission factors for fertilizers, pesticides, and energy use.

1)

Data preprocessing, especially the calculation of carbon emissions is shown in Table 1.

Table 1 Carbon sources, carbon emission coefficients, and references for cultivated land use.
Full size table

The research highlights the close relationship between river basin divisions and grain-producing regions. Variations in climate, soil, and water resources across basins create unique opportunities and challenges for grain production. Based on geographical characteristics and State Council guidelines, the study categorizes 13 provinces into three river basins: the Yellow River basin (Hebei, Henan, Inner Mongolia, Shandong), the Yangtze River basin (Jiangxi, Anhui, Hubei, Hunan, Jiangsu, Sichuan), and the Songhua River basin (Heilongjiang, Jilin, Liaoning), covering 180 prefecture-level cities.

Figure 2 presents a 2020 land-use map of these areas, sourced from the CLCD dataset. Economic and social data were collected from provincial statistical yearbooks, bulletins, and rural statistics, with missing data estimated using interpolation or trend methods. Administrative changes, such as Laiwu City’s merger into Jinan City in 2019, required data adjustments for consistency. Similarly, Heilongjiang’s 2018 agricultural reclamation reform affected Shuangyashan City’s data, necessitating the use of sown crop areas as a proxy for ploughing data.

The agricultural labor force was calculated as follows61: Employees in the primary industry × (agricultural output value/total output value of agriculture, forestry, animal husbandry, and fishery). To account for price fluctuations, the agricultural output value was adjusted using the agricultural product price index, with 2011 as the base year. These methods ensure data consistency and reliability for analyzing ecological efficiency trends.

Fig. 2
figure 2

Land-use map of major grain producing areas in China in 2020. Note: (1) Fig. 2 was created using ArcGIS 10.8.2 visualization software (link: https://soft.youxishen.cn/soft/106402.html?bd_vid=13452434789428261120); (2) The base map was sourced from the Standard Map Service of China’s Ministry of Natural Resources (Approval No.: GS(2019)1822) and remains unmodified; (3) Land-use data in Fig. 2 comes from the CLCD (China Land Cover Dataset), Reference: Yang, J. & Huang, X. (2021). The 30 m annual land cover dataset and its dynamics in China from 1990 to 2019. Earth System Science Data, 13, 3907–3925.

Full size image

Research methods

Non-expected output super-efficiency EBM model

DEA has been widely used to measure efficiency in systems characterized by input–output relationships. Traditional DEA models are typically categorized into radial models, such as the CCR model (assuming constant returns to scale) and the BCC model (assuming variable returns to scale), and non-radial models, with the Slack-Based Measure (SBM) model being a representative example.

Radial models require all input and output factors to change proportionally during calculation, which overlooks the effects of non-radial slack variables and fails to account for the negative externalities of undesirable outputs, deviating from the complexities of real-world economic systems. In contrast, non-radial models like SBM incorporate slack variables, but often at the cost of losing proportional information between expected and actual input-output values.

To address these shortcomings, Tone and Tsutsui (2010) introduced the Epsilon-Based Measure (EBM) model. This model effectively integrates both radial proportional relationships and non-radial slack variables across different input and output indicators. As a result, it mitigates the bias associated with single-distance function approaches and improves the comparability among Decision Making Units (DMUs).

When evaluations involve multiple inputs and outputs, the number of DMUs identified as efficient often increases, complicating the comparison of performance among them. To further distinguish the efficiency levels among these efficient units, this study incorporates the super-efficiency DEA model proposed by Andersen and Petersen (1993) into the EBM framework, forming a super-efficiency EBM model.

In the context of agricultural production, efficiency must account not only for positive outputs, such as grain yield and total agricultural output, but also for negative externalities, notably CO₂ emissions arising from processes such as planting, irrigation, fertilization, and crop growth. Incorporating the concept of undesirable outputs62, this study treats carbon emissions from grain production as an undesirable output within the super-efficiency EBM framework.

The super-efficiency EBM model63 with undesirable outputs offers a more nuanced evaluation of DMU performance by simultaneously capturing radial and non-radial inefficiencies, reducing the limitations of traditional DEA models. This enhanced model is particularly well-suited for agricultural systems, where the balance between productivity and environmental impact is critical64. In this study, we consider both positive outputs and negative externalities, particularly carbon emissions generated during food production. These emissions are treated as unintended outputs65, shown in Function (1) as follows.

$$\begin{gathered} {\gamma ^*}=\hbox{min} \frac{{\theta - {\varepsilon ^ - }\sum\limits_{{i=1}}^{m} {\frac{{w_{i}^{ - }s_{i}^{ - }}}{{{x_{ik}}}}} }}{{\varphi +{\varepsilon ^+}\left( {\sum\limits_{{r=1}}^{s} {\frac{{w_{r}^{+}s_{r}^{+}}}{{{y_{rk}}}}+\sum\limits_{{p=1}}^{q} {\frac{{w_{p}^{{u - }}s_{p}^{{u - }}}}{{{y_{pk}}}}} } } \right)}} \hfill \\ s.t.\left\{ {\begin{array}{*{20}{c}} {\sum\limits_{{j=1}}^{n} {{\lambda _j}{x_{ij}}} +s_{i}^{ - }=\theta {x_{ik}},i=1,2,...,m} \\ {\sum\limits_{{j=1}}^{n} {{\lambda _j}{y_{ij}}} - s_{r}^{+}=\phi {y_{rk}},r=1,2,...,s} \\ {\sum\limits_{{j=1}}^{n} {{\lambda _j}{u_{pj}}} +s_{p}^{ - }=\phi {u_{pk}},p=1,2,...,q} \\ \begin{gathered} j \ne {j_0} \hfill \\ j=1,2,...,n \hfill \\ \end{gathered} \\ {{\lambda _j} \geqslant 0,s_{i}^{ - },s_{r}^{+},s_{p}^{ - } \geqslant 0} \end{array}} \right. \hfill \\ \end{gathered}$$
(1)

Kernel density estimates

Kernel Density Estimation (KDE), also known as the Parzen window method, estimates the probability density function of random variables without requiring assumptions about the data distribution. This approach highlights the intrinsic structure of the sample data and effectively captures its distribution dynamics. KDE is particularly valuable in economic analysis, as it allows for the examination of spatial imbalances. By generating a smooth, continuous density curve, KDE illustrates regional dynamics, distribution trends, and polarization effects3.

Kernel density estimation of space

Spatial kernel density estimation enhances traditional kernel density estimation by incorporating spatial factors, allowing for a more detailed analysis of data distribution. This method employs a conditional kernel density approach to evaluate cropland eco-efficiency over time and space, revealing spatial effects and long-term trends. The relevant formula for this analysis is as follows:

$$g\left( {\left. y \right|x} \right)=\frac{{f\left( {x,y} \right)}}{{f\left( x \right)}}$$
(2)
$$f\left( {x,y} \right)=\frac{1}{{n{h_x}{h_y}}}\sum\limits_{{i=1}}^{n} {{K_x}\left( {\frac{{{X_i}-x}}{{{h_x}}}} \right)} {K_y}\left( {\frac{{{Y_i}-x}}{{{h_y}}}} \right)$$
(3)

Where \(g\left( {\left. y \right|x} \right)\) denotes the distribution state of \(\:y\) under the condition \(\:x\), \(f\left( {x,y} \right)\) represents the joint kernel density function; and\(f\left( x \right)\) denotes the marginal kernel density function of \(\:x\).

Dagum Gini coefficient

The analysis utilizes the Dagum Gini coefficient and its decomposition to examine the spatial differentiation and sources of ecological efficiency in arable land use across China’s major grain-producing regions. The overall difference G is decomposed into three components: intra-group differences \(\:{G}_{w}\), net inter-group differences \(\:{G}_{nb}\), and the contribution of supervariable density\(\:\:{G}_{t}\)62. These components reflect the internal differences within the major grain-producing regions of the Yangtze River Basin, Songhua River Basin, and Yellow River Basin, as well as the net differences between these regions and the contribution rates of cross terms among subgroups to the overall difference, the three satisfy the relationship: \(G=Gw+Gnb+Gt\).

Construction of index system

This paper constructs a comprehensive index system to evaluate the ecological functions of cultivated land, focusing on three key dimensions: agricultural production, socioeconomic benefits, and ecological functionality. Building on this framework, an input-output model is developed to capture the complex interactions among resource inputs, socioeconomic outputs, and the ecological environment.

Input indicators (land, capital, labor, energy) reflect agricultural production inputs. Expected outputs include economic benefits (total agricultural output value) and social benefits (grain output, critical for food security, competitiveness, and stability).

In terms of undesirable outputs, this study addresses the challenges associated with measuring agricultural non-point source pollution, which remains highly uncertain and lacks a unified standard66. Moreover, the loss coefficients of agrochemicals such as pesticides and fertilizers are significantly influenced by factors like topography, soil type, and crop variety, making generalized estimations problematic. Therefore, this paper selects carbon emissions from cultivated land use, particularly relevant under the dual carbon goals, as the indicator of undesirable output. The full evaluation index system is presented in Table 2.

Table 2 Evaluation index system for the ecological efficiency of cultivated land use.
Full size table

To measure CO2 emissions associated with economic activities, such as the use of fertilizers, pesticides, agricultural films, machinery, irrigation, and ploughing during agricultural production, we apply a systematic method for quantifying carbon emissions from cultivated land use with the estimation formula as follows:

$$E=\sum {Ei} =\sum {Ti{\text{ }} \times {\text{ }}} \delta i$$
(4)

Where\(\:\:E\) represents the total carbon emissions, \(\:{E}_{i}\:\)denotes the carbon emissions from various sources, \(\:{T}_{i}\) refers to original amount of each type of carbon emission source, and \(\:{\delta\:}_{i}\) is the carbon emission coefficient of each type of carbon source. According to the main sources and referring to existing studies67, the carbon emission coefficients of chemical fertilizer, pesticide, agricultural film, agricultural diesel oil, agricultural irrigation and arable land ploughing are 0.8956 kg/kg, 4.9341 kg/kg, 5.18 kg/kg, 0.5927 kg/kg, 18.51 kg/hm2 and 312.6 kg/km2, respectively (shown in Table 1).

Results

Ecological efficiency of cultivated land in china’s grain heartlands

Spatio-temporal variation law

Overall, the eco-efficiency of cultivated land use in major grain-producing areas of China increased from 2011 to 2020, following a trend of “long-term increase and short-term decline”. The average eco-efficiency rose from 0.561 to 0.647 before dropping slightly to 0.636, indicating that there is still significant room for improvement. Scale efficiency consistently exceeded both comprehensive efficiency and pure technical efficiency, showing a relatively stable but slight decline. The trends in pure technical efficiency closely mirrored those of comprehensive efficiency, with a correlation coefficient of 99.75%. This strong correlation suggests that pure technical efficiency significantly influences changes in comprehensive efficiency and plays a crucial role in overall eco-efficiency (see Fig. 3).

Fig. 3
figure 3

Cultivated land eco-efficiency: mean values and decomposition.

Full size image

Despite a modest increase in total grain production in 2020, the level of ecological efficiency experienced a temporary decline, decreasing by 1.7% from the previous year. This downturn may be attributed to the impacts of COVID-19, which disrupted the circulation of production inputs and curtailed food supply and consumption, negatively affecting production activities.

Significant regional differences in ecological efficiency were observed across China’s grain-producing areas. The eco-efficiency of cultivated land use in the Songhua River Basin was notably higher than in both the Yellow River Basin and the Yangtze River Basin, with an average annual eco-efficiency of 0.718. Within the Songhua River Basin, Heilongjiang Province exhibited the highest eco-efficiency (0.797), surpassing Liaoning Province (0.674) and Jilin Province (0.673). Notably, cities like Siping and Baishan in Jilin maintained consistently high efficiency values, with Siping exceeding 0.95 each year throughout the study period.

In the Yangtze River Basin, eco-efficiency was slightly higher than in the Yellow River Basin, with an average annual efficiency of 0.587. From 2011 to 2018, eco-efficiency increased consistently, with a slight decline in 2016, but began to show a downward trend after 2018. Jiangsu Province led the Yangtze River Basin with an average eco-efficiency of 0.663, followed by Sichuan (0.654) and Hubei (0.609). Cities like Guang’an and Ya’an achieved high efficiency values consistently, indicating positive trends in efficiency improvements.

The Yellow River Basin exhibited the lowest ecological efficiency in cultivated land use, with an average annual efficiency of 0.559. Although there was a brief decline in efficiency in 2017, the efficiency level improved significantly during the study period, rising from 0.511 to 0.609. By 2020, the eco-efficiency in the Yellow River Basin surpassed that of the Yangtze River Basin. Among the provinces in the Yellow River Basin, Shandong ranked highest with an average annual eco-efficiency of 0.575, followed closely by Henan (0.571), Inner Mongolia Autonomous Region (0.544), and Hebei (0.534) (see Fig. 4).

Fig. 4
figure 4

Eco-efficiency of cultivated land in major grain-producing areas (2011–2020).

Full size image

Decomposition of inefficiency sources

A deeper analysis of the sources of inefficiency reveals substantial redundancy in the input factors for cultivated land use in China’s grain-producing areas. The redundancy and insufficiency rates, calculated by analyzing the ratio of relaxation variables to original values of each index, provide insight into the areas contributing to inefficiency (see Table 3).

Table 3 Input-output redundancy and deficiency rates (%) in key grain regions.
Full size table

The overall input indices for cultivated land in China’s major grain-producing areas exhibit significant redundancy, contributing to high levels of carbon emissions. The overall deficiency rate for expected output is relatively low, and the redundancy and deficiency rates for each index have generally declined over the study period, albeit at varying paces. Notably, the redundancy rates for land, irrigation, mechanization, fertilizer, and agricultural film saw a slight increase in 2020. Among the input factors, labor redundancy remains the highest, followed by fertilizer redundancy. The Songhua River Basin stands out for having a significantly lower redundancy rate compared to the other regions, while the Yangtze River Basin has a notably lower energy redundancy rate.

Despite carbon emissions continuing to be a limiting factor for improving ecological efficiency, the conflict between carbon emissions and eco-efficiency is decreasing. This aligns with findings from the Report on Low-carbon Development in China’s Agriculture and Rural Areas 2023, which highlights the declining carbon emission intensity of major crops in recent years.

Spatial patterns of ecological efficiency

Spatial distribution pattern

To illustrate the dynamic evolution of local spatio-temporal differentiation in the ecological efficiency of cultivated land use in major grain-producing areas, we selected data from 2011, 2014, 2017, and 2020. Using the ArcGIS 10.6 visualization platform and the natural break point method, we classified the ecological efficiency into four levels: high efficiency, second-high efficiency, second-low efficiency, and low efficiency (see Fig. 5).

Fig. 5
figure 5

Spatial distribution of eco-efficiency in cultivated land use. Note: Figure 5 was created using ArcGIS 10.8.2 visualization software (official link: https://soft.youxishen.cn/soft/106402.html?bd_vid=13452434789428261120)

Full size image
Analysis of trends and regional differences

The overall distribution pattern across regions is relatively stable, but significant local evolution is evident, displaying a characteristic of high versus low agglomeration. High- and second-high-efficiency areas are primarily concentrated in the Northeast Plain, with some scattered regions in the Chengdu Plain and Hetao Area. The second-high-efficiency areas are also clustered in the North China Plain, Yangtze River Delta, and Chengdu Plain, while low-efficiency areas are mostly contiguous in the Inner Mongolia Autonomous Region and sporadically located in other areas.

Regional variations and influencing factors

These regional differences may be attributed to local climatic, economic, and policy factors68. For example, Heilongjiang Province benefits from its black soil resources, which are conducive to high ecological efficiency. Similarly, the Songhua River Basin’s higher eco-efficiency is largely influenced by more favorable natural conditions and effective policy interventions compared to the Yellow and Yangtze River basins, which face greater challenges such as water scarcity and soil degradation. Analysis reveals that technological innovation, resource efficiency, and local policies drive regional variations. For example, Sichuan’s high efficiency in cities like Ya’an and Guang’an reflects successful adoption of advanced farming techniques and resource-sharing.

Differences and sources of Spatial structure

To better understand the spatial disparities in eco-efficiency across major grain-producing areas of China, we used the Dagum Gini coefficient and its decomposition method (see Figs. 6 and 7). The Gini coefficient for eco-efficiency in cultivated land use has remained relatively stable, averaging 0.133 over the study period. Specifically, the coefficients for the Songhua River Basin, Yangtze River Basin, and Yellow River Basin are 0.145, 0.120, and 0.103, respectively. Notably, the Songhua River basin exhibits the highest Gini coefficient among these regions. The Gini coefficient for efficiency in the Songhua River basin displays a pronounced “M” shape, indicating significant fluctuations. In contrast, the Yangtze River basin shows an inverted U-shaped trend with relatively minor fluctuations. The Yellow River basin, with the lowest Gini coefficient, demonstrates a “V-shaped” change trend, suggesting an increase in intra-regional disparities.

Fig. 6
figure 6

Trend of overall and intra-regional differences in eco-efficiency.

Full size image
Fig. 7
figure 7

Trend of net differences in eco-efficiency among regions.

Full size image

The Gini coefficient for eco-efficiency in the Yellow and Yangtze River basins averaged 0.114, indicating strong efficiency coordination within these regions. However, the coefficients for the Yangtze-Songhua and Yellow-Songhua basins were 0.163 and 0.167, respectively, revealing fluctuations and a declining trend over time.

The spatial differences in eco-efficiency of cultivated land use can primarily be attributed to inter-regional disparities, including net differences and super-variable density. From 2011 to 2015, net differences accounted for 36.18% of the spatial disparity, which narrowed to 25.61% by 2020. This reduction can be attributed to improved efficiency in the Yellow and Yangtze River Basins, which benefited from better transportation infrastructure, technological advancements, and more integrated regional policies. These factors facilitated stronger regional connections and contributed to more even distribution of resources and productivity improvements.

The improvement in eco-efficiency across these basins can be linked to both policy interventions and technological advances. For instance, policies promoting sustainable agricultural practices and better resource management in the Yangtze and Yellow River basins have proven effective in improving eco-efficiency. From an economic perspective, further investments in technology and infrastructure are essential to continue bridging the efficiency gap between regions. It is important for policymakers to focus on enhancing collaboration between regions with varying efficiency levels to ensure the sustainable development of agriculture in China.

From 2016 to 2020, the contribution of super-variable density to spatial differences rose to 33.21%, making it the primary driver of regional disparities. This shift occurred as high-efficiency areas in lower-efficiency regions began outperforming low-efficiency areas in higher-efficiency regions. By 2020, cities with high eco-efficiency in regions with traditionally low efficiency, such as Wuhai, Panzhihua, and Nanjing, saw larger efficiency gains compared to inefficient cities in more efficient regions. This trend highlights the potential for these cities to influence surrounding areas positively, contributing to regional cohesion and economic integration.

Spatial disparities in agricultural eco-efficiency are a global phenomenon, with regions like India and Brazil underperforming despite high potential due to infrastructure gaps. Comparative studies between China and other agricultural leaders could reveal how policies and economic models shape eco-efficiency. Facilitating exchanges between China’s grain-producing regions across river basins may help optimize disparities, align policies, and boost national production.

Dynamic evolution and long-term trend of cultivated land use efficiency

Dynamic shifts in cultivated land ecological efficiency

To further explore the differences, we applied Kernel density estimation (KDE) to examine the dynamics of eco-efficiency and regional disparities (see Fig. 8). The KDE maps reveal significant regional polarization over the period, with most regions initially exhibiting low ecological efficiency. However, from 2011 to 2019, eco-efficiency improved steadily across the major grain-producing regions, leading to a gradual reduction in regional disparities. Despite a slight decline in 2020, there were signs of convergence in the Yangtze River Basin, which showed a pronounced polarization initially but exhibited dynamic convergence thereafter.

Fig. 8
figure 8

Dynamic evolution of cultivated land use eco-efficiency distribution.

Full size image

Overall, the pattern in these areas was polarized, with most regions exhibiting low ecological efficiency in cultivated land use. However, eco-efficiency improved steadily from 2011 to 2019, leading to a gradual reduction in regional disparities, although there was a slight decline in 2020. In the Yangtze River Basin, polarization was pronounced, but the region showed signs of dynamic convergence. The Songhua River Basin exhibited relatively high eco-efficiency, following a “double peak-single peak-double peak” trend, indicating an increasing disparity in regional efficiency levels. Conversely, the Yellow River Basin experienced a transition from “single-peak” to “multi-peak” to “double-peak,” with initial convergence followed by differentiation; the inter-regional efficiency gap narrowed before increasing again.

Long-term transfer trends of cultivated land use eco - efficiency

Static kernel density estimation reflects spatial correlations and convergence patterns among cities during the same period, while dynamic estimation incorporates time, demonstrating how neighboring areas’ ecological efficiency impacts a region’s efficiency over time. In the kernel density maps, the X-axis and Y-axis represent eco-efficiency values of neighboring and local areas, respectively, while the Z-axis indicates kernel density. Density contours represent varying density values, with denser areas signifying greater changes in probability density69.

  1. 1)

    Static Kernel Density Estimation (shown in Fig. 9): The density contours predominantly align along the positive 45° diagonal, indicating that eco-efficiency in major grain-producing areas does not converge. Instead, there is a pattern of high and low agglomeration, with a positive spatial correlation and clear segmentation (0.4 and 0.8 as key demarcation points). Areas with efficiency below 0.4 have minimal impact on regional eco-efficiency. However, neighboring areas with efficiency between 0.4 and 0.8 exhibit a positive spatial correlation, facilitating a diffusion effect through the flow of resources, technology, and capital, ultimately enhancing collaborative ecological efficiency. In contrast, when neighboring areas exceed an efficiency of 0.8, regions must improve their eco-efficiency through internal resource optimization, industrial upgrades, and technological innovation.

  2. 2)

    Dynamic Kernel Density Estimation (shown in Fig. 10): The kernel density maps considering time factors show general similarities but slight differences. Overall, the spatial correlation between regions diminishes over time, and spatial spillover effects are weak for both low (below 0.4) and high efficiency levels (above 0.9). Conversely, regions with efficiency levels between 0.4 and 0.9 benefit from positive spatial spillover effects, positively impacting the improvement of ecological efficiency in cultivated land use.

Fig. 9
figure 9

Static Kernel density and density contour of spatial conditions for the distribution of ecological efficiency in cultivated land use.

Full size image
Fig. 10
figure 10

Spatial static Kernel density and density contours of cultivated land use eco-efficiency distribution.

Full size image

In conclusion, promoting more equitable distribution of resources, technology, and policy interventions can significantly improve eco-efficiency in China’s grain-producing regions. This approach has both national and international implications, as similar agricultural efficiency disparities are seen globally. Moving forward, fostering inter-regional cooperation, technology sharing, and policy harmonization will be essential to address these disparities and optimize the sustainability of agricultural production.

Conclusion

This research introduces an innovative approach by adopting a super-efficiency EBM model that integrates both radial and non-radial perspectives. By incorporating weight parameters to balance the contributions of different input factors, our model not only resolves the issue of variable redundancy but also aligns more closely with the practical requirements of evaluating the ecological efficiency of cultivated land use.

The selection of the non-desirable super-efficiency EBM model presents several key advantages over the super-efficiency SBM model which has been widely applied in existing literature. These advantages include greater comprehensiveness in addressing input-output factors, improved comparability of effective decision-making units, enhanced flexibility in adapting to different data types and business scenarios, and greater accuracy and reliability. This methodological innovation allows for a more nuanced and robust assessment of ecological efficiency, making it better suited to the complexities of land use evaluation.

Key findings

This study analyzes data from 180 prefecture-level cities across 13 provinces (autonomous regions) in China’s major grain-producing areas. Using the super-efficiency EBM model, which incorporates non-desired outputs, it estimates the ecological efficiency of cultivated land use under carbon emission constraints. The research examines spatio-temporal patterns, agglomeration degrees, dynamic evolution, and long-term trends. Key findings include:

  1. 1)

    Temporal trends: From 2011 to 2020, ecological efficiency showed a pattern of “long-term increase and short-term decline,” with significant room for improvement. Technological advancements were the primary driver, contributing 99.75% to efficiency gains. The Songhua River Basin had the highest average efficiency (0.718), surpassing the Yellow River Basin (0.559) and Yangtze River Basin (0.587). Regional trends varied: the Songhua Basin exhibited an “M”-shaped trend, the Yangtze Basin an inverted U-shape, and the Yellow Basin a V-shape, reflecting growing disequilibrium.

  2. 2)

    Carbon emissions impact: Carbon emissions remain a constraint on efficiency, though their influence is diminishing. Improving efficiency requires more than reducing resource inputs or expanding food acreage. Strategies should focus on resource conservation, technological advancement, and environmentally friendly practices, such as optimizing fertilizer and pesticide use, to promote sustainable agriculture.

  3. 3)

    Spatial dynamics: Ecological efficiency displays a complex spatial structure with areas of high and low concentration. Long-term evolution reveals a bipolar pattern, with intermediate efficiency regions (0.4 to 0.9) showing positive spatial spillover effects. The Songhua Basin exhibited “bimodal” evolution, the Yangtze Basin displayed polarization and convergence, and the Yellow Basin shifted from convergence to differentiation, highlighting changing regional disparities.

  4. 4)

    Sources of spatial differences: Spatial differences primarily arise from inter-group disparities. From 2011 to 2015, net variations between groups accounted for 36.18% of differences. However, from 2016 to 2020, high variable density within groups became dominant, contributing 34.8%. This shift reflects varying policy effectiveness and implementation across regions.

Implications

The key implications from this research include:

  1. 1)

    Improvement potential: China’s major grain-producing areas have significant potential to improve ecological efficiency. Policymakers should focus on implementing strategies tailored to local conditions. It is crucial to establish robust land conservation measures to prevent the conversion of farmland to non-agricultural uses and to mitigate the degradation of existing prime farmland. Investing in high-standard farmland and enhancing agricultural production conditions will create a foundation for sustainable practices that ensure both food security and ecological health;

  2. 2)

    Technological advancements: Enhancing ecological efficiency requires not only strategic planning but also substantial advancements in technological innovation. Increased investment in regional agricultural research and the dissemination of advanced production technologies are essential. Integrating modern tools–such as drones, smart irrigation systems, and big data platforms–can optimize resource utilization, reduce labor costs, and improve overall efficiency. Building a strong support network for new farmers and technologists will facilitate the adoption of these innovations, ushering in a new era of agricultural development;

  3. 3)

    Sustainable resource use: Improving resource use and reducing carbon emissions are vital for sustainable agricultural practices. Promoting land transfer can effectively integrate fragmented farmland, enhancing management and mechanization. This approach addresses the issue of land abandonment due to labor shifts to the non-agricultural sector and encourages large-scale modern farming practices that can boost productivity. Facilitating this transition is essential for improving the ecological efficiency of arable land and achieving long-term sustainability.

Limitations

The limitations of this paper are as follows:

  1. 1)

    Data constraints: This study covers 2011–2020, which may miss recent ecological efficiency trends, including potential impacts from the COVID-19 pandemic. Given that environmental and economic conditions have likely shifted since then, future research should incorporate more recent data to provide a comprehensive view of current trends in eco-efficiency.

  2. 2)

    Model assumptions: The analysis assumes that carbon emissions are the primary limiting factor to efficiency, though other external factors, such as climate change and major policy shifts, may also play significant roles in influencing agricultural practices and eco-efficiency metrics. Future research should explore how these additional factors, including regional environmental and socio-economic conditions, may impact ecological efficiency.

  3. 3)

    External context: While spatial kernel density estimation was used to assess regional dynamics, this study lacks a deeper examination of inter-regional policy, socio-economic contexts, and their interplay with eco-efficiency. Future research should incorporate a more detailed analysis of these external factors to better understand why certain regions outperform others in terms of eco-efficiency. This would allow for more targeted policy interventions that can help reduce inefficiencies in underperforming regions.

Future research directions

Given the limitations outlined above, future research can explore the following directions:

  1. 1)

    Recent trends: Incorporating more recent data, particularly from the post-2020 period, to capture the impacts of ongoing policy reforms, technological advancements, and external shocks like the COVID-19 pandemic on ecological efficiency.

  2. 2)

    Broader factors: Expanding the model to account for additional variables such as climate change, policy shifts, and regional socio-economic changes that may influence eco-efficiency beyond carbon emissions. This would help paint a more holistic picture of the forces driving agricultural sustainability.

  3. 3)

    Policy analysis: Conducting more in-depth studies on the role of regional policy interventions in shaping eco-efficiency trends70 By examining case studies of regions with varying policy landscapes, future research could identify best practices and more effective approaches to policy implementation.

  4. 4)

    Comparative studies: Broadening the scope of the research to include comparative analyses with other countries, which would provide a more global perspective on eco-efficiency in agricultural land use. Such studies could reveal common trends and strategies that could be applied to China’s context.