Energy Budgeting, Data Envelopment Analysis and Greenhouse Gas Emission from Rice Production System: A Case Study from Puddled Transplanted Rice and Direct-Seeded Rice System of Karnataka, India

: The energy consumption pattern and greenhouse gas (GHG) emission of any rice production system is important to know the sustainability of varied cultivation and establishment technique. This study was conducted to determine the energy use pattern, GHG emission and e ﬃ ciency of rice farms in puddled transplanted (PTR, rainfed) and direct-seeded rice (DSR, irrigated) production systems in Karnataka, India. The energy indices and GHG emission of di ﬀ erent input and output in a rice production system were assessed by using energy and carbon equivalence. The e ﬃ ciency of PTR and DSR farms were identiﬁed using data envelopment analysis productivity (0.3 kg MJ − 1 ) and energy proﬁtability (6.3) were mainly attributed to the large decrease in energy inputs under DSR. The DEA showed e ﬃ ciency for 26 PTR farms in comparison for 87 DSR farms. The mean technical e ﬃ ciency value highlighted the scope for saving energy by 6% and 2% in PTR and DSR, respectively and showed an economic reduction of $405.5 / ha with PTR versus $163.3 / ha with the DSR method if these ine ﬃ cient farms perform e ﬃ ciently. The GHG emissions revealed that the total emissions for PTR versus DSR production caused by on-farm emissions were 86% and 65%, respectively. The DSR method also had a higher carbon e ﬃ ciency ratio and carbon sustainability index (10.1 and 9.1, respectively). Thus, adoption of DSR method is imperative for reduction of energy consumption and GHG emissions to achieve the carbon sustainability.


Introduction
The demand for food production in developing countries has increased tremendously concurrent with technological advancement using non-renewable energy resources such as fossil fuel, machinery, fertilizers and pesticides [1].This increased production has put constant pressure on natural resources, which in turn has jeopardized the agricultural sustainability.To achieve sustainability in agricultural production, improving the efficiency of energy (input) use is one of the prerequisites because it reduces production costs and environmental pollution [2].Rice is the staple crop of Karnataka, India, and the rice farmers have adopted extensive use of non-renewable inputs and generally cultivate nutrientand irrigation-responsive high-yield varieties.High rainfall in the hilly region of Karnataka only supports the cultivation of rice during Kharif (the monsoon season), and the continuous anaerobic waterlogged condition creates favorable situations for greenhouse gas (GHG) emissions from rice fields [3].The GHG emission is mainly influenced by the type of inputs used for production, such as fertilizers, pesticides, organic manure, fossil fuels, machinery and irrigation methods [4].The rice cultivation in hilly region of Karnataka is characterized by either direct-seeded rice (DSR, irrigated) or puddled transplanted rice (PTR, rainfall), whereas the approach to paddy cultivation is typically a PTR-based system with heavier use of agricultural inputs than in the hilly region.
In general, the cultivation of rice is energy-intensive with respect to use of fertilizers, fossil fuel for machinery and pesticides [5] and thus has led to a negative effect on the environment through the emission of GHGs.These gases are released into the atmosphere and then absorb and re-emit infra-red radiation, and they are the main cause for global warming [6].Agricultural practices are one of the potential sources of GHG emissions, with ~18% of carbon dioxide (CO 2 ) emissions from this sector in India [7] and a rate of ~14% net global emissions [7,8].India's contribution to global GHG emissions rose by an alarming 4.7% in 2016, when India became the third largest GHG emitter after the United States and China [9].Most GHG emissions from Indian agricultural sector are occurring at the input manufacturing stage, followed by usage, farm mechanization and irrigation practices [10,11].Recently, global awareness is increasing about environmental safety and the need for cleaner agricultural systems [12].Therefore, together with energy efficiency, agricultural practices affecting the surrounding environment should be carefully considered [13].Several studies have reported about the energy consumption and GHG emission under different agricultural crop systems [14][15][16].For example, a potato production system produced 993 kg CO 2 -equivalent (eq.)/ha of GHGs and consumed 47 GJ/ha of energy from different inputs [17].Soni et al. [4] reported that transplanted rice produced the highest GHGs (1112 kg CO 2 eq./ha) compared with other crops.Firouzi et al. [18] studied the input consumption in sole groundnut and groundnut-bean intercropping systems and found that the total GHG emissions were 636.14 and 657.36 kg CO 2 eq./ha, respectively.However, very little is known about energy inputs responsible for GHG emissions from the rice ecosystem in this region to develop mitigation strategies.
The analysis of energy flow in the paddy agroecosystem under different cultivation methods is useful for policymakers, administrators and researchers in understanding the functions of these systems and the imperative for energy efficiency and production planning.The inappropriate energy usage in the rice production system can be reduced by analyzing the farm-level energy consumption pattern.This practice will provide a definite advantage for making energy guidelines and working to reduce environmental burden.An understanding of GHG emission from critical agricultural inputs expressed in kilograms of carbon equivalent (kg CO 2 eq./ha) for different field operations, nutrients and chemical use, and irrigation practices and for post-harvesting is essential to identifying carbon-efficient cultivation methods for seedbed preparation, soil fertility management, pest control and other farm operations [19].Furthermore, the efficiency of different methods and technologies are generally analyzed using data envelopment analysis (DEA).This analysis is a nonparametric statistical tool that helps to compare the production efficiency of different farmers for various set of inputs and outputs [20].Because of these multiple advantages, DEA has been used in agriculture to determine the technical efficiency (TE) of tea [21], orange [22], areca nut [23] and rice [24] production systems.However, the information on energy consumption and the effect on GHG emission from varied rice production systems is meagre.In this context, these authors hypothesized that variation in the method of cultivation and efficient use of external non-renewable inputs will lead to lower GHG emissions and improve energy efficiency.Therefore, this study was conducted with the following goals: (1) assess the energy consumption pattern and energy use efficiency (EUE) of PTR and DSR crops; (2) identify the efficiency of cultivation methods using the DEA approach; and (3) determine the impact of different cultivation methods on estimated average GHG emissions.

Experimental Location and Survey Details
The survey was conducted in Siruguppa Taluka in the Bellary district and Gonikoppal Taluka in the Kodagu District of Karnataka, India.The soil of Siruguppa Taluka (situated at 15.6 N latitude and 76.8 E longitude, Figure 1) was black with the following characteristics: pH 8.04, electrical conductivity 0.47 ds/m, medium organic carbon content (0.41%), low nitrogen content (189 kg/ha), medium phosphorus content (58.5 kg/ha) and potassium 287.5 kg/ha.Maximum and minimum temperatures recorded in Siruguppa Taluka were 34.5 • C and 22.3 • C, respectively, with an average rainfall of 285 mm, compared with 25.6 • C and 16.3 • C, respectively, with mean rainfall of 5820.9 mm in Gonikoppal Taluka.The input and output data were collected from the 200 rice farmers of PTR and DSR system through a questionnaire administered in 2019.The basic data were provided as supplementary material (Tables S1 and S2).The questionnaire was designed based on the type of the inputs most of the farmers are using and on method of cultivation (PTR/DSR).These two system represent most of the rice cultivation methodology in South East Asia.The survey was conducted through face-to-face interview during 2018-19.The farmer's data were validated through agricultural department officers and extension workers, those are actively monitoring all the farming activities of these regions.Moreover, the data were collected throughout the season by making frequent field visits.The farmers those who are having minimum one hectare of land with 10 years of farming experience were selected for the survey.The system boundary of rice production is depicted in Figure 2.

Energy Analysis
The different input and output data were converted to be energy equivalent by following standard procedures [25] (Table 1).The key inputs used for rice production were as follows: labor, machinery, diesel, farmyard manure (FYM), fertilizers, seeds and pesticides, whereas grain and straw yield were considered as outputs.The following equation was used for computation of different energy indices [26][27][28][29].

Energy Analysis
The different input and output data were converted to be energy equivalent by following standard procedures [25] (Table 1).The key inputs used for rice production were as follows: labor, machinery, diesel, farmyard manure (FYM), fertilizers, seeds and pesticides, whereas grain and straw yield were considered as outputs.The following equation was used for computation of different energy indices [26][27][28][29].

Data Envelopment Analysis Approach
Two DEA models were used to analyze the efficiency of different decision-making units (DMUs): the Charnes, Cooper and Rhodes (CCR) and Banker, Charnes and Cooper (BCC) models.The CCR model is based on constant returns to scale, whereas while BCC model is based on variable returns to scale.The CCR and BCC models were used to calculate TE and pure technical efficiency (PTE), respectively.Scale efficiency computes the variation of efficiency scores between the CCR and BCC models and is calculated as the ratio of TE to PTE.For this study, DMUs are defined as set of farmers using similar inputs in rice production system.The TE and PTE are calculated using standard formulae [30][31][32].In general, the TE is always lower than PTE [33], and the value of TE and PTE varies between 0 and 1, in which a value of 1 indicates the DMU is the highly efficient.The values <1 implies that the DMU is inefficient.

Carbon Indicators
In this study, the GHG emissions from different inputs were determined by multiplying their corresponding coefficients and expressing the result in terms of CO 2 -equivalent per hectare (CO 2 eq./ha) as given by Lal [19] and West and Marland [34] (Table 2).The GHG emissions were computed on hectare basis.The indices such as carbon input, carbon output, carbon sustainability index, and carbon efficiency ratio were estimated using following equations as mentioned by Chaudhary et al. [35]: Carbon output = Total biomass × 0.4 Carbon Carbon efficiency ratio = C output C input (13) estimated average GHG emissions kg CO 2 eq.kg

Energy Balance
The estimated energy inputs and outputs for PTR and DSR systems are depicted in Table 3.In the PTR system, use of human labor was quite high because of the manual approach to transplanting rise compared with the DSR system.The PTR rice cultivation used a higher amount of diesel for the preparation of land and puddling than the DSR method.Mechanization was a significant share of input in PTR, and more tillage operations were needed in this transplanted method of cultivation, which contributed to higher share of input in PTR versus DSR method.For PTR farmers, FYM (on an average 4326.4kg/ha) application had only a 4% share in the total energy, whereas in the DSR system the FYM (11%) and leaf manure (15%) application together contributed to 26% of total energy (Figure 3).The consumption of chemical fertilizers was found to be higher under the PTR system.The use of nitrogen fertilizers under PTR, which is higher than national average of 144.4 kg/ha, contributed to about 31% of the total input energy.The chemical fertilizers altogether accounted for 36% and 38% of energy input under the PTR and DSR systems, respectively.The consumption of nitrogen, phosphorus and potassium in PTR was 398%, 224% and 126% higher than DSR, respectively, indicating very high use of chemical fertilizers in PTR conditions.The use of herbicides, pesticides, and fungicides were found to be higher with PTR, whereas they were negligible amount under the DSR system.This finding shows that the incidence of pests and diseases were comparatively higher under PTR conditions than the DSR system because farmers were intensively using agrochemicals to manage pest and disease outbreak under the PTR system.The higher amount of irrigation as water energy in PTR was a result of the water used for puddling and for maintaining submerged conditions compared with the DSR method.The share of irrigation water energy to the total energy was about 39% in PTR cultivation.However, the DSR crop was grown under rainfed condition, so no external irrigation was supplemented.For comparison in the literature, Alipour [39] also reported higher irrigation energy in rice production, and another study on the rice production system in India revealed that irrigation and fertilizers have a higher share of energy input [35].In addition, in another study, the amount of energy inputs consumed in different rice-based cropping systems were reported to be 44-54% for chemical fertilizers and 11-14% of for in India [40].The energy input and output estimated for PTR were 35,605 and 155,776 MJ/ha and for DSR were 7823 and 53,227 MJ/ha, respectively (Table 3).The contribution of direct energy and indirect energy towards total energy input were 13% and 87%, respectively, in the PTR system compared with 44% and 66%, respectively in the DSR system (Table 4).Likewise, the consumption of non-renewable energy was higher with the PTR (92.4%) versus DSR system (60.3%).Among non-renewable energy sources, the fertilizers were the maximum share, followed by diesel and machinery in both systems; however, the use of irrigation water contributed more to PTR conditions.In the DSR system, the consumption of renewable energy was found to be higher compared with the PTR system.For context in the literature, Bockari-Gevao et al. [41] reported an energy input of 12400 MJ/ha in rice crops with major contributions from chemical fertilizers (7700 MJ/ha).Agha-Alikhani et al. [42], also reported higher percent share of energy from non-renewable resources (43%) in rice crops.A study in India revealed that irrigation and fertilizers have the higher percent share of energy in rice production system [35].Paramesh et al. [43] highlighted the higher consumption of nitrogen (25-33%), diesel fuel (6.8-18.2%)and irrigation water (8.6-23.7%)under PTR conditions in the Indo-Gangetic Plain.The different energy indices studied in two rice production system are presented in Table 4.The average EUE recorded was 4.4 and 7.3 in the PTR and DSR systems, respectively.This finding indicates a higher amount of input energy is required to produce a unit of output energy under the PTR system and further highlights the efficient use of inputs under the DSR system.In the literature, Bockari-Gevao et al. [41] reported an EUE of 8.86 in Malaysia for rice cultivation.In Iran, the irrigated maize cultivation recorded an EUE of 1.86 [44].Pahlavan et al. [45] reported lower EUE of 0.01 in greenhouse tomato production because of the higher energy input under greenhouse conditions.Paramesh et al. [23] wrote that lower EUE in areca nut gardens was attributed to higher energy consumption, especially from labor, machinery and diesel fuel.Similarly, the higher energy input in the PTR system, especially in terms of fertilizers and irrigation water, led to lower EUE compared with the DSR system.
The energy productivity and energy profitability were found to be higher in DSR than PTR systems, which shows that there is room for improving energy productivity of rice crops in both cultivation methods.The lower energy productivity in PTR was largely owing to more use of diesel fuel, machinery and fertilizers [45][46][47].The lower energy productivity (0.27 and 0.33 kg MJ −1 ) was also observed under PTR conditions in China.Likewise, the energy profitability was also found higher in DSR (6.3) than PTR (3.4) systems.The higher EUE, EP and EFP were primarily attributed to the large decrease in energy inputs (i.e., energy inputs from fertilizer, irrigation water and human labor) but were also attributed to a reduction in energy output under DSR conditions.However, the higher mean net energy in PTR system (120171 MJ ha −1 ) is attributed to higher output energy, and it increases with an increase in energy output.These results indicated that energy being saved in both systems; the positive net energy in DSR was mainly due to use of renewable energy sources such as application of FYM.In Gilan Province of Iran, net energy from paddy production recorded was only 36928 MJ/ha [48].The nutrient energy ratio was found higher in DSR conditions (13.0), indicating more output per unit of nutrient with DSR than PTR.Although the energy output was higher in PTR, it consumed a sizeable amount of nutrients to achieve this production, which implies indiscriminate use of fertilizers in the region.The human energy profitability was lower (98.7) for DSR because of the lower crop yield and more use of labor for weed control because weeds are more of a problem under the DSR system.Paramesh et al. [43] reported human energy profitability of 162.9 and 125.4 for wheat and PTR, respectively.

Identification of Efficient and Inefficient Rice Farmers
The average technical, pure and scale efficiency of PTR and DSR farms is presented in Table 5.The analyzed data indicated that, of 100 farmers tested in PTR farms, 26 farms were found efficient with a score of 1, and the remaining 74 farms had a score <1, which thus were found relatively inefficient in using different energy inputs.The mean value of TE for PTR was 0.94 with a standard deviation of 0.05 (Table 5).The mean value of scale efficiency for inefficient PTR farms was 0.87.This finding implies that there is further room for improving the farm cultivation practices increasing the crop yield and energy conservation.However, under the DSR system, only 13 farms were inefficient and remaining 87 farms were found efficient in utilizing available resources.The average SE of DSR was found to be 0.99, which indicated efficient management of inputs by DSR farms to achieve the targeted yield.Furthermore, the mean TE value highlighted the possible range for saving energy by 6% and 2% in the PTR and DSR methods, respectively, if these inefficient farms were to perform more efficiently.
The energy values that convert an inefficient farm to being an efficient farm for reducing the present level of energy consumption for DSR and PTR production systems are presented in Tables 6 and 7. Based on results of TE, the realistic energy values to obtain the same level of output were determined using DEA.The mean operational reduction values indicated an excessive use of irrigation water (1136.7 MJ ha −1 ) and FYM (358.3MJ ha −1 ) in PTR, whereas in DSR excessive use of FYM (599.5 MJ ha −1 ) was observed.Therefore, adoption of the suitable water management system and soil test-based fertilizer application will help to reduce the energy consumption in both methods of rice cultivation.Similarly, Pahlavan et al. [45] reported that reducing consumption of diesel fuel, electricity and fertilizers is important for energy saving in greenhouse tomato production.Based on DEA results, Mohammadi et al. [2] suggested that reduction in the use of nitrogen-based fertilizers, irrigation water and electricity are necessary to achieve the efficiency on soybean farms.

Economic Indices
The study results revealed that major savings may be possible by reduction of labor ($126/ha), irrigation ($116/ha) and FYM ($91.4/ha) with a total mean economic saving of $472.9/haunder PTR conditions (Tables 8 and 9).Similarly, in DSR, the inefficient farmers can save $163/ha by a reduction in use of FYM ($141/ha).The economic estimation was made by considering the average market price of the inputs in 2019.The results clearly indicate the advantages of DSR for saving capital for the farmers compared with PTR farms.The DSR system is more efficient in reducing the consumption of scarce natural resources such as irrigation water.Although the yield advantage under DSR conditions was found low, it was sustainable with limited resource use over PTR conditions.In other economic studies, Vázquez-Rowe et al. [49] reported an average annual savings of 400 € to 5150 € for inefficient vine farms.Paramesh et al. [23] wrote that adoption of an improved strategy of practices can save resources up to 11% with a mean economic saving of $413/ha.These economic savings would be possible if the farmers with inefficient rice systems adopt the improved management practices by considering soil test values, farm mechanization, suitable method of crop establishment and better pest and disease management.

Greenhouse Gas (GHG) Emission
The GHG emissions data revealed that 86% of the total emissions under PTR conditions were due to on-farm methane emissions followed by nitrogen (6%), diesel fuel (3.9%) and machinery (1.3%).Compared with the DSR method of cultivation, the GHG emission from on-farm methane emissions were 65%, followed by nitrogen (8.3%), diesel fuel (8.9%) and machinery (6.1%).Similar results were reported by earlier studies [50][51][52].Chaudhary et al. [35] reported that 70-75% of the total emissions under DSR conditions were CO 2 emissions only, mainly due to field operations, followed by fertilizer application and the remainder (10%) was methane emissions.In comparison, in the PTR method of cultivation, methane emission-based CO 2 equivalent emissions were higher (57%).In the current study, there was a significant reduction in total GHG emission in DSR (574.1 kg CO 2 equivalent) versus the PTR system (3954.8kg CO 2 equivalent).Pathak [53] reported that continuous flooding, nitrogen fertilizers and machinery are responsible for the higher GHG emissions from PTR farms.Puddling and continuous flooding of rice fields encourages the activity of methanogenesis, thereby increasing methane emission.In contrast, the aerobic conditions of DSR server to reduce methane emissions effectively.Fertilizer-responsive high-yield varieties and lower soil fertility are leading to an increased application of chemical fertilizers, which in turn leads to higher GHG emission.Conventional tillage practices and improved mechanization are responsible for the higher use of diesel fuel, leading to more GHG emission, which in turn primes the farm to have methane emissions under reduced conditions.Periodic soil testing and the use of organic sources of nutrients such as green manure, Azolla cultivation and FYM application can reduce the indiscriminate use of fertilizers [54].This result highlights the scope of conservation tillage (in DSR conditions) to conserve energy and reduce GHG emission by lowering the use of machinery and fossil fuel combustion.
The data show a higher carbon input and carbon output in the PTR system (Table 10).This trend in PTR farming is mainly due to higher consumption of chemical fertilizers, machinery and diesel fuel for field operations compared with the DSR system.The PTR method recorded lower carbon efficiency ratio and carbon sustainability index (4.3 and 3.3, respectively) than DSR method (10.1 and 9.1, respectively).The higher carbon efficiency ratio indicated more carbon efficiency in the DSR method.The higher carbon sustainability index DSR was due to reduced carbon input and reduced number of tillage operations.The estimated average GHG emissions was higher in the DSR method; this trend was mainly based on lower carbon emission, especially from fertilizers and fuel.Chaudhary et al. [35] reported that conservation tillage practices such as minimum tillage and zero tillage have profound effect in reducing GHG emission.The present study highlights the importance of the DSR method as a conservation tillage practice in reducing the carbon input and GHG emission, particularly compared with the PTR method.

Figure 1 .
Figure 1.Map of the study area.Gonikoppal represents direct-seeded rice and Siruguppa represents puddled transplanted rice.

Figure 2 .
Figure 2. System boundary of rice production.

Figure 2 .
Figure 2. System boundary of rice production.

Table 1 .
Energy equivalents of different energy inputs in puddled transplanted and direct-seeded rice production systems.

Table 2 .
Carbon coefficients of different inputs used in a rice production system.

Table 3 .
Energy consumption pattern of puddled transplanted rice and direct-seeded rice systems per hectare.

Table 3 .
Energy consumption pattern of puddled transplanted rice and direct-seeded rice systems per hectare.

Table 4 .
Energy indices in rice production system.
a Includes electricity, diesel fuel and human labor.b Includes biocide, chemical fertilizer, irrigation water, farmyard manure, organic fertilizers and machinery.c Includes human labor and farmyard manure, irrigation water and organic fertilizers.d Includes electricity, machinery, diesel fuel, chemical fertilizer and biocide.

Table 5 .
Average technical, pure and scale efficiency of paddy farmers (200 units).

Table 6 .
Technical efficiency (TE) and operational reduction of energy inputs (MJ) for the inefficient puddled transplanted rice system.

Table 7 .
Technical efficiency (TE) and operational reduction of energy inputs (MJ) for the inefficient rice fields in using the direct-seeded rice method.

Table 8 .
Total economic saving (TES) linked to the accomplishment of operational targets in puddled transplanted rice.

Table 9 .
Total economic saving (TES) linked to the accomplishment of operational targets in direct-seeded rice.