Is the Ratoon Rice System More Sustainable? An Environmental Efficiency Evaluation Considering Carbon Emissions and Non-Point Source Pollution

Abstract

The sustainability of rice-cropping systems hinges on balancing resources, output, and environmental impacts. China is revitalizing the ancient ratoon rice (RR) system for input savings and environmental benefits. Prior research has explored the RR system’s performance using various individual indicators, but few studies have focused on its overall balance of these factors. Environmental efficiency (EE) analysis addresses this gap. Using field survey data from Hunan Province in China and the slacks-based data envelopment analysis method, we quantified the EE of the RR, double-season rice (DR), and single-season rice (SR) systems. Key findings include: (1) the RR system outperforms in carbon emissions and non-point source pollution; (2) the RR system’s EE is 0.67, significantly higher than the DR (0.58) and SR (0.57) systems, indicating superior performance; and (3) despite its relatively high EE, the RR system can still improve, mainly due to input redundancy and production value shortfall. These findings provide strategies for optimizing RR systems to enhance agricultural sustainability.

1. Introduction

Transitioning to sustainable agricultural systems is critical for mitigating environmental issues and achieving long-term sustainability [1]. While essential for sustaining life and fostering economic development, agricultural production significantly impacts environmental sustainability. The Intergovernmental Panel on Climate Change (IPCC) reports that agricultural and land-use activities contributed to 23% of net anthropogenic greenhouse gas emissions between 2007 and 2016 [2]. Agriculture, particularly that which uses intensive methods, is also a major source of non-point source pollution (NSP) due to the inefficient use of inputs such as fertilizers [3]. With the global population expected to reach 10 billion by 2050, necessitating a 50% increase in food production from 2010 levels, the environmental pressure from agriculture is set to intensify [2]. This increasing demand poses a significant challenge, especially for developing countries, which need to balance meeting substantial food requirements with environmental sustainability. Agricultural strategies must ensure a basic equilibrium between supply and demand within the constraints of the available resources, carefully considering the potential environmental impacts [4].

In light of these challenges, the cultivation of rice, the predominant staple crop globally, becomes a focal point for the development of sustainable agricultural systems. Rice is primarily cultivated using two systems: single-season rice (SR) and double-season rice (DR) systems. SR refers to a cultivation practice where rice is grown in a single crop per year, typically in areas with limited water resources or shorter growing seasons. DR refers to a cropping system where two rice crops are grown per year, often in areas with abundant water and longer growing seasons, allowing for multiple harvests.

In China, the ancient ratoon rice (RR) system is being revitalized for its input savings and potential environmental advantages as it can compensate for the yield reduction caused by shifting from DR to SR with minimal environmental impacts [5]. RR is grown from the seedlings of buds that remain at the nodes of rice stubble following the harvest of the main rice crop [6]. Compared to DR, RR can be harvested twice annually without the need for secondary tillage, seedling cultivation, and transplanting, thereby reducing the inputs of agrochemicals and labor [7]. For instance, based on the survey data, the fertilizer input per unit yield for RR is observed to be 14.40% less than that for DR. Similarly, the labor input per unit yield is reduced by 3.70%, and the mechanical input is decreased by 35.80% in comparison to DR. Compared with SR, RR only needs to be sown once a year and can be harvested twice. Recently, local governments in China have actively promoted the cultivation of RR to enhance productivity while mitigating environmental impacts.

However, RR has been historically limited due to unsuitable cultivars and ineffective management, resulting in low and unstable yields [8,9,10]. Firstly, there is a lack of varieties that have strong regenerative capacities. Secondly, the progress made in hybrid rice technology has markedly enhanced grain yields, thereby allowing farmers to attain satisfactory harvests through the implementation of the SR system. Finally, the lack of specialized harvesting machinery, coupled with labor scarcity and rising costs, made the cultivation of RR economically unviable. However, in-depth studies on the integration and application of high-yield and efficient cultivation technology of mechanically-harvested ratoon rice resulted in an expansion from 2015 onwards [6]. Now, the RR system has been extended to several provinces in central China [9]. Based on our field survey, Hunan Province currently boasts an RR cultivation area of 333,333 ha, Hubei Province has over 200,000 ha, Chongqing Municipality features more than 66,667 ha, and scattered distributions are also found in other provinces. Taking these figures into account, the total cultivated area for RR nationwide can reach approximately 800,000 ha.

Compared with the DR and SR systems, is the cultivation of RR a more sustainable rice production system? Recent studies have adopted various indicators to assess the impacts of the rice-cropping systems, with a primary focus on the resource use efficiency [11,12], yield and productivity [5,13], economic performance [14], and environmental impacts [5,13,14]. Specifically, when compared to the DR system, the RR system yields statistically similar net energy while doubling the net economic return with 32–42% lower energy input, production costs, and global warming potential (GWP) [12]. On average, the yield of the RR system exceeds that of the SR system by 25.3% [5]. Additionally, the income-to-costs ratio of RR increases by 25.5% relative to DR [14]. However, it is noteworthy that the environmental impacts of the RR system have increased, resulting in pollutant emissions that are 23.5% to 35.1% higher than the SR system [5]. The sustainability of the rice-cropping system is determined by the integrated effects of input utilization, agricultural outputs, and environmental impacts. While the aforementioned studies provide valuable insights, they tend to isolate individual indicators, lacking a comprehensive analysis of the overall performance. To effectively inform policy initiatives aimed at balancing grain production, environmental sustainability, and farmer well-being, it is essential to develop a more holistic understanding of the impacts of the rice-cropping systems.

This study simultaneously collected data related to the three rice-cropping systems across three dimensions: input utilization, agricultural outputs, and environmental impacts. By incorporating indicators from these three domains into an efficiency analysis framework, we derived a comprehensive metric: environmental efficiency (EE). Additionally, we employed non-efficiency decomposition methods to identify pathways leading to EE losses, thereby identifying strategies to enhance the sustainability of the rice-cropping systems.

The objectives of this study were threefold: (1) to develop an EE framework that concurrently considered input utilization, agricultural outputs, and environmental impacts to comprehensively assess the rice-cropping systems; (2) to employ non-efficiency decomposition methods to identify factors contributing to EE losses within the rice-cropping systems; and (3) to offer specific policy insights aimed at optimizing and promoting rice-cropping systems based on the findings. A primary contribution of this study is the advancement of an integrated framework that comprehensively considers the input utilization, agricultural outputs, and environmental impacts, distinguishing it from previous research efforts. This comprehensive approach enhances our understanding of how rice-cropping systems can be optimized to meet future food demands while minimizing environmental impacts and enhancing farmer welfare, offering a model for sustainable agricultural practices worldwide.

2. Materials and Methods

2.1. Study Area and Data Collection

Hunan Province was selected as the study area due to its status as the province with the largest cultivated area for rice in China. With its fertile lands, abundant water resources, and favorable climatic conditions, Hunan is well-suited for rice farming. During the sample period, Hunan Province accounted for approximately 13.3% of the total rice cultivation area and contributed nearly 12.6% of the total rice production in China. Additionally, Hunan is a leading producer of RR in China, with a cultivated area of 333,333 ha, representing approximately 42% of the national RR production. Hunan Province, situated in south-central China, spans longitudes 108°47′ to 114°15′ E and latitudes 24°38′ to 30°08′ N. The region experiences a continental subtropical monsoon climate with abundant light, heat, and water resources. The annual average temperature generally ranges from 16 °C to 19 °C.

Given the absence of official yield statistics for RR, our study selected 35 townships across 7 cities in Hunan Province based on recommendations from the Department of Agricultural and Rural Affairs of Hunan Province. The seven cities included in our study are as follows: Changde (110°40′–112°17′ E, 28°24′–30°07′ N), Hengyang (110°32′–113°16′ E, 26°07′–27°28′ N), Huaihua (108°47′–111°06′ E, 25°52′–29°01′ N), Loudi (110°45′–112°31′ E, 27°12′–28°14′ N), Yiyang (110°43′–112°55′ E, 27°58′–29°31′ N), Yueyang (112°18′–114°09′ E, 28°25′–29°51′ N), and Changsha (111°53′–114°15′ E, 27°51′–28°40′ N). From July to September 2019, we conducted a household survey involving 219 rice farmers selected randomly. The survey encompassed farmers employing single as well as multiple rice-cropping systems across different plots. A total of 307 plots were analyzed to assess inputs, outputs, and their corresponding environmental and economic impacts over all the rice production cycles in the preceding year. Subsequently, after excluding 8 plots with missing data, our analysis focused on 299 plots to compute the EE of rice production, comprising 110 plots with RR, 86 plots with DR, and 103 plots with SR.

2.2. Estimation of the Environmental Efficiency

Efficiency analysis constitutes a crucial research agenda in the realm of economics. The data envelopment analysis (DEA) method serves as a potent tool for conducting such analysis [15]. The DEA model deploys linear programming techniques to define the production frontier, allowing for the assessment of the technical efficiency of decision-making units by measuring their proximity to this frontier, thereby aiming to maximize outputs while minimizing resource inputs. Over time, the DEA method has undergone significant refinement and expansion. Notably, Tone [16] developed the slack-based measure (SBM) model, which addresses the limitations of traditional DEA models by incorporating slack variables for both inputs and outputs into the objective function, creating a non-radial model. Furthermore, Tone [17] enhanced the SBM model by introducing slack variables for undesirable outputs and adding constraints for these outputs. This adaptation is particularly relevant to the advancement of sustainable and green agriculture practices. In light of these developments, this study employed the SBM-DEA model to calculate the EE of various rice-cropping systems, taking into account undesirable outputs in the efficiency evaluation process.

In this study, the decision-making units where input–output practices occur were the plots of land where farmers cultivate various rice systems. The technical relationship between input and output, including the environmental pollution byproducts, is defined as environmental technology [18]. Adopting the SBM-DEA model [16,17], we assumed the existence of n decision-making units, each with m types of inputs $x=\left(x_1,\dots ,x_m\right)\in R_m^{+}$, $s_1$ types of desirable outputs $y^g=\left(y_1^g,\dots ,y_{s_1}^g\right)\in R_{s_1}^{+}$, and $s_2$ types of undesirable outputs $y^b=\left(y_1^b,\dots ,y_{s_2}^b\right)\in R_{s_2}^{+}$. The production possibility set, representing the environmental technology, can be expressed accordingly:

$$P=\left\{\left(x,y^g,y^b\right)|x\ge X\lambda ,y^g\le Y^g\lambda ,y^b\ge Y^b\lambda ,\lambda \ge 0\right\}$$

(1)

Drawing on the characterization of environmental technology; the input; desirable output; and undesirable output of the oth decision-making unit $(o=1;\dots ,n)$ are denoted by $x_o$; $y_o^g$; and $y_o^b$; respectively. The EE can be expressed as follows:

$$\begin{gathered} {\rho }^{*}=\min {\displaystyle \frac{1-\left(\frac{1}{m}\right){\sum }_{i=1}^m\left(\frac{s_{\mathit{io}}^{-}}{x_{\mathit{io}}}\right)}{1+\left[1/{(s}_1+s_2)\right]\left[{\sum }_{l=1}^{s_1}\left(\frac{s_{\mathit{lo}}^g}{y_{\mathit{lo}}^g}\right)+{\sum }_{k=1}^{s_2}\left(\frac{s_{\mathit{ko}}^b}{y_{\mathit{ko}}^b}\right)\right]}} \\ \begin{array}{cc}\hfill s.t.& x_o=X\lambda +s_o^{-}\hfill \\ & y_o^g=Y^g\lambda -s_o^g\hfill \\ & y_o^b=Y^b\lambda +s_o^b\hfill \\ & s_o^{-}\ge 0,s_o^g\ge 0,s_o^b\ge 0,\lambda \ge 0\hfill \end{array} \end{gathered}$$

(2)

where the slack variables $s_o^{-}$, $s_o^g$, and $s_o^b$ represent the extent of the input overuse, the desirable output shortfall, and the undesirable output over-emission, respectively. The indices $i$, $l$, and $k$ correspond to the inputs, desirable outputs, and undesirable outputs. The EE indicator ${\rho }^{*}$ ranges from 0 to 1, with higher values indicating greater efficiency. As the slack variables increase, ${\rho }^{*}$ decreases. When all the slack variables are 0, the efficiency indicator reaches its maximum value of 1, signifying that the assessed decision-making unit is efficient and situated on the production frontier. Conversely, a value of ${\rho }^{*}<1$ indicates a loss of EE in the decision-making unit.

In the event of EE loss, the inefficiency can be decomposed into two components: the input inefficiency and output inefficiency. The decomposition can be represented as follows:

Input redundancy rate ${IE}_x$:

$${\mathit{IE}}_x={\displaystyle \frac{1}{m}}{\sum }_{i=1}^m{\displaystyle \frac{s_{\mathit{io}}^{-}}{{\mathrm{x}}_{\mathrm{io}}}},(i=1,\dots ,\mathrm{m})$$

(3)

Desirable output shortfall rate ${IE}_g$:

$${\mathit{IE}}_g={\displaystyle \frac{1}{s_1}}{\sum }_{l=1}^{s_1}{\displaystyle \frac{s_{\mathit{lo}}^g}{y_{\mathit{lo}}^g}},(l=1,\dots ,{\mathrm{s}}_1)$$

(4)

Undesirable output redundancy rate ${IE}_b$:

$${\mathit{IE}}_b={\displaystyle \frac{1}{{\mathrm{s}}_2}}{\sum }_{k=1}^{s_2}{\displaystyle \frac{s_{\mathit{ko}}^b}{y_{\mathit{ko}}^b}},(k=1,\dots ,{\mathrm{s}}_2)$$

(5)

where the input redundancy rate denotes the proportion of input that can be reduced; the desirable output shortfall rate indicates the proportion of desirable output that can be expanded; and the undesirable output redundancy rate represents the proportion of undesirable output that can be reduced.

The evaluation of rice-cropping systems must consider the trade-off among the input utilization, agricultural outputs, and environmental impacts. Therefore, based on the aims of this research, the following specific indicators were selected to calculate EE:

(1)

Input indicators. Based on the primary inputs and pollution sources associated with rice production, we selected the following input indicators: land, fertilizer, pesticide, labor, and machinery.

(2)

Desirable output indicators. Given that grain production yields both social and economic benefits, it is important to account for these dual advantages. Therefore, the yield and production value of rice were included as desirable output indicators for each decision-making unit.

(3)

Undesirable output indicators. In rice production, alongside the attainment of desirable outputs, there is also the generation of undesirable outputs that may result in adverse environmental impacts on air, soil, and water quality. Accurately assessing the environmental impacts of rice-cropping systems necessitates the careful selection of appropriate indicators to gauge these undesirable outputs. These indicators should encompass a comprehensive range of typical pollutants associated with the rice production process, and their measurement should accurately reflect the extent of pollution generated.

Rice production is known to exert adverse environmental impacts, primarily through two mechanisms: gas emissions and nutrient loss. To assess these impacts comprehensively, this study employed the carbon emissions (CEs) and NSP indicators to measure the undesirable outputs within rice-cropping systems. CEs occur throughout the entire process, from input production to field management [19]. This metric offers a comprehensive assessment of greenhouse gas emissions during rice production, crucial for monitoring environmental impact, sustainability, and agricultural efficiency [20]. Agricultural NSP predominantly originates from agrochemical inputs and agricultural residues [21]. These pollutants contribute to nutrient imbalances in the environment through leaching and runoff, leading to a range of environmental challenges. Figure 1 shows the framework for estimating the EE of the rice-cropping system. The methods for calculating the CEs and NSP are detailed in the following subsections.

2.3. Estimation of Carbon Emissions

The estimation of CE is mostly conducted within the framework of life cycle assessment (LCA), which evaluates the environmental impact throughout all lifecycle stages [20,22,23,24,25]. This process necessitates initially defining the system boundary, compiling the emission inventory, and specifying the functional unit.

2.3.1. System Boundary

The system boundary delineates the scope of activities and materials considered in calculating the CEs [24]. Variations in its definition across studies contribute to differing environmental impact assessments [26]. Due to data limitations and practical constraints, many studies adopt a “from cradle to gate” approach, encompassing activities from agricultural input production to crop harvesting [26,27]. Continuing in this vein, we applied the “from cradle to gate” framework to set the system boundary for rice production. Specifically, the system boundaries for DR and RR conclude at the final seasonal crop harvest within a given year.

2.3.2. Emission Inventory

To establish the emission inventory, it is essential to identify the primary sources contributing to CEs within the defined system boundary. We systematically assessed key environmental impact stages associated with rice production, encompassing (1) carbon dioxide (CO₂) resulting from the production, transportation, and utilization of various agricultural inputs; (2) methane (CH₄) emissions from paddy fields during rice cultivation; and (3) direct and indirect nitrous oxide (N₂O) emissions arising from nitrogen application [12,27,28,29,30]. It is pertinent to note that straw burning can also contribute to CEs. However, based on our field investigations, we confirmed that Chinese farmers do not engage in straw burning due to stringent government regulations, hence, it was not factored into the CE calculation. Consequently, the pollution emission inventory related to rice production encompassed CO₂, CH₄, and N₂O.

Below are the calculation methods for CEs:

CO₂ emissions. The calculation of CO₂ emissions primarily considered agricultural inputs such as fertilizers, pesticides, diesel fuel consumption by agricultural machinery (tillage, planting/transplanting, and harvesting), and electricity consumption for irrigation. The most authoritative method for estimating CO₂ from these agricultural inputs is the CE factor approach provided by the IPCC [27,31], which was also employed in this analysis. The specific formula is as follows:

$$E_{\mathit{input}}=\sum (A_i\times {\delta }_i)$$

(6)

$${\delta }_i={\delta }_{i1}+{\delta }_{i2}+{\delta }_{i3}$$

(7)

Here, $E_{input}$ indicates the CO₂ emission from the production, transportation and utilization of various agricultural inputs; $A_i$ denotes total amount of the $i$th agricultural input throughout the entire life cycle of rice production; ${\delta }_i$ is the coefficient factor for the $i$th agricultural input; and ${\delta }_{i1}$, ${\delta }_{i2}$, and ${\delta }_{i3}$ correspond to CO₂ from the production, transportation, and utilization of the $i$th agricultural input, respectively. The CO₂ produced from input utilization primarily originates from diesel combustion.

Given the diversity of fertilizer types, which is influenced by factors such as regional geography, climate conditions, crop growth characteristics, and farmer operational habits, accurate statistics on their usage can be challenging. We convert various fertilizers into their active components (N, P₂O₅, K₂O) for unified measurement [22,26,27], and the corresponding carbon emission coefficients we used are based on [26]. Given the diversity of pesticides and their varying active components, precise calculations of their CO₂ emissions are challenging. Therefore, we employed a unified comprehensive factor to estimate the emissions resulting from the various pesticides used by farmers [26]. Additionally, we considered diesel fuel consumption, including the use of farm machinery both owned and rented by farmers. This calculation accounted for the diesel expenditure, diesel price, machinery operating time, and the unit fuel consumption, converting these into diesel fuel consumption based on a diesel density of 0.84 kg/L [32]. Furthermore, we estimated the electricity consumption for irrigation by considering the irrigation fees paid by farmers and the corresponding agricultural electricity prices in the region. Notably, during the survey period, poverty-stricken counties in Hunan Province implemented discounted electricity prices for agricultural irrigation and drainage. Consequently, we matched the relevant irrigation and drainage electricity prices based on whether the sample household was located in a poverty-stricken county.

CH₄ emissions. The CH₄ emissions were calculated using the methodologies detailed in the literature [33,34,35] and in accordance with the ISO/TS14067 [36] accounting standard. To account for variations in the reproductive characteristics among different rice varieties and the operational disparities during production, we adjusted the calculation method and conventional parameter values. The specific formula is as follows:

$$E_{{\mathit{CH}}_4}={\mathit{EF}}_{i,j,k}\times A\times t_{i,j,k}$$

(8)

$${\mathit{EF}}_{i,j,k}={\mathit{EF}}_c\times {\mathit{SF}}_w\times {\mathit{SF}}_p\times {\mathit{SF}}_o$$

(9)

$${\mathit{SF}}_o={\left(1+\sum _i{\mathit{ROA}}_i\times {\mathit{CFOA}}_i\right)}^{0.59}$$

(10)

In Equation (8), $E_{{CH}_4}$ represents the CH₄ emissions within the system boundary of rice production (kg); ${EF}_{i,j,k}$ is the daily CH₄ emission factor (kg/(ha × d)), where $i,j,k$ denote factors affecting CH₄ emissions from rice fields, such as ecosystems, water conditions, and organic matter input; $A$ is the harvest area of the rice field (ha); and $t_{i,j,k}$ is the total number of days in the rice growth period. In Equations (9) and (10), ${EF}_c$ is the baseline CH₄ emission factor, representing the CH₄ emissions during the continuous irrigation of rice fields without organic matter input, with a default value of 1.3 kg/(ha × d); ${SF}_w$ and ${SF}_p$ are the conversion factors for water conditions during the planting and pre-planting periods, respectively; ${SF}_o$ is the conversion factor for the type and quantity of organic matter input; and ${CFOA}_i$ is the conversion coefficient for different varieties of soil organic matter input. The values of ${SF}_w$, ${SF}_p$ and ${CFOA}_i$ for different rice cultivation techniques in different growing seasons were derived from [12] to highlight the disparities among various rice production methods. Finally, ${ROA}_i$ is the input density (t/ha) of organic matter, primarily referring to the amount of straw return, and its calculation formula is as follows:

$${\mathit{ROA}}_i=Y\times S_G\times {\mathit{ISR}}_p\times 0.85\times 0.001$$

(11)

$$S_G={\displaystyle \frac{1}{G_H}}-1$$

(12)

where $Y$ represents the yield of rice (kg/ha); $S_G$ denotes the straw-to-grain ratio of rice; ${ISR}_p$ indicates the straw return rate of rice; and 0.85 reflects the proportion of dry weight to fresh weight of rice straw. To accurately capture the heterogeneity in rice production processes, we calculated the value of $S_G$ based on the harvest index $G_H$ specific to different rice-cropping systems, as opposed to using conventional uniform values, following the methodology outlined by [12]. Additionally, considering the unique harvesting characteristics of the RR system during the main season, we estimated the straw return rate of RR during the main season by comparing the stubble height between RR and conventional rice.

N₂O emissions. N₂O emissions from paddy fields encompass both direct and indirect emissions. Direct N₂O emissions occur from the soil due to the nitrification and denitrification of nitrogen sources. Indirect N₂O emissions result from the deposition of ammonia volatilized in the forms of NH₃-N and NOx-N, as well as nitrate leaching and runoff. The quantity of N₂O emissions, denoted as $E_{N_{2}O}$, can be calculated as follows:

$$E_{N_{2}O}=E_{N_{2}O}^{\mathit{di}}+E_{N_{2}O}^{\mathit{in}}$$

(13)

where $E_{N_{2}O}^{di}$ indicates the direct emissions and $E_{N_{2}O}^{in}$ represents the indirect emissions. The formula for calculating direct N₂O emissions is as follows:

$$E_{N_{2}O}^{\mathit{di}}=\left(N_f+N_s\right)\times {\mathit{EF}}_{\mathit{di}}\times 48/28$$

(14)

$$N_s=N_{\mathit{su}}+N_{\mathit{sd}}=({\displaystyle \frac{Y}{G_H}}-Y)\times {\mathit{ISR}}_p\times C_s+{\displaystyle \frac{Y}{G_H}}\times \mathit{RSR}\times C_r$$

(15)

Based on the research findings, fertilizers and straw are the primary nitrogen sources contributing to N₂O emissions during rice cultivation. Therefore, in Equation (14), $N_f$ and $N_s$ represent the nitrogen inputs from fertilizers and straw to the rice field, respectively. ${EF}_{di}$ denotes the direct N₂O emission factor (kg N₂O-N/kg × N). The value of $N_f$ is derived from the active components in the fertilizers, while $N_s$ comprises the nitrogen from straw returned to the field $N_{su}$ and the root nitrogen content $N_{sd}$. Equation (15) outlines the calculation method for $N_s$. Here, $Y$, $G_H$, and ${ISR}_p$ retain their previously defined meanings; $RSR$ is the root-to-shoot ratio taken from the Guidelines for Provincial Greenhouse Gas Inventories compiled by research institutions under the Climate Change Department of the National Development and Reform Commission of China; and $C_s$ and $C_r$ represent the nitrogen content of straw and roots, respectively, based on the values documented in [12].

The formula for calculating the indirect N₂O emissions is as follows:

$$E_{N_{2}o}^{\mathit{in}}={(N}_d+N_l)\times {\displaystyle \frac{48}{28}}$$

(16)

$$N_d=\left(N_f+N_s\right)\times \mathit{VR}\times {\mathit{EF}}_d$$

(17)

$$N_l=\left(N_f+N_s\right)\times \mathit{LR}\times {\mathit{EF}}_l$$

(18)

Equation (16) describes the two main contributors to indirect N₂O emissions: nitrogen volatilization leading to atmospheric deposition, $N_d$, and nitrogen leaching or runoff, $N_l$. In Equation (17), $VR$ represents the rate of nitrogen volatilization, and ${EF}_d$ denotes the N₂O emission factor (kg N₂O-N/kg × N) that results from atmospheric nitrogen deposition. Equation (18) defines $LR$ as the nitrogen leaching and runoff rate, and ${EF}_l$ as the indirect N₂O emission factor (kg N₂O-N/kg × N) resulting from nitrogen leaching and runoff losses. The values of the emission factors $VR$ and $LR$ used in the above equations were sourced from the Guidelines for Provincial Greenhouse Gas Inventories.

After quantifying the CEs originating from agricultural inputs, as well as CH₄ and N₂O emissions associated with rice cultivation, the quantity of CEs from rice production, denoted as $E_{total}^{{\mathrm{CO}}_2\mathrm{eq}}$, was computed using the following formula:

$$E_{\mathit{total}}^{{\mathrm{CO}}_2\mathrm{eq}}={\mathrm{E}}_{\mathrm{input}}+28{\mathrm{E}}_{{\mathrm{CH}}_4}+265{\mathrm{E}}_{{\mathbf{N}}_2\mathrm{O}}$$

(19)

where CH₄ and N₂O emissions were converted to CO₂ equivalents (CO₂ eq) according to their respective GWP (kg ha⁻¹). The GWP values utilized in this study accorded with the latest 100-year GWP coefficients of CO₂, CH₄, and N₂O (1:28:265) recommended by the IPCC [22].

2.3.3. Functional Unit

The functional unit refers to the standardized measurement utilized for quantifying CEs. Ensuring the environmental assessment of rice-cropping systems within the same system boundary relies on employing a consistent functional unit [26,37]. Previous studies have employed various functional units [19,20,22,26,30]. In this study, we adopted both CEs per unit area (kg CO₂-eq/ha) and CEs per unit weight (kg CO₂-eq/1000 kg) for comprehensive assessment.

2.4. Estimation of Non-Point Source Pollution

In the present study, the unit survey and assessment method was employed to analyze and estimate the NSP generated during the rice production process [38,39]. This methodology comprised an initial analysis of the primary sources of NSP, followed by the identification of pollution units within each source. Subsequently, the pollutant emission coefficients for these units were determined [38].

The NSP in rice production is primarily associated with the loss of fertilizers through water runoff [3]. Consequently, this study focused on NSP arising from the use of fertilizers in the rice cultivation process. The pollution unit refers to the minimum independent unit that can reasonably quantify the pollution, determined by the pollution source and its characteristics. In the context of rice production, the relevant pollution units include nitrogen fertilizers, phosphate fertilizers, and compound fertilizers. After obtaining the application amount of each pollution unit, the pollutant emission coefficients were applied to calculate the emission of each unit. This approach enabled the estimation of the potential magnitude of the NSP [38]. The calculation formula is as follows:

$$E_j=\sum _{\mathrm{i}}{\mathit{EU}}_i(1-{\eta }_i)C_{\mathit{ij}}$$

(20)

where ${EU}_i$ represents the input amount of pollution unit I and ${\eta }_i$ is the resource utilization rate coefficient; thus, ${EU}_i\left(1-{\eta }_i\right)$ indicates the amount of unabsorbed pollution unit. $C_{ij}$ denotes the emission coefficient of the pollutant indicator j within the pollution unit i. Based on rice production practices, the pollutant indicators j are total nitrogen (TN) and total phosphorus (TP). Therefore, $E_j$ represents the emissions of TN or TP.

The Handbook of Fertilizer Loss Coefficients, published by the Leading Group of China’s State Council during the country’s first national pollution source census, provides crucial technical support for evaluating NSP levels using the unit survey and evaluation method. This coefficient incorporates the effects of the planting region, crop type, planting method, tillage method, farmland type, soil type, and topography on the fertilizer loss. Utilizing the fertilizer loss coefficient is a typical practice in pollution source studies. By integrating the unit survey and evaluation method for calculating the pollution production with the fertilizer loss coefficient, the formula for calculating the NSP caused by rice production is as follows:

$$E_j={\sum }_{\mathrm{i}}{\mathit{AI}}_i\times {\mathit{FC}}_{\mathit{ij}}$$

(21)

where $E_j$ represents the amounts of TN or TP emitted; ${AI}_i$ denotes the equivalent active component amount (N and P₂O₅) of the input for each pollution unit; and ${FC}_{ij}$ is the loss coefficient for TN and TP, which can be sourced from the Fertilizer Loss Coefficient Handbook.

The description of the indicators for the EE estimation is shown in

Table 1. Description of the input–output indicators for the EE estimation.

Indicator	Variable	Variable Description (Unit)
Desirable outputs	Rice yield	Total rice yield (kg)
	Rice production value	Total rice production value (CNY)
Undesirable outputs	CEs	(kg CO2 eq)
	NSP	TN and TP emissions (kg)
Inputs	Land	Harvested area (ha)
	Fertilizer	Total fertilizer expenditure (CNY)
	Pesticide	Total pesticide expenditure (CNY)
	Labor	Labor inputs of household and employed workers (Day)
	Machinery	Total expenditure on machinery operations (CNY)

.

3. Results

3.1. Carbon Emission Estimation Results

3.1.1. Carbon Emissions of the RR, DR, and SR Systems

Table 2. The CEs of different rice-cropping systems and their composition.

Emission Source	CEs per Unit Area (kg CO2-eq/ha)			CEs per Unit Weight (kg CO2-eq/1000 kg)
	RR	DR	SR	RR	DR	SR
N fertilizer (Pure N)	2663.48	3384.72	1814.78	232.52	256.39	268.62
P fertilizer (P2O5)	67.41	119.39	64.26	5.83	8.98	9.42
K fertilizer (K2O)	47.33	71.00	47.82	4.14	5.36	7.04
Pesticides	19.17	53.59	33.88	1.62	3.86	4.82
Irrigation electricity	1088.41	1660.12	870.76	93.44	125.46	126.54
Machinery fuel	1414.87	2946.30	1177.58	124.45	221.23	166.68
CH4	4611.51	7685.48	2820.59	400.99	574.74	405.78
N2O	2193.05	2703.05	1472.56	190.80	203.84	215.65
CEs	12,105.23	18,623.65	8302.23	1053.79	1399.86	1204.55

presents the CEs of different rice-cropping systems. When using the unit area as the functional unit, the CEs were highest for the DR system (18,623.65 CO₂-eq/ha), followed by the RR system (12,105.23 kg CO₂-eq/ha), with the SR system having the lowest CE (8302.23 kg CO₂-eq/ha). The presence of two growing seasons and the correspondingly higher agricultural inputs within the system boundary of the DR and RR systems is one of the reasons for this difference [12].

Table 3. The Kruskal–Wallis rank-sum test for the CEs.

Nonparametric Test	Rice-Cropping System	CEs per Unit Area			CEs per Unit Weight
		Mean Rank	Chi-Squared	p	Mean Rank	Chi-Squared	p
Kruskal–Wallis rank-sum test	RR	158.65			105.28
	DR	256.76	248.124	0.000	212.22	73.23	0.000
	SR	56.22			152.39
		Mean Rank-Difference	Critical Value	p	Mean Rank-Difference	Critical Value	p
	RR and DR	98.11	29.92	0.000	106.94	29.92	0.000
	RR and SR	102.42	28.57	0.000	47.11	28.57	0.000
	DR and SR	200.54	30.54	0.000	59.83	30.54	0.000

displays the results of the Kruskal–Wallis rank-sum test, indicating significant distribution differences in the CEs per unit area among the three rice-cropping systems. Additionally, pairwise comparisons showed statistically significant differences in the CEs per unit area between the RR, DR, and SR systems. Consequently, the distribution of unit area CEs across the three rice-cropping systems follows the order DR > RR > SR.

When using the unit weight as the functional unit, the RR system demonstrated environmental advantages. Among the three rice-cropping systems, RR exhibited the lowest CE (1053.79 kg CO₂-eq/1000 kg). Compared to the DR (1399.86 kg CO₂-eq/1000 kg) and SR (1204.55 kg CO₂-eq/1000 kg) systems, the quantity of CEs from the RR system was reduced by 24.72% and 12.52%, respectively. The Kruskal–Wallis rank-sum test results, presented in Table 3, indicated significant distribution differences in the CEs per unit weight among the three rice-cropping systems. Furthermore, pairwise comparisons revealed statistically significant differences in the CEs per unit weight between the RR, DR, and SR systems. The CEs per weight among the three rice-cropping systems follow the order DR > SR > RR.

3.1.2. Contribution of Different Sources to Carbon Emissions

The contributions of various sources to the CEs of the three rice-cropping systems are detailed in Figure 2. The CE contribution structures from different sources were generally consistent across the three rice-cropping systems, regardless of whether the unit area or unit weight was considered. The CH₄ from rice fields was the highest contributor to the CEs of rice production, followed by nitrogen fertilizer, rice field N₂O, machinery fuel, and electricity for irrigation. A slight difference was observed in the DR system, where the contribution of machinery fuel exceeded that of the rice field N₂O emissions, although the values were very close. Collectively, these five key sources accounted for more than 98% of the total CEs. The contributions of phosphorus fertilizer, potassium fertilizer, and pesticides to the CEs in rice production were relatively low.

Utilizing unit area as a functional unit has limitations due to the cropping scenarios within the system boundaries that influence CEs. Thus, using the unit weight as a functional unit was more meaningful for this analysis. The subsequent results are illustrated based on the unit weight measurements. Overall, the RR system demonstrated a reduction in CEs from key sources compared to the DR and SR systems. Specifically, the RR system significantly reduced the CEs compared to the DR system, with notable decreases in emissions from pesticides (58.03%), machinery fuel (43.75%), phosphate fertilizer (35.08%), CH₄ emissions in rice fields (30.23%), irrigation electricity (25.52%), potash fertilizer (22.76%), nitrogen fertilizer (9.31%), and N₂O emissions (6.40%) (all p < 0.01). In comparison to the SR system, the RR system also showed substantial reductions in CEs: 66.39% from pesticides, 41.19% from potassium fertilizer, 38.11% from phosphate fertilizer, 26.16% from irrigation electricity, 25.34% from machinery fuel, 13.44% from nitrogen fertilizer, and 11.52% from N₂O emissions (all p < 0.01).

3.2. Non-Point Source Pollution Estimation Results

Table 4. NSP in different rice-cropping systems.

NSP Indicator	NSP per Unit Area NSP (kg/ha)			NSP per Unit Weight (kg/1000 kg)
	RR	DR	SR	RR	DR	SR
TN	3.46	4.39	2.29	0.30	0.33	0.34
TP	0.60	1.06	0.34	0.05	0.08	0.05

presents the NSP for different rice-cropping systems. When using unit area as the functional unit, the TN emissions from the production process were highest in the DR system (4.39 kg/ha), followed by the RR system (3.46 kg/ha), with the SR system having the lowest TN emissions (2.29 kg/ha). Similarly, the TP emissions were highest in the DR system (1.06 kg/ha), followed by the RR system (0.60 kg/ha), and they were lowest in the SR system (0.34 kg/ha).

The Kruskal–Wallis rank-sum test results in

Table 5. The Kruskal–Wallis rank-sum test for NSP (unit area).

Nonparametric Test	Rice-Cropping System	TN			TP
		Mean Rank	Chi-Squared	p	Mean Rank	Chi-Squared	p
Kruskal–Wallis rank-sum test	RR	175.58			160.12
	DR	239.02	205.25	0.000	257.58	235.51	0.000
	SR	60.41			61.48
		Mean Rank-Difference	Critical Value	p	Mean Rank-Difference	Critical Value	p
	RR and DR	63.44	30.21	0.000	97.46	30.21	0.000
	RR and SR	115.18	28.73	0.000	98.64	28.73	0.000
	DR and SR	178.62	30.65	0.000	196.10	30.65	0.000

indicate significant differences in the distribution of TN and TP emissions per unit area across the three rice-cropping systems. Pairwise comparisons further reveal statistically significant differences in the TN and TP emissions per unit area among the RR, DR, and SR systems. Consequently, the distribution of the TN and TP emissions per unit area follows the order DR > RR > SR. The higher TN and TP emissions in the DR and RR systems can be attributed to the higher agricultural inputs per unit area due to longer growth periods within these systems.

As shown in Table 4, when using unit weight as the functional unit, the RR system demonstrated superior environmental sustainability in terms of the NSP. Among the three rice-cropping systems, RR had the lowest TN emissions (0.30 kg/1000 kg), which were 9.09% and 11.76% lower than those of DR (0.33 kg/1000 kg) and SR (0.34 kg/1000 kg), respectively. The TP emissions from the RR system were equivalent to those from SR, with both being 0.05 kg/1000 kg. In comparison to DR (0.08 kg/1000 kg), the TP emissions during RR cultivation were reduced by 37.50%.

The Kruskal–Wallis rank-sum test results in

Table 6. The Kruskal–Wallis rank-sum test for NSP (unit weight).

Nonparametric Test	Rice-Cropping System	TN			TP
		Mean Rank	Chi-Squared	p	Mean Rank	Chi-Squared	p
Kruskal–Wallis rank-sum test	RR	136.55			118.02
	DR	169.91	7.586	0.023	244.80	129.07	0.000
	SR	159.40			116.98
		Mean Rank-Difference	Critical Value	p	Mean Rank-Difference	Critical Value	p
	RR and DR	33.36	30.21	0.004	126.77	30.21	0.000
	RR and SR	22.85	28.73	0.028	1.05	28.73	0.465
	DR and SR	10.51	30.65	0.206	127.82	30.65	0.000

indicate significant differences in the distribution of TN and TP emissions per unit weight across the three rice-cropping systems. Pairwise comparisons further reveal statistically significant differences in the TN emissions per unit weight between RR and both DR and SR, and in the TP emissions per unit weight between RR and DR, as well as between DR and SR.

3.3. Analysis of the Environmental Efficiency

3.3.1. Measurement Results and Comparison

The EE of the decision-making units in the sample had an overall mean of 0.61. Among the three rice-cropping systems, RR exhibited the highest average EE at 0.67, followed by DR at 0.58, and SR at 0.57.

The Kruskal-Wallis rank-sum test results presented in

Table 7. The Kruskal-Wallis rank-sum test for EE.

Nonparametric Test	Rice-Cropping System	Mean Rank	Chi-Squared	p
Kruskal–Wallis rank-sum test	RR	179.48
	DR	134.94	20.33	0.000
	SR	131.09
		Mean Rank-Difference	Critical Value	p
	RR and DR	44.54	29.79	0.000
	RR and SR	48.39	28.38	0.000
	DR and SR	3.85	30.23	0.380

indicate a significant difference in EE across the three rice-cropping systems, as the null hypothesis was rejected. The pairwise comparisons further highlighted that these differences were primarily between RR and the other two rice systems, with RR demonstrating a higher mean EE than both DR and SR. No significant difference was found between the DR and SR systems. These findings suggest that the EE of the RR system is closer to the production frontier than the EEs of the other two rice-cropping systems. Based on the SBM-DEA model test, two key implications emerge: first, the RR system maximizes social and economic output with minimal economic input; second, it effectively reduces undesirable outputs through more efficient resource consumption. Consequently, RR cropping systems facilitate a positive coupling of input utilization, agricultural outputs, and environmental impacts.

3.3.2. Inefficiency Decomposition and Comparison

To gain a deeper understanding of the causes and magnitude of EE loss, we evaluated the efficiency loss from both the input and output perspectives for the three rice-cropping systems. Specifically, we calculated the input redundancy rate, desirable output shortfall rate, and undesirable output redundancy rate. This analysis is crucial for identifying the sources of inefficiencies in rice production. The results are shown in

Table 8. The inefficiency decomposition of different rice-cropping systems.

Rice-Cropping System	Input Redundancy Rate (%)					Desirable Output Shortfall Rate (%)		Undesirable Output Redundancy Rate (%)
	Land	Fertilizer	Pesticide	Labor	Machinery	Yield	Value	CEs	NSP
RR	9.54	13.85	36.31	27.60	40.31	0.15	31.91	12.05	10.08
DR	4.29	30.06	49.35	36.67	44.32	0.00	34.82	29.37	16.64
SR	33.72	28.39	49.57	30.78	45.16	0.29	13.38	23.38	16.13

.

The RR system had the lowest input redundancy rate, followed by the SR system, with the DR system having the highest. Upon decomposing the input redundancy rate, it was observed that, with the exception of the land input in the RR and DR systems, all three rice-cropping systems demonstrated significant redundancy across various inputs. This indicates that the high-input extensive agricultural development model continues to be predominant.

Analyzing the desirable output shortfall rate revealed that the shortfall in production value is a more critical factor influencing EE than the shortfall in yield. Additionally, the desirable output shortfall rate for RR did not reach the lowest level, primarily due to a higher shortfall rate for the production value. Specifically, the production value shortfall rates for RR and DR were approximately two and a half times that of SR.

Based on the undesirable output redundancy rate, the RR system exhibited the lowest CE and NSP redundancy rate. This suggests that the DR and SR systems have inadequate control over undesirable outputs compared to the RR system, resulting in excessive CEs and NSP.

4. Discussion

4.1. The Distribution of Environmental Efficiency

Table 9. The distribution of EE in different rice-cropping systems.

EE	RR (%)	DR (%)	SR (%)
[0–0.3)	0.91	1.16	3.88
[0.3–0.5)	22.73	53.49	50.49
[0.5–0.7)	40.91	18.60	18.45
[0.7–0.9)	10.00	2.33	6.80
[0.9–1]	25.45	24.42	20.39

outlines the distribution of EE across three rice-cropping systems. The EE of the RR predominantly fell within the interval [0.3–0.7), comprising 63.64% of the sample analyzed. The efficiencies of the DR and SR systems were chiefly within the interval [0.3–0.5), accounting for 53.49% and 50.49% of their respective samples. Although the RR system exhibited significantly higher EE than the DR and SR systems, all three systems demonstrate considerable potential for enhancement, as they remain distant from the production frontier. Collectively, these findings indicate that the current rice production process continues to experience significant challenges in reconciling the input utilization, agricultural outputs, and environmental impacts.

4.2. Redundancy Rate of Rice-Cropping Systems

The elevated input redundancy observed in the RR system may stem from the entrenched practice of utilizing high-input planting methods during the main season. Notably, the mechanical input redundancy rate was highest in the RR system, which can be attributed to two primary factors. Firstly, there is a lack of specialized machinery tailored for main-season harvesting in the RR system. Mechanical harvesting has been documented to decrease yields in the ratoon season, with a more pronounced yield reduction observed in areas compacted with harvesting machinery compared to areas not compacted. Secondly, mechanical harvesting adversely impacts the rice quality, which in turn affects its market value. During the main season, areas traversed by machinery may exhibit continued growth; however, the growth in compacted areas differs from that in non-compacted areas, leading to inconsistent maturity of the harvested rice. Especially in mountainous and hilly terrains, this phenomenon is even more evident due to the unsuitability of large and medium-sized harvesters [40]. This inconsistency results in a lower selling price for the rice. Therefore, the excessive deployment of mechanical inputs not only affects the yield but also diminishes the market value of RR.

In the DR and SR systems, pesticides significantly contribute to the input redundancy rate. Despite China’s rigorous policy efforts to reduce pesticide and fertilizer use, the findings of this study indicate a substantial scope for further reductions. Moreover, it is crucial to note that the land redundancy in SR production is markedly higher compared to that in RR and DR systems. This suggests that land inputs in the SR system are underutilized, leading to increased redundancy. As China’s shift from double to single-season rice cropping intensifies, the inefficient use of land in the SR system may continue to hinder advancements in EE across rice-cropping systems. Should this trend persist, it could diminish the overall EE in rice production, not only affecting national agricultural sustainability but also having broader implications for global food security and environmental health.

The production value shortfall rates for RR and DR are approximately two and a half times greater than that of SR. This discrepancy can be attributed to two primary factors: firstly, the main-season rice for RR and the early-season rice for DR mature during peak temperature periods, which adversely affects the grain quality, subsequently resulting in lower market prices; secondly, although the public’s growing negative perception of pesticides in agricultural production has increased consumer interest in ratoon rice [5], the underdeveloped market mechanism often results in ratoon crops being sold together with late-season rice from DR, leading to insufficient returns. The similar value shortfall rates observed in the RR and DR cropping systems underscore this issue.

Furthermore, the exceptionally low yield shortfall rates across the three rice-cropping systems, coupled with a high level of input redundancy, suggest that the scope for enhancing rice yields through increased inputs is limited. This indicates that excessive inputs do not necessarily correlate with higher outputs and may instead have detrimental effects on the environment.

5. Conclusions

This study employed the SBM-DEA model to assess the EE of rice-cropping systems. With an EE of 0.67, the RR system not only surpasses the other rice-cropping systems, reflecting its superior integration of input utilization, agricultural outputs, and environmental impacts, but also shows potential for global replication in efforts to achieve sustainable agricultural systems. Despite its high EE, the RR system has room for improvement, as 63.64% of the RR samples exhibited EE values falling between 0.3 and 0.7. The shortfall in production value and the redundancy across various inputs (excluding land) are the primary causes of efficiency losses in RR.

The findings underscore several policy recommendations to enhance the sustainability of the RR system. Firstly, government agricultural technology extension departments and scientific research institutions should continue to strengthen their research in developing environmentally friendly and efficient operational procedures for the RR system. The emphasis should be on the efficient use of inputs and field water management to minimize undesired outputs [41,42], thereby further enhancing the sustainability of the RR system. To ensure the effective implementation of these procedures, government agricultural technology extension departments should organize regular workshops and field demonstrations to train farmers. Secondly, robust market mechanisms should be implemented to clearly distinguish and price the ratoon-season rice separately from other rice-cropping systems. Establishing specific marketing channels, certification processes, and promotional campaigns will emphasize the unique qualities and benefits of RR. Thirdly, grain quality should be improved through technological advancements. This includes developing heat-resistant rice varieties and introducing specialized harvesting machinery to maintain the grain quality and enhance market competitiveness. Collectively, these measures emphasize the importance of integrated efforts by governments, market entities, and research institutions to foster innovation and promote sustainable agricultural practices in the ratoon rice system.

For future research, two key directions are envisioned. Firstly, from a micro-perspective centered on farmers, there is a need to delve deeper into the diffusion mechanism of the RR system, examining the obstacles hindering its spread and identifying measures that can facilitate its adoption. With the Chinese Ministry of Agriculture already emphasizing and promoting ratoon rice this year and its cultivation expanding to more regions, there are ample opportunities to gather micro-data from diverse locations to enrich this research perspective. Secondly, in the macro domain, recognizing the economic, social, and environmental benefits of ratoon rice, future studies can explore its overall value to society from the perspective of social welfare.