A quantitative framework for characterizing the current and obtaining a future sustainable agricultural production mix meeting environmental, nutritional, and economic goals

Transitioning toward sustainable agriculture production is key in achieving sustainable development goals. However, quantifying the sustainability status of current agricultural mix and figuring out a future sustainable crop and livestock production mix that is nutritionally adequate, low in environmental impact and profitable to farmers, is a formidable challenge for any country. Here through a case-study on Indian agriculture, we demonstrate the application of a framework to first characterize the current agricultural mix by comparing 31 sustainability indicators with respective benchmarks across nutrition (social), environment and economic dimensions in each of the 36 Indian states. Next, we demonstrate the application of mathematical optimization algorithms to calculate the sustainable production mix in each state with the objective of maximizing the farmers’ profit under the constraint that the total national agricultural production meets the nutritional requirements of population without exceeding the current environmental footprint levels. Characterization results show that India does not produce enough of certain micronutrients (vitamin-E, and choline) and the carbon, nitrogen and freshwater footprints exceed their planetary boundaries while the farmers’ income remains below national targets. Optimization algorithm generated the sustainable production amounts of 138 crop and livestock items in each state and showed that fruits, vegetables, legumes, and oilseeds production needs to increase by 50%–100% while the cereals production amounts need to reduce to 50% of current levels in India. This will result in an increase in farmer’s income by 25% on average, fulfil the nutritional requirements of population and reduce the environmental footprint by 10%. Our approach can act as a template for other countries in collecting necessary data on sustainability indicators and determining optimum crop and livestock mix.


Agricultural production sector and SDGs: general background
Agriculture is one such sector which can significantly contribute towards achievement of multiple sustainable development goals (SDGs) such as poverty reduction (SDG 1), hunger eradication (SDG 2), human health and wellbeing (SDG 3), climate change adaptation and mitigation (SDG 13), life under water (SDG 14), life on land (SDG 15), and others (Brown et al 2021).However, current agriculture sector of most nations (especially the low-income ones) is not sustainable as it does not produce enough essential micronutrients, causes high environmental impacts and keeps farmers' income low (Willett et al 2019, Zhang et al 2021).Transitioning from currently unsustainable crop and livestock production mix to a sustainable agricultural production mix therefore is fundamental to make progress on multiple SDGs for most countries.
A sustainable agricultural mix will ensure quantitative targets on population macro and micro nutrition requirements, environmental planetary boundaries, farmers' profit and cultural food preferences are met (Willett et al 2019).The need of the hour is policies that take a holistic approach in deciding a sustainabile agriculture production mix by considering impacts on multiple indicators of social, environmental and economic dimensions of sustainability simultaneously by involving all stakeholders (FABLE 2019, Davis et al 2022b, Mukherjee 2022, Chaudhary 2023) rather than in silos as is often done (Damerau et al 2020).
In the absence of a holistic approach, agricultural policies designed in silos will not achieve sustainability as it can lead to achievement of one goal (e.g.nutritional security) at the cost of others (e.g., environmental degradation).For example, increased cereals production through increased fertilizers and irrigation application (e.g., during green revolution) combined with policies providing subsidies to farmers on fertilizer and irrigation water withdrawal, can lead to caloric sufficiency at the national level but can generate environmental externalities such as freshwater scarcity, eutrophication, land degradation etc. (Pingali 2012, Birkenholtz 2018, Jain et al 2021).Other examples include how policies aimed at mitigating one environmental problem (groundwater depletion) inadvertently exacerbated the other (air pollution) (Singh et al 2019).

Key research gaps and past literature
A first step towards implementing policies to enable a transition towards sustainable agricultural production mix in any country is to characterize the sustainability status of current agricultural production mix by comparing (juxtaposing) the current indicator scores with their respective benchmarks or targets.Once this is done, the next step would be to come up with a sustainable (or optimized) agricultural mix in which all the sustainability indicator scores are higher than or equal to the benchmark values while ensuring no trade-offs occur across the indicators.
The first step of characterization entails quantification of damage done to different domains of the environment due to crop and livestock production in a country vis-à-vis global environmental planetary boundaries (Steffen et al 2015), calculating what amounts of different nutrients (calories, protein, vitamins, minerals etc.) essential for human health are being produced vis-à-vis national population requirements (Chen et al 2021) and what monetary profit farmers derive from their produce vis-à-vis national target for farmers income.Moreover, since the climate, soil, growing conditions and other farm sector policies often vary from state to state especially for large countries, it is necessary to conduct such an analysis at a sub-national rather than at a national level so as to take into account regional production characteristics and constraints in achieving a sustainable mix (Das et al 2018, Zhang et al 2021, Mukherjee 2022).
However, generating such quantitative information on how well a country's different regions are doing in terms of agricultural sustainability is challenging primarily because it entails connecting the data on crop and livestock production statistics, food composition tables containing nutrient density of food items, environmental footprints of individual foods, cost of cultivation, food market price and farmer profit per food item.These data are scattered across a wide range of publications and websites, requiring massive efforts to synthesize them.A handful of past studies have attempted to characterize the sustainability status of national food systems and have found that each country needs to improve on certain indicators and no country is perfect (Chaudhary et al 2018, Willet et al 2019, Zhang et al 2021).However, the scale of assessment of these studies is at a national rather than sub-national level and often they do not consider the economic indicators (farmers' income).
Moreover, even when such data on sustainability indicators is compiled, the second step of obtaining a sustainable production mix is a formidable challenge as it entails ensuring several constraints on different sustainability indicators are met and trade-offs across indicators is avoided.Past studies attempting this are limited to designing a limited number of future scenarios focusing on improvement in a limited number of sustainability indicators often at a national level only (Davis et al 2019, Damerau et al 2020, Ashok et al 2021, Devineni et al 2022).

Goal and scope
This study aims to fill the above research gaps by demonstrating a quantitative framework that first characterizes the sustainability status agricultural mix through 31 indicators and then leverages linear optimization algorithm to come up with an optimum (sustainable) crop mix in each state of India with the objective of maximizing the farmers' profit under the constraint that the total national agricultural production meets the daily requirements of country's population for 25 essential macro and micro nutrients without exceeding the current environmental footprint (for greenhouse gas emissions, freshwater use, cropland use, nitrogen, and phosphorus application).The optimization algorithm generates the necessary changes in production amounts (in tons) of each of 138 crops and livestock items at a sub-national level in all 36 states and Union Territories of India compared with their current production levels.
We chose India as case-study to demonstrate this approach because it is the most populous country in the world and home to over 1.4 billion people with just 2.4% of world's land and 4% of global water resources.On the economic dimension of sustainability, around 50% of country's population is directly dependent upon agriculture sector for their livelihoods, yet the sector contributes less than 18% to the country's total economic output with majority of farmers being marginal or small land owners often living in poverty with stagnant incomes (Binswanger-Mkhize 2013, Ranganathan 2015, Chandrasekhar and Mehrotra 2016, Chand 2017, Gulati et al 2021).
On the contribution to social (nutrition security) dimension, the Indian agriculture sector performs even worse than many sub-Saharan Africa and Asian countries whose per capita incomes are lower than India as it suffers from triple burden of malnutrition where majority of population suffers from either undernourishment (prevalance of wasted, stunted, and underweight children), micronutrient deficiency (not enough intake of essential vitamins and minerals), or overweight/obesity issues (Meenakshi 2016, Das et al 2018, Pingali et al 2019).
Finally, Indian agriculture sector is responsible for high levels of air pollution, water pollution (Aditya et

Crop and livestock area, production and yield data per state data
We obtained the state-level statistics on the area (ha), production (tonnes), and yields (tonnes/ha) of 126 crops and 12 livestock items (average over 2014-2017; data for recent years is not yet available) from the records of the Ministry of Agriculture and Farmers Welfare, Government of India (Directorate of Economics and Statistics 2024).

Environmental footprint per item per state data
Next we compiled the environmental characterization factors (CFs) of each crop and livestock item from existing literature (supplementary table S1).The Indian-specific crop and livestock-specific CFs for GHG emissions (kgCO 2 e/kg), and freshwater consumption (L kg −1 ) were obtained from Damerau et al (2020).For certain crops, the freshwater CFs were not available through Damerau et al (2020) and we therefore used the values provided by Mekonnen and Hoekstra (2010).The CFs for nitrogen use (gN kg −1 ) and phosphorus use (gP kg −1 ) per item were obtained from Springmann et al (2018).The CFs for each of the 138 crop and livestock items were multiplied with their production amounts (in tons) per state to calculate the environmental footprint of each crop and livestock item produced in each of the 36 state and Union Territories of India (a total of 2505 combinations).

Nutrient production per item per state data
The amounts of 25 essential macro and micro nutrient content per 100 g of each of the 138 crop and livestock item produced in India was obtained from The Indian Food Composition Tables (Longvah et al 2017).The nutrient content of each crop and livestock items were multiplied with their production amounts (in tons) per state to calculate the nutrient supplied by each item in each of the 36 state and Union Territories of India (a total of 2505 combinations).
Since not all crop and livestock production amount is consumed by humans and a part of them is used for other purposes such as animal feed, seed or non-edible products or lost during different stages such as during farm stage (food loss), processing, or consumption stage, we calculated the net nutrient production per crop and state destined for human consumption by applying the necessary correction factors for each crop (Chen et al 2021).

Cost of cultivation, revenue and farmers' profit per crop per state data
We obtained the cost of cultivation (production) and the selling price of each agricultural item per state (averaged for the years 2014-17) from the Ministry of Agriculture and Farmers Welfare and Indiastat websites (Directorate of Economics and Statistic 2024, Directorate of Marketing & Inspection 2024, Indiastat 2024).Multiplying the cost of cultivation and selling prices data for each item with their respective production amounts, we obtained the total cost incurred and total revenue received by farmers in each state.As an indicator of economic sustainability, we calculated the farmers' profit (total revenue from sales minus total cost divided by total cost) in % for each state.Supplementary figure S2 shows the ratio of selling price (received by farmer for selling one ton of crop) and cost of cultivation (cost incurred by farmer to produce one ton of crop) for 15 broad crop categories in India (averaged over 2014-17).

Characterizing the sustainability status of current agricultural production mix
Once the above raw data on different indicators was collected, we characterized the sustainability status of current agricultural production mix of the country by comparing these indicators with respective benchmarks.
Regarding nutrition indicators, the total annual national production of each of 25 essential nutrients were compared with their respective population requirements (table 1) as per India's national institute of nutrition (NIN 2011, Chen et al 2021).We also carried out this comparison after subtracting food exports and adding imports to the total nutrient production (FAO 2024).
For environmental benchmarking, the total environmental footprint of country's agricultural production were compared with the respective Indian food production-related planetary boundaries that are: 0.885 GtCO2eq.yr −1 for GHG emissions; 373 km 3 yr −1 for Bluewater use; 2.377 million km 2 for land use; 13 TgN yr −1 for nitrogen and 3.013 TgP yr −1 for phosphorus application.These values were calculated by scaling down the available global food-related boundaries (Springmann et al 2018) based on global population share of India (O'Neill et al 2018).
Regarding the sustainability of economic dimension, there is not any acceptable benchmark at the global level per se.We therefore adopted the highest farmers' profit (in %) currently achieved in any major state (population > 1 million) among the 36 states as a national benchmark.

Optimized (sustainable) agricultural production mix
To calculate the optimized production mix for each of 36 Indian states, we start with current production amounts of 138 crop and livestock items and use linear optimization algorithm that aims to maximise farmers' profit while meeting nutritional, environmental and cultural constraints (table 1).
In this study, the objective function of our linear programming model was defined as the maximization of the farmers' profit and was expressed mathematically by equation (1) for each state: max where Q opt,i is the annual optimum production amounts of agriculture item i and p i is the profit derived from item i (selling price-cost of cultivation).The optimizations were carried out in the R (3.5.1) statistical software environment.The model was coded in the Rsymphony package using the Rsymphony_solve_LP function (Harter et al 2017).We carried out 36 such optimization runs (one for each state).We assumed that both the export and import amounts of each item in the optimized scenario will remain the same at their respective current levels (Food and Agricultural Organization of the United Nations 2024).The above objective needs to be achieved while meeting the nutritional, environmental and cultural acceptability constraints listed in table 1 below.See supplementary text A for full details of our optimization model.

Current agricultural production mix
In terms of annual production mass-cereals (rice, wheat, maize mainly), milk, vegetables (potato, onion, tomato mainly), fruits (banana, mango), nuts (groundnut and coconut mainly) and cash crops (sugarcane, cotton mainly) are the dominant agricultural commodities produced in India.
North Indian states (Uttar Pradesh, Haryana, Punjab, Rajasthan, Himachal Pradesh, Uttarakhand, Jammu and Kashmir) produce majority of cereals, cash crops, oilseeds and milk while South Indian states (Andhra Pradesh, Karnataka, Kerela, Telengana and Tamil Nadu) produce the highest amounts of condiments, spices, nuts, and eggs, fish and poultry.Almost half of country's beans and peas are grown in the two central states (Madhya Pradesh and Chhattisgarh) followed by western states of Gujarat and Maharashtra.The uneven distribution of agricultural commodity production across the country is primarily due to suitability of climate, soil characteristics and local dietary preferences.Fruits, vegetable, lentils and pulses production distribution is much evenly distributed across India.We also found that the yield (tons/ha) of a particular crop also varies substantially across India.Higher yields are usually found in southern and western states compared with other four regions.
The state-specific scores of all 31 nutrition (social), environmental and economic sustainability indicators are presented in supplementary excel table S1.

Characterizing the nutritional sustainability of current agricultural production mix
Despite having world's second largest arable area (after the United States) of around 180 million hectares and multiple crops per year in many regions, we found that the total annual agricultural production of India is unable to fulfill the population requirements (benchmark) for four nutrients-polyunsaturated fatty acids (PUFA), Vitamin A, Vitamin E and Choline.Even after considering the imports of agricultural items, the requirements of Vitamin A and Choline remain unfulfilled (table 1).This is because the foods with high density of these nutrients (e.g.eggs, meat for choline and carrots, pumpkin, sweet potato for vitamin A) are not produced and available in sufficient quantities nationwide.In addition, the production amounts of Vitamin B12, folate (Vitamin B9), and calcium are also borderline when compared with their annual population requirements.The production amounts of all other 19 essential macro and micro nutrients are sufficient with respect to their benchmark amounts needed to fulfill the annual population requirements (table 1). Figure 1(A) shows the contribution (%) of each state to the total Indian nutrient production.Consistent with the production amounts of crops, northern states produce the majority of any of the 25 given essential nutrients (∼37% of total) followed by southern states (20%) while the eastern, western, and central states produce ∼14% of the total of any given nutrient.There were some exceptions though such as for vitamin K whose production is concentrated in southern states (35%) instead of north (16%) or selenium whose production is disproportinely high in northern states (53% of total) and low in southern states (11%).Supplementary table S1 provides the amounts of each of 25 nutrients currently produced in each of the 36 states.

Characterizing the environmental sustainability of current agricultural production mix
We found that the carbon, freshwater and nitrogen footprint of current Indian agriculture production mix is 18%, 86% and 7% higher than the national benchmark planetary boundaries respectively while the land and phosphorus footprints are within their benchmark thresholds.
Supplementary figure S1 shows the crop categories with high or low environmental characterization factors (CFs; environmental footprint per kg) relatively.Red meat products have the highest CFs among all 15 crop categories for GHG emissions and land use domains of the environment.S1 for indicator values per state and supplementary figures S3 and S4 for % contribution of each state to the total national environmental footprint and nutrient production.
Nuts have the highest CF for bluewater footprint followed by red meat, poultry, lentils, eggs and cash crops.Eggs have highest nitrogen footprint followed by poultry, red meat, oilseeds, and cereals.The poultry products have highest phosphorus footprint per kg among the 15 crop categories followed by eggs, red meat, oilseeds and cereals.
Multiplying the characterization factors of individual crops with their production amounts, we obtained the environmental footprint of their production in each state (figures 1(B)-(F)).We found that current annual agricultural production in India results in a total of 1.042 billion CO 2 eq.tons of GHG emissions consuming 695 km 3 of blue water, 14 million tons of nitrogen application and 2.35 million tons of phosphorus application.In terms of hotspots and regional distributon of impacts, current agricultural production in North zone is responsible for ∼40% of the total national agricultural environmental footprint, followed by south zone (20%), east zone (14%), west zone (14%), central zone (10%) and north-east zone (2%).Figures 1(B)-(F) show the environmental footprints per state (see supplementary table S1 for values).
Breaking down the total footprint by item type, we found that milk, cereals and cash crops are the major contributors to the total environmental fooprint of Indian agriculture production sector for almost all four categories.Cash crops and cereals alone account for around 90% of total national agricultural freshwater footprint, 60% of nitrogen and phosphorus use and 40% of GHG emissions (see supplementary table S2 for footprints of all 138 agricultural items).
In terms of individual items-sugarcane, buffalo milk, cow milk, rice, and wheat are the major contributors to the total footprint for all environmental domains.Sugarcane alone is responsible for 58% of freshwater footprint and 17% of the total GHG emissions while wheat and rice combined are responsible for ∼50% of the total nitrogen and phosphorus footprints, 27% of freshwater and 20% of GHG emissions.

Characterizing the economic sustainability of current agricultural production mix
We found that at present among all the major states, the highest farmers' profit (benchmark) is achieved in the southern state of Karnataka (108%) while the lowest in Rajasthan (63%) while the national median profit stood at 81% (figure 1(G)).The farmers in western and southern regions derive profits amounting to on average 100% while farmers in northern and eastern states achieve an average profit of 75% from their current agricultural production.Although these profit numbers seem high, the actual income of most farmers in India remains low and government is trying to double their current income over next few years (Chand 2017).Therefore the objective of our optimization algorithm was to maximize the farmers' profit (equation ( 1)).
Supplementary figure S2 shows that the selling price received by farmers for nuts, roots/tubers, and fruits are around 3.5 times more than the cost incurred in growing them.Producing condiments/spices, vegetables, fish and lentils/pulses is also much more profitable to farmers (ratio > 2) relative to other crop categories such as poultry, eggs and milk (ratio < 1.5).

Optimized (sustainable) agricultural production mix
We found that the optimized agricultural production mix for each state obtained using linear programming (equation (1)) scores substantially higher on the nutrition, environment and economic indicators of sustainability than the current production mix.First, most states achieved the national benchmark profit (108%) in the optimized mix (figure 1(H)).The national average farmers' profit (total revenue obtained from sales minus total cost divided by total cost) increased by ∼25% (from 85% currently to 109% in the optimized scenario) with some states such as Odisha achieving as much as 52% increase (supplementary table S1).
Second, on the nutritional front, the supply of all 25 nutrients meets the population requirements (benchmark) after accounting for imports and exports although vitamin A and choline supply levels are at the borderline.Compared to current, the optimized production mix produced 50% and 21% higher amounts of vitamin A and choline respectively (whose current production was falling short from their population requirements in India).
Third, the total environmental footprints of the optimized production mix are ∼10% less than their current footprint levels.This corresponds to a reduction of 62.59 million tons CO2eq., 50.27 km 3 of freshwater, 12.41 Tg N, and 1.72 Tg P annualy.The environmental footprints in the optimized scenario also meet their respective benchmarks (i.e., they are below respective planetary boundaries).
To achieve this sustainable production mix, Indian fruits and vegetables production needs to increase by 105% and 134% respectively while the production of cereals (wheat, rice, maize in particular) needs to decrease by 51%.Other items whose production needs to increase are: lentils and other pulses (65%), beans and peas (49%), oilseeds (48%), roots and tubers (38%), nuts (20%) and fish (20%).
Table 2 shows the geographic distribution of required changes in different item categories across India (see supplementary table S3 for required changes in all 138 agricultural items per state).For a particular crop or livestock item, the required changes are not uniform throughout India.For example, the lentil/pulses production needs to increase in all regions of India except in the north where it needs to decrease by 51%.These variations are a result of yield differences across India as the optimization algorithm proposes to shift their production from low-yield states to high-yield states.Oilseeds production needs to increase in central and northern parts of India and decrease elsewhere.
To achieve a sustainable production mix, a substantial percentage departure from current production across all states is needed (calculated using equation ( 10) in supplementary text A).Highest departure (>115%) needs to take place in Assam, Telangana, Haryana, Madhya Pradesh and Rajasthan (figure 1(I)).Approximately one-third of the states necessitate a substantial departure from current production levels (exceeding 107%) to achieve a sustainable production mix (see supplementary table S1 for departure value per state).Relatively low departure is needed in the northern state of Uttar Pradesh (57%).See supplementary table S1 for departure value per state.

Discussion
Identifying what are the constraints in shifting to an optimal production mix calculated here and which kind of efforts the local governments and farmers need to put to achieve this departure from current mix would be the next research step.To incentivize the farmers to make such a shift, the government needs to work on reducing their production costs and increasing their profits through actions such as-building inclusive supply chains, promoting efficient markets and removing intermediaries where farm goods are sold, reforming land laws, improving access to essential farm inputs, increasing investments in agriculture research and development, infrastructure like cold storage (for fruits, vegetables and perishable items) and warehouses to reduce food loss (Spiker et  Overall, our results agree with past studies who also found a need to increase the production of fruits, vegetables, pulses and nuts and reduce the production of currently popular staples such as rice and wheat to achieve nutritional security and reduce environmental footprint (Davis et al 2019, Willett et al 2019) while increasing farmers' profit in India (Damerau et al 2020).The need to increase production volumes of above food groups identified in this study complements the results of past studies who also highlighted this from a consumption (dietary intake) rather than production perspective (Rao et al 2018, Chaudhary andKrishna 2021).Our results revealing the heterogeneity of production changes needed in different states across India and the finding that shifting production of certain crops from one state to another (table 2) can lead to environmental savings also corroborates with past studies (Devineni et al 2022).The novelty of our analysis above existing literature lies in the simultaneous consideration of 31 sustainability indicators while deriving the optimum production mix.
Several data gaps and limitations must be kept in mind while interpreting our results presented above and addressing them represent important future research fronts.First, we call for launching publicly accessible national database portal documenting annual data at a state or district (county) level on area, yields, production, farmers profit, environmental emissions and nutritional composition of agricultural production mix to conduct multi-indicator sustainability characterization for all countries.Currently such data is scattered across numerous platforms and to compile them is very challenging.Other environmental impacts associated with agricultural activities such as biodiversity loss, air pollution, human health risk, etc. should also be reported in addition to the five domains considered in our study.
Second, the changes in production amounts calculated by optimization algorithm might not be feasible or acceptable to stakeholders.For example, replacing the wheat and rice (major staple crops of India) harvest area with fruits, vegetables, legumes, and roots might not be acceptable to certain farmers as it might entail increased labor work or alternative skills/machinery (Rajkhowa and Kubik 2021).On the other hand, increased profit and improved nutritional profile through such replacements can be enough of an incentive for certain farmers to change their current behaviors.Future research should quantify the extents of readiness to changes or inertia that farmers and consumers of agricultural commodities have.This research can then be used to better formulate the cultural acceptability constraints used in optimization algorithms (table 1) and obtain more realistic results.
Third, the % production changes presented here (supplementary table S3) are not meant to be adopted directly as they are based on a certain set of assumptions regarding constraints (table 1) and objective function (equation ( 1)) and the quality of underlying input data.Our goal was to demonstrate the application of multi-indicator framework for characterizing the sustainability status current agriculture mix and the use of optimization algorithm for obtaining a mix that performs better than the current one on all sustainability indicators.It was beyond the scope of this study to consider all possible objective functions, constraints and underlying data uncertainty to provide a suite of % production changes.
Future research should explore alternative optimization objectives (e.g.minimizing environmental footprints, maximizing nutrition production, minimizing deviation from current mix, achieving multiple objectives etc) employing different mathematical algorithms and acceptability constraints (e.g.widening or narrowing the variation bounds for crop and livestock items listed in table 1) that are informed through behavioral data collected after consultation with relevant stakeholders (farmers, consumers, government policy makers, businesses).
Further, more nuanced and high resolution underlying data can also change optimization results.For example, we used country level average environmental characterization factors but ideally the factors should be state-specific as the fertilizer and other inputs and farm management practices vary across the country, resulting in different footprints for the same crop grown in different places (Goel et al 2021).
One can also explore alternative scenarios of improved production efficiencies through sustainable intensification (Khatri-Chhetri et al 2023, Sapkota and Takele 2023) that results in higher crop yields without harming the environment (Rosa et al 2018, Jat et al 2020), and running the algorithm accounting for such possibilities.Higher land use and fertilizer input efficiency can ensure that the production of cereals need not be reduced for accomodating increased production of other crop categories like fruits and vegetables.Results from these alternative optimization runs will enable a more comprehensive understanding of trade-offs across different indicators.

Conclusions
Our quantitative indicator framework leveraging 31 sustainability indicators and optimization algorithm demonstrated here can act as a template for other countries and guide them in collecting necessary data and determining how much the production amounts of different crops and livestock items need to be increased or decreased in order to achieve win-winwin outcomes for nutrition, environmental and economic dimensions of sustainability.
For India, we found that increasing farmers' income by 25% above current levels is possible while simultaneously reducing the environmental impacts by 10% and meeting the nutritional demand of the population.This will entail doubling the fruits and vegetable production and increasing the legumes, oilseeds and roots production by ∼50% while reducing the cereals production.Another insight gained from our analysis is that the required changes in production amounts for a particular crop is not uniform throughout India, thus underscoring the need to conduct sustainability assessments at a subnational level.Our results can guide policy makers and act as a first step in designing future strategies for solving India's farmer poverty problem without jeopardizing the environment and nutritional goals.
We call for breaking the silos and increased information sharing and engagement between environmental researchers, social scientists, economists, mathematical modelers, government departments, food producers, consumers and other stakeholders for collecting of high quality data on agricultural sustainability indicators and cultural acceptability at a state and district (county) levels throughout the world and making it publicly available to enable, apply and improve future assessments like ours globally.

Figure 1 .
Figure 1.Sustainability indicator scores for each of the 36 Indian states in the current agricultural production mix (unless otherwise stated).(A) Essential nutrients production contribution (average of % of national total for each nutrient).(B) GHG emissions (tons CO2eq.yr −1 ).(C) Freshwater use (km 3 yr −1 ).(D) Land use (km 2 yr −1 ).(E) Nitrogen application (tons N yr −1 ).(F) Phosphorus application (tons P/year).(G) Farmers' profit % in the current mix.(H) Farmers' profit % in the optimized mix.(I) Departure of current agricultural mix from optimized mix (supplementary text A.6). Also see supplementary tableS1for indicator values per state and supplementary figures S3 and S4 for % contribution of each state to the total national environmental footprint and nutrient production.

Table 1 .
Nutritional, environmental, economic and cultural acceptability constraints applied in linear optimization (equation (1)) for obtaining sustainable (optimized) agricultural production mix in India.Current and optimized values of each sustainability indicators is also shown.Values are annual(2014-17 average).Nutrient values are based on national production minus export plus imports.

Table 2 .
% difference between the optimized and current production amounts of major agricultural item categories for six regions in India.A positive value indicates higher production in the optimized scenario.Values of spices, condiments and cash crop remain the same for current and optimized scenario, see supplementary tableS3for % difference values for all 138 items per state (2505 combinations).