Eco-efficiency and technical efficiency of different integrated farming systems in eastern India

ABSTRACT

Integrated farming system (IFS), comprising various enterprises such as crop, horticulture, dairy, poultry and fishery optimally, uses farm resources to suit small and medium farms of India. In this study, 55 IFSs with different sizes and several enterprises were assessed in terms of their environment impact, eco-efficiency and technical efficiency (TE). The households, which own these IFS, were grouped into four major categories based on farm size viz., marginal (<0.8 ha), small (0.8–1.2 ha), medium (1.2–2.0 ha) and large (>2.0 ha). Among the farm size category, the highest eco-efficiency was recorded for marginal farms, whereas the lowest eco-efficiency was observed for small farms. The highest eco-efficiency was recorded for farm having five enterprises (INR195 kg CO₂eq. ha⁻¹) followed by four enterprises (INR190 kg CO₂eq.ha⁻¹). Among the enterprises, the highest eco-efficiency was recorded in fruits, whereas the lowest eco-efficiency was observed in dairy. The TE scores estimated using stochastic frontier analysis decreased with increasing farm size and the highest TE score was recorded for marginal farms, whereas medium and large farms recorded the lowest TE scores. The findings indicated the necessity for imparting training and demonstrations and funding support as well as subsidy for larger adoption of IFS to reap higher returns.

KEYWORDS:

• Acidification
• eco-efficiency
• eutrophication
• farming system
• global warming potential
• integrated life cycle analysis
• technical efficiency

1. Introduction

India’s agriculture is facing multiple and complex challenges of increasing population, climate change, decline in factor productivity and declining per capita operational land holding size, etc. Continuous fragmentation of agricultural land due to increasing population declined the average operational land holding size from 1.23 ha in 2005–2006 to 1.15 ha in 2010–2011 (ASG, 2015). In India, small and marginal farms are the dominant groups engaged in agriculture and allied enterprises. They represent 80% of operational farm holdings (ASG, 2015) and about 64 and 18.5% of total landholders (129.22 million) are marginal farmers holding <1 ha and small farmers with 1–2 ha, respectively. The proportion of marginal and small farmers in eastern India is about 70% (GOI, 2009). They are economically poor with low-resource endowment and engaged in farming operation and risk-prone environments (Khandwal, 2015). Greenhouse gas (GHG) emissions from energy, industry, agriculture, road transportation and other sectors of India account for 39, 25, 14 and 6%, respectively (Busby & Shidore, 2017). The main source of GHG emission in agriculture is enteric fermentation, which is close to 60%, followed by methane emission from rice cultivation close to 23% (Chaudhary, 2010). Rice is a dominant crop of this region, mostly cultivated in wet season under rainfed situation; hence, there is also a risk of crop losses due to flood/drought, insect pest and diseases due to the adverse effect of monsoonal climate. The integrated farming system (IFS), taking rice with livestock, fishery and other enterprises, can enhance adaptation to climate change, which would also reduce emission (Babu et al., 2021; Little & Edwards, 1999; Reddy et al., 2020), with significant co-benefits of rural livelihood and ensuring food security (Behera et al., 2010) of these farmers. In IFS, livestock acts as a risk minimization strategy along with multiple cropping and enhances the overall farm production and economic sustainability (Reijntjes et al., 1992). IFS mostly relies on renewable energy and resources from their own production system which are considered neutral with regard to GHG emission where emissions are balanced by the absorption of CO₂ in the process of photosynthesis and humification. GHG emissions from energy derived from wind, solar and water are considered close to negligible (Amponsah et al., 2014) while energy derived from biomass considered mostly as neutral with regard to greenhouse gas emissions during their operation. Choice of farm enterprises, optimum enterprise number and appropriate size of operational holding within the permissible land holding when managed properly can enhance productivity, livelihood, carbon storage and minimizing negative environmental consequences.

Life cycle assessment (LCA) has been widely used for eco-efficiency analysis of different production systems (Tyteca, 1996), including in agriculture (Van der Werf & Petit, 2002). Environmental impacts due to a production process in its life cycle are generally assessed through LCA. The LCA helps in identifying the options to improve the environmental performance of a production system (ISO, 2009a, 2009b). Renewable energy sources are considered to be zero (wind, solar and water), low (geothermal) or neutral (biomass) with regard to greenhouse gas emissions during their operation. A neutral source has emissions that are balanced by the amount of carbon dioxide absorbed during the growing process. Eco-efficiency can be defined as the ratio of created value per unit of environmental impact. (Thanawong et al., 2014; Van Passel et al., 2007). Whereas, the technical efficiency (TE) represents the firm’s ability to use input resources in the best possible way (Pradhan, 2018). A farm is technically efficient if it produces a larger output from the same input or the same output with less of one or more inputs without increasing the amount of other input. Shortfalls in efficiency indicate that the productivity of a system can be increased with the same level of input and technologies. In such cases, measures of efficiency are essential for determining the quantum of the gain possible by improving performance in production with the given technology (Binam et al., 2004), so that the farm productivity can be enhanced by focussing on improving efficiency rather than relying on new technologies. In this paper, we have attempted to characterize the different IFS models differing in farm size, number and type of enterprises in eastern India by computing eco-efficiency using partial life cycle analysis and technical efficiency by using stochastic production frontier with Cobb–Douglas function for optimizing the size of the operational farm size and the number of enterprises.

2. Materials and methods

2.1. Study area and its characteristics

Four districts in eastern state Odisha having a relatively higher number of operational integrated farming systems (IFS) were selected for the study. In the present study, only those IFS models were considered which are scientifically managed with proper guidance provided by extension functionaries and the state agricultural university. Some of the IFS models operated by farmers may not be considered as an IFS model because the farmers managing these models are neither guided by the Department/Extension functionaries nor having the basic components included in these models. The study area included three districts (Puri, Kendrapara and Khordha) coming under East and South East coastal table land and one district (Angul) coming under mid-central table land as per agro-climatic zone classification. A list of operational integrated farming systems were collected from the extension agencies of the state. A random household survey was conducted from the farmers that constitute about 5% of the operational farm holding of the study area and those who adopted the IFS model taking advisory from the extension agencies and the state agricultural university. We grouped the IFS owning farm households into four major categories based on the size of IFS i.e. marginal (<0.8 ha), small (0.8–1.2 ha), medium (1.2–2 ha) and large (>2 ha) farms. Of the total 55 farms identified for the survey, 22 were marginal, 21 were small, 9 were medium and 3 were of large categories. The number of enterprises under different IFS studied, ranged from four to seven.

2.2. Life cycle inventory (LCI) analysis

From the perspective of a whole farm enterprise, the primary data of various IFS were collected through personal interviews with the household head (farmer), usually in the presence of family members who are directly or indirectly associated with the farming systems in all the three cropping seasons (kharif, rabi and summer). Though the year to year variation of weather can have a significant effect on the agricultural practices, we collected information from a good year of operations during the 2013–2014 field survey. The functional land management units were subdivided on the basis of specific crops grown, the seasonality and multiple products coming from the same unit area. The survey was conducted using a structured questionnaire for the information regarding specific input-output data and other management practices the farmers have used in the immediate past. All the farmers owning selected IFSs (55) were covered during the survey. In addition, data on rainfall, minimum support price (MSP) of different crops were collected from the state- and national-level databases.

2.3. System boundary and inventories of datasets

Defining the system boundary line is an important component of LCA analysis and the output of LCA will not be valid in the absence of properly defined system boundary (Manfredi & Vignali, 2014; Nabavi-Pelesaraei et al., 2018). We used a farm gate approach as system boundary, where the production goes to the point of material leaving the farm, no processing and value addition was done within the farm gate. The conceptual diagram of system boundary is presented in the supplementary file (Figure S1). In this study, the potential environmental impacts were computed taking unit as ha of area for agricultural crops and fish, and per animal for dairy and poultry. The minimum data, required for each enterprise for the analysis of partial LCA and techno-economic analyses, are presented in the supplementary file (Table S1a).

2.3.1. GHG emission (direct and indirect) from IFS

To estimate the carbon footprint (CF), the GHG (CH₄, N₂O) emissions from different agricultural inputs, farm operations and enterprises were computed using the partial life cycle assessment (LCA) method (Dubey & Lal, 2009; Hillier et al., 2009). The GHG emissions from different IFS were modelled by following the standard norms as described by the International Panel on Climate Change (IPCC, 2006) and adjusted to regional scale emissions using relevant literatures (Dubey & Lal, 2009; Hillier et al., 2009; Yan et al., 2003; Yan et al., 2015). The emission of GHGs was classified into direct and indirect for calculating total GHG emission. Emission of N₂O from nitrogen fertilizer use, CH₄ from rice cultivation and CO₂ from manual and mechanical operations (tillage, sowing or transplanting, irrigation, weeding, fertilizer, pesticide application and harvesting, etc.) account for the direct emission (IPCC, 2006; Meera et al., 2019; Yan et al., 2003; Yan et al., 2015; Zou et al., 2007) whereas, GHG emitted during the agrochemical production accounts for the indirect emissions. The GHG emissions resulting from the use of agro-inputs, such as fertilizers, pesticides and farm mechanical operation, were estimated as per the methodology given by Yan et al. (2015). The emission of CH₄ and N₂O from paddy field was estimated using standard procedures with the reference values provided by IPCC (IPCC, 2006, 2014) (Supplementary file Table S1b). Similarly, GHG emissions from fossil fuel combustion by machinery were also computed by using all the individual emission and were expressed as t CO₂ eq.ha⁻¹. The GHG emission from dairy enterprise was calculated using secondary data and by following the methodology of Chhabra et al. (2013) and IPCC (2006). The emission of CH₄ from enteric fermentation and manure management was estimated and reported in the IPCC Tier-1 report (Supplementary file Table S1b). The GHG emission from pisciculture was calculated by using emission factor for all the input used (Adhikari et al., 2013). Emission from poultry enterprise was calculated using secondary data (GHG emission for per unit production of poultry meat) as given in the supplementary file (Table S1b) (Hillier et al., 2009; Pathak et al., 2010).

2.3.2. Energy input and output analyses

All the input energy resources used in the farm were taken into account for computing the energy balance. Labour (man days) and fossil fuel (diesel, petrol and electricity) consumed were categorized as direct energy inputs and all other activities such as seeds, chemical fertilizers, FYM, poultry manures, pesticides, machinery and transportation were categorized as indirect energy inputs. The total energy input was computed, by taking the sum of direct and indirect energy resources used in IFS. All the physical activities involved in the IFS were converted into energy units (MJ) by taking the standard conversion factors given in the supplementary file (Table S1b) (Devasenapathy et al., 2009; Heidari & Omid, 2011; Singh et al., 2008; Singh & Mittal, 1992,).

2.3.3. Eco-efficiency and impact assessment

The eco-efficiency was calculated by taking the gross income per environmental impact. The gross income from every individual enterprise was calculated by multiplying the total production with the minimum support price (MSP) of rice and maize during 2013–2014. MSP is fixed by the Government of India for certain specific agricultural commodities. In this study, we have used MSP only for rice and maize. For other commodities (Oil seed, Pulses, Vegetable, Fruits, Fishery, Dairy and Poultry) produced from the IFS for which MSPs were not available, we used market price for calculating gross return. The net return was computed by subtracting the cost of total production from gross income, which signifies eco-efficiency from the farmers’ perspective (Baum & Bieńkowski, 2020; Gustafson et al., 2014; Thanawong et al., 2014). There is no appropriate methodology for measuring the environmental impact created by different farm enterprises (Blengini & Busto, 2009; Thanawong et al., 2014). However, we estimated energy use (EU), freshwater use (WU) as input indicators and eutrophication potential (EP), acidification potential (AP) and global warming potential (GWP₁₀₀) as environmental impact (mid-point) indicators (Thanawong et al., 2014).

The GHG emissions were converted to a global warming potential for a 100-year time horizon (GWP₁₀₀) using 1, 34 and 298 equivalent factors for CO₂, CH₄ and N₂O, respectively in kg CO₂eq. (IPCC, 2014). EP and AP were estimated in terms of kg PO₄eqha⁻¹.and kg SO₂eq.ha⁻¹, respectively, using the generic methodology proposed by Heijungs et al. (1992) for the rice-based crop production system (Thanawong et al., 2014). Urea-induced NH₃ emissions from the rice were calculated using reference data from Sharma et al. (2014). Similarly, EP and AP for dairy, fishery and poultry were calculated using the secondary data reported by Kalhor et al. (2016), and Henriksson et al. (2018), respectively. The total amount of water used for different crops in the IFS system was calculated using the reference value reported by Mekonnen and Hoekstra (2011). Water used in dairy, fishery and poultry was computed from the reference data on the basis of water use per L or kg product reported by Mekonnen and Hoekstra (2012), Henriksson et al. (2018), and Mekonnen and Hoekstra (2012), respectively. Land use denotes the net land area utilized for the particular crop to the total area of IFS and being temporarily unavailable for other purposes. The land use was calculated in terms of the cropping intensity index (CII) and expressed in percentage.

2.3.4. Estimation of technical efficiency

2.3.4.1. Empirical model

The magnitude of technical efficiency (TE) represents the possibility of increasing the farm yield by adopting better management practices and technology (Hoang-Khac et al., 2021). Several studies in agriculture have been conducted to measure the efficiency of the system using either parametric Stochastic Frontier Analysis (SFA) or Data Envelopment Analysis (DEA), which is a nonparametric approach. The DEA requires only the information related to input and output quantities and the efficiency is measured relative to the highest observed performance rather than an average. However, a DEA-based estimate is sensitive to measurement errors or other noises in the data, because the DEA is deterministic and attributes all deviations from the frontier to inefficiencies. Mostly DEA is used for systems having a single enterprise like rice (Basavalingaiah et al., 2020; Krasachat, 2004; Pradhan, 2018; Tipi et al., 2009), wheat (Bhushan, 2005), sorghum (EtichChepng et al., 2014), cotton (Shafiq & Rehman, 2000) and apple (Ahmadabad et al., 2013). The integrated farming system we have taken in this study involves many enterprises; a collection of input- and output-related data from these enterprises is likely to result in varying extents of measurement errors and noise. Hence, we used SFA, which has strength to consider stochastic noise in data and also allows for the statistical testing of hypotheses concerning the production structure and degree of inefficiency. In this study, we attempted to calculate the technical efficiency of Integrated Farming Systems (IFSs), which have many enterprises as their components such as crops (food grain, fruits and vegetables), dairy, poultry and fishery. Mango et al. (2015) even used the stochastic frontier production model and the Cobb–Douglas production function to evaluate the technical efficiency of a single enterprise, i.e. maize production in Zimbabwe’s smallholder farming communities. In this study, the TE was computed using stochastic production frontier with the Cobb–Douglas function proposed by (1977) and Meeusen and van Den Broeck (1977) as it is the most widely used method to estimate TE of agricultural production (Battese, 1992). The enterprise-wise TE is computed using the stochastic production function given below.

ln = represents the natural logarithm (i.e. to base e), Y_ij: product output from jth enterprise of ith farm [yield (kg ha⁻¹), milk production (litre), fingerlings (number), poultry meat production (kg), fish production (kg ha⁻¹); i = 1, 2, … ., m (m: the number of farms, total number of farms are 55); j = 1,2, …  … k (k: total number of enterprises (10), the number of enterprises under different IFS studied ranged from four to seven) (Number of enterprises used in the different farms is presented in Table S3); α = unknown parameter to be estimated; β_jl = the parameters pertaining to the lthinput of jth enterprise; X_ij: inputs representing the ith farm and jth enterprise (Rice, Maize, Cereal, Pulses, Oil seed, Vegetables, Fruits, Dairy, Fisheries and Poultry); X_jl: inputs representing the lthinput for jth enterprise [(area (ha), seed rate (kg ha⁻¹), N (kg ha⁻¹), P (kg ha⁻¹), K (kg ha⁻¹), poultry manure (kg ha⁻¹), pesticides (ml ha⁻¹), tractor (hr), bullock (hr), fuel (litre), cow dung (kg), seed treatment (yes/no), irrigation (m³ ha⁻¹), number of irrigation, feed (kg), medicine (kg), fodder (kg), chicks (number), lime (kg ha⁻¹),] (the enterprise-wise variables are listed in Table S2); v_ij: random errors (normally and independently distributed) having zero mean and constant variance [N (0, σ_v²)] and independent of u_ij; u_ij is the technical inefficiency effects, associated with technical efficiency of production of farmers involved. The parameters of the stochastic frontier production function model were estimated by the maximum likelihood estimation (MLE) method using FRONTIER Version 4.1 (Coelli, 1996). The farm-wise TE of all the enterprises was calculated and averaged to represent the TE of individual farms. Furthermore, TE was also averaged for different farm size categories. Technically efficient or inefficient farms were identified based on the computed value of TE ranging between zero and one. A farm with a higher value of TE is technically more efficient than w farm with low TE.

A common and important methodological problem that arises in fitting stochastic frontier models is that the least-squares residuals estimated from these models may display skewness in the ‘wrong’ direction. Using the Likelihood ratio test, the CD function was significant, hence, the translog specification was not tested. Although the translog function is more appropriate for SFA, the effect on the technical efficiency was thought of as mild and negligible (Marie Simpachova and Simpach, 2020).

3. Results

3.1. Resource utilization for production and techno-economic performances

The quantity of inputs consumed in each farm size category is presented in Table 1. Among the categories, the highest input (fertilizer and pesticides) was used in the large farms, whereas the least input was used in the small farms. The average total fertilizer consumption was 70 and 95 kg NPK ha⁻¹ in small and large farms, respectively for crop production. There was also a wide variation in N fertilizer use which varied from 33 to 58 kg ha⁻¹ in marginal and large farms, respectively. The average consumption of N fertilizer followed the order: large > medium > small > marginal farms. Similarly, the average pesticide consumption was 1200 ml ha⁻¹ in large and 815 ml ha⁻¹ in small farms. The small farms required lesser off-farm inputs as most of them are mobilized from within. Whereas, the trend for labour use (man-days unit⁻¹) was exactly the reverse to that of off-farm input use. The decreasing trend for labour use followed the order: marginal > small > medium > large farms. Among the farm size categories, the highest labour (115 man-days unit⁻¹) was used in marginal farms and the lowest (83 man-days unit⁻¹) in large farms for crop production.

Table 1. Production factor use and techno-economic performances in selected rice-based integrated farming systems of eastern India.

Land holding (ha)	Production factors and performance	Rice	Maize	Cereals	Oil seed	Pulses	Vegetable	Fruits	Mean	Fishery	Dairy	Poultry	Mean
	(Reference unit)	Crop (ha)								Animal Unit
										ha		Head
<0.8	N (kg Nha^−1)	53	31	42	22	13	57	33	33	–	–	–	–
	P (kg P2O5 ha^−1)	26	17	22	28	27	28	17	24	–	–	–	–
	K (kg K2Oha^−1)	18	4	11	17	14	25	9	15	–	–	–	–
	Pesticides (ml ha^−1)	828	692	760	699	671	1990	170	858	–	–	–	–
	Man days (days unit^−1)	127	142	135	144	85	93	117	115	333	35	0	123
	Total energy (MJ unit^−1)	7918	6103	7011	6071	3602	7080	5052	5763	104,844	11,214	48	38,702
	Gross incomeha^−1 (INR)	60,832	294,735	177,783	69,582	29,337	245,625	491,677	202,801	11,400	16,342	18,236	15,326
	Cost of production ha^−1 (INR)	26,424	42,277	34,351	37,311	9456	28,142	36,426	29,137	3500	6794	8011	6102
	Net income ha^−1 (INR)	33,677	252,458	143,068	32,271	19,882	217,483	455,251	173,591	7900	9548	10,225	9224
	Return per rupee investment (INR)	2	7	5	2	3	9	13	6	3.3	2	2	2
0.8–1.2	N (kg Nha^−1)	50	29	39	33	12	58	25	33	–	–	–	–
	P (kg P2O5ha^−1)	25	15	20	22	25	29	13	22	–	–	–	–
	K (kg K2Oha^−1)	30	5	18	13	15	21	8	15	–	–	–	–
	Pesticides (mlha^−1)	780	619	699	550	621	2046	158	815	–	–	–	–
	Man days (days unit^−1)	105	140	122	96	68	78	143	101	288	33	0	107
	Total energy (MJ unit^−1)	7348	5920	6634	5060	3219	6856	5138	5381	105,324	11,437	43	38,935
	Gross income ha^−1 (INR)	58,632	361,827	210,230	37,366	26,465	224,136	413,245	182,288	16,000	11,870	21,833	16,568
	Cost of production ha^−1 (INR)	24,799	36,638	30,718	4253	10,888	27,495	31,680	21,007	5100	3804	12,406	7103
	Net income ha^−1 (INR)	32,431	325,190	178,810	33,113	15,577	196,641	381,565	161,141	10,900	8066	9427	9464
	Return per INR investment	2	10	6	9	2	8	13	8	3.1	3	2	3
1.2–2	N (kg Nha^−1)	55	32	44	28	17	62	28	36	–	–	–	–
	P (kg P2O5ha^−1)	28	18	23	29	34	31	14	26	–	–	–	–
	K (kg K2Oha^−1)	24	5	14	18	20	30	10	18	–	–	–	–
	Pesticides (mlha^−1)	864	702	783	730	850	2174	178	943	–	–	–	–
	Man days (days unit^−1)	126	158	142	124	80	68	83	99	251	37	0	96
	Total energy (MJ unit^−1)	7894	6703	7299	5948	3844	6713	4686	5698	115,828	11,506	39	42,458
	Gross income ha^−1 (INR)	54,194	415,188	234,691	64,190	28,800	195,425	628,562	230,334	18,000	9191	30,546	19,246
	Cost of production ha^−1 (INR)	25,260	37,972	31,616	32,508	6000	30,104	51,470	30,340	4250	4648	17,047	8648
	Net income ha^−1 (INR)	28,934	377,215	203,075	31,682	22,800	165,321	577,092	199,994	13,750	4543	13,499	10,597
	Return per INR investment	2	11	7	2	5	6	12	6	4	2	2	3
>2	N (kg Nha^−1)	79	–	79	–	–	62	34	58	–	–	–	–
	P (kg P2O5ha^−1)	39	–	39	–	–	31	17	29	–	–	–	–
	K (kg K2Oha^−1)	15	–	15	–	–	7	2	8	–	–	–	–
	Pesticides (mlha^−1)	1229	–	1229	–	–	2158	213	1200	–	–	–	–
	Man days (days unit^−1)	113	–	113	–	–	63	74	83	240	31	0	90
	Total energy (MJ unit^−1)	9987	–	9987	–	–	6338	4853	7059	115,310	11,351	29	42,230
	Gross income ha^−1 (INR)	86,110	–	86,110	–	–	127,741	383,200	199,017	19,000	20,108	47,505	28,871
	Cost of production ha^−1 (INR)	49,795	–	49,795	–	–	67,327	38,113	51,745	7500	12,671	17,000	12,390
	Net income ha^−1 (INR)	36,315	–	36,315	–	–	60,414	345,087	147,272	11,500	7437	30,505	16,481
	Return per INR investment	2	–	2	–	–	2	10	5	2.5	2	3	3

Large farms used higher input energy than smaller farms (Table 1). Large farms used the highest input energy (7059 MJ ha⁻¹) and small farms the lowest (5381 MJha⁻¹). The use of input energy followed the order: large > medium > marginal > small farms. Similarly, the average production cost was the highest (INR 51,745 ha⁻¹) (INR: Indian Rupee) for large farms and the lowest (INR 21,007 ha⁻¹) for small farms. The production cost followed the order: large > medium > marginal > small farms. The highest gross return was recorded in the medium and the lowest in small farms. The gross return from crop production followed the order: medium > marginal > large > small farms. The highest net return and the highest return per rupee investment were found in the marginal (INR 19,9994 ha⁻¹ and INR8) farms, and the lowest in the large farms (INR1 47,272 ha⁻¹ and INR5). The return per rupee investment followed the order: marginal > medium > small > large farms.

3.2. Environmental impact indicators

The environmental impact created by different enterprises according to farm size is presented in Table 2. Among the farm size category, the highest total GWP (4486 kg CO₂eq.ha⁻¹) was recorded from the medium and the lowest (3049 kg CO₂eq.ha⁻¹) from large farms. The GWP emission followed the order: medium > marginal > small > large farms. Among the different crop enterprises, the highest GWP was recorded in rice (1081–1381 kg CO₂ eq. ha⁻¹)followed by maize (919–1012 kg CO₂ eq. ha⁻¹) and vegetables (851–897 kg CO₂ eq. ha⁻¹).

Table 2. Environmental Impact indicators in selected rice-based integrated farming systems of eastern India.

Land categories (ha)	Unit	Impact indicator (Reference unit)	Rice	Maize	Cereals	Oil seed	Pulses	Vegetable	Fruits	Mean	Fishery	Dairy	Poultry	Mean
			Crops (ha)								Animal unit
											ha	head
<0.8	Output-related indicators	GWP100 (Total kg CO2 eq. unit^−1)	1236	919	1078	355	207	897	518	611	126	2.6	0.04	43
		AP (kg SO4 eq. unit^−1)	11	7	9	5	3	12	4	7	2202	12.9	0.05	738
		EP (kg PO4 eq. unit^−1)	9	8	9	8	8	10	4	8	600	25.9	0.02	209
	Input related indicators	WU [Water use (10^3 m^3) unit^−1]	6.6	4.7	5.6	4.1	2.4	3.7	8.2	4.9	14.2	0.42	0.01	146
		LU (CII %)	–	–	–	–	–	–	–	178	–	–	–	–
0.8–1.2	Output-related indicators	GWP100 (Total kg CO2 eq. unit^−1)	1081	948	1015	366	186	851	565	597	119	2.4	0.04	40
		AP (kg SO4 eq. unit^−1)	10	7	9	7	3	12	5	7	2349	13.2	0.06	787
		EP (kg PO4 eq. unit^−1)	9	9	9	7	7	10	4	7	684	27.1	0.02	237
	Input- related indicators	WU [Water use (10^3 m^3) unit^−1]	6.3	4.6	5.5	2.2	2.2	3.4	6.9	4.3	15.9	0.44	0.01	152
		LU (CII %)	–	–	–	–	–	–	–	179	–	–	–	–
1.2–2	Output- related indicators	GWP100 (Total kg CO2 eq. unit^−1)	1253	1012	1133	370	238	866	611	644	133	3.3	0.06	45
		AP (kg SO4 eq. unit^−1)	12	8	10	6	4	13	7	8	2743	17.5	0.06	920
		EP (kg PO4 eq. unit^−1)	10	9	10	8	10	11	8	9	775	28.8	0.02	268
	Input related indicators	WU [Water use (10^3 m^3) unit^−1]	5.8	4.3	5.1	3.7	2.4	2.9	10.5	4.9	17.2	0.49	0.01	171
		LU (CII %)	–	–	–	–	–	–	–	165	–	–	–	–
>2	Output- related indicators	GWP100 (Total kg CO2 eq. unit^−1)	1381	–	1381	–	–	888	628	966	149	3.6	0.07	51
		AP (kg SO4 eq. unit^−1)	17	–	17	–	–	13	8	13	2990	20.2	0.08	1003
		EP (kg PO4eq. unit^−1)	14	–	14	–	–	11	9	11	808	31.8	0.03	280
	Input- related indicators	WU [Water use (10^3 m^3) unit^−1]	9.3	–	9.3	–	–	1.9	6.4	5.9	17.7	0.57	0.02	197
		LU (CII %)	–	–	–	–	–	–	–	158	–	–	–	–

where GWP₁₀₀ is the Global warming potential, EP is the Eutrophication Potential, AP is the Acidification Potential, WU is the freshwater resource use, LU is the land use and CII is the cropping intensity index.

The data on acidification potential (AP) and eutrophication potential (EP) are presented in Table 2. The highest (3048 kg SO₄ eq. ha⁻¹) and the lowest AP (2256 kg SO₄ eq. ha⁻¹) were recorded in large and marginal farms, respectively. In contrast to AP, the EP was highest (280 kgPO₄ eq. ha⁻¹) in the large and the lowest (209 kg PO₄ eq. ha⁻¹) in marginal farms for crop production. Among the enterprises, the highest AP and EP were recorded in fishery (2202–2990 kg SO₄ eq.ha⁻¹, 600–808 kg PO₄ eq.ha⁻¹), followed by dairy (12.9–20.2 kg SO₄ eq. per animal, 25.9–31.8 kg PO₄ eq. per animal) and rice (11–17 kg SO₄ eq. ha⁻¹, 9–14 kg PO₄ eq. ha⁻¹).

The intensity of land use represented by the cropping intensity index (CII %) and water use of different enterprises are presented in Table 2. Among the farm size category, small farms recorded the highest cropping intensity (179%), and large farms had the lowest (158%). The intensity of land use followed the order of small > marginal > medium > large farms. In terms of cropping intensity, land use between small (179%) and marginal (178%) farmers was comparable. Water use data showed that enterprise-wise water requirements increased with increasing farm size (Table 2). Among the farm size category, the highest crop water use (5.9 × 10³ m³ ha⁻¹) was recorded for the large and the lowest (4.3 × 10³ m³ ha⁻¹) for the marginal farms. The water use of farms followed the order large > medium > small > marginal.

3.3. Eco-efficiency

The eco-efficiency of different IFS based on farm size category (Table 3) showed the highest eco-efficiency (INR201kg CO₂eq.ha⁻¹) for marginal and the lowest (INR166 kg CO₂eq.ha⁻¹) in small farms. The eco-efficiency followed the order: marginal > medium > large > small farm size category. The highest eco-efficiency (INR195 kg CO₂eq.ha⁻¹) was recorded when five enterprises were taken followed by four enterprises (INR190 kgCO₂ eq. ha⁻¹). The eco-efficiency was reduced when the number of enterprise combinations increased beyond five (Table 3). Among the enterprises, the highest eco-efficiency was recorded for fruits (INR539-1112 kg CO₂eq. ha⁻¹), followed by maize (INR152-600 kg CO₂eq. ha⁻¹) and vegetables (INR221-271 kg CO₂eq. ha⁻¹). The lowest eco-efficiency (INR6-19 kg CO₂eq.ha⁻¹) was recorded for dairy. In addition, the AP and EP were higher in the large farm than other farm sizes.

Table 3. Impact of farm size and number of enterprises on Eco-efficiency and technical efficiency of the integrated farming system.

		Eco-efficiency (Rs./kg CO2 eq)			Technical efficiency
		Mean	SD	SE	Mean	SD	SE
Farm size	<0.8 ha	201	124.69	26.58	0.77	0.10	0.022
	0.8–1.2 ha	166	114.64	25.00	0.75	0.07	0.021
	1.2–2 ha	179	33.39	11.1	0.74	0.06	0.021
	>2 ha	189	47.40	27.4	0.74	0.02	0.009
Enterprise number	4 (10)	190	99.42	31.4	0.77	0.11	0.033
	5 (25)	195	136.51	29.1	0.74	0.06	0.013
	6 (16)	179	105.12	24.8	0.76	0.11	0.025
	≥7 (4)	177	34.35	15.4	0.77	0.09	0.038

Note: The data in the parenthesis is the number of Integrated farming system models for the corresponding enterprise number.

SD: standard deviation; SE: standard error.

3.4. Technical efficiency

The TE scores of different IFS based on the enterprise combinations were computed using stochastic frontier analysis and averaged over the farm sizes. Among the farm size categories, the highest technical efficiency (77%) (Table 3) was recorded for the marginal and the lowest (74%) for the medium farms. The TE was found to decrease with increasing farm sizes which followed the order: marginal > small > large > medium. Technical efficiency calculated based on the numbe of enterprises indicated that the combination of seven enterprises recorded the highest TE (Table 3).

4. Discussions

4.1. Resource consumption and environmental impact

The owners of the bigger farms used more off-farm input such as energy, fertilizers and pesticides because they have access to the credit market or capital market and can get a loan from a bank or other financial institution, while owners of the small farms are credit-constrained because of lacking securities. Bigger farms are favourable for machine operation while small farms are mostly cultivated manually using on-farm inputs. Thus, smaller farms are more efficient in terms of GWP and energy. Farming practices, such as tillage operations, type of crops, method and time of fertilizer application and crop residue use (Goglio et al., 2014; Pappa et al., 2011; Saggar, 2010), affect the GHG emission and energy consumption. Poultry and fishery consumed more energy than the crop and dairy enterprise due to the extensive use of electricity in poultry and fishery. Small farms use more energy in crop production, whereas large farms use more energy for poultry and fishery. Using LCA, Goglio et al. (2014) and Yan et al. (2015) reported that GHG emission from crop production was affected by crop management practices and farm size. The mixedcrop–livestock system, practised mainly in small farms, utilizes crop by-products and residues for livestock feeding, thus minimizing the GWP emission on the acquisition of external feeds (Eshel, 2021; Garg et al., 2016). Zhang et al. (2011) reported that rice treated with chemical fertilizer with duck emitted 6.7% less CH₄ than the rice with chemical fertilizer alone. The GWP of rice obtained from our study is lower than the value reported by Kumar et al. (2016). Among the enterprises, the crop contributed the highest percentage to GWP followed by fishery, irrespective of land-holding size.

The acidification potential (AP) expressed as kg SO₂-equivalents unit⁻¹ (per ha in crop and fishery and animal for dairy and poultry), was higher in the large farm system applied with higher inorganic nitrogenous fertilizers than marginal and small farms. Urea constitutes a higher percentage of nitrogen fertilizer used by farmers, gets converted to nitrate in the soil through nitrification and hydrogen ions are released into the soil and increase the soil acidity over a long period. Researchers have also reported that the application of N fertilizers lowers the soil pH (Chien et al., 2008; Malhi et al., 2000; Schroder et al., 2011) and decreases soil pH followed the order ammonium sulphate < ammonium nitrate = urea (Chien et al., 2008). Moreover, other fertilizers, such as single super phosphate (SSP) and di-ammonium phosphate (DAP) mostly used in larger farmers, are converted to sulphate and nitrate through microbial-mediated chemical reaction increasing acidity (i.e. free protons, H⁺) into the soil (Barak et al., 1997). The application of N to the soil also decreases the saturation of Ca and Mg which leads to the reduction of soil acidity (Fageria et al., 2010). There are reports that the cultivation of legume crops also increases soil acidity since it uptakes more basic cations (Ca, Mg and K) (Slattery et al., 1991). In our study, higher soil acidification was recorded in large farms. Among the individual components, the fish component recorded higher AP than the crops. Comparison with the animal component is not possible here because of different allocation methods/characterization factors used. Irrespective of land holding size shows, fishery contributes higher AP followed by poultry.

The eutrophication potential (EP) was worked out by computing the loss of nutrients from chemical fertilizer use, FYM, poultry droppings and fish feeds. The EP was 1.3 times higher in large farms compared to small and marginal farms. Fishery contributed a higher percentage (42–87%) to EP compared to other enterprises of IFS. Higher use of fertilisers and larger size of the animal component increases the nutrient load in the system (Smith, 2003) and enhances the EP. Integration of fish and/or animal components in farming increases the N and P load through surface runoff that carries manures to the fish system since the feed used in animal production is rich in these nutrients (Sharpley et al., 1992). The higher concentration of fertilizers like N and P promotes the growth and development of algae and vascular plants in freshwater ecosystems (Smith, 1998). Dwivedi and Panday (2001) reported phosphorus along with nitrogen is the main cause of eutrophication in India. Nitrate enrichment in the surface and groundwater is mostly due to leaching from the agricultural system (Robertson & Vitousek, 2009). Small changes in N fertilization (Goglio et al., 2014) and the use of rice straw for livestock fodder (Soam et al., 2017) will reduce EP with minimal impact on crop productivity.

On large farms, 70% of the land is used for rice cultivation, which requires huge quantities of water (Vijayakumar, 2017), in addition, these farms adopt water-intensive enterprises such as dairy, poultry and fishery for commercial purposes. On the contrary, crops like oilseed and pulses are grown in small and marginal farms by using the residual moisture in the dry season and rice crops in the wet season along with fish on a dug-out pond on a small fraction of land and some animals such as cattle and poultry on the pond dyke. In such IFS, the same water is used for multiple enterprises such as crops, fishery, dairy and poultry (Singh & Gautam, 2002) and enhances water productivity. Thus, the total use of water is higher in large farms compared to small farm size. The water footprint of sole rice cultivation in India is 2020 m³/t with an additional 1403 m³/t as percolation loss (Chapagain & Hoekstra, 2011). The water productivity of rice production can be enhanced by pond-based rice-fish farming with vegetables on the dyke (Samra et al., 2003). In rainfed areas, the pond is used to collect the rainwater which can subsequently be used as lifesaving irrigation for different enterprises cultivated in the Rabi season in the IFS (Ansari et al., 2014). The total water use among the individual components of the IFS does not reflect its absolute use since the same water also contributes directly or indirectly to the production and productivity of other components. Thus, the direct comparison needs the development of different allocation methods/ characterization factors for IFS.

4.2. Efficiency of environmental, technical and economic performance

A farming system that generates higher economic returns with minimum negative impacts on the environment is considered more eco-efficient. In this study, the highest average eco-efficiency score was recorded in marginal farms (Table 2), which may be due to higher productivity with minimum use of external inputs. There are complementary and synergistic interactions between different enterprises in marginal farms to the maximum possible limit by efficient utilization of on-farm waste (a by-product of inter-related enterprises) such as dairy and poultry and require less amount of external inputs. In these farms, operations are performed manually involving the minimum use of fossil fuels (petrol, diesel and kerosene). In contrast, large farms depend more on external inputs (fertilizers and pesticides) and machinery that increase fossil fuel consumption and CO₂ emission. All internal expenses were integrated to calculate net returns as followed by Gustafson et al. (2014), Thanawong et al. (2014), and Baum and Bieńkowski (2020). However, external costs related to environmental externalities were ignored during the calculation of profitability. Moreover, marginal farms use the land to take more crops in a year cultivating in both wet and dry seasons (higher cropping intensity) using family labour that ensures higher production in a year. While working on a cotton-based farming system, Ullah et al. (2016) reported higher eco-efficiency for small farms. Yiridoe et al. (2013) reported a lower level of N (eco-efficient fertilization rate) substantially reduced the eco-efficiency; nevertheless, it involved a certain degree of trade-off with crop yield. The inclusion of cash crops and high biomass-producing crops in IFS increase eco-efficiency in comparison to low-value traditional crops (Yiridoe et al., 2013). We found fiveenterprise combinations were more eco-efficient than the other combinations. The ideal combination of crops spatially and temporally links appropriately with contrasting enterprises such as dairy, fishery and poultry within the system, which can enhance the eco-efficiency of IFS by exploiting the positive interaction and synergy among the enterprises (Wilkins, 2008). Unlike direct N₂O-N emissions, indirect emissions per Mg milk decreased with decreasing intensity, caused by the lower reliance on external inputs, especially concentrates and on the conventional farms, mineral N-fertilizer. Because of this, indirect N₂O-N emissions per Mg milk were approximately 10–20% lower on organic dairy farms. Combining direct and indirect N₂O-N emissions, organic dairy farms generated about 10–15% lower emissions per Mg milk (Bos et al., 2014).

The highest average TE was observed in marginal farms than in other farm size categories, where inputs were mobilized from within the system. Such recycling of resources enables farmers to reduce the use of off-farm chemical fertilizers and other inputs by efficient utilization of resources. In such farms, most of the farm operations, such as irrigation, fertilizer and pesticide application, weeding and harvesting, were carried out by the family members which reduced the overall cost of cultivation. While working on rice productivity using DEA, Pradhan (2018) reported that the size of the operational holdings and efficient utilization of resources are some of the factors that govern the TE. Whereas in large farms, the major inputs, such as chemical fertilizer, pesticides and labour input for various agricultural operations, were outsourced. But in some cases, small farms with a monocrop-based system, such as cereals and horticulture crops, are technically inefficient due to low productivity and low resource use efficiency (Bhatt & Bhat, 2014; Rajendran, 2014). Our findings suggest that most of the IFS were operating with a medium-to-high level of technical efficiency and no major variation in technical efficiency (TE) was found among different enterprises. The integrated farming system was more profitable and more efficient than non-integrated farming methods due to the combination of enterprises, where the output of both enterprises is used as the input for each other (Bonaudo et al., 2014; Dao & Lewis, 2013). Bhatt and Bhat (2014) reported that an increase in farm size increased the marginal returns and lowered the marginal cost. However, beyond a certain size, marginal returns will decrease and marginal cost will increase (but not necessarily simultaneously), at the optimum size, the marginal returns equals marginal costs. In this study, TE and eco-efficiency were estimated from the pooled data only and thereafter averaged to factor out the differences where statistical validation is not required.

Cost of cultivation, gross return and net return were recorded as the highest for large farms. However, the return per rupee investment was lowest in large farms due to over-dependence on off-farm inputs and higher investment in farm mechanization that increased the production cost. In contrast, small farms, mostly engaged family labourers for farm operations with less dependence on heavy machinery which reduces the cost of cultivation. The highest gross return was obtained in medium farm size due to the adoption of enterprises such as fruits, vegetables, pulses and oilseed crops that fetched a higher market price, whereas in large farms, a major portion of land is used for rice cultivation.

4.3. Uncertainty of LCA for IFS

The environmental impact, accounting for the input and output resources in food production systems by using LCA, has been extensively studied (Goossens et al., 2017; Nemecek et al., 2016; Notarnicola et al., 2017; Soussana, 2014);. The LCA study requires detailed input and output data in the crop-livestock system (Caffrey & Veal, 2013). In a developing country like India, conducting a full LCA study is very challenging due to the lack of data and higher regional variability with respect to different farms, farming practices and inputs used. Moreover, these systems are evolving at a faster pace in recent time, and published literature is becoming outdated and not providing robust information on the specific production system.

We used the partial LCA to estimate the GHGs emitted from the crop and livestock while considering its impact on the surrounding environment. The GHG emissions were calculated by using the IPCC Tier I methodology based on global level data. We used well-established reference values (factors) and regional scale CH₄ emission factors for GWP calculation to reduce the uncertainty level. As per IPCC guidelines, the default values for N₂O emission factors of livestock waste management systems are still in consideration and they widely vary from region to region. For calculating acidification and eutrophication potential, the large-scale national averages were used which may increase the uncertainty. The sampling error could be higher due to inadequate samples, time frame and changes in management practices over time (Cooper et al., 2011). Uncertainty-related GHG emission from farms is also associated with both spatial and temporal variations (Gibbons et al., 2006). Allocative efficiency was ignored due to imperfections in the market; for dissimilarities in the type and quality of produce as well as price (MSP for few products and these MSP was not available, prevalent local/market rice was considered); lack of data on consumers’ preferences, many items are not traded frequently and many other reasons.

5. Conclusion

The eco-efficiency of several IFS in eastern India differed across farm size categories. Due to their greater profitability and less environmental impact, marginal and medium-sized farms are substantially more eco-efficient. Integrating multiple businesses minimizes the demand for off-farm inputs by recycling on-farm waste, making a sustainable living in eastern India. Marginal farms work well because they cluster many enterprises into a small area. However, all farms have a lot of room to increase their technical efficiency by switching to a more scientific approach. Based on the results of this research, it is recommended that farmers be provided with training, demonstrations and credit facilitation to encourage them to adopt an IFS strategy, which is more profitable, efficient and environment-friendly than a non-integrated mono-cropping agricultural system.

Acknowledgements

The authors acknowledge the financial help provided by the Ministry of Earth Sciences, Govt. of India and also thank the Director General of the Indian Council of Agricultural Research (ICAR) and the Director of ICAR-National Rice Research Institute (NRRI) for giving all the necessary help in executing the work. This work is a part of the farming system project (Project No 2.4) running in ICAR-NRRI, Cuttack and partially funded by DST, Govt. of India (Grant No DST/CCP/MRDP/143/2018 (G)). The help provided by Odisha University of Agricultural Technology, Bhubaneswar, Odisha for sharing the data is gratefully acknowledged. This study is a part of the project entitled ‘Delivering food security on limited land (DEVIL; Belmont Forum / FACCE-JPI via NERC: NE/M021327/1)’.

Disclosure statement

No potential conflict of interest was reported by the author(s).