Facebook Google LinkedIn Twitter Blogger

Our corporate sponsors (learn more about corporate sponsorship)


Our Partners

Seed quality attributes of food-grade soybeans from the U.S. and Asia


Diversity of food-grade soybean is critical for utilization of genetic resources in cultivar development, germ plasm  enhancement,  and end product commercialization .The objective of this study was to assess seed quality attributes and phenotypic variability among 54 U.S. and 51 Asian food-grade cultivars and breeding lines. The results showed greater genetic diversity of protein content, calcium content, seed hardness, and seed size uniformity than other quality traits in both small and large-seeded genotypes evaluated in this study. Among the small-seeded soybeans, the U.S. genotypes were more diverse and exhibited higher swell ratio and oil content but lower stone seed ratio and protein content than Asian accessions. Among the large-seeded accessions, U.S. genotypes had higherstone seed ratio and oil content but lower swell ratio



Soybean [Glycine max (L.) Merr.] is the second largest crop grown in the United States (National Agricultural Statistics Service 2006). According to Soy Stats (2007), the soybean acreage in the U.S. reached 75.5 million acres in 2006, which accounted for 32% of the total world soybean production. In the same year, the average on-farm price was $6.20 per bushel for regular soybeans, whereas the price of specialty soybeans was approximately $9.0 per bushel (Soy Stats 2008).


With an average of 20% oil and 40% protein content, soybean is consumed as sources of both edible vegetable oil for human and high-protein feed supplements for livestock (Smith and Huyser 1987). Soyfood has been part of the diet in Asian culture for several centuries (Cui et al. 2004). Recent research has revealed that soyfood has nutritional qualities that reduce human blood serum cholesterol levels and lower the risk of cardiovascular diseases (Omoni and Aluko 2005). Soyfood is a natural source of isoflavones (daidzein, genistein, and glycitein). It is also a good source of calcium, which is important for the formation and maintenance of bones and teeth (US soyfoods directory 2006). Soyfood available in the market is classified into two groups based on the soybean seed size. Soyfoods made of large seeds ([20 g/100 seeds) include tofu (soybean curd), edamame (green vegetable soybeans), miso (fermented soup-based paste), and soymilk (soybeans, soaked, fine-ground and strained). Soyfoods made of small seeds (\12 g/100 seeds) include natto (fermented whole soybeans), soy sauce (tamari, shoyu, and teriyaki), tempeh (made of whole cooked soybeans), and bean sprouts. The global market for soyfood has increased considerably in the last 20 years.


Soybean seed need to satisfy specific physical and chemical requirements for soyfood production. In addition to seed size, Graef and Specht (1989) reported that visual appearance is the main consideration of food-grade soybeans for most buyers. For example, the seed must be round in shape, uniform, with light-colored hilum, and with yellow seed coat without physical defects such as mottling, splits, shriveling, purple stain, and insect damage. Poysa et al. (2002) reported that seed size uniformity affected water absorption and the quality of the final soy product. Bachman et al. (2001) affirmed that broken, shrunken, or discolored seed were undesirable due to the consumer requirements.


Stone seeds that do not absorb water during soaking cause serious problems for food processing because extra cost is required to remove them. The stone seed rate affects the texture and consistency of the soy products particularly for fermented soy food such as natto (Mullin and Xu 2001). According to Saio (1976), seed texture or seed hardness was positively associated with calcium content and negatively correlated with water absorption. Seed with harder texture have higher calcium content and absorbs less water. Seed hardness could be estimated by the seed swell ratio that was related to seed weight or water volume change before and after soaking

(Chen 2004). Taira (1990) reported that high water absorption was required to obtain soft steamed seeds.


Thus, processors aim to have a rapid and high water uptake in order to provide more products per unit of time (Poysa et al. 2002).


Seed composition requirements vary according to the type of soyfood. Seed with high protein content ([45%), low oil content, high sucrose, and low raffinose and stachyose content are suitable for making tofu. For soyfood such as natto made through short fermentation process, soybean seed with a high content of carbohydrates are preferred for the purpose of getting a quick conversion to simple sugars (Bachman et al. 2001).


Assessment of agronomic traits has been used to evaluate phenotypic diversity. Based on 15 qualitative and quantitative traits in 20,570 Chinese soybean accessions, Dong et al. (2004) reported that seed coat color had the highest diversity index among the qualitative traits. They also found that the plant height had the most variation among quantitative traits, and followed by seed size, protein content, growth period, and oil content. The seed size of those accessions ranged from smaller than 2 to as large as 46 g/100 seeds. The protein content ranged from 30 to 53%; and oil content ranged from 10 to 25%. Isozyme data was also used to estimate soybean diversity (Gorman 1984). In the evaluation of wild soybeans, the Korean collections were more diverse than the Chinese collections. However in cultivated soybeans, the Chinese collections had a higher level of diversity than Korean soybeans based on isozyme data (Gorman 1984). Cui et al. (2001) used leaf, stem, and seed traits including seed size, and protein and oil content to compare the diversity of modern Chinese and U.S. soybean cultivars. The variances of seed size, protein content, and oil content of U.S. cultivars were lower than those of Chinese cultivars. The Southern U.S. soybeans were more variable in protein and oil contents and less variable in seed size than the Northern U.S. soybeans.


Food-grade soybean breeding aims to increase the nutritional content and quality of protein and oil (Liu 1997). However, little research has been done on characterization of seed quality traits of food-grade soybeans. The objectives of this study were to characterize the physical and chemical properties of 105 food-type soybean lines from the U.S. and Asia for human consumption and to evaluate their phenotypic diversity for breeding purposes.


Materials and methods


Plant materials


One hundred and five specialty soybeans (53 small seeded and 52 large-seeded genotypes), including 54 U.S. and 51 Asian cultivars and breeding lines, were selected based on seed size for this study (Table 1).The U.S. genotypes were collected from six states, including Iowa, Ohio, Missouri, Virginia, Kansas, and North Dakota, and Asian genotypes were from South Korea and Japan. The experiment was conducted in Fayetteville, AR, in 2004 and 2005. The 105 entries were grown in the field in a randomized complete block design with two replications. Each entry was grown in a single-row plot of 3.05 m in length with 0.95-m row spacing. At maturity, each plot was harvested in bulk with a plot combine.


Trait evaluation


Eight seed quantitative traits were evaluated including seed size, seed uniformity, swell ratio, stone seed ratio, seed hardness, protein, oil, and calcium contents. The qualitative trait evaluated was hilum color, which was recorded as yellow, buff, brown, gray, imperfect black, and black.


The 100-seed weight of the 105 entries was measured when they were collected in 2003, and again after harvest in 2004 and 2005. Seed size of each entry did not consistently fall in the required range for soyfood in 2004 and 2005, although all entries were originally collected based on their seed size (either \12 g/100 seeds or[20 g/100 seeds). Although seed size heritability is as high as 0.92 (Cicek et al. 2006), seed size is affected significantly by location and genotype 9 location interaction (Chandler et al. 2000; Cicek et al. 2006). Therefore, we defined the entries with average seed size in 3 years of\14 g/100 seeds as small seeds and[19 g/100 seeds as large seeds.


Seed size uniformity was determined by a sieve procedure. A sample of 200 g seed from each entry was sieved sequentially on a set of slotted sieves. Sieves of no. 8 (3.2 mm), 10 (4.0 mm), 11 (4.4 mm), 12 (4.8 mm), 13 (5.2 mm), 14 (5.6 mm), 15 (6.0 mm), and 16 (6.4 mm) were used to screen small seeds; and sieves of no. 14 (5.6 mm), 15 (6.0 mm), 16 (6.4 mm), 17 (6.7 mm), 18 (7.1 mm), 19 (7.5 mm), 20 (7.9 mm), and 21 (8.3 mm) were used to screen large seeds. The sieve size numbers were used to represent the seed size classes. Seeds staying on each sieve were weighed, and the percentage of seed retained by each sieve were calculated as the weight divided by 200 g. Seed larger than no. 16 sieve are considered non-acceptable for the small-seeded entries and seeds smaller than no. 14 sieve were considered non-acceptable for the large-seeded entries. If there were no seeds in a seed class, 0.01% was used instead of 0 for data standardization purpose. The uniformity product was calculated by multiplying seed percentage of each size class within the defined seed size range requirement. The smaller value of the uniformity product represented higher uniformity of the seed size (Doehlert et al. 2004).


Seed swell ratio was represented by the ratio of (the weight of soaked soybean seeds)/(the weight of 50 g dry seeds - stone seed weight) (Zhang et al. 2008a, b). Fifty grams of seeds from each entry were weighed and soaked in heat-resistant plastic boxes with 250 ml water at ambient temperature for 16 h. Then, seeds were recovered from the soaking water with a sieve and blot-dried with paper towels. Stone seeds were picked out and weighed as stone-seed weight. Stone seed ratio was calculated as (stone seed weight/

50 g) 9 100%. Additionally, soaked seed samples were then pressure-cooked at 121.1C and 1.2 kg/cm2 for 20 min. Hardness of 30 g cooked seeds from each entry was measured in two replications using a TMS Texture System equipped with a 16-blade shear cell (TMS-2000, Food Technology Corp.). The maximum force in Newtons (N) to shear cooked beans was determined and reported as seed hardness (Song et al. 2003; Zhang et al. 2008a, b).


The HNO3 method was used to determine calcium content (Campbell and Plank 1991; Zhang et al. 2009). Briefly, 10 g of dry seeds from each entry were ground by a Knifetec 1095 sample mill (Foss,Inc), and 0.25 g of each ground sample was digested by 2.5 ml HNO3. The samples were added to 1 ml H2O2, gradually heated to 60C for 45 min, and then slowly increased to 120C for 1 h. Cooled samples were mixed with deionized water, filtered through #41 quantitative paper, and analyzed for calcium content using a Spectrometer Model CIROSCCE (Spectro Analytical Instruments, Inc.). A 25-g seed sample was analyzed for protein and oil content with a near infra-red (NIR) analyzer at USDA Northern Regional Research Center in Peoria, IL.




Statistical analysis


JMP 5.0 (SAS Institute Inc.) was used to analyze variance components and determine the least significance difference (LSD) of each trait for mean separation among 105 accessions (JMP 2005). Each trait was assigned various classes separately (Dong et al. 2004) based on the range of variation for the trait (Table 2). The Shannon diversity index (H) of each trait and the average of the Shannon diversity indice of one genotype were calculated based on H = -PPi ln (Pi) (Dong et al. 2004). Pi was the frequency of the ith class of traits. The pairwise genetic similarity (S) between pairs of genotypes was calculated using similarity coefficient (Nei and Li 1979), and S matrix was used for the unweighted pair group method using arithmetic averages (UPGMA) method to generate the cluster tree using NTSYS pc 2.1 software (Rohlf 2000).


Results and discussion


The U.S. and Asian food-grade soybeans differed in

some of the physical and chemical traits evaluated in

this study (Table 3). In small-seeded category, U.S.

and Asian accessions had similar seed size (100-seed

weight), seed hardness, seed size uniformity, swell

ratio, and calcium content as shown in ranges and

means of each trait. However, the U.S. small-seeded

soybeans exhibited higher oil content, lower protein

content, and lower stone seed ratio than Asian

accessions, indicating US small-seeded soybeans

were more desirable for natto production due to their

lower stone seed ratio. Evidently, the Asian smallseeded

soybeans were selected for higher protein

content, and their high stone-seed rates might be due

to lack of adaptation to the U.S. environment. The

smaller seed size of Asian accessions might be the

cause of, although non-significant, harder seeds,

lower water absorption, more calcium, and more

stone seeds. This association has been reported in

previous studies (Mullin and Xu 2001; Zhang et al.

2008a, b). In large-seeded category, U.S. soybeans

had, although similar seed size and calcium content,

higher stone seed ratio, higher oil content, lower

swell ratio, lower protein content, and higher seed

uniformity product than Asian soybeans. Asian largeseeded

soybeans were more uniform in size than U.S.

soybeans because most Asian entries in this study

were from South Korea, and Korean and Japanese

soybean accessions are closely related (Abe et al.

2003). The higher protein and lower oil content in the

Asian accessions compared to the U.S. accessions

The U.S. and Asian food-grade soybeans differed in some of the physical and chemical traits evaluated in this study (Table 3). In small-seeded category, U.S. and Asian accessions had similar seed size (100-seed weight), seed hardness, seed size uniformity, swell ratio, and calcium content as shown in ranges and means of each trait. However, the U.S. small-seeded soybeans exhibited higher oil content, lower protein content, and lower stone seed ratio than Asian accessions, indicating US small-seeded soybeans were more desirable for natto production due to their lower stone seed ratio. Evidently, the Asian small seeded soybeans were selected for higher protein content, and their high stone-seed rates might be due to lack of adaptation to the U.S. environment. The smaller seed size of Asian accessions might be the cause of, although non-significant, harder seeds, lower water absorption, more calcium, and more stone seeds. This association has been reported in previous studies (Mullin and Xu 2001; Zhang et al. 2008a, b). In large-seeded category, U.S. soybeans had, although similar seed size and calcium content, higher stone seed ratio, higher oil content, lower swell ratio, lower protein content, and higher seed uniformity product than Asian soybeans. Asian large seeded soybeans were more uniform in size than U.S. soybeans because most Asian entries in this study were from South Korea, and Korean and Japanese

soybean accessions are closely related (Abe et al. 2003). The higher protein and lower oil content in the Asian accessions compared to the U.S. accessions



were also observed in other studies (Cui et al. 2001; Wang et al. 2006). The Asian large-seeded collections may be a good choice as parents in breeding for high protein beans for tofu and milk. In this research, a negative correlation between protein and oil content

in both small-seeded and large-seeded soybeans was confirmed as in other studies (Hartwig 1994; Wilcox and Shibles 2001).


Table 4 shows the genetic diversity of all genotypes based on nine seed quality traits evaluated in this study. Among all small-seeded genotypes, protein content had the highest variation (2.28), followed by seed hardness (2.18), calcium content (2.17), uniformity product (2.12), oil content (1.80), swell ratio (1.61), stone seed ratio (1.48), 100-seed weight (1.02), and hilum color (0.78). Similarly, both Asian and U.S. small-seeded genotypes showed consistently high diversity indices for protein content, calcium content, seed hardness and seed uniformity. These result indicated that there are adequate genetic variation for these seed quality traits and that these genotypes evaluated can serve as germplasm pool for cross-breeding. Despite of the fact that seed coat color affects food-type soybean grading as reported previously (CGC 2004), all entries tested in this study had yellow seed coat and met the requirement of the



soyfood market. In addition, approximately 45% of the small-seeded lines tested in this study had yellow hilum, whereas 39% were buff, meeting the requirement by the soyfood market for a light hilum color. Seed uniformity product, seed hardness, and calcium content were highly diverse in small-seeded soybean category. These traits have not been emphasized in small-seeded soybean breeding programs, perhaps because the testing methods are time-consuming and labor-intensive. Therefore, improving seed uniformity product, seed texture, and calcium content could potentially increase the market value for small seeded soybeans. It’s worth noting that the Asian small-seeded lines exhibited high diversity indices than the U.S. lines for seed hardness, calcium content, and stone seed rate. In addition, the average genetic diversity of U.S. small-seeded soybeans (1.48) was lower than that of Asian small-seeded soybeans (1.57), suggesting narrower genetic base in the U.S. lines. Therefore, it is possible to enhance genetic diversity in the U.S. germplasm when using PIs from Asia.


For the large-seeded genotypes, the greatest diversity existed in seed uniformity followed by protein content, seed hardness, calcium content, oil content, swell ratio, 100-seed weight, stone seed ratio, and hilum color. As with small-seeded soybeans, 67% large-seeded genotypes had yellow hilum and 12% had buff hilum. The U.S. large seeded soybeans (1.61) had a higher average genetic diversity

index than Asian soybeans (1.32), particularly in protein, oil, calcium, hardness, swell ratio, and stone seed ratio. Seed uniformity, hardness, protein, and calcium content appeared to be relatively high in diversity index for both U.S. and Asian large-seeded lines. The average genetic diversity index in this study was much lower than what was reported previously probably due to the different genotypes

used (Dong et al. 2004). In this study, we used only food-grade soybeans and evaluated only traits that are related to soyfood quality.


Small-seeded soybeans were grouped into four clusters (A to D) with three outliers (Fig. 1). The two outliers, S02-9031 and PI 407788A, had the lowest similarity coefficient (0.71) and the other outlier PI 407805B had a low average similarity (0.82) because it had very low oil (16.9%), high protein content (45.9%) and high stone seed ratio (24.2%) with gray hilum. Cluster A was composed of most genotypes with seed size less than 9.0 g/100 seeds except for Nornatto with 12.1 g/100 seeds. Cluster A was further divided into three sub-clusters: A-1, A-2 and A-3. Sub-clusters A-1 and A-2 consisted of all genotypes from U.S. with A-2 having all the genotypes

from Iowa, whereas A-3 consisted of genotypes from Asia. The U.S. genotypes in A-1 and A-2 had lower protein content (38.3–41.9%) than Asian genotypes in sub-cluster A-3 (43.6–46.0%). The genotypes in cluster A-2 had higher stone seed ratio and lower swell ratio than those in sub-cluster A-1.


Cluster B contained 20 genotypes, most of which had a seed size of 9.2–10.2 g/100 seeds and a low stone seed ratio (less than 10%). Cluster B was further divided into two sub-clusters B-1 and B-2 and one outlier, PI 398368 with a gray hilum color. Most accessions in sub-cluster B-1 were PIs from South Korea, while most accessions in sub-cluster B-2 were from the U.S. Sub-cluster B-1 accessions contained higher protein content (over 40%) than sub-cluster B-2 (less than 40%) except for PI 424413 (47.1%). Cluster C contained seven genotypes including both



PIs from South Korea, Japan, and U.S. accessions, all of which hydrated 2.05–2.19 times of their dry seed weight in water after 16 h soaking. Stone seed ratio was also low in cluster C with an average of 8.5% (ranging from 7.2 to 10.1%). Cluster D was composed

of three PIs with an average swell ratio of only 1.48 and high protein content (44.6%, ranging from 43.7 to 46.0%).


Large-seeded soybeans were grouped into six clusters (I–VI) with five outliers (Fig. 2). V98-2711 was an outlier due to its lowest protein content (39.0%) and PI 408052C was an outlier due to its low stone seed number. Cluster I was the largest cluster consisting of 31 accessions. The seed of cluster I genotypes were harder in texture and lower in stone seed ratio in comparison with the other five clusters. Cluster I was further divided into five sub-clusters I-1 to I-5. Sub-cluster I-1 was composed of seven U.S. accessions: one from Iowa, one from Virginia, and five from Ohio with average seed-size uniformity product of 7,574,807.6 and fairly uniform swell ratio

from 2.28 to 2.32. Sub-cluster I-2 was mainly composed of South Korean PIs containing higher protein content, higher swell ratio, and lower stone seed ratio than genotypes in other sub-clusters. Both sub-clusters I-3 and I-4 included accessions from Iowa, Ohio, and PIs from South Korea and Japan. Sub-cluster I-3 accessions had similar seed size (23.3–23.9 g/100 seeds), while sub-cluster I-4 had a similar seed size uniformity of class 12 except for PI 398475. Sub-cluster I-5 consisted of Asian genotypes that had a seed size from 23.1 to 25 g/100 seeds and a low stone seed ratio less than 5%. There were three to five accessions in each of clusters II to VI. Cluster II, III, and IV had seed hardness from 351 to 380, less than 290, and from 321 to 350 N, respectively. Only cluster VI had all four accessions from the U.S. with 25 g/100 seeds. Cluster V had similar stone seed ratio and 100-seed weight.


The results from this research suggested that U.S. small-seeded soybeans were desirable for natto production because of their softer texture with higher water absorption capacity and lower stone seed ratio. However, Asian large-seeded soybeans had a lower


stone seed ratio and a higher water absorption capacity. Therefore, using Asian large-seeded genotypes may potentially improve seed quality for tofu and soymilk. A weakness in large-seeded U.S. soybeans is the low protein content that is especially important for firm tofu yield (Griffis and Wieder mann 1990). Therefore, the Asian soybean gene pool may serve as valuable genetic source for increasing protein content of U.S. food-grade soybeans. Because seed of both small- and large-seeded types in this study were low in size uniformity, it is important to develop uniform seed size for food-grade soybean market, which may potentially improve the crop value. Genetic diversity among and within U.S. and Asia small-seeded and large-seeded soybeans identified in this research will help breeders with selection of parents in food-grade soybean breeding. Research is underway to characterize the genetic diversity in

food-grade soybean germplasm at the molecular level using SSR markers.