Soybean [Glycine max (L.) Merr.] is a major protein source for monogastric livestock and comprises 70% of the world’s protein-meal consumption (FAO Faostat, 2022). Surprisingly, only 2% of soybean protein is consumed directly as human food, with the remaining 98% used for animal feed (Goldsmith et al., 2004). Driven by the growing demand for oil and animal protein (Godfray and Garnett, 2014), soybean production in the Americas has doubled in the last two decades, mainly through an expansion in the cropping area (Zalles et al., 2019; Song et al., 2021; Marin et al., 2022). Currently, the United States, Brazil, and the Southern Cone (Argentina, Paraguay, and Uruguay) account for 95% of soybean production in the continent. It is worth noting that almost one-third of the arable land in the Western Hemisphere (95 million ha) is used to produce over 85% of the world’s soybean grain (OECD and FAO, 2022).

According to the relevance of the feed end-use of soybean and because modern monogastric rations are oriented to maximize feed conversion efficiency, the market demands high-protein concentration soybeans (Hertsgaard et al., 2019; Pope et al., 2023). In addition, the ultimate nutritional value of soybean protein is determined by the proportion of each amino acid in total protein, the amino acid profile (Osborne et al., 1914; Tessari et al., 2016). Optimizing the amino acid profile can lead to less excess amino acid deamination and excretion of nitrogen, resulting in increased conversion efficiency of vegetable into animal protein (Waguespack et al., 2009; Attia et al., 2020). Soybean protein analysis typically reports 18 amino acids (Kovalenko et al., 2006), with lysine, threonine, tryptophan, cysteine, and methionine considered the most critical amino acids as they frequently limit the conversion efficiency of soybean meal in monogastric animal products (Thakur and Hurburgh, 2007; Ravindran et al., 2014).

Soybean grain composition is determined by a combination of environmental resources such as radiation, water, and nutrients, as well as regulators like temperature, all of which interact with the genotype (Piper and Boote, 1999; Carrera et al., 2009, 2022; Rotundo and Westgate, 2009; Anthony et al., 2012; Pfarr et al., 2018; Bosaz et al., 2019). The spatial structure of these factors varies due to differences in soil fertility, precipitation, temperature, cultivar adoption, and other factors. As a result of these combined effects, soybean grain protein concentration exhibits geographical variability. Moreover, year-to-year variability also exists due to weather-determined environmental factors (Rotundo et al., 2016).

There is some limited evidence that suggest that changes in the amino acid profile are linked to variations in protein concentration, and thus may be also influenced by the environment (Park and Hurburgh, 2002; Song et al., 2016; Assefa et al., 2018). Controlled experiments have shown that as grain protein concentration increases, the abundance of critical amino acids decreases (Pfarr et al., 2018). This is particularly relevant because regions with lower protein concentration may produce soybean meal of better quality (William et al., 2020).

Knowledge about the occurrence of spatial patterns in grain composition across the Western Hemisphere would be relevant for the market, trading, and the industry (Hertsgaard et al., 2019). Defining spatial patterns would help in the segregation of production according to protein concentration or amino acid profile, which would allow for the optimization of quality for different end-uses (Pope et al., 2023). Different soybeans could even be combined in a way that the amino acid profile results in a higher protein score (Messina, 2022). Additionally, it is essential to consider spatial variability in the estimation of current and future protein and amino acids production capacities, especially since the future increase in production may not be evenly distributed among regions (OECD and FAO, 2022; Ricard et al., 2022).

Previous studies have demonstrated regional variations in soybean protein concentrations in the U.S., Brazil, and the Southern Cone (Breene et al., 1988; Rotundo et al., 2014; Assefa et al., 2019; Bosaz et al., 2019). While attempts have been made to assess the protein profile of some amino acids in the U.S. (Assefa et al., 2018), there is no comprehensive analysis of the geographical patterns for protein concentration and amino acid profiles across the entire Western Hemisphere. Therefore, in this study, we aim to (i) investigate the spatial patterns of soybean protein concentration and amino acid profiles, (ii) explore the interrelationships between amino acids and protein content in soybean grain, and (iii) project the soybean protein and amino acid production capacity of the Western Hemisphere.

Our analysis of 1817 soybean seed samples revealed an average crude protein concentration of 39.6%, ranging from 36.1 to 42.6% defined by the fifth and 95th percentiles (P5-P95), indicating a 16% variation relative to the average. On average, the sum of the 18 primary amino acids (net protein) accounted for 97.2% of grain crude protein, while the remaining 2.8% was represented by apparent non-protein N. The fraction of the most critical amino acids in protein averaged 14.7%, ranging from 14.1 to 15.2% (P5-P95), indicating only a 7% variation relative to the average. The complete compositional information at a subregion level is presented in Supplementary Tables S1, S2.

Major soybean cropping regions and subregions were sampled according to their contribution to total grain production. This original method allowed us to create realistic comparisons of grain composition. In this way, the study stands out as the first comprehensive examination of protein and amino acid profiles in soybeans across the entire Americas, surpassing previous research that primarily focused on individual country-level analyzes. Furthermore, this research pioneered the exploration of novel relationships among the most relevant amino acids and protein for soybean production systems, shedding new light on their complex interactions. Additionally, our study breaks new ground by providing valuable insights into the continent’s protein production capacity, an essential component of global soybean production, which has not been addressed in prior literature. This crucial perspective offers unique implications for the agricultural industry and food security on a global scale.

Our study revealed a consistent spatial pattern for crude protein concentration in soybeans across the United States, Brazil, and the Southern Cone soybean-producing regions. Brazil produced 2.0 and 5.4% more protein per ton of grain than the United States and the Southern Cone, respectively (Figure 2). Within the Southern Cone, we observed notable differences in protein concentrations among subregions, with a consistently higher crude protein concentration in the Northern Pampas, Paraguay, and Uruguay compared to the Southern and Central Pampas (Figure 3). In contrast, a relatively lower variability among subregions was found within the United States and Brazil.

It’s worth noting that the market perceived soybeans from the U.S. Northern Cornbelt to have lower protein concentrations, leading to trading discounts for soybeans shipped at Pacific Northwest (Hertsgaard et al., 2019; William et al., 2020). This concept is also supported by an older study by Hurburgh et al. (1990) which found that protein for soybean from the northern states were 1.5 to 2 percentage points lower compared to that from central states. However, from our dataset the effect was not consistent across years, and on average, soybeans produced in the Northern Corn Belt region had only 0.5 percentage points lower protein concentration than the other regions examined (Table 1 and Figure 3). This last difference aligns with more recent studies on soybean protein patterns across the United States (Rotundo et al., 2016; Assefa et al., 2019).

Crude protein content of commercial soybeans is typically determined by nitrogen combustion and multiplying the result by a conversion factor of 6.25 (Krul, 2019). Our analysis showed that this method overestimated the net protein content by an average of 3%, due to the presence of non-protein nitrogen fractions (Figure 2 and Supplementary Table S1). This finding is consistent with previous studies that were based on a more limited number of samples (Krober and Gibbons, 1962; Thakur and Hurburgh, 2007). Relaying solely on crude protein content can be misleading when assessing soybean meal quality, especially for monogastric animals that cannot utilize non-protein nitrogen fractions. Additional methods for evaluating the quality of soybean meal should be considered.

The protein amino acid profile in soybean exhibited consistent differences across regions, as shown in Table 1 and Figure 3. Among the assessed samples, the abundance of the most limiting amino acids varied up to 8.5% (Figures 3, 5), with the Southern Cone having the highest abundance. The ranking among subregions was season-dependent, but averaged across years, the Central and Southern Pampas subregions showed the highest limiting amino acids abundance (Figure 3). Limiting amino acids concentration variability was also highest for the Southern Cone (Figures 3, 4), suggesting that soybeans from the United States and Brazil may be more consistent and predictable in their amino acid profile. Understanding and predicting this variability could be relevant to adequate production according to market or industry requirements.

While being aware of the potential correlations between subregions and regions, our initial analysis, as presented in Figure 5, provided a valuable starting point for understanding relevant relationships between grain components. Critical amino acids abundance in protein was reduced when protein content increased. This dilution effect has been previously reported (Thakur and Hurburgh, 2007; Medic et al., 2014; Pfarr et al., 2018) but never using such a wide spectrum of samples. The implications of this dilution process on the nutritional value of the protein (Tessari et al., 2016) are significant and warrant further investigation. A higher abundance of the most limiting amino acids in protein would lead to a higher conversion efficiency of protein in meat (kg meat/kg protein; Selle et al., 2020). Soybean meal made from low protein soybeans produced in the Central and Southern Pampas would likely have a higher efficiency to convert protein in meat than a meal made from high protein soybeans produced in Brazil or the United States, which could be relevant from a sustainability standpoint (Pomar et al., 2021). Conversely our data showed that increasing protein concentration, though more diluted, also increased the amount of the most critical amino acids per ton of grain (Figure 5). Thus, meals made from higher protein concentration soybeans, like those produced in Brazil or the United States, would likely present a higher conversion efficiency of meal in meat (kg meat/kg meal) than meals made from lower protein concentration soybeans like those from the Central and Southern Pampas.

Our data revealed a strong negative association (R² = −0.55, p < 0.001) between oxaloacetate path-derived amino acids (lysine, methionine, threonine, isoleucine, asparagine) and alpha-ketoglutarate path-derived amino acids (arginine, glutamine, proline, histidine) in protein (Figure 6). Increasing protein concentration resulted in a greater increase in alpha-ketoglutarate pathway-derived amino acids compared to amino acids derived from the aspartate pathway. This association was primarily driven by an increase in arginine abundance and a decrease in lysine (Supplementary Figure S1) and is consistent with results from controlled experiments on soybean seed protein concentration alteration (Pfarr et al., 2018).

Arginine, with the highest N content per mass unit (32%), is a major nitrogen storage and transport molecule in plants (Winter et al., 2015). Conditions of abundant reduced N may promote the synthesis of this amino acid, which can store large quantities of N per unit of protein to support future N requirements of seedling. What is relevant from a feed end-use standpoint is that using the reduced nitrogen to produce a gram of arginine would allow the plant to synthesize almost twice the amount of lysine and tryptophane, three times the threonine and cysteine, and four times more methionine (Tillman, 2019). Understanding the physiological basis, limits, and control of these relationships among amino acids through breeding, genetic engineering, or crop management could provide a via to increase soybean protein value by adapting the amino acid profile according to its end-use.

This study represents a significant step towards a comprehensive understanding of the variation in soybean seed composition at a regional scale in the Western Hemisphere. However, to achieve a more complete and accurate prediction of soybean composition, further research is required to investigate the physiological processes involved. While recent studies have shed light on the factors that impact soybean seed protein concentration (Piper and Boote, 1999; Carrera et al., 2009; Rotundo and Westgate, 2009; Anthony et al., 2012; Pfarr et al., 2018; Bosaz et al., 2019), more research is needed to fully elucidate the control of the amino acid profile. Nonetheless, the extensive and high-quality dataset of grain composition presented in this study (Supplementary Table S2) can be a valuable resource for validating emerging conceptual frameworks and modeling tools in this field.

By 2030, soybean production in the Western Hemisphere is expected to reach 352 Mt, accounting for 86% of global soybean production (OECD and FAO, 2022). Based on this forecast, we projected the protein production capacity of the region to be approximately 120 Mt, containing only 17 Mt. of the most limiting amino acids for monogastric diets. While our estimations are subject to uncertainties, we used the best available data to estimate the protein and amino acid production capacity of the Western Hemisphere. Brazil is projected to be the most important region for soybean protein production (Table 2), driven by high soybean grain production and protein concentration (Figure 3). Notably, the Cerrado subregion alone could produce almost 20% of the soybean protein in the Western Hemisphere. These findings have important implications for addressing global challenges related to producing enough protein for feed and food protein markets (Kim et al., 2019; Guthman et al., 2022).

Finally, while our study focused on analyzing the protein concentration and amino acid profile of soybeans in the Western Hemisphere, it’s important to note that the complete nutritional and economic value of soybean grain includes not only the protein fraction but also the carbohydrate and the oil fractions (Medic et al., 2014). Therefore, to fully assess the value of soybean grain produced in different regions or its potential applications in various industries, it’s essential to consider not only the protein but also the oil and carbohydrates fractions of the grain.

This study presented a novel method for assessing soybean grain composition across the Western Hemisphere. Spatial processes played a relevant role in determining protein concentration, which exhibit a consistent pattern across the major soybean-producing regions. The analysis also uncovered differences in the amino acid profile across regions and established relationships among relevant amino acids which could be relevant for markets and industries. The generated dataset should be relevant for researchers seeking to understand processes involved in grain composition determination. Lastly, the study projected global soybean protein and amino acid production capacities for this major region by 2030, which is highly relevant to analyzing the global challenges of producing enough food. Overall, this research provides valuable information for industry, markets, and researchers useful as a benchmark for soybean grain composition across the Western Hemisphere.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

AC: conceptualization, data curation, formal analysis, investigation, methodology, visualization, and writing – original draft. JM-G: writing – review & editing. SN: conceptualization, investigation, methodology, funding acquisition, and writing – review & editing. All authors contributed to the article and approved the submitted version.