Areas Available for the Potential Sustainable Expansion of Soy in Brazil: A Geospatial Assessment Using the SAFmaps Database

Abstract

Recently, soybean production almost doubled in Brazil, reaching 122 million tonnes, and it is expected to increase even more. Brazil is the world’s largest producer and is primarily an exporter. From a sustainability point of view, soy production has been strongly criticized mainly in relation to deforestation, albeit for indirect effects. Soybean oil is a potential feedstock for the production of bio-jet fuels, which needs to be sustainable according to international criteria (sustainable aviation fuels—SAF). This paper aims to estimate the areas still available for soy expansion in Brazil, considering conditions that would allow the production of SAF. We used the SAFmaps platform, a geospatial database with information on the most promising bioenergy crops for SAF and their supply chains. Just by displacing pastures and observing a set of constraints, the total area available for expansion was estimated at 192.8 thousand km², of which 43% is of high suitability. These areas are concentrated in the Center-West region. Assuming a vertical supply chain, the results of the case studies of SAF production indicate potential feasibility, but some hypotheses considered are optimistic. Moreover, the results indicate that there can be sustainable production of soybean oil and contribution to the production of SAF.

1. Introduction

Remote sensing techniques are widely used to map the soybean areas and the dynamic of land use and land cover changes due to its expansion in Brazil [1,2,3,4,5,6]. In general, the monitoring of soy area is done by using thousands of images from a set of remote sensing satellites, like Landsat 5 (sensor TM), Landsat 7 (ETM+), Landsat 8 (OLI), and Sentinel-2 (MSI) [1,2,3,4,5,6]. The support of vegetation indexes of the MODIS sensor, such as Normalized Difference Vegetation Index (NDVI) [2], Enhanced Vegetation Index (EVI) [3,6], and Perpendicular Vegetation Index (PVI) [4], is also applied to analyze the dynamics of the agricultural areas. A large number of images is necessary (i) to obtain images free of clouds during the soy’s growth and development periods that coincide with the rainy season, (ii) due to the short growing cycle of the annual crops, for some varieties less than 100 days, and thus, in some cases, the culture identification is only possible during relatively short periods, and (iii) to take into account a temporal sowing window that can vary around two months among the suitable Brazilian states for soy, according to the agricultural calendars for each region [6].

Soybean is by far the largest agricultural crop in Brazil [7] and, currently, the country is responsible for one-third of worldwide soy production [8]. From 2010 to 2020, the cultivated area with soy expanded 60%, reaching almost 371,900 km² in 2020. In the same period, soy production increased from 68.8 to 121.8 million tonnes; in 2022, production is expected to be 17% higher compared to 2020 [7,9]. Mato Grosso (MT) state is the main soy producer with 30% of the national production, and recently, the culture has expanded in the region called MATOPIBA, comprising the states of Maranhão, Tocantins, Piauí, and Bahia.

Soy production is predominantly in the Cerrado biome (50% in 2020), one of the most threatened biodiversity hotspots in the world. Around 15% of soy is cropped in the Amazon biome (in 2019) [1], concentrated in the northern part of the MT state. The Amazon Soy Moratorium, a voluntary agreement implemented in 2006 that prohibits commercialization of soy grown in areas of the Amazon biome that have been deforested since 2008, reduced pressure for direct forest conversion in the Amazon. However, soy expansion over native vegetation in the Cerrado has increased and this conversion has not necessarily been legal [2,10,11].

Besides, the potential soy expansion over pastures can also induce indirect land-use changes (iLUC) [12,13], displacing livestock areas to other regions and eventually causing deforestation [10,11,14,15]. In order to mitigate iLUC, an option is to prioritize soy production on degraded pasturelands, a practice that has been growing due to restrictions on opening new agricultural areas in regions of Brazil, as São Paulo (SP) state and MATOPIBA, and that can be made possible with no-tillage, crop rotation, and adequate nitrogen management [16,17,18]. In this way, the demand for more pasture areas could be reduced, without compromising the production of cattle and meat, avoiding competition with food production and also reducing GHG emissions [19,20]. A second low-iLUC risk option is to apply a double-cropping system, characterized by the sequential harvest of two commercial crops in the same field, as the corn production in association with soy, which is common in the Center-West (CW) region of Brazil [7,9]. Double cultivation can enlarge the profitability of land, reduce adverse environmental impacts due to cropland expansion [21] and, specifically for biofuels production, can reduce costs and diversify the supply.

Recently, soybean oil has been considered a potential feedstock for bio-jet fuel production. International civil aviation aims to reduce CO₂ emissions by 50% in 2050, relative to 2005 levels [22], and in the short term, the only significant contribution can come from the development and commercialization of fuels produced from renewable crop-based biomasses or residues [23]. The aim is the production of certified Sustainable Aviation Fuels (SAF).

According to the Carbon Offsetting and Reduction Scheme for International Aviation (CORSIA) guidelines, at present, an alternative jet fuel may be eligible for SAF only if it meets the following requirements: (i) generates at least 10% less carbon emissions in its life cycle compared to the reference fossil fuel, and (ii) the feedstock for SAF production cannot be produced on lands with a high carbon stock and be associated with deforestation after 1 January 2008. After 2024, the sustainability of SAF production must also be verified regarding other issues, such as minimization of impacts on water, soil and air quality, conservation of biodiversity, and respect for social and labor rights [23].

Among the eight already certified routes of alternative bio-jet fuel production, the HEFA-SPK (synthetic paraffinic kerosene from hydroprocessed esters and fatty acids) pathway was approved by ASTM D7566 in 2011 and is currently commercially available [24]. Current production of bio-jet fuels is primarily through the HEFA-SPK route and it is anticipated that it will still be dominant in the years to come [25,26].

Due to the initial stage of production of bio-jet fuels and the requirements for being recognized as sustainable, an important issue is to assess the potential in specific regions and under what conditions this potential can be achieved and, in this sense, the SAFmaps platform was created. SAFmaps is an open access platform with a geospatial database about promising feedstocks for the production of SAF in Brazil [27]. The biomasses addressed are eucalyptus, soybean, palm, macaw palm, sugarcane and corn, and the geographic scope corresponds to the areas with the greatest potential for their production. For all these crops, SAFmaps provide maps of suitability, estimated yields, and predicted production costs. All the maps were validated with literature information and real crop cultivation data available for Brazil [28]. This innovative platform is the only one that makes geospatialized information publicly available for the Brazilian geographic scope. Based on the data available, the users can combine feedstock information with infrastructure data (i.e., roads, railways, pipelines, energy conversion units, etc.), and incorporate environmental and socio-economic restrictions into the analysis, such as legal reserves, sensitive biomes, specific land uses and land covers, and regions where violations to land- use and water-use rights have recently been reported. The applicability of the SAFmaps database is large as, besides SAF, the reported bioenergy crops could also be used for the production of, for instance, ethanol, biodiesel, hydrotreated vegetable oil (HVO), and pellets.

Based on the SAFmaps platform, an assessment of the appropriate conditions for the production of SAF was carried out in Brazil [28], followed by an assessment of three certified routes of SAF production, also in Brazil [29] (FT-SPK, Fischer–Tropsch process to produce synthetic paraffinic kerosene, from gasified biomass, based on planted eucalyptus; ATJ-SPK, i.e., alcohol-to-jet, based on anhydrous ethanol produced from sugarcane and corn, and HEFA-SPK, based on soy oil). An evaluation of the production of SAF from macaw oil by the HEFA-SPK route was also carried out [30].

Several studies have evaluated soy expansion [10,11,31] and its role in driving deforestation in Brazil, as well as the effectiveness of forest protection policies (Song [2] cites 14–22). However, there is still no broader national assessment of the areas available for the sustainable expansion of soy cropping, mainly aimed at its energy use. Sustainable agricultural expansion must combine information on available areas with geospatial environmental, economic, and social data, aiming to reduce pressure on natural resources and impacts on biodiversity, in addition to minimizing production and transport costs [32,33].

The main objective of this paper is to estimate the areas still available for soybean expansion in Brazil, but considering specific conditions to enable its sustainable production. For this, cultivation was considered only over anthropized pastures and, in addition, protected areas and sensitive biomes were also excluded. The restrictions imposed are typical of those that could allow the production of SAF with a lower carbon footprint. Since the place of cultivation depends on the existing infrastructure and the use of biomass, case studies are illustrated by the assumption that soybean oil would be used in the production of SAF, in industrial plants that are located close to international airports. For reasons of simplicity and also to keep reference with other studies, only soybean oil was considered for the production of SAF. In CORSIA, no default life cycle emissions values are presented for the production of SAF from the corn grain (only from co-products of ethanol production from corn), besides the fact that the oil content in the corn grain is relatively small when compared to soybean (in general, less than 6%, mass basis [34,35], compared to 21% for the soy grain [36]). In cases in which the combined production of soybean and corn was assumed, the net revenue from the sale of corn was considered to reduce the cost of soybean oil.

2. Materials and Methods

The procedure for estimating areas potentially available for soy expansion in Brazil is based on the database presented on the SAFmaps platform.

2.1. Suitability, Costs, and Yield Database

The maps of suitability for cultivation, estimated yield and production cost were developed for twelve Brazilian states, which present the most suitable conditions for the production of soybeans. The region addressed includes MATOPIBA—states of Maranhão (MA), Tocantins (TO), Piauí (PI), and Bahia (BA); the Center-West region—Mato Grosso do Sul (MS), Goiás (GO), Mato Grosso (MT), and Federal District (DF); part of Southeast region—the states of São Paulo (SP) and Minas Gerais (MG), and the whole South region—Paraná (PR), Santa Catarina (SC), and Rio Grande do Sul (RS)).

The procedures and criteria adopted for the production of these maps are detailed in Walter et al. [28,37] and are summarized in Table 1. All these maps were developed using the software QGIS 3.10. The geodatabase was validated against crop information currently available in Brazil, including yields and production costs of conventional and transgenic soy [1,7,9,36]. Irrigation was not considered, first because the practice is not usual in soybean cultivation in Brazil and, second, with the objective of mitigating potential impacts on water resources and minimizing production costs.

2.2. Estimative of Areas Available for the Potential Expansion of Soy in Brazil

Some specific conditions were imposed to assess the viability of sustainable soy expansion. First, the exclusion of (i) legally protected areas (i.e., conservation units [43,44], indigenous lands [45], and areas where quilombola communities are settled [46]), (ii) entire Pantanal and Amazon biomes, assumed as sensitive biomes, and (iii) areas covered by natural vegetation in January 2008 aiming to attend the Principle 2 of CORSIA’s sustainability criteria for SAF [23]). In a conservative approach, excluded areas of natural vegetation corresponded to the following classes, as presented in Mapbiomas—Collection 4.1, according to land use mapping in December, 2007 [1]: Forest Formation, Savanna Formation, Mangrove, Wetland, Grassland, Other Non-Forest Natural Formation. Second, agricultural expansion was assumed just considering the displacement of pastures in 2018. The information about the land use and land cover is based on the Mapbiomas’ mapping—Collection 4.1 [1].

Since soy production is highly mechanized, only areas with a slope lower than 13% [47] and modules of at least 100 hectares were considered. Soybean is traditionally produced in very large areas to enable full mechanization. In addition, 96% of the soybean produced in 2017 in Brazil was cultivated in farms larger than 100 hectares [7,48,49]. To exclude areas with a surface smaller than 100 hectares, pixels were aggregated in clusters using the Landscape Ecology Statistics (LecoS) plugin for QGIS [50]. As the classification of pastures is given in a raster layer [1], small isolated pixels within continuous pasturelands limit the aggregation of several areas available for production. Therefore, the pixels of the available area (before aggregation) were smoothed using a low-pass filter in the software ArcGis 10.8.

The resulting area was combined with the maps of predicted yields and estimated costs [27], for both soybean and corn, when the combined production was considered. In addition, the level of pasture degradation in 2018 in the selected areas was also explored, using the pasture degradation level map developed by LAPIG [51], which characterize pasture into four classes: no degradation, slight, moderate, and severe degradation. The classification of pasture degradation level, developed by LAPIG [51], was done based on the NDVI values from Landsat images. Median NDVI images were normalized for each biome and the results were stratified into pasture degradation state classes: no degradation (≥0.6), slight (0.5–0.6), moderate (0.4–0.5), and severe degradation (≤0.4) [51,52]. Finally, the existing and planned infrastructure (e.g., roads, railways, oil refineries, airports) were considered, aiming to choose the priority areas for SAF production.

Figure 1 summarizes the procedure adopted to estimate the area available for soybean expansion and describes the stages of construction of the case studies, which are detailed in the next section.

2.3. Construction of Case Studies

Based on the area available for expansion of soybean cultivation, two case studies were developed assuming a vertical supply chain of its oil for the production of SAF. In both cases, the soy oil production was estimated based on the potential biomass production within a zone created based on a radius of 200 km around two new processing poles located in the municipalities of Paranaíba, in the state of Mato Grosso do Sul, and Presidente Venceslau, in the state of São Paulo (Figure 2). These two poles were defined based on the following criteria: areas of high or medium suitability for soy and corn (whenever considered, corn would be the second crop), with potentially low costs, relatively far from areas where land-use and water-use rights have been violated in recent years, close to railways, not far from the largest oil refinery in the country (REPLAN; see Figure 2), where SAF would be produced, and from where it would be possible to easily transport SAF to the main international airports in Brazil. For simplicity, the oil would be extracted at the center of the pole, processing exclusively the soy produced within each circle shown in Figure 2.

The reported violations of land-use and water-use rights were compiled by the CPT (Comissão Pastoral da Terra) during the period 2016–2018 [28,53]. Reported cases include threats of homicide associated with disputes over land use and tenure, threats in general, illegal procedures, destruction of socio-cultural heritage, pollution, and reduced access to water bodies.

In Case 1, only soy production was considered, and no additional restrictions were imposed in relation to those mentioned above. Alternatively, two variants were considered, assuming that the actions could reduce the perception and, effectively, the potential risk of iLUC due to the expansion of soy production. In Case 2a, a double cropping system with corn produced as a second crop was assumed. Soybeans would be sown between September and December, and corn would be sown after the soybean harvest, between mid-December and late February. Suitability maps, estimated yield and expected costs for corn as a second crop are also available in the SAFmaps database [27,54]. In a second variant, Case 2b, soy and corn production were considered only on degraded pasturelands (i.e., pastures with moderate and severe degradation from [51]), in line with the understanding that cultivation under these conditions is an alternative to mitigate the iLUC risk. Here, the production of soybean and corn was evaluated in pasturelands classified as moderate and severe degradation. As mentioned, both in Cases 2a and 2b, the production of SAF from corn oil was not considered; the net revenue of corn production (i.e., the after taxes difference between the average market price in the producing regions in 2018 [36], in the harvest months, and the estimated costs of production), was deducted from the estimated cost of soybean oil.

2.4. Estimative of Feedstocks’ Transport Costs

It was assumed that soy would be processed in a unit near the crop areas and the extracted oil would be transported to the location of SAF production. The transport of biomass from the field, where the crop is produced, to the processing units, for feedstock oil extraction, would be by trucks.

To estimate the total distances from the field to the processing units (d_fu), it was supposed that several grain warehouses would be installed spread within the influence zone of each pole, located close to the existing roads. The parameter (d_fu) was calculated based on two components: (i) distance field—warehouse (d_fw) and (ii) distance warehouse—processing units (d_wu). The d_fw was obtained using the tool Proximity (Raster Distances) from the QGIS, which calculates the distance of each pixel (field) to the warehouse closer. The d_wu was calculated using the Network analyst extension from ArcGIS, which required, as inputs, the location of the warehouse and the map of roads [55,56]. Thiessen polygons were created based on the location of the warehouse, and the values of (d_wu) were attributed to the polygon associated to the respective warehouse, using ArcGIS. The value of (d_wu) for each Thiessen polygon was converted to raster format. Finally, the raster map of (d_fw) was summed to the raster map (d_wu) to compound the total distance (d_fu).

The transportation cost by truck (Cost_t,truck), in Brazilian currency per tonne (BRL⋅t⁻¹), as a function of the distance traveled (d), in km, was calculated based on Equation (1), which has been adjusted for different estimates presented by [57]. The equation leads to representative Brazilian grain freight costs in 2018.

(1)$Cos{t}_{t,truck}\left(\mathrm{BRL}·{\mathrm{t}}^{-1}\right)=\left[1.3322 · d^{-0.3076} \right] · d$

It was assumed that bio-jet fuel production should take place next to an oil refinery due to the hydrogen required in large quantities (needed for the HEFA route), the infrastructure related to blending terminals and storage capacity, and also the proximity to major international airports [58]; these aspects are essentials to the reduction of costs of SAF production. The refinery chosen, Refinaria do Planalto—REPLAN, in the municipality of Paulínia in São Paulo state, is the largest Brazilian refinery in oil processing capacity and is located near two international airports.

Soybean oil would be transported by rail to the refinery. The rail distances from the processing units to REPLAN are around 650 km in the case of Paranaíba and 750 km in the case of Presidente Venceslau. Transport costs were calculated from more accurate estimates of the cost of road transport, for a rail/road freight cost ratio of 0.5 which is valid for distances close to 1000 km [59,60,61]; this resulted in costs in the range between 0.31 and 0.74 USD$·$t⁻¹$·$km⁻¹.

The costs of feedstock production and transport were originally estimated in Reais (BRL) (2018) and posteriorly converted to Euros (EUR) using an average exchange rate for that year (4.41 BRL/EUR).

2.5. Industrial Parameters

The industrial parameters assumed are based on [62,63], which provide a review of performance factors and costs for different pathways for SAF production units (Table 2). According to the literature, 0.83 tonne of hydrocarbons (diesel and LPG, besides bio-jet, being the share of diesel production—mass basis—much higher than bio-jet) could be produced from one tonne of soy oil. The production of bio-jet fuels would be equal to 300.1 tonnes of bio-jet per day, operating all over the year with a 90% capacity factor. The total adjusted investment cost for the considered industrial plant would be EUR 662.1 million (2018), which is an estimate for the nth plant, i.e., after learning effects.

Industrial SAF production capacity considered is around three quarters of the few commercial plants in operation, but the units that will come into operation from 2022 onwards are expected to be significantly larger [26].

At the end, based on hypotheses presented by [62], the minimum selling price (MSP) of the SAF was calculated for all cases considered. Information about the hypotheses on calculating the MSP are presented in the Supplementary Materials, based on [62,63,64].

3. Results

3.1. Areas Available for the Potential Expansion of Soy

Table 3 presents details of the estimated area available for soy expansion in the four geographic regions evaluated, after imposing the set of restrictions here considered and pixel aggregation in clusters of 100 ha; the geospatial location of these areas is presented in Figure 3. The same information, with details for the 12 states, are presented in the Supplementary Materials.

In a first classification, the area available sums up 696,187 km², from the total 4427 × 10³ km² (total area of the 12 states). More than one constraint is applicable to some areas. The most restrictive aspect is related to areas covered by natural vegetation as of January 2008 (as a proxy, to be in accordance with CORSIA’s Principle 2), which would justify the exclusion of 53.6% of the total area. By the exclusion of portions of Pantanal and Amazon biomes, 17.7% of the total area would not be suitable for cropping. Areas that were legally protected (in 2018) correspond to 13% of the total, while areas with a slope higher than 13% justify the exclusion of 16% of the total area. In 2018, pasturelands corresponded to 30% of the total assessed area.

By imposing aggregation in clusters of at least 100 ha, the estimated available area for mechanized cultivation is restricted to less than 28% of the previously estimated area (or 4.4% of the total). Figure 3b,c exemplify the resulting areas classified by the LecoS plugin (in black). Comparing both figures, one can observe the effect of the low-pass filter that smooths the occurrence of non-pasture discontinuous pixels circumscribed by extensive pastures and, consequently, allows some areas previously discarded by the LecoS plugin to be considered in the aggregation process. From now on, the results are related to the estimated areas after smoothing and aggregating in pixels of at least 100 ha.

From the total estimated available area (192,800 km²), 83,476 km² (43%) have high suitability and 47,674 km² (25%) medium suitability for soybean cultivation. These areas with medium-high agricultural suitability represent 35% of the total area cropped with soy in 2020, with estimated yields higher than 4.0 t$·$ha⁻¹ (see Table 3 and Figure 4). For yields between 4.0–4.5 t$·$ha⁻¹, the potential soy production could achieve 52.5–59.0 Mt only in sites with medium-high suitability.

The Center-West region (CW) concentrates 62% of the total area available for crop expansion, according to the set of restrictions considered here. The bulk of the potential is in the MS state, with 73,325 km² being 65% in areas of high suitability for soybean (Figure 5). Around 45% of the areas in the CW are pastures characterized by moderate and severe levels of degradation. The median expected costs and yields are similar between CW, SE, and S regions, varying according to local specific aspects (in the case of costs) (Figure 4) and regional climatic parameters (in the case of yields [28]). Although the South region is responsible for 30% of the current national soy production, the non-availability of anthropized pasturelands suitable for mechanization restricted the potential for expansion of soy.

About 25% of the estimated area is in the MATOPIBA region, but these areas predominantly have medium-low suitability for soybeans (Figure 4 and Figure 5), lower yields, and higher production costs compared to other regions (Table 3). Areas identified for potential expansion are not concentrated near the current soy frontier in this region (blue rectangle in Figure 3a), as the recent expansion region is located in western Bahia and in the central region of MATOPIBA [65], where the estimated yields are high. This indicates that keeping the expansion of soybean cultivation close to the current agricultural frontier in MATOPIBA may be associated with the suppression of natural vegetation, a point already mentioned in other studies [10,11,66,67]. It is also estimated that the expansion of soybeans in MATOPIBA may be restricted by agricultural suitability and by physical and climatic parameters that are not suitable for soybean cultivation [65], as can be verified in Figure 4a and Figure 5. However, agrotechnological advances and irrigation may allow production in areas that in this paper are assessed as inadequate.

Figure 6 shows the existing and planned infrastructure that can be considered to guide strategies for the production and transport of SAF; the background of the map indicates the expected costs of soybean production in areas available for crop expansion. Considering that the production of SAF must be close to oil refineries, in addition to the proximity to the main airports, it can be seen in the figure that the suitable places for biomass production, given the conditions imposed here, are relatively distant. To reduce costs and the carbon footprint of SAF, transporting feedstock by rail would be essential. This justifies the consideration of new soybean production in the states of MS and PR, and in the west of SP, as explored in the case study reported below.

3.2. Results of Case Studies

Figure 7 shows the available areas for soy and corn production in Case 1 (only soy) (Figure 7a), Case 2a (Figure 7a,b), Case 2b (Figure 7c), and the expected biomass production costs, including the transport costs from the field (within each circle) to the respective feedstock processing and storage pole (triangle).

The costs of production are similar between the available areas and, depending on the distances, the transport costs contribute with around 1–12% of the total cost in the case of soy, and 1–23%, in the case of corn. The total area available for production in Cases 1 and 2a is 20,468 km², in Paranaíba, and 21,124 km², in Presidente Venceslau (Table 4). For Case 2b, in which production is estimated only in pastures with moderate and severe levels of degradation, the areas are reduced to 3972 km² and 3517 km², respectively in Paranaíba and Presidente Venceslau poles (Figure 7c). Table 4 synthesizes the main results for Cases 1, 2a, and 2b in both regions. The available areas, maximum production, and average CIF costs of grain in the processing poles are quite similar.

Around the poles in Paranaíba and Presidente Venceslau, the estimated maximum soy production would be 9.3 and 9.5 million tonnes per year, with a weighted average cost of 146.9 and 150.4 EUR$·$t⁻¹, respectively (at the processing unit; standard deviation 9.5 and 15.9 EUR$·$t⁻¹); the average yield would be 4.5 t$·$ha⁻¹ in both cases. The maximum soybean oil production in each pole would be 1.95 and 2.0 million t$·$year⁻¹, respectively, which is more than enough to supply the industry considered in this study.

The production of corn as second crop would be 12.5 and 12.8 million tonnes per year in Paranaíba and Presidente Venceslau, respectively, with a weighted average cost of 87.4 and 93.5 EUR$·$t⁻¹ at the center of the circle (standard deviation 8.9 and 39.7 EUR$·$t⁻¹); the average yield would be quite similar in both cases, 6.1 t$·$ha⁻¹.

Figure 8 shows the supply curves of soy and corn at the center of both producing regions for Cases 1, 2a, and 2b.

Considering the production of soybean and corn only in pasturelands with a reasonable level of degradation, after soil recovery, the impacts would be large in potential (due to the lower area), but not in costs, assuming the dilution of these recovery costs in several years of continuous production. The results can be seen in Table 4 and Figure 8.

Figure 9 shows the estimated supply curves of soy oil at REPLAN, for Cases 1 and 2a, for oil being transported by rail from the center of the producing region to the SAF industrial plant. For Case 1 and for the annual production that assures industrial production under the conditions considered (requirement of 830 kt of oil per year), the oil costs at REPLAN are quite similar for production in either of the two regions, with an average of 5.08 EUR$·$t⁻¹ for Paranaíba and 5.13 EUR$·$t ⁻¹ for Presidente Venceslau.

For Case 2a, assuming that the net revenue from the sale of corn production (produced as a second crop)—only in the area necessary for the production of soybean that allows the production of oil—is used to reduce the cost of oil, there is a small advantage for Paranaíba (MS) (on average, 2.23 EUR$·$GJ⁻¹ for Paranaíba and 2.52 EUR$·$GJ⁻¹ for Presidente Venceslau).

In Case 2b, due to the lower availability of land, soybean production is lower and, consequently, soybean oil production. Even adding the maximum production of the two evaluated sites (686 kt$·$year⁻¹), it is not possible to reach the annual requirement of the SAF industrial plant (a little less than 830 kt$·$year⁻¹). Then, for the production of SAF without restrictions, it would also be necessary to produce soy in other regions, under the same conditions, what was not explored in this paper. Figure 10 shows the soybean oil supply curve in REPLAN from the two producing regions. The cost estimate is based on the same procedure applied to Case 2a, but here, the net revenue from the sale of all corn produced as a second crop was considered. Comparing the results for Cases 2a and 2b, the higher cost of corn production, combined with the lower production, has a greater impact on the cost of soy oil produced in Presidente Venceslau (SP). The average CIF cost of soybean oil in REPLAN was estimated at 2.98 EUR$·$GJ⁻¹ for production around Paranaíba (MS) (standard deviation 0.11 EUR$·$GJ⁻¹), and 4.20 EUR$·$GJ⁻¹ for the Presidente Vencesalau case (standard deviation 0.23 EUR$·$GJ⁻¹).

Table 5 presents the estimated MSP of SAF production in each of the cases. Comparing the two feedstock producing regions, it can be seen that the results are quite similar and this indicates that from an economic point of view, there is no advantage of one over the other. In Case 2b, as the production of feedstock in only one area is not enough to feed the industrial plant, so it was assumed the combined supply, and even then it would not be possible to operate with the predicted annual capacity factor (90%). In this case, the available soybean oil would allow the industrial plant to operate with an annual capacity factor of 75%, with an obvious impact on the MSP.

4. Discussion

4.1. The Expansion of Soy Production and the Perception of Deforestation

Historically, the dynamics of soybean expansion indicates the prioritization of areas close to existing production areas and mainly displacing pastures [2,14,68]. In 2020, about 20% of all soybean planted area was in areas that in 2008 were anthropized pastures [1]. In the 2000s, soybean expansion occurred predominantly in the Center-West (CW) and MATOPIBA regions [2,7], which can be explained by a set of factors, especially in the CW case: the availability of extensive anthropized pastures, the existence of consolidated areas of soy production, the low price of land (in relative terms), and the high potential yields. In practice, prioritizing expansion close to existing crops is the main reason for deforestation associated with recent soy expansion in MATOPIBA region [2,10,11,66,67,69].

Branco et al. [32] evaluated the possible expansion of soybeans together with corn (as second crop) in Brazil, prioritizing the proximity to railroads and waterways to facilitate the flow of production. Except for the areas that were assumed to have some restriction in our study, the results of that study are very similar to those presented here, mainly in the regions CW, SE, and S.

Even recently, soy production is still associated with direct and indirect conversion of native vegetation [2,10,11,14]. Song et al. [2] studied the expansion of soybean cultivation between 2001 and 2016, which occurred mainly (about 40%) in the Brazilian Amazon (9820 km²) and in the Cerrado (35% of the total, i.e., 8850 km²). The authors defined direct drive when soy cultivation was identified within three years after forest clearing, and latent drive when cultivation occurred more than three years after deforestation. Their results show that in 5% of the deforested area in the period (which in absolute terms occurred mainly in the Cerrado—60%, followed by the Amazon—28%), soybean was planted less than three years later; evidencing the latent drive aspect (more than three years after deforestation), soybean was later identified in 4% of the deforested area in the period.

After the Amazon Soy Moratorium, which defines that soy cannot be produced on land deforested after mid-2008, the direct conversion of native vegetation into soy cropping reduced in the Amazon biome, but it continues in Cerrado biome [2,10,11]. From 2005 to 2017, 55% of the loss of native vegetation that occurred in the Cerrado was in the MATOPIBA region [11], concomitant with the growing agriculture in the region [70].

As can be seen in Figure 11a, most of the areas available for low-cost soy production are in the Cerrado, which could result in the expansion of the crop in this biome. Only 8.3% of the Cerrado biome is legally protected [71], and over 30% of the remaining vegetation outside protected areas could be legally converted under the Forest Code (totaling 380,000 km²) [11]. Forest Code is the national legislation that defines limits of conversion of areas with natural cover and what must be recovered to preserve ecosystem services. According to study [11], around 80,000 km² of Cerrado’s native vegetation is on areas highly suitable for soy, while 70,000 km² is on potentially suitable land. Scholars suggest an agreement aimed at zero conversion of native vegetation, similar to Moratorium, in order to preserve the Cerrado’s ecosystem services and biodiversity [10]. Results of studies such as those presented here can be useful in this regard.

Mainly in the MATOPIBA region, speculation and rising land prices due to soy production have put pressure on indigenous peoples, quilombolas, and local communities; smallholders who do not have land titles lost the right to use their land [53,72,73]. Regarding water-use rights, soy farms have physically restricted access to water sources to several local communities, and many cities have been harmed by the contamination of water bodies due to the use of fertilizers and pesticides applied to soy plantations. Although rain-fed soy cultivation predominates in MATOPIBA, the use of irrigation has increased, and there have been reports of excessive water use by farmers, without considering limits established by public licenses, while local communities have restrictions [53,72,73].

The perception that soybean cultivation is directly and/or indirectly associated with deforestation resulted in a high emission factor for the HEFA-SPK route based on its oil, as can be seen from the assessment made in the context of CORSIA: in the case of SAF production in Brazil, 67.4 gCO₂eq$·$MJ⁻¹ is the factor, with 27.0 gCO₂eq$·$MJ⁻¹ being the component due to iLUC, which would greatly reduce avoided emissions in comparison to 89 gCO₂eq$·$MJ⁻¹ of the fossil jet-fuel [23]. In addition to this aspect, other sustainability criteria will be considered after 2024 to allow the certification of bio-jet fuels as effectively sustainable, and this should reinforce the concern to seek sustainable conditions in the production of the feedstock.

4.2. SAF Results and the Conditioning of Sustainability

IRENA [25] claims that currently, bio-jet fuels produced by the HEFA-SPK route cost three to six times more than conventional ones, which obviously depends on the international price of oil and the feedstock used to produce the jet fuel. In the same report, it is mentioned that the price of HEFA-SPK in September 2020 was USD 2124 per tonne.

From an economic point of view, production by the HEFA-SPK route, from soybean oil, has the disadvantage that the oil is a commodity, and its market price is too high for this purpose. For instance, assuming the average international price of soybean oil in 2018 (15.53 EUR$·$GJ⁻¹), the MSP for the industrial unit considered here would be 24.63 EUR$·$GJ⁻¹ (1054 EUR$·$GJ⁻¹), assuming CAPEX and OPEX of the nth plant. The estimated MSPs could be barely reduced, as the reported cases already correspond to a very favorable situation in relation to the cost of the raw material. Taking Case 1 as a reference, the raw material contributes 45% of the estimated MSP. However, some scale effect can be expected, as the considered industrial plant has a nominal capacity of 62–84% of the two existing plants and is much smaller than the units that are expected to come into operation from 2022 onwards [26].

From the point of view of sustainability, the disadvantages are related to all the issues mentioned here, starting from the perception that soy causes, directly or indirectly, deforestation. Here, it is important to draw attention to the fact that sustainability criteria in the context of CORSIA are only part of the whole issue, as aircraft operators will want to reduce their exposure to the risk that the fuel they use could be considered unsustainable.

The hypothesis that SAF production is associated with a vertical supply chain, here explored aiming to reduce the feedstock cost, is unlikely due to the need to be deeply involved in agricultural activities. In any case, the estimated MSPs presented in this paper are quite low compared to what is currently happening, which raises the question of how soybean oil can contribute to boosting SAF production.

In practice, one possible approach is to combine soybean oil with other feedstocks, such as residues (or by-products) like used cooking oil (UCO) or beef tallow, and non-edible oils, such as macaw oil. Moreover, soy production needs to be as sustainable as possible, which refers to all the conditions imposed in this paper.

Ensuring non-deforestation is essential, but not enough, and adopting low iLUC risk practices are equally necessary [74,75]. The combined production of soy with second-crop corn reduces iLUC perception, but also allows cost reductions (even if it is not as favorable as the case explored here) and can also contribute to oil supply (a possibility not explored in this paper). However, despite the importance given in the literature [74,75,76], the issue of producing in degraded land is much more complex than what was presented here, because it is necessary to deepen the evaluation of the costs of soil recovery, the carbon stocks that can be achieved, and the yields after such actions.

Another important aspect of sustainability is to avoid production based on extensive monoculture, which the SAF producer will have difficulties to avoid if the supply chain is not verticalized. On the other hand, the supply curves for soybeans and corn, which are very flat in a large extent, suggest that it is possible to produce at low cost in different areas.

It is worth mentioning that large-scale production of bio-jet fuels based on HEFA-SPK from soybean oil would require an expressive amount of the feedstock. Only one plant able to produce 300 t.day-1 of bio-jet fuel (contributing to no more than 1.3% of the total Brazilian consumption in 2018 [77]; 3.0% of the consumption due to international flights departing from Brazil that year) would require an amount equivalent to 9.3% of the soy oil production in 2018 [36]. This suggests that the production of SAF from soybean oil would have an impact on the market, even considering a verticalized supply chain.

5. Conclusions

In this paper, we estimate the areas still available for soybean expansion in Brazil, considering conditions that would allow its sustainable production. The restrictions imposed are those that are understood to be necessary for bio-jet fuels to be considered sustainable (SAF—Sustainable Aviation Fuels).

The results show that significant soy expansion can still occur, without further suppression of natural vegetation and enforcing environmental law. The total available area was estimated at 192,800 km², of which 83,476 km² is on areas highly suitable and 47,674 km² with medium suitability for soy cultivation. The available areas of high-medium suitability are concentrated in the Brazilian Center-West region, with prospected average yields higher than 4.0 t$·$ha⁻¹ and expected costs of production between 600–1000 BRL$·$t⁻¹. It is also concluded that keeping crop expansion close to the current soybean frontier, in the region known as MATOPIBA, may be associated with the suppression of natural vegetation.

As an illustration, case studies were carried out, assuming the use of soybean oil for the production of SAF. A vertical supply chain was assumed in order to estimate a favorable case of lower production costs, with soybean production in areas where it never took place. In this sense, two producing regions were considered, both suitable for double cropping of soybean and corn. The economic viability was evaluated by the minimum selling price of the SAF. The results show MSP ranging from 9.7 to 12.9 EUR$·$GJ⁻¹ (414 to 553 EUR$·$t⁻¹), which is quite good compared to future SAF production if soybean oil were paid by market prices (1054 EUR$·$t⁻¹), or with the current price of HEFA-SPK (USD 2124 per tonne). Lowest results were obtained when the net revenue from the sale of corn (produced as a second crop) was used to reduce the cost of soybean oil. In the case that combined soy and corn cultivation only in degraded pastures, oil supply would not be sufficient to guarantee the production of SAF with the predicted annual capacity factor.

The results can be seen as an indication that soybean oil, if sustainably produced, can be used to complement a blend of other feedstocks, such as waste (e.g., used cooking oil) or inedible oils. It is important to remark that the risks of producing soy oil are low and, under appropriate conditions, sustainability can be ensured. In addition, if produced in a vertical supply chain, their costs can be low.

Author Contributions

Conceptualization, M.M.G., A.W., J.E.A.S. and J.V.R.; Data curation, M.M.G., N.V., D.D. and J.L.S.; Formal analysis, M.M.G. and A.W.; Funding acquisition, A.W. and J.E.A.S.; Methodology, M.M.G., A.W., J.E.A.S., J.V.R. and J.L.S.; Project administration, A.W. and J.E.A.S.; Supervision, A.W.; Writing—original draft, M.M.G. and A.W.; Writing—review and editing, M.M.G., A.W., J.E.A.S. and J.V.R. All authors have read and agreed to the published version of the manuscript.

Funding

The database used in this paper, available at www.safmaps.com, accessed on 2 February 2021, was developed in the context of a project supported by Boeing Research & Technology, a Division of the Boeing Company. The project was conceived as a collaborative between the University of Campinas (Unicamp) and the Boeing–Embraer Joint Research Center for Sustainable Aviation Fuels (SAF): n. 5389—BOEING/FEM/Biocombustíveis.

Informed Consent Statement

Not applicable.

Data Availability Statement

Supplementary Material is available in electronic format. Maps, reports and data can be accessed through www.safmaps.com (accessed on 2 February 2021). Moreover, detailed information can be accessed through the following links: SAFmaps—Soybean: doi.org/10.17632/jpwggmp9zy, SAFmaps—Corn: doi.org/10.17632/g25wt3t7k5, and SAFmaps—Infrastructure: doi.org/10.17632/kwdd5mbg4h.

Acknowledgments

The authors are grateful to The Boeing Company (Boeing Research & Technology division) for the financial support to the project Development of Database Management System (DBMS) for Sustainable Aviation Biofuel in Brazil.

Conflicts of Interest

The authors declare no conflict of interest.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures and Tables

Figure 1. Framework used to estimate the available land for soy expansion in the 12 Brazilian states, and to define the case studies.

Figure 2. Location of the case studies of soy production (the blue triangles indicate the poles), the available and predicted infrastructure, areas with land-use and water-use rights constraints, and the estimated suitability of soybean and corn in pasturelands (cluster of 100 ha).

Figure 3. (a) Available area for soy cropping expansion with (black) and without (green) pixel aggregation; the current new soy frontier in the 2000s is highlighted. Effects of (b) no smoothing and (c) after smoothing in the aggregation process.

Figure 4. (a) Suitability map, (b) expected costs, and (c) estimated yields for soybean production in the 12 Brazilian states assessed, indicating only contiguous areas with at least 100 hectares, in which soy cropping could be full mechanized.

Figure 5. (a) Share of the estimated available areas according to the suitability for soy, in percentage, in the assessed states; (b) Total available area in each state. No available area was identified in one of the states, RS, after smoothing and aggregation of pixels.

Figure 6. Expected costs of soybean production in available areas for agriculture expansion and required infrastructure for SAF production.

Figure 7. Expected costs of (a) soybean and (b) corn production in available areas for agriculture expansion, in clusters of 100 ha, considering transport costs from the field to the feedstock processing units (triangles); (c) Available areas for agriculture expansion only in pasturelands with moderate and severe degradation.

Figure 8. Estimated supply curves for soybean and corn in the center of the two new producing regions: Paranaíba (MS) and Presidente Venceslau (SP). Case 1: only soy production; Case 2a: double cropping system of soy and corn, and Case 2b: double cropping system, only displacing degraded pasturelands.

Figure 9. Estimated supply curves of soy oil at REPLAN, for Cases 1 and 2a, from two new producing regions: Paranaíba (MS) and Presidente Venceslau (SP).

Figure 10. Estimated supply curves of soy oil at REPLAN, for Case 2b, from two new producing regions: Paranaíba (MS) and Presidente Venceslau (SP).

Figure 11. Soybean cultivation in 2020 in the Brazilian Cerrado and (a) the expected costs of soybean production in available areas for agriculture expansion. In this study, restrictions were imposed on (b) biomes, (c) legally protected areas, (d) CORSIA sustainability criteria, and (e) municipalities with reported violations on land-use and water-use rights.

Table 1

General procedures and main assumptions considered in the development of maps of suitability, yields, and costs of soybean production.

Base Map	Procedures and Criteria
Suitability	Based on climatic conditions (mainly rainfall and atmospheric temperature), altitude, soil suitability and slope, using a Boolean classification. The climatic suitability criteria were determined according to the crop requirements database of [38], using climatic data from [39], considering the planting period from September to January. The altitude criterion was defined based on the location of the largest soybean areas in Brazil, using data presented in [39]. The slopes considered suitable are less than 13%, taking into account the objective of total mechanization; a digital elevation model presented in [40] was used to obtain the slope of the terrain. The soil classification according to the suitability for agriculture was defined based on soil characteristics described in [40] and the soil map available in [41] was used to spatialize the information. Each parameter was classified into three groups (low, medium, and high suitability). All the information was presented in raster format, with spatial resolution of 30 m × 30 m, and the maps were combined spatially. A pixel was classified as “low suitability” if at least one parameter presented low conditions to cultivation, “medium” if at least one parameter was classified as marginal (and not low), and “high” if the suitability of the pixel was adequate for all parameters. The final map presents the areas classified in low, medium, and high suitability for soybean production. Municipal cultivation data [7] and the map of soybean production in 2018 [1] were also used in the validation procedure.
Estimatedsoybean yield	A statistical regression model on a municipal basis was constructed between current soybean yields [7] and a set of explanatory variables (e.g., rainfall and temperature over the period of growth). For SAF production, only transgenic soy was considered. Dummy variables were introduced into the model to differentiate the best from worst average yield values. The municipal data were combined with the suitability map and the validation procedure was carried out with current data at the municipal level (average values) [7].
Expected costs of soybeanproduction	The agricultural costs of soybean production, in BRL (2018), were estimated considering new production areas and that the expansion of the soybean crop would occur only with the displacement of pastures. The costs are based on cost structures presented by [36] for different producing regions. Costs comprise sowing, crop management, harvest, short-term grain storage, and land prices (land used as pastures). The estimated values were compared to the information available in [36] and to market prices presented in [36,42], for validation.

Source: Adapted from Walter et al. [28,37].

Table 2

Technical parameters for SAF production based on HEFA-SPK route, assumed in the case studies, based on Walter [29].

Parameters	Units 1	Values
Industrial yield	$({\mathrm{t}}_{\mathrm{HC}}·$tf−1)	0.83
Industrial yield	$({\mathrm{kg}}_{\mathrm{SAF}}·$tf−1)	120.0
Input capacity	$({\mathrm{t}}_{\mathrm{f}}·$day−1)	2500
Hydrocarbon production	$({\mathrm{t}}_{\mathrm{HC}}·$day−1)	2075
SAF production	$({\mathrm{t}}_{\mathrm{SAF}}·$day−1)	300.1

¹ t_HC: tonnes of hydrocarbons; t_f: tonnes of the feedstock, in dry basis. Sources: [29,62,63].

Table 3

Available area for soy expansion after imposing restrictions, information about the average estimated costs and yields, and share of degraded pasturelands, by level of degradation.

Procedure	No Aggregation	Remaining Areas after Pixel Smoothing and Aggregation (100 ha) Procedure
	Available Area	Available Area		Estimated Costs		Estimated Yield		Shares ofDegraded Land a
Region	Area(km2)	Area(km2)	(%) a	Average(BRL$·$t−1)	Median(BRL$·$t−1)	Average(BRL$·$t−1)	Median(BRL$·$t−1)	Moderate (%)	Severe (%)
Center-West (CW)	288,232	119,580	62	811	604	3.8	4.5	24	21
MATOPIBA	192,819	47,274	25	980	1264	3.2	2.2	14	64
Southeast (SE)	180,749	22,326	12	665	604	4.3	4.5	22	38
South (S)	34,387	3617	2	677	604	4.6	4.5	28	19
Total	696,187	192,797	100

^a Related to the total area after smoothing procedure and aggregation of pixels in clusters of 100 ha.

Table 4

Summary of results for the two processing poles regarding soybean and corn production.

	No Additional Restriction				Only in Degraded Pasturelands a
	Paranaíba		PresidenteVenceslau		Paranaíba		PresidenteVenceslau
	Soy	Corn	Soy	Corn	Soy	Corn	Soy	Corn
Case studies	1 and 2a	2a	1 and 2a	2a	2b	2b	2b	2b
Area available for production (km2)	20,468	20,468	21,124	21,124	3972	3972	3517	3517
Cultivable area compared to total (%)	18%	18%	17%	17%	3%	3%	3%	3%
$\mathrm{Maximum}\mathrm{production}({10}^3\mathrm{t}·$year−1) (grain)	9270	12,536	9515	12,808	1798	2433	1583	2145
$\mathrm{Maximum}\mathrm{production}({10}^3\mathrm{t}·$year−1) (soy oil) b	1948	–	1999	–	365	–	321	–
$\mathrm{Weighted}\mathrm{average}\mathrm{yield}(\mathrm{t}·$ha−1·year−1) (grain)	4.53	6.12	4.50	6.11	4.53	6.13	4.50	6.10
$\mathrm{Weighted}\mathrm{CIF}\cos \mathrm{t}\mathrm{at}\mathrm{the}\mathrm{center}\mathrm{of}\mathrm{the}\mathrm{area}(\mathrm{EUR}·$t−1)	146.9	87.4	150.4	93.5	146.8	86.8	147.6	94.5

^a Moderate and severe degradation levels; ^b oil content 21%, mass basis.

Table 5

Summary of results: SAF production based on HEFA-SPK, using soybean oil produced in two self-dedicated poles.

Regions of Soy (and Corn) Production	Paranaíba (MS)		Presidente Venceslau (SP)		CombinedProduction a
Case studies	1	2a	1	2a	2b
$\mathrm{CIF}\cos \mathrm{t}\mathrm{of}\mathrm{soybean}\mathrm{oil}(\mathrm{EUR}·$GJ−1)	5.08	2.23	5.13	2.52	3.56
$\mathrm{SAF}\mathrm{MSP}(\mathrm{EUR}·$GJ−1)	12.87	9.66	12.93	9.90	12.13
$\mathrm{SAF}\mathrm{MSP}(\mathrm{EUR}·$t−1)	550.86	413.64	553.27	427.60	519.03
$\mathrm{SAF}\mathrm{annual}\mathrm{production}(1000\mathrm{t}·$year−1)	98.6	98.6	98.6	98.6	74.0
$\mathrm{SAF}\mathrm{annual}\mathrm{production}(1000{\mathrm{m}}^3·$year−1)	79.3	79.3	79.3	79.3	59.5

^a Combined production of Paranaíba and Presidente Venceslau.

Supplementary Materials

The following are available online at https://www.mdpi.com/article/10.3390/rs14071628/s1, Table S1: Available area for soy expansion after imposing restrictions, information about the average estimated costs and yields, detailed for the 12 states, Table S2: Co-products in the FT-SPK route, Table S3: Assumed prices of product and co-products, Table S4: Assumed prices of the required inputs, Table S5: Assumptions for calculating the MSP of SAF.

References

1. Mapbiomas. Map and Data Platform. Brasil Collections 4.1, 5.0 and 6.0. Available online: https://mapbiomas.org/ (accessed on 2 October 2020).

2. Song, X.-P.; Hansen, M.C.; Potapov, P.; Adusei, B.; Pickering, J.; Adami, M.; Lima, A.; Zalles, V.; Stehman, S.V.; Di Bella, C.M. et al. Massive Soybean Expansion in South America since 2000 and Implications for Conservation. Nat. Sustain.; 2021; 4, pp. 784-792. [DOI: https://dx.doi.org/10.1038/s41893-021-00729-z] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34377840]

3. Agrosatélite Geotecnologia Aplicada Ltda. Análise Geoespacial da Soja No BIOMA Cerrado: Dinâmica da Expansão | Aptidão Agrícola da Soja | Sistema de Avaliação para Compensação Financeira: 2001 a 2019; Agrosatélite Geotecnologia Aplicada Ltda: Florianópolis, Brazil, 2020; 60.

4. da Silva Junior, C.A.; Leonel-Junior, A.H.S.; Rossi, F.S.; Correia Filho, W.L.F.; de Barros Santiago, D.; Oliveira-Júnior, J.F.D.; Teodoro, P.E.; Lima, M.; Capristo-Silva, G.F. Mapping Soybean Planting Area in Midwest Brazil with Remotely Sensed Images and Phenology-Based Algorithm Using the Google Earth Engine Platform. Comput. Electron. Agric.; 2020; 169, 105194. [DOI: https://dx.doi.org/10.1016/j.compag.2019.105194]

5. Paludo, A.; Becker, W.R.; Richetti, J.; Silva, L.C.D.A.; Johann, J.A. Mapping Summer Soybean and Corn with Remote Sensing on Google Earth Engine Cloud Computing in Parana State—Brazil. Int. J. Digit. Earth; 2020; 13, pp. 1624-1636. [DOI: https://dx.doi.org/10.1080/17538947.2020.1772893]

6. Rudorff, B.; Risso, J.; Baldi, C.; Aguiar, D.; Salgado, M.; Perrut, J.; Oliveira, L.; Virtuoso, M.; Cabral, G.; Rosa, O. et al. Geospatial Analysis of Soy Expansion, Associated Land Use and Land Cover Change, and Agricultural Suitability in the Brazilian Amazon Biome—2000 to 2017; Agrosatélite Applied Geotechnology Ltda: Florianópolis, Brazil, 2018; 40.

7. IBGE—Instituto Brasileiro de Geografia e Estatística. Sistema IBGE de Recuperação Automática—SIDRA. Available online: https://sidra.ibge.gov.br/home/cnt/brasil (accessed on 2 November 2021).

8. FAO Statistics. FAOSTAT. Available online: https://www.fao.org/faostat/en/#home (accessed on 2 December 2021).

9. CONAB—Companhia Nacional de Abastecimento. Available online: https://www.conab.gov.br/ (accessed on 2 December 2021).

10. Soterroni, A.C.; Ramos, F.M.; Mosnier, A.; Fargione, J.; Andrade, P.R.; Baumgarten, L.; Pirker, J.; Obersteiner, M.; Kraxner, F.; Câmara, G. et al. Expanding the Soy Moratorium to Brazil’s Cerrado. Sci. Adv.; 2019; 5, eaav7336. [DOI: https://dx.doi.org/10.1126/sciadv.aav7336] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31328157]

11. Rausch, L.; Gibbs, H.; Schelly, I.; Brandão Junior, A.; Morton, D.; Carneiro Filho, A.; Strassburg, B.; Walker, N.; Noojipady, P.; Barreto, P. et al. Soy Expansion in Brazil’s Cerrado. Conserv. Lett.; 2019; 12, e12671. [DOI: https://dx.doi.org/10.1111/conl.12671]

12. Lapola, D.M.; Schaldach, R.; Alcamo, J.; Bondeau, A.; Koch, J.; Koelking, C.; Priess, J.A. Indirect Land-Use Changes Can Overcome Carbon Savings from Biofuels in Brazil. Proc. Natl. Acad. Sci. USA; 2010; 107, pp. 3388-3393. [DOI: https://dx.doi.org/10.1073/pnas.0907318107]

13. Berndes, G.; Bird, N.; Cowie, A. Bioenergy, Land Use Change and Climate Change Mitigation: Background Technical Report (No. ExCo:2011:04); Background Technical Report IEA Bioenergy: Rotorua, New Zealand, 2011; 62.

14. Picoli, M.C.A.; Rorato, A.; Leitão, P.; Camara, G.; Maciel, A.; Hostert, P.; Sanches, I.D. Impacts of Public and Private Sector Policies on Soybean and Pasture Expansion in Mato Grosso—Brazil from 2001 to 2017. Land; 2020; 9, 20. [DOI: https://dx.doi.org/10.3390/land9010020]

15. Trase Insights—Yearbook. Available online: https://insights.trase.earth/yearbook/contexts/brazil-soy (accessed on 2 January 2022).

16. da Silva PC, G.; Tiritan, C.S.; Echer, F.R.; dos Santos Cordeiro, C.F.; Rebonatti, M.D.; dos Santos, C.H. No-Tillage and Crop Rotation Increase Crop Yields and Nitrogen Stocks in Sandy Soils under Agroclimatic Risk. Field Crops Res.; 2020; 258, 107947. [DOI: https://dx.doi.org/10.1016/j.fcr.2020.107947]

17. dos SCordeiro, C.F.; Batista, G.D.; Lopes, B.P.; Echer, F.R. Interactive Effects of Nitrogen-Fixing Bacteria Inoculation and Nitrogen Fertilization on Soybean Yield in Unfavorable Edaphoclimatic Environments. Sci. Rep.; 2019; 9, 15606. [DOI: https://dx.doi.org/10.1038/s41598-019-52131-7]

18. Cordeiro CF, D.S.; Batista, G.D.; Lopes, B.P.; Echer, F.R. Cover Crop Increases Soybean Yield Cropped after Degraded Pasture in Sandy Soil. Rev. Bras. Eng. Agrícola Ambient.; 2021; 25, pp. 514-521. [DOI: https://dx.doi.org/10.1590/1807-1929/agriambi.v25n8p514-521]

19. Alkimim, A.; Sparovek, G.; Clarke, K.C. Converting Brazil’s Pastures to Cropland: An Alternative Way to Meet Sugarcane Demand and to Spare Forestlands. Appl. Geogr.; 2015; 62, pp. 75-84. [DOI: https://dx.doi.org/10.1016/j.apgeog.2015.04.008]

20. Bergtold, J.S.; Caldas, M.M.; Sant’anna, A.C.; Granco, G.; Rickenbrode, V. Indirect Land Use Change from Ethanol Production: The Case of Sugarcane Expansion at the Farm Level on the Brazilian Cerrado. J. Land Use Sci.; 2017; 12, pp. 442-456. [DOI: https://dx.doi.org/10.1080/1747423X.2017.1354937]

21. Borchers, A.; Truex-Powell, E.; Wallander, S.; Nickerson, C. Multi-Cropping Practices: Recent Trends in Double-Cropping; EIB-125 U.S. Department of Agriculture, Economic Research Service: Washington, DC, USA, 2014; pp. 1-22.

22. IATA—International Air Transport Association. Climate Change. Available online: www.iata.org/en/programs/environment/climate-change (accessed on 3 August 2021).

23. ICAO—International Civil Aviation Organization. ICAO Document Series. Available online: https://www.icao.int/environmental-protection/CORSIA/Pages/CORSIA-Eligible-Fuels.aspx (accessed on 2 December 2020).

24. ASTM D7566-21. Standard Specification for Aviation Turbine Fuel Containing Synthesized Hydrocarbons; ASTM International: West Conshohocken, PA, USA, 2020.

25. IRENA. Reaching Zero with Renewables: Biojet Fuels; International Renewable Energy Agency: Abu Dhabi, United Arab Emirates, 2021; pp. 1-96.

26. van Dyk, S.; Saddler, J. Progress in Commercialization of Biojet/Sustainable Aviation Fuels (SAF): Technologies, Potential and Challenges. IEA Bioenergy Task 39; IEA Bioenergy: Rotorua, New Zealand, 2021; pp. 1-95.

27. SAFmaps. Project Web Page. Available online: http://www.safmaps.com/ (accessed on 2 December 2020).

28. Walter, A.; Seabra, J.; Rocha, J.; Guarenghi, M.; Vieira, N.; Damame, D.; Santos, J.L. Spatially Explicit Assessment of Suitable Conditions for the Sustainable Production of Aviation Fuels in Brazil. Land; 2021; 10, 705. [DOI: https://dx.doi.org/10.3390/land10070705]

29. Walter, A.; Seabra, J.; Rocha, J.; Guarenghi, M.; Vieira, N.; Damame, D.; Santos, J.L. Spatially Explicit Assessment of the Feasibility of Sustainable Aviation Fuels Production in Brazil: Results of Three Case Studies. Energies; 2021; 14, 4972. [DOI: https://dx.doi.org/10.3390/en14164972]

30. Walter, A.; Seabra, J.; Rocha, J.; Guarenghi, M.; Vieira, N.; Damame, D.; Santos, J. Bio-Jet Fuels Production from Macaw Oil Palm in Brazil: An Assessment Based on a Comprehensive Database of Feedstocks. Proceedings of the European Biomass Conference and Exhibition, Virtual Conference, ETA; Florence, Italy, 26–29 April 2021; pp. 45-56. [DOI: https://dx.doi.org/10.5071/29thEUBCE2021-1AO.7.4]

31. Chaves, M.E.D.; De Carvalho Alves, M.; De Oliveira, M.S.; Sáfadi, T. A Geostatistical Approach for Modeling Soybean Crop Area and Yield Based on Census and Remote Sensing Data. Remote Sens.; 2018; 10, 680. [DOI: https://dx.doi.org/10.3390/rs10050680]

32. Branco, J.E.H.; Bartholomeu, D.B.; Alves Junior, P.N.; Caixeta Filho, J.V. Mutual Analyses of Agriculture Land Use and Transportation Networks: The Future Location of Soybean and Corn Production in Brazil. Agric. Syst.; 2021; 194, 103264. [DOI: https://dx.doi.org/10.1016/j.agsy.2021.103264]

33. FAO—Food and Agriculture Organization of the United Nations. The Future of Food and Agriculture: Trends and Challenges; FAO: Rome, Italy, 2017; pp. 1-180.

34. Duarte, A.P.; Carvalho, C.R.L.; Cavichioli, J.C. Densidade, teor de óleo e produtividade de grãos em híbridos de milho. Bragantia; 2008; 67, pp. 759-767. [DOI: https://dx.doi.org/10.1590/S0006-87052008000300026]

35. Mittelmann, A.; de Filho, J.B.M.; de Lima, G.J.M.M.; Hara-Klein, C.; da Silva, R.M.; Tanaka, R.T. Análise Dialética do teor de óleo e milho. Curr. Agric. Sci. Technol.; 2006; 12, pp. 139-143. [DOI: https://dx.doi.org/10.18539/cast.v12i2.4509]

36. Agrianual. Private Database on Agribusiness. Available online: http://www.agrianual.com.br/ (accessed on 12 March 2020).

37. Walter, A.; Seabra, J.; Rocha, J.; Guarenghi, M.; Vieira, N.; Dalmane, D.; Santos, J.L. SAFmaps-Soybean, Mendeley Data, 2021, 1. Available online: https://data.mendeley.com/datasets/jpwggmp9zy/3 (accessed on 2 December 2020). [DOI: https://dx.doi.org/10.17632/jpwggmp9zy]

38. FAO—Food and Agriculture Organization of the United Nations. Crop Ecological Requirements Database (ECOCROP); FAO: Rome, Italy, Available online: https://www.fao.org/land-water/land/land-governance/land-resources-planning-toolbox/category/details/en/c/1027491/ (accessed on 18 April 2021).

39. Alvares, C.A.; Stape, J.L.; Sentelhas, P.C.; de Moraes Gonçalves, J.L.; Sparovek, G. Köppen’s Climate Classification Map for Brazil. Meteorol. Z.; 2013; 22, pp. 711-728. [DOI: https://dx.doi.org/10.1127/0941-2948/2013/0507]

40. Manzatto, C.; Freitas-Junior, E.; Peres, J.R.R. Uso Agrícola dos Solos Brasileiros; Portal Embrapa; Embrapa Solos: Rio de Janeiro, Brazil, 2002; pp. 1-184.

41. IBGE—Instituto Brasileiro de Geografia e Estatística. Mapeamento de Recursos Naturais Do Brasil. Escala 1:250.000. Pedologia. Versão 2019. Available online: https://geoftp.ibge.gov.br/informacoes_ambientais/pedologia/vetores/escala_250_mil/ (accessed on 3 February 2020).

42. ABIOVE—Associação Brasileira das Indústrias Óleos Vegetais. Statistics Database. Available online: https://abiove.org.br/ (accessed on 3 February 2020).

43. BRAZIL-Ministério do Meio Ambiente. Download de Dados Geográficos. Available online: http://mapas.mma.gov.br/i3geo/datadownload.htm (accessed on 12 February 2020).

44. ICMBio-SIMRPPM. Reservas Particulares do Patrimônio Natural. Available online: https://sistemas.icmbio.gov.br/simrppn/publico/ (accessed on 12 February 2020).

45. FUNAI—Fundação Nacional do Índio. Geoprocessamento e Mapas. Available online: https://www.gov.br/funai/pt-br/atuacao/terras-indigenas/geoprocessamento-e-mapas (accessed on 12 March 2020).

46. INCRA—Instituto Nacional de Colonização e Reforma Agrária. Acervo Fundiário. Available online: https://acervofundiario.incra.gov.br/i3geo/interface/openlayers.htm (accessed on 12 March 2020).

47. TOPODATA. Banco de Dados Geomorfométricos Do Brasil. Available online: http://www.dsr.inpe.br/topodata/ (accessed on 3 February 2020).

48. Lima, M.; Skutsch, M.; de Medeiros Costa, G. Deforestation and the Social Impacts of Soy for Biodiesel: Perspectives of Farmers in the South Brazilian Amazon. Ecol. Soc.; 2011; 16, 4. [DOI: https://dx.doi.org/10.5751/ES-04366-160404]

49. Rodrigues, M.; Campos, I. Soybean Cropping by Family Farmers: A New Institutional Path for Rural Development in Brazilian Central-West. Italian Rev. Agric. Econ.; 2019; 74, pp. 29-39. [DOI: https://dx.doi.org/10.13128/rea-10851]

50. Jung, M. LecoS—A Python Plugin for Automated Landscape Ecology Analysis. Ecol. Inform.; 2016; 31, pp. 18-21. [DOI: https://dx.doi.org/10.1016/j.ecoinf.2015.11.006]

51. LAPIG—Laboratório de Processamento de Imagens e Geoprocessamento. Atlas Digital das Pastagens Brasileiras. Available online: https://pastagem.org/map (accessed on 20 September 2020).

52. Oliveira-Santos, C.; Mesquita, V.V.; Parente, L.L.; de Siqueira Pinto, A.; Ferreira, L.G. Assessing the Wall-to-Wall Spatial and Qualitative Dynamics of the Brazilian Pasturelands 2010–2018, Based on the Analysis of the Landsat Data Archive. Remote Sens.; 2022; 14, 1024. [DOI: https://dx.doi.org/10.3390/rs14041024]

53. CPT—Comissão Pastoral da Terra. Massacres No Campo. Available online: https://www.cptnacional.org.br/ (accessed on 12 March 2020).

54. Walter, A.; Seabra, J.; Rocha, J.; Guarenghi, M.; Vieira, N.; Dalmane, D.; Santos, J.L. SAFmaps—Corn, Mendeley Data, 2021, 1. Available online: https://data.mendeley.com/datasets/g25wt3t7k5/1 (accessed on 10 March 2020). [DOI: https://dx.doi.org/10.17632/g25wt3t7k5]

55. BRAZIL. Cartas e Mapas. Bases Cartográficas Contínuas. Versao 2019. Available online: http://geoftp.ibge.gov.br/cartas_e_mapas/bases_cartograficas_continuas/bc250/versao2019/shapefile/ (accessed on 10 March 2020).

56. BRAZIL. Mapas e Bases Dos Modos de Transportes—Português (Brasil). Available online: https://www.gov.br/infraestrutura/pt-br/assuntos/dados-de-transportes/bit/bitmodosmapas (accessed on 10 March 2020).

57. IEMA—Instituto Energia e Meio Ambiente. Reservatórios Verdes. São Paulo, Brazil, 2017. Available online: https://Energiaeambiente.Org.Br/Produto/Estudo-de-Pre-Viabilidade-Potencial-Do-Uso-de-Florestas-de-Eucalyptus-Na-Geracao-de-Eletricidade-No-Brasil (accessed on 10 March 2020).

58. de Jong, S.; Hoefnagels, R.; Wetterlund, E.; Pettersson, K.; Faaij, A.; Junginger, M. Cost Optimization of Biofuel Production—The Impact of Scale, Integration, Transport and Supply Chain Configurations. Appl. Energy; 2017; 195, pp. 1055-1070. [DOI: https://dx.doi.org/10.1016/j.apenergy.2017.03.109]

59. Lemos, R. Custos do Transporte Ferroviário; Instituto Brasil Logístico: São Paulo, Brazil, 2020; Available online: https://institutobrasillogistico.com.br/2020/01/29/custos-do-transporte-ferroviario/ (accessed on 12 March 2020).

60. Forkenbrock, D.J. External Costs of Truck and Rail Freight Transportation; Public Policy Center, University of Iowa Press: Iowa City, LA, USA, 1998.

61. Leite, C.; Bittencourt, J.; Pereira, L.; Marinho, C. Análise Comparativa Custos Entre Os Meios de Transporte Rodoviário e Ferroviário. Proceedings of the Congresso Nacional de Excelência em Gestão; São Paulo, Brazil, 29–30 September 2016.

62. de Jong, S.; Hoefnagels, R.; Faaij, A.; Slade, R.; Mawhood, R.; Junginger, M. The Feasibility of Short-Term Production Strategies for Renewable Jet Fuels—A Comprehensive Techno-Economic Comparison. Biofuels Bioprod. Biorefin.; 2015; 9, pp. 778-800. [DOI: https://dx.doi.org/10.1002/bbb.1613]

63. de Jong, S.; Antonissen, K.; Hoefnagels, R.; Lonza, L.; Wang, M.; Faaij, A.; Junginger, M. Life-Cycle Analysis of Greenhouse Gas Emissions from Renewable Jet Fuel Production. Biotechnol. Biofuels; 2017; 10, 64. [DOI: https://dx.doi.org/10.1186/s13068-017-0739-7]

64. Chemical Engineering. Chemical Engineering Price Cost Index. Available online: https://www.chemengonline.com/2019-cepci-updates-january-prelim-and-december-2018-final/ (accessed on 10 December 2019).

65. Araújo, M.L.S.; de Sano, E.E.; Bolfe, É.L.; Santos, J.R.N.; dos Santos, J.S.; Silva, F.B. Spatiotemporal Dynamics of Soybean Crop in the Matopiba Region, Brazil (1990–2015). Land Use Policy; 2019; 80, pp. 57-67. [DOI: https://dx.doi.org/10.1016/j.landusepol.2018.09.040]

66. Zalles, V.; Hansen, M.C.; Potapov, P.V.; Stehman, S.V.; Tyukavina, A.; Pickens, A.; Song, X.-P.; Adusei, B.; Okpa, C.; Aguilar, R. et al. Near Doubling of Brazil’s Intensive Row Crop Area since 2000. Proc. Natl. Acad. Sci. USA; 2019; 116, pp. 428-435. [DOI: https://dx.doi.org/10.1073/pnas.1810301115]

67. Alencar, A.; Shimbo, J.Z.; Lenti, F.; Balzani Marques, C.; Zimbres, B.; Rosa, M.; Arruda, V.; Castro, I.; Fernandes Márcico Ribeiro, J.P.; Varela, V. et al. Mapping Three Decades of Changes in the Brazilian Savanna Native Vegetation Using Landsat Data Processed in the Google Earth Engine Platform. Remote Sens.; 2020; 12, 924. [DOI: https://dx.doi.org/10.3390/rs12060924]

68. de Espindola, G.M.; de Aguiar, A.P.D.; Pebesma, E.; Câmara, G.; Fonseca, L. Agricultural Land Use Dynamics in the Brazilian Amazon Based on Remote Sensing and Census Data. Appl. Geography; 2012; 32, pp. 240-252. [DOI: https://dx.doi.org/10.1016/j.apgeog.2011.04.003]

69. Polizel, S.P.; Vieira RM DS, P.; Pompeu, J.; da Cruz Ferreira, Y.; de Sousa-Neto, E.R.; Barbosa, A.A.; Ometto, J.P.H.B. Analysing the Dynamics of Land Use in the Context of Current Conservation Policies and Land Tenure in the Cerrado—MATOPIBA Region (Brazil). Land Use Policy; 2021; 109, 105713. [DOI: https://dx.doi.org/10.1016/j.landusepol.2021.105713]

70. Carneiro-Filho, A.; Costa, K. The Expansion of Soybean Production in the Cerrado: Paths to Sustainable Territorial Occupation, Land Use and Production; INPUT and Agroicone: São Paulo, Brazil, 2016; pp. 1-30.

71. Sawyer, D.; Mesquita, B.; Coutinho, B.; Almeida, F.V.; Figueiredo, I.; Lamas, I.; Pereira, L.E.; Pinto, L.P.; Pires, M.O.; Kasecker, T. et al. Ecosystem Profile of the Cerrado Ecosystem Biodiversity Hotspot: Full Report. Critical Ecosystem Partnership Fund; Supernova: Brasília, Brazil, 2017.

72. Russo Lopes, G.; Bastos Lima, M.G.; Reis, T.N.P. dos Maldevelopment Revisited: Inclusiveness and Social Impacts of Soy Expansion over Brazil’s Cerrado in Matopiba. World Dev.; 2021; 139, 105316. [DOI: https://dx.doi.org/10.1016/j.worlddev.2020.105316]

73. Nakagawa, L.; Pó, M.; Seifer, P.; Kleeb, S. Entre Chapadas e Baixões do MATOPIBA-Dinâmicas Territoriais e Impactos Socioeconômicos na Fronteira da Expansão Agropecuária no Cerrado; Favareto, A. Greenpeace and Ilustre Editora: São Paulo, Brazil, 2019.

74. Wiegmann, K.; Hennenberg, K.; Fritsche, U.R. Degraded Land and Sustainable Bioenergy Feedstock Production: Issue Paper; Öko-Institut: Darmstadt, Germany, 2008; pp. 1-12.

75. Sumfleth, B.; Majer, S.; Thrän, D. Recent Developments in Low ILUC Policies and Certification in the EU Biobased Economy. Sustainability; 2020; 12, 8147. [DOI: https://dx.doi.org/10.3390/su12198147]

76. Ben Aoun, W.; Gabrielle, B. Chapter 8—Life Cycle Assessment and Land-Use Changes: Effectiveness and Limitations. Life-Cycle Assessment of Biorefineries; Elsevier: Amsterdam, The Netherlands, Academic Press: Cambridge, MA, USA, 2017; pp. 221-231.

77. ANP—Agência Nacional de Petróleo, Gás Natural e Biocombustíveis. Statistic Database. Available online: http://www.anp.gov.br/dados-estatisticos (accessed on 12 April 2021).