Main

Multiple interactive challenges facing society described by the sustainable development goals (SDGs)1,2 will require substantial complementary changes involving technology and production, shifts in consumer perceptions of consumer behaviour and redesign of institutions and policies. Analysis at both global and local scales provides needed, complementary insights helpful in formulating strategies responsive to these challenges3,4.

Modern bioenergy, often combined with carbon capture and storage, plays a prominent role in projected climate stabilization pathways4. As described herein, most of the SDGs (for example, 12 of 17) are meaningfully impacted by bioenergy. This reality gives rise to the possibility of several benefits—highlighted by advocates—as well as trade-offs and unintended consequences—highlighted by critics. Bioenergy thus provides strong motivation to develop tools to assess the complex web of potential positive and negative consequences of deployment across the spectrum of SDGs.

Several modelling approaches and methods have been used to assess bioenergy pathways and their socio-economic and environmental impacts5, such as bottom-up models, life cycle and social life cycle assessment (LCA), system dynamics models, computable general equilibrium (CGE) models, integrated assessment models and hybrid methods combining some of these tools. They have been applied at several aggregation levels and considered an alternative range of socio-economic indicators. However, most assessments usually focus on a limited spectrum of impacts, instead of the full spectrum of SDG goals. This means that there are few, if any, integrated analyses that include broad macro-economic effects, environmental impacts beyond greenhouse gas (GHG) emissions and impacts on food security and low-income populations. Holistic and integrated frameworks based on the SDGs to evaluate projects and policies using LCA and bridging global and regional effects have been developed lately6,7,8,9,10,11,12. However, the application of these frameworks is thus far restricted to a few projects and technologies and is still nascent for bioenergy7,10. Integrated analysis encompassing several SDG is needed to avoid trading a benefit in one dimension for liability in another13,14,15,16.

Bioenergy feedstock production can either be primary or ancillary land use, with each strategy having advantages and challenges. Double crops—producing two crops on land that formerly produced one crop—are a prominent example in the ancillary category. The development of double cropping has led to the rapid expansion of grain production in the Central-West region of Brazil, particularly in Mato Grosso state, where second-crop maize area has shown significant growth in recent years, rising from 3.2 million ha in 2013/14 to 7.4 million ha in 2022/23. Essentially all (99%) of this additional maize is produced as a second crop in a sequential system after soybean using the same area. The expansion of second-crop maize can be attributed to several drivers, including the adoption of better technologies and cultivation practices which increased production and productivity in double-crop areas17 and optimized land use in the off-season cultivation system18,19 (Supplementary Information Section 1).

Producing ethanol from second-crop maize on land that formerly produced only soybeans in Brazil is the fastest-growing example of the double-crop bioenergy feedstock strategy in the world today and is also likely to be the first application of bioenergy with carbon capture and storage (BECCS) in Brazil20,21,22. In the last season (2022/23), nearly 100% of Brazilian maize ethanol was produced in the Central-West region and 73% in Mato Grosso state. Previous investigation23 has shown large reductions in GHG emissions when comparing double-cropped maize ethanol with gasoline in Brazil, in addition to economic and employment benefits at local and national levels and reduced direct land-use changes. However, previous efforts to evaluate the maize ethanol system in Brazil did not investigate several SDGs and indirect land-use changes, nor have they accounted for impacts on household income and food security.

Here, we offer a comprehensive analysis of the emergent double-cropped maize ethanol production system in Brazil. We provide an ‘on-the-ground’, context-specific evaluation of a system producing food and energy using a combination of environmental and socio-economic methods and metrics, extending the methodological frontier on bioenergy analysis concerning both scope and approach. Our goal is to assess the socio-economic and environmental impacts of the fast-expanding energy–food production system based on double-crop maize in Brazil, asking if the impacts of this production system are well aligned with the SDGs challenges.

Results

Environmental impacts from LCA analysis

Net LCA indicators associated with the displacement of the conventional fossil-based system by the multicropping maize-based system are summarized in Table 1. Among the midpoint and endpoint categories considered by the ReCiPe 2016 method24, it can be noted that crop-based systems typically feature greater local/regional environmental impacts due to agrichemicals and water use, while traditional/fossil-based systems show greater global impacts (for example, climate change) because of fossil resource exploitation (see Supplementary Information for further details through midpoint indicators of ethanol in Supplementary Table 2 and Supplementary Fig. 2 and a sensitivity analysis in Supplementary Fig. 3). However, it is worth highlighting the peculiar behaviours (negative emissions in Supplementary Fig. 2) in terms of human toxicity due to non-carcinogens and global warming, which result, respectively, from the removal of copper and zinc from the soil by maize cultivation and biogenic CO2 capture from fermentation in the configuration with BECCS. This arises from the Canals model25,26 for estimating the emissions of heavy metals to soil and surface water, in which the fraction of heavy metals exported to the crop is subtracted from the total value added to the system. Therefore, some of the emissions to the soil can be negative, meaning that these metals are transferred either to the water bodies or to the harvested product27. However, one should note that the continuous removal of metals from the soil may eventually lead to consequences to crop yield, which would thus require additional inputs of metals in the future.

Table 1 Net LCA (endpoint level) associated with implementing the multicropping maize-based system
Full size table

For most environmental impact categories, the main contributions derive from the life-cycle stages before fuel combustion, with only two important exceptions for gasoline. In the case of global warming, the use of gasoline in internal combustion engines is responsible for most of the CO2 emissions, while for human toxicity by carcinogens, aldehyde emissions (evaporative and incomplete burning) stand out. Therefore, the substitution of gasoline for ethanol in internal combustion engines is particularly beneficial for these categories, considering the relatively small contributions from maize cultivation and processing.

Although there are trade-offs among the impact categories, overall, it can be concluded that implementing the product system based on second-crop maize to replace equivalent products leads to net benefits from human health, ecosystem and resource perspectives, especially due to the GHG mitigation resulting from the displacement of gasoline with ethanol.

In the case of human health, for example, for each tonne of maize, impact mitigations were estimated at 7.91 × 10−4 and 1.07 × 10−3 disability-adjusted life years (DALY)21 for systems without and with BECCS, respectively. The difference between these systems lies in their effect on global warming, while the results for the other categories are practically unchanged. In fact, including BECCS leads to a slight increase in the impact on the other categories given the additional resources required for CO2 capturing and logistics and reduction in the generation of surplus power to provide energy for CO2 compression. Nevertheless, the remarkable effect of BECCS on GHG emissions translates into significant benefits to ecosystems and human health, thereby amplifying the preference for the maize-based system. It may be noted that the net benefits of the maize-based systems are not necessarily linear so different production scales could eventually lead to equivalences between products that are different from those considered in this study (Supplementary Fig. 5).

This broad dominance of GHG mitigation on the overall benefit of the maize-based system makes the general conclusions robust against the uncertainties of the analysis. As illustrated in Supplementary Fig. 3, even in cases where there is a significant reduction in industrial yields or the displacement ratio between alternative products, net human health, ecosystem and resource benefits are still anticipated.

Socio-economic and overall GHG emissions results

The socio-economic modelling tools deployed in this study (Economic Projection and Policy Analysis—EPPA model28 and the General Equilibrium Analysis Project—PAEG model29,30; Methods) allow us to assess impacts on consumer prices and incomes across ten different household income levels in Brazilian regions, considering changes in markets and prices of goods and services.

The direct economic impact of multicropping maize ethanol production expansion from existing 2.9 billion to 8 billion l (expected production for the year 2030, according to Empresa de Pesquisa Energética—EPE)20 is on energy and food markets. Besides the increase in ethanol production, the system also adds 600 GWh of electricity, 4 million t of distiller's dried grains (DDG) and 0.16 million t of maize oil (Methods). Increased production of maize ethanol and coproducts reduces energy prices (fuel and power) and energy-dependent industry prices (transportation) in the region (from −0.2% and −1.5%), while energy prices barely change (less than ±0.05%) in other regions (Supplementary Figs. 7 and 8).

Overall, food prices decreased in the region by 0.21%, increased by 0.02% at the national level and changed by less than ±0.01% in other countries and regions. Food production in the Central-West region decreased by 1.97%, up to 0.04% at the national level and less than ±0.01% in other world regions. Household maize consumption in the Central-West region drops by 12%, while consumption of other primary agricultural goods and processed food increases between 0.07% and 1.7%. The share of maize in family spending on food and agricultural products accounts for less than 0.5% for all income classes in the region, while processed food accounts for 38% to 49%, depending on the income class.

Income increases more in relative terms for the three poorer income classes than for other families (Fig. 1). The utility index, an aggregate measure of well-being translates how all changes in prices (including changes in salaries and returns to capital and land) impact consumer choices, increases for all family groups in the Central-West region and represents the net income gains and improvements in aggregate consumption. With expansion of maize ethanol, all household classes in the Central-West region increase food consumption by between 0.26% and 0.68%. These results indicate small but positive indirect impacts from the double-cropped bioenergy system concerning poverty reduction, economic growth and susceptibility to food insecurity in the region. As for the other regions, changes in income, utilities and total consumption are less than ±0.01% and are less than ±1% in terms of consumed quantities and prices of each good and service. This means that families in other regions are not negatively affected by the double-cropped maize food–energy system expansion in the Central-West region of Brazil. These positive impacts on lower-income households are robust to changes in the key parameters of the model, while middle-income households (F06 to F09; Methods) may face null or slightly negative impacts under a specific parameter choice (Supplementary Figs. 9–11).

Fig. 1: Changes in household income, utility index and consumption.
figure 1

Results from alternative scenarios and assumptions of the PAEG model due to a 5 billion l increase in multicropping maize ethanol in the Central-West region of Brazil. Household F01 is the lowest income class; F10 is the highest (F01, income up to US$240; F02, >US$240 up to US$360; F03, >US$360 up to US$599; F04, >US$599 up to US$719; F05, >US$719 up to US$958; F06, >US$958 up to US$1,198; F07, >US$1,198 up to US$1,796; F08, >US$1,796 up to US$2,395; F09, >US$2,395 up to US$3,593; F10, >US$3,593). Each bar represents model output assuming default model parameters (central values). Vertical bars represent the range of impacts and dot plots represent several contemporary outputs from the same scenario under sensitivity analysis of key model parameters (n = 20) (Methods; Supplementary Information Section 4 and Supplementary Figs. 9–11 for full sensitivity assumptions and outcomes). Ind. util., changes in household utility index; val. cons., changes in total value of consumer expenses.

Full size image

Impacts on land use from the socio-economic modelling

Increased production of multicropping maize ethanol expands the physical planted area of second-crop maize in Brazil by 600,000 ha on an existing area of soybean, reducing the area devoted to single-cropped soybean. Coproduction of distiller’s dried grains with solubles (DDGS) leads to a lower need for soybean and pasture areas, as it provides a greater volume of feed within the region and contributes to livestock intensification. As a result, the total agricultural area in Brazil is reduced by 25,000 ha and pasture areas decrease by 168,000 ha, which frees up the area for 50,000 ha of reforestation or reduced deforestation (Fig. 2). Higher demand for biomass in the maize ethanol industry contributes to increase in planted forests since the additional biomass requirement is equivalent to 12% of current biomass output in the Central-West of Brazil. The volume of secondary vegetation increases relative to the business-as-usual scenario but opportunity costs of transforming managed areas back to their natural state are not low enough to allow natural areas to recover. Changes in natural forests and other natural vegetation are negligible and hardly visible in Fig. 2. Small differences in land-use changes among scenarios mean the overall magnitude and direction of the results are not sensitive to the addition of BECCS to the maize food–bioenergy system.

Fig. 2: Land-use changes in Brazil in 2030.
figure 2

Results from alternative scenarios and assumptions of the recursive dynamic general equilibrium model EPPA, which explicitly assumes costs of conversion and land competition among alternative uses (Methods), due to a 5 billion l increase in multicropping maize ethanol in the Brazilian Central-West region. Secondary vegetation refers to previously deforested areas in the process of recovering original vegetation. Each bar represents model output assuming default model parameters (central values). Vertical bars represent the range of impacts and dot plots represent several contemporary outputs from the same scenario under sensitivity analysis of key model parameters (n = 20) (Methods; Supplementary Information Section 4 and Supplementary Fig. 12 give full sensitivity assumptions and outcomes).

Full size image

Impacts on overall GHG emissions

Expanding multicropping maize ethanol in Brazil by 5 billion l contributes to reducing the GHG emissions of Brazil by 12 million tCO2e in 2030 (Fig. 3), considering all potential substitution among fuels and changes in economic activities and land-use changes induced by market-mediated responses, as projected by the socio-economic modelling. Assuming a conservative scenario of BECCS adoption on only 0.78 billion l of ethanol reduces the GHG emissions of the country by 0.6 million tCO2e. A more optimistic scenario of BECCS adoption on the total 5 billion l of ethanol expansion would reduce the GHG emissions of Brazil by 15.9 million tCO2e, which means an additional reduction of 3.8 million tCO2e compared to the scenario without BECCS.

Fig. 3: Global changes in GHG emissions.
figure 3

Results from alternative scenarios and assumptions of the recursive dynamic general equilibrium model EPPA due to a 5 billion l increase in multicropping maize ethanol in the Central-West region of Brazil. Each bar represents model output assuming default model parameters (central values). Vertical bars represent the range of impacts and dot plots represent several contemporary outputs from the same scenario under sensitivity analysis of key model parameters (n = 20) (Methods; Supplementary Information Section 4 and Supplementary Fig. 13 for full sensitivity assumptions).

Full size image

Impacts on SDGs

Considering the results of the LCA, based on the models and parameters used in this study, the system based on multicropping maize brings important contributions to different SDGs (Fig. 4), especially due to GHG mitigation. Despite increased demand for land by LCA and the increasing trend of acidification, mitigating GHG emissions not only contributes to SDG 13 but also brings great human health and ecosystem benefits. Reducing the use of fossil resources would be in line with SDG 7. Expansion of the double-cropped bioenergy system does not result in significant positive or negative effects for SDG 14, while SDG 6 tends to be negatively affected when environmental impacts are considered in isolation.

Fig. 4: Multicropping maize-based production system contributions to SDGs.
figure 4

a,b, Results of the LCA without BECCS (a) and with BECCS (b).Colour scale refers to the percentage changes of the environmental indicators due to the displacement of conventional products with maize-based products. Green colours represent positive impacts on SDGs, red colours represent negative impacts. Darker colours represent higher impacts. Since all indicators for a given SDG are in the same units (at the endpoint level), the assessment of the overall effect was based on the direct summation of the indicator results of each contributing environmental category. For some SDGs the negative impacts of red indicators may be more than counterbalanced by the benefits of green indicators. Credit: icons from the United Nations Sustainable Development Goals (https://www.un.org/sustainabledevelopment).

Full size image

The colour scale in Fig. 4 refers to the percentage changes of the environmental indicators due to the displacement of conventional products with maize-based products. Since all indicators for a given SDG are in the same units at the endpoint level, the assessment of the overall effect was based on the direct summation of the indicator results of each contributing environmental category. This means that for some SDGs the negative impacts of red indicators may be more than counterbalanced by the benefits of green indicators. This was the case of SDG 2, for example, because of the remarkable contribution of the multicropping maize for GHG emissions mitigation, whose contributions in terms of DALY are orders of magnitude greater than the impacts from acidification and ecotoxicity (Table 1).

Socio-economic and land-use change results allow us to evaluate the effects on SDG 1, 2, 7, 8, 10 and 13 (Fig. 5). Most of the impacts are positive at the regional level, while the effects are close to zero in other regions. Thus poverty, hunger and regional income inequality are reduced locally, without prejudice to other regions. Results from the socio-economic analysis complement those of the environmental analysis. In the case of the investigated food–energy system, complementarity between maize and soybean crops combined with the benefit of DDG production reduces agricultural area. Changes in household income and prices in the region under analysis improve family well-being, contributing to SDG 2 (zero hunger).

Fig. 5: Multicropping maize-based product system’s contribution to the SDGs, results of the socio-economic analysis.
figure 5

Colour scale refers to changes in indicators relative to a reference level. Green colours represent positive impacts on SDGs, red colours represent negative impacts. Darker colours represent higher impacts. Assumptions about the intensity scale are: indicators in percentage variation (%) are compared relatively to the percentage variation in biofuel production (12%), except income, consumption, utility and economic growth indicators, whose relative comparison was a 1% marginal increase; quantitative energy indicators (fuel and power) were assumed increasing by 100% since these are direct products of the maize ethanol plant; emission indicators were compared relatively to emissions from passenger transportation industry; area indicators were compared relatively to the land area currently in use. Credit: icons from the United Nations Sustainable Development Goals (https://www.un.org/sustainabledevelopment).

Full size image

Discussion

We have assessed double-cropped maize ethanol production in the Central-West region of Brazil concerning the environment, socio-economic indicators and land use. Life-cycle assessment (LCA) is used to evaluate environmental impacts with particular attention to GHG emissions. Two CGE models, EPPA and PAEG, are used to analyse socio-economic effects and land-use changes with particular attention to effects across a wide range of household incomes. An impact relationship was established between the environmental and socio-economic indicators and the twelve SDGs, with SDG 2 (zero hunger) being analysed in detail at local and global levels. This enabled a comprehensive sustainability analysis of a biomass-based food and energy system.

The use of products derived from maize as substitutes for conventional products leads to a net reduction in environmental impacts for the three areas of protection (human health, ecosystems and resources), especially regarding GHG emissions due to gasoline displacement. With BECCS, this benefit becomes larger as a result of CO2 removal from the atmosphere. Although there are trade-offs when the set of environmental impacts is considered, the large effect of climate change results in a net benefit to ecosystem quality and human health, whereas the results for the other dimensions are practically unchanged. Results from attributional LCA are aligned with those of refs. 23,31, apart from the scenario involving BECCS considered here. Others32 found much higher GHG emissions due to differences concerning land-use change. Terrestrial acidification and photochemical oxidant formation were also higher in ref. 32, which can be largely explained by the differences in terms of industrial yields considered by that study.

Socio-economic impacts are positive at the regional level with negligible spillovers to other regions. Economies of scope (producing several goods more cost-effectively than each good individually), efficiency gains in land use and accelerated economic activity are the key regional drivers. The maize ethanol industry expansion in the Central-West region increases the supply of bioenergy, DDG, maize oil and electricity, reducing prices and inducing higher consumption. The system demands inputs from other sectors and resources (capital, labour and land), supporting local activities and raising returns to local businesses, landowners and employees. Income grows for all household groups, boosting consumption and creating a second wave of positive economic effects. Agricultural and food prices are minimized as a result of the high local absorption of the goods produced by the ethanol industry and the efficiency gains in double-cropping farming. These economic impacts are aligned with those found by ref. 23. However, we advance the socio-economic analysis by capturing changes in economic behaviour due to changes in relative prices and including impacts on food demand and food security, which was not investigated in previous studies. Concerns about food security and hunger due to biomass production have been raised in several studies33,34,35. However, changes in prices and food output, taken alone, are limited impact indicators on food security. Food–energy systems can impact economic returns from labour, capital, land and indirectly affect other markets and consumer choices. The resulting impact on food security must consider how income and consumption of lower-income families will be hit by the bioenergy shock and how consumers will change their choices among alternative options of agricultural and food items36. We found that positive changes in household income were relevant in increasing food consumption and the well-being of poor families in the Central-West region, consistent with previous studies such as refs. 37,38. Together with lower food prices and an increase in food consumption, maize ethanol production as considered here improves regional food security.

Harnessing the second-crop maize production potential allows DDG to substitute grains used for animal feed, reducing land requirements for livestock and risks of land-use changes. Our results show larger secondary vegetation areas in 2030 by 148,000 ha under multicropping maize ethanol expansion compared to the baseline, which could contribute to reducing the cost of compliance with the Brazilian land-use law, the Forest Code. These impacts go in the opposite direction of what has been found in literature so far because most existing agricultural-economic models dealing with land use change (LUC) do not explicitly represent double-cropping soybean–maize systems in Brazil39,40,41. By introducing this system in the socio-economic models, we represent farmers’ trade-offs and opportunities when deciding how to better allocate land among single-cropping and double-cropping systems. This statement is reinforced by ref. 42, which finds robust negative induced land use changes (ILUC) for sustainable aviation fuels produced from second-crop maize in Brazil, under different scenarios. Examples of other double-crop schemes, also called second crops, are increasingly being noted in scientific literature43,44,45,46,47,48,49.

The double-crop maize ethanol system considered here contributes to SDG 1, 2, 7, 8, 9 and 13 (and possibly 10 and 11). These results are a consequence of the specific configuration of the system (multicropping maize, biofuel production, bioelectricity, DDG, maize oil, using eucalyptus chips and BECCS). The results also indicate the need for attention regarding SDG 6 and should assess whether there are public or private policies for avoiding undesirable impacts on water quality. Facing the challenges associated with SDGs requires developing and applying similar or customized systems adjusted to local realities pursuing synergy in different dimensions (inputs, products, chains, logistics and so on). Replicability to other regions cannot be assumed and merits study.

The characteristics of the Central-West region and particular features of the maize ethanol production system are key factors underlying our results. The region has large soybean areas and favourable edaphoclimatic conditions to produce double-cropped maize. While soybean international demand counterbalances high logistic costs to deliver this commodity overseas, maize markets do not, hindering the potential increase in second maize production in the region. The maize ethanol energy–food production system takes advantage of these circumstances and sparks a socio-economic prosperity cycle in the region. Besides, assessing a cheap feedstock to generate biofuel the industry also benefits from selling DDGS to the livestock sector of the region, one of the largest in the country. Farmers produce maize as a second crop instead of leaving the soybean land uncultivated for more than half of the year.

Future analysis may consider some issues related to land-use changes, such as land tenure, corruption, policy enforcement and rural development policies, among others, which are not explicitly represented in our modelling tools. As such, our analysis assumes that the tested scenarios would not change the existing decisions and behaviour related to such complex issues.

We conclude that the current production of energy and food based on second-cropping maize in the Central-West region of Brazil is decisively beneficial compared to the status quo without such production in terms of environmental and social impacts considered both independently and across the spectrum of SDGs. Looking beyond current practices, there is potential for such production to achieve net removal of atmospheric CO2 if carbon capture and storage are added. Fermentation off-gas is essentially pure CO2, thus avoiding the largest cost associated with most other CO2 removal approaches50,51. It is promising to identify critical features underlying this positive assessment and to explore whether they can be replicated in other bioenergy systems. One such feature is that process energy for second-crop maize ethanol production is provided by renewable biomass. Renewable process energy is used today for sugarcane ethanol but not for most maize ethanol produced in the United States. This difference is the main reason for the lower carbon intensity of current production of first- and second-generation sugarcane ethanol compared to maize ethanol52. A second key feature is that second-crop maize ethanol production integrates feedstock supply into existing land use such that the net agricultural footprint decreases23. Such integration is often not featured in bioenergy assessments, although advantages of this strategy have been noted previously by refs. 45,53. Finally, we recognize that domestic and international policy decisions around the world on first-generation biofuels have been driven by sustainability issues. Although some regions opted to avoid them as transportation fuel options (such as the European Union), other geographies and institutional frameworks recognize environmental footprint differences among alternative biofuels (such as in the United States and at the ICAO-CORSIA). Our assessment contributes to this discussion by highlighting how the configuration of a biofuel supply chain determines its capability to meet sustainability objectives.

Methods

Analysis systems

This study performs a multimodel analysis based on different science areas, seeking to provide a broad view from different perspectives. Socio-economic and LCA models are generally adopted for achieving the proposed objectives.

Socio-economic models assess major changes in the system, such as introducing or expanding an industry/sector. Here, we selected two different models to address several SDG dimensions. One model covers regional markets and several household impacts, while the other tracks major sources and sinks of GHGs, including land-use changes. Combining both models, we cover changes in prices and quantities at global and regional levels.

LCA allows for in-depth evaluation of specific cradle-to-grave indicators. Initially, LCA models are applied to a specific predefined system with limited reactions from the surrounding environment (particularly economic agents). Complementary studies consider that reactions from the surroundings can be relevant and should be recognized. This study combines these approaches in a complementary manner.

We simulate the same double-cropped maize ethanol system and its technical-economic characteristics in LCA and socio-economic models, which assures consistency among models and SDGs assessments, respecting the capabilities of models and their specific simulation strategies, which are described next. While the LCA analysis measures the impacts of replacing current fossil fuel production and use by the maize-based system following a consequential approach, the socio-economic models simulate all direct and indirect market effects from a given level of ethanol production and use. Details of scenarios implemented in each model are provided below and results discussed in the text were identified by model deployed. The LCA model is used to evaluate several environmental impacts associated with SDGs 2, 3, 6, 7 and 11–15. The socio-economic models are deployed to assess impacts on SDG 1, 2, 7, 8, 10 and 13 related to food and energy quantities and prices, household income and consumption, economic growth, land-use changes and overall GHG emissions from all market-mediated responses (energy substitution, agricultural production and changes in land use). Although both LCA and socio-economic models evaluate GHG emissions, the differences in their nature lead to different results, which should not be directly contrasted but rather seen as complementary indications. It may be noted that all approaches have reached the same qualitative conclusions regarding the potential for mitigating GHG emissions.

LCA and scenarios

The environmental analysis consisted of evaluating the comparative performance of the product system based on the double-crop (multicropping) maize use in Brazil dedicated to producing ethanol, DDG, oil, power and capturing CO2 (Supplementary Fig. 4) against its conventional counterpart, which relies on traditional supply chains to deliver the same functions in the form of fossil gasoline, animal feed (maize and soybean meal), soybean oil and thermoelectricity. To assess the effect of the carbon capture, two configurations of the second-crop maize product system were modelled: one with the deployment of BECCS and one case without BECCS (reference). In both cases, energy inputs (steam and power) are provided by an adjacent thermoelectric (cogeneration) plant using biomass (eucalyptus chips) as fuel. The energy system is designed so that the power generated by the cogeneration plant is greater than the demand for maize processing. The information used in this study was provided by the fuelling sustainability facility. More information is available in Supplementary Information Section 3 and Supplementary Fig. 6.

The analysis followed the LCA technique, using 1 t (1 Mg) of maize as the reference flow. For both maize-based product systems (with and without BECCS) the environmental life-cycle indicators were expressed as the difference compared to the conventional system featuring a single soybean crop, assuming the displacement of the traditional products (according to the equivalences shown in Supplementary Fig. 5). For this study, the 18 environmental impact categories considered by the ReCiPe Method 2016 (H) were evaluated at the endpoint level, which were then grouped into the three endpoint indicators on the basis of impacts on ecosystems (in species year), human health (in DALY) and resource availability (in US$2003). To facilitate interpretation, an alternative approach based on the well-to-wheel performance of maize ethanol (the main maize product) was also developed and compared to fossil gasoline using 1 MJ as a functional unit. The midpoint-level indicators for these products are presented in Supplementary Information Section 3.

For foreground data, the study considers information from specific plant processes (Supplementary Fig. 4) and, in addition, data that reflect and are close to agricultural practices featured in Mato Grosso State (the main geographic scope) were selected (from RenovaCalc54 and SICV55). Discharges to the environment associated with maize cultivation were quantified26,56 and vehicle emissions running on ethanol and gasoline were taken from Cetesb57. Infrastructure requirements for CO2 sequestration and storage were estimated from the literature58 assuming the characteristics of the central part of the Parecis Basin, which occupies most of the state of Mato Grosso and features a clear potential for CO2 storage59. The area hosts sediment deposits >5,600 m thick and it is possible to drill injection points with sandstone layers >200 m thick at depths of 2,000 m, capped by >80 m of shale which could serve as a seal60. As for the background systems, the study relied on datasets from Ecoinvent, OpenLCA and Excel spreadsheets were used as auxiliary tools to compile the final life-cycle inventories and indicators.

Once the impact indicators related to each environmental impact category were quantified at the endpoint level, a relationship was established between these indicators and the SDGs, as shown in Supplementary Table 4. Supplementary Table 5 presents a list of environmental impact categories. This relationship followed an arbitrary classification criterion, adapting the correlations used in ref. 10 (Supplementary Information Section 3). As each category has contributions of different magnitude, the total contribution to each SDG was considered as the sum of endpoint-level indicators for each associated category.

Socio-economic and land-use change analysis and scenarios

The socio-economic analysis consists of evaluating results of scenarios with increased multicropping maize ethanol system production in economic-wide models. Two simulation models used in this study are capable of projecting: (1) land-use dynamics (including second crop); (2) access to food (income, family spending, purchasing power and so on); (3) coproducts impact (DDG and maize oil); (4) prices; (5) GHG emissions; and (6) production and demand.

The EPPA model28 and the PAEG model29,30 were enhanced and deployed. Both are CGE models and the EPPA model is a component of the Integrated Global System Modelling (IGSM) framework61. Both models were improved to represent Brazilian double-cropped maize ethanol technology with its coproducts, which use second-crop maize planted in the same area and after a single soybean crop.

In the EPPA model, carbon capture and sequestration technology from the maize ethanol fermentation process was also introduced. It captures capital and labour costs with investments, equipment and operation of carbon capture and storage facilities. Both models are multiregional and multisectoral and the EPPA model is dynamic recursive, while the PAEG model is static and presents the five disaggregated macroeconomic regions of Brazil, in addition to several regions of interest in the rest of the world (Supplementary Tables 8 and 9 give industries and regions represented in each model). Multicropping maize ethanol technology in the PAEG model was considered only for the Central-West macroregion of Brazil, which currently produces 99% of Brazilian maize ethanol, while in the EPPA model, this technology was considered for Brazil as a whole.

CGE models represent the exchange relations between different economic agents (producers, consumers and government) through explicit formulation and resolution of supply and demand equations in different domestic and international markets for goods, services and productive factors. The quantities and prices that provide market equilibrium are endogenously established. The model assumes optimizing producer and consumer behaviour when choosing inputs, factors, goods and services. Behavioural parameters (elasticities) establish to which extent the agents can substitute inputs, productive factors, goods and services, in the face of changes in relative prices.

Both the PAEG and EPPA models are built from the global trade analysis project (GTAP) economic database62. GTAP provides a consistent representation of regional production and demand, bilateral trade flows between countries and regions of the world and their markets.

Physical information is added in the case of energy, agriculture and natural resource markets, based on data from the United Nations Food and Agriculture Organization (FAO), the Organization for Economic Co-operation and Development (OECD), the International Energy Agency (IEA) and domestic statistics (in Brazil, the Companhia Nacional de Abastecimento (CONAB), the Instituto Brasileiro de Geografia e Estatística (IBGE) and Empresa de Pesquisa Energética (EPE)20). The production and consumption sectors of the models are represented by constant elasticity substitution (CES) of production functions. The macroeconomic projections by the International Monetary Fund (IMF) are used as a reference for projecting future scenarios. See Supplementary Information Section 6 for references to these and following data sources. The EPPA model is recursively resolved in 5 year steps as of 2020, while the PAEG model is used for comparative static analyses. The models are calibrated for projecting long-term trends. Additional information on advanced technology (such as carbon capture and sequestration) costs, yield, GHG and air pollutant emissions, taxes and details of select economic sectors are also used. In addition to representing the Brazilian economy disaggregated into five economic macroregions, for each of these regions, the PAEG model also considers ten different classes of households by average household income. This enables assessing impacts on different family classes, inequality and income distribution30. The household income classes are grouped by monthly household income, based on the Brazilian IBGE household budget survey. These are defined as: household F01, income up to US$240; F02, >US$240 up to US$360; F03, >US$360 up to US$599; F04, >US$599 up to US$719; F05, >US$719 up to US$958; F06, >US$958 up to US$1,198; F07, >US$1,198 up to US$1,796; F08, >US$1,796 up to US$2,395; F09, >US$2,395 up to US$3,593; F10, >US$3,593. Per capita monthly income in Brazil in the benchmark year was US$1,006 at current prices and market exchange rate. Per capita monthly income at power parity purchase was US$1,305. Power parity purchase equalizes the purchasing power of different currencies by eliminating the differences in price levels between countries allowing comparison among them. There are different institutions defining income groups in Brazil. Using the definition by Brazil-SAE, F01 to F03 are low-income households, F04 to F07 are middle-income households and F08 to F10 are high-income households.

The general equilibrium models used in this work capture the direct and indirect effects on salaries, land rents, employment and income in the economy. These effects include not only those directly related to the biofuel refinery and the additional maize production, such as the recruitment of labour in the region and demand for inputs from local businesses, but also the chain of indirect effects due to income being spent by those economic agents, which ultimately may affect all income classes. As such, small local farmers and the poorest income classes may still benefit from higher demand for their products or services provided by their employers once more income is generated and spent in the region due to the additional maize and biofuel production.

Regarding changes in land use, the PAEG model considers that the total land supply resources are allocated to different uses on the basis of a constant elasticity of transformation function, as in the original GTAP model63. Each region has a representative agent who owns land resources and chooses how much of them will be deployed in each agricultural sector, according to financial return or relative rental amount paid by farmers for different uses. This approach captures the possibility of converting areas from one use to another, considering that there are edaphoclimatic and temporal limitations for such a conversion, so there is imperfect substitution between the different uses.

In the EPPA model, changes in land use are explicitly represented by conversion cost between different uses, maintaining physical consistency in terms of converted hectares and economic consistency in profit functions. Potential conversion of natural vegetation areas is also considered in the EPPA model, in contrast to most of the CGE models64. Further details on formulating land-use changes are discussed elsewhere65.

Because the original versions of the models used did not consider the possibility of producing two crops in the same area, both models have been modified to represent a new agricultural sector producing two crops, soybean and maize, in the same area in a single crop year. To perform this modification, the soybean-producing sector was subdivided into two other sectors: one sector producing only soybeans and the other producing soybeans and maize in a two-crops-per-year system. Both compete for the same land-use area, which means that, if the double-cropping system expands, farmers who previously were single-soy producers are now producing soybean and maize. This update is essential for this study because it enables adequate representation of the local productive system and is an innovation from a methodological standpoint since no other land-use agricultural-economic computable general equilibrium model has explicitly represented this soy–maize double-cropping system to date. More details are provided in Supplementary Information Section 7.

Economic impacts of the multicropping maize ethanol production expansion system were evaluated by alternative shocks in the PAEG and EPPA models. Supplementary Fig. 17 illustrates the shock applied to PAEG, a 5 billion l multicropping maize ethanol increase. Production in 2020 (before the shock was 2.9 billion l, reaching 8 billion l in 2030, the expected production according to EPE20. Additionally, DDG, maize oil and bioelectricity coproduction are considered.

Supplementary Table 11 shows shocks applied to the EPPA model, as well as assumptions and fundamentals. Four different scenarios were evaluated: (1) in the first one, it was assumed that multicropping maize ethanol production would be kept constant through 2030, according to the levels seen in 2020 (2.9 billion l), while maintaining the projection for sugarcane ethanol established by Ministério de Minas e Energia (MME), 38.6 billion l in 2030; (2) in the second scenario, multicropping maize ethanol production increases from 2.9 to 8 billion l by 2030, through increased efficiency, whereas sugarcane ethanol production is kept at the same level as in the previous scenario and no carbon capture and sequestration from the fermentation process capacity is installed; (3) in the third scenario, the multicropping maize ethanol volume was as in scenario (2), adding BECCS from 2025, considering a project under implementation with the capacity to capture CO2 emissions from the fermentation of 0.53 billion l of ethanol in 2025 and 0.78 billion l of ethanol in 2030 (information provided by an industry company); and, finally, (4) in the fourth scenario, the projected multicropping maize ethanol volumes are the same as in scenarios (2) and (3) but an expansion of facilities using BECCS capable to meet the entire multicropping maize ethanol expansion volume is assumed, which is an optimistic scenario for a potential BECCS growth.

It is worth mentioning that, in the case of the third scenario, 8billion+BECCS (conservative), growth in carbon sequestration capacity through BECCS is assumed for 2030, in proportion to the growth in multicropping maize ethanol production from 2025 to 2030 projected by EPE. This means that sequestration would reach 0.59 million tCO2 associated with fermentation from producing 0.78 billion l of multicropping maize ethanol.

In the fourth scenario named 8billion+BECCS (optimistic), it is hypothetically assumed that the installed capacity of BECCS in 2025 and 2030 will be increased in proportion to the additional multicropping maize ethanol production volume (5.05 billion l in 2030). That is, all additional multicropping maize ethanol production will be associated with CO2 sequestration, assuming that the carbon credits considered in the study are capable of covering the installation costs of the BECCS projects.

The scenario simulated in the PAEG model allows assessment of impacts on household income and consumption (Fig. 1) in the region expanding double-maize ethanol production, while scenarios in the EPPA model capture changes in land use (Fig. 2) and in GHG emissions (Fig. 3) at country and world levels. Both models also provide changes in price and quantities in several markets and regions, as well as globally (Fig. 5). Sensitivity analyses concerning key model parameters were performed (Supplementary Information Section 4).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.