Introduction

Saponins are secondary metabolites that are abundant in the soybean [Glycine max (L.) Merrill], where they amount to 0.6–6.5% of dry weight (Shiraiwa et al. 1991b). Soybean saponins have a lot of health functional activities (Tsukamoto and Yoshiki 2006; Fenwick et al. 1991; Murata et al. 2006; Kuzuhara et al. 2000; Rowlands et al. 2002; Ellington et al. 2006; Kang et al. 2005; Lee et al. 2005; Ishii and Tanizawa 2006; Yang et al. 2015). Recent studies have reported that group-A saponins can prevent memory impairment (Hong et al. 2014). However, group-A saponins have a bitter and astringent taste owing to the acetylation in hydroxyl groups of the terminal sugar of the sugar chain attached at the C-22 position (Okubo et al. 1992), and there is a demand for group-A saponins to be genetically decreased or deleted from seeds in soybean breeding for soy milk (Kato et al. 2007).

Saponin composition has been investigated in wild soybean (Glycine soja Sieb. & Zucc.) and cultivated soybeans (Kudou et al. 1992, 1993; Shiraiwa et al. 1991a, c; Tsukamoto et al. 1993; Krishnamurthy et al. 2014b), and soybean saponin composition are classified into several groups, including group A, DDMP (and their decomposition), and group α saponins (formerly Sg-6 saponin, here designated as the α saponins). Disintegrated DDMP saponins would create group B or E (Kudou et al. 1993; Tsukamoto et al. 2009). The group α saponins are a newly identified group in wild soybean accessions (Honda et al. 2009; Takahashi et al. 2016c) and are reported to be genetically controlled by the Sg-6 locus (Krishnamurthy et al. 2014b), but group α saponins have not yet been found in cultivated soybeans. The group α has been found to have at least four components H, I, J, and K, with only a single sugar chain at the C-3 position of their aglycones (Takahashi et al. 2016b). Recently, a new group β saponin (KA-αg component) was found in a semi-wild soybean accession lacking group A saponins in China (Takahashi et al. 2016a, 2017).

Thus far, six loci Sg-1 (Shiraiwa et al. 1990; Tsukamoto et al. 1993), Sg-3, Sg-4 (Tsukamoto et al. 1993; Takada et al. 2012), Sg-5 (Tsukamoto et al. 1998; Takada et al. 2013), Sg-6 (Krishnamurthy et al. 2014b), and Sg-7 (Takahashi et al. 2016a) have been found to be involved in saponin biosynthesis in soybeans, and thirteen alleles (Sg-1 a/Sg-1 b/sg-1 0, Sg-3/sg-3, Sg-4/sg-4, Sg-5/sg-5, Sg-6/sg-6, Sg-7/sg-7) have been identified. Saponin composition phenotypes for soybean varieties or accessions can be described by the combination of saponin components, which depends on the combination of dominant and recessive alleles. Tsukamoto et al. (1993) analyzed the saponin composition of 800 Japanese soybean varieties and 329 wild soybeans and identified eight phenotypes (Aa, Ab, AcAf, AaAb, AaBc, AbBc, AaAbBc and A0Bc). Based on the components of group A and α saponins and the absence of Aa or Ab of group A, or loss of the whole group A, eleven types of saponin phenotypes (Aa, Ab, AaBc, AbBc, Aa + α, AaBc + α, AbBc + α, AuAeBc, A0-S, A0Bc-S, and A0Bc) were distinguished in Korean wild soybeans (Krishnamurthy et al. 2014b).

China is thought to be the area of origin of cultivated soybeans, and investigation of saponin polymorphisms in Chinese cultivated soybeans has not been carried out yet. The group α saponins have been newly identified in wild soybeans. Little is known about the functional activities of α saponins at present so the public are suspicious of their safety as α saponins do not exist in cultivated soybeans. If cultivated soybeans contain the group α saponins, there is no need to worry at all about this group of saponins. Our objective in this work was to clarify the general situation of the polymorphisms of saponins in Chinese soybeans and their geographical distribution, and to examine whether α saponins exist in cultivated soybeans in the area of origin, China. We also would like to probe into the area of the origin of soybeans considering taking another otherwise indicator of chemical composition for exploration of this issue such as saponins.

Materials and methods

Plant materials and chemicals

A total of 2959 cultivated soybean accessions, stored in the national crop gene bank at the Chinese Academy of Agricultural Sciences (CAAS), were used in this study. These accessions all were landraces, randomly taken from all the administrative regions of China (Table 1).

Table 1 Saponin phenotypes and their geographical distribution in Chinese cultivated soybeans
Full size table

To assess the distribution among geographical regions for saponin composition and phenotypes, the accessions were zoned into eight geographical regions (Northeast, Northwest, North, Centre, Eastern North, Southeast, Southwest, South) according their provenances (Table 1; Fig. 3). For insight into the frequency differences of alleles relevant to saponin phenotypes in different character types in morphology, agronomy and ecology, we partitioned the accessions into thirteen character groups including seven qualitative characters [seed-coat colours (yellow, green, brown, bicolour and black), cotyledon colours (yellow and green), stem erectness (erect, weak-erect, semi-prostrate and prostrate), pubescence colours (brown and grey), flower colours (purple and white), leaf shapes (rotundifolius, ovoid, elliptic, and lanceolate), and growth habits (definite, semi-definite and indefinite)], four quantitative characters [100-seed weight (g), plant height (cm), protein and oil contents (%)] and an ecological character (three sowing-ecotypes: spring-sowing, summer-sowing, and autumn-sowing) (Table 2).

Table 2 Frequency distribution of saponin phenotypes for botanically qualitative and quantitative character types in Chinese cultivated soybeans
Full size table

The quantitative characters were divided into different classes. The 100-seed weight was grouped into four classes (≤10.0, 10.01–20.0, 20.01–30.0, >30.0 g), plant height into five classes (≤50.0, 50.01–100.0, 100.01–150.0, 150.01–200.0, >200.0 cm), protein content into four classes (≤40.0, 40.01–45.0, 45.01–50.0, >50.0%), and oil content into five classes (≤16.0, 16.01–18.0, 18.01–20.0, 20.01–22.0, >22.0%). All morphological and quantitative characters data were adopted from the soybean collection records preserved in the national crop gene bank, CAAS.

Japanese variety “Suzuyutaka” (saponin type: Ab) was used as the standard of saponin composition in thin layer chromatography (TLC). All chemicals (chloroform, methanol, acetonitrile, and sulfuric acid) were purchased from Beijing Chemical Works (Beijing, China).

Extraction of saponins

A mature dry seed from each accession was divided into hypocotyls, cotyledons and seed coats with a utility knife. The saponin components were extracted from the seed hypocotyl with 80% (v/v) methanol (MeOH) at room temperature for 12 h.

TLC and LC–MS analysis

The identification of saponin composition was carried out with TLC analysis according to our previous method (Takahashi et al. 2016a). Silica gel (SiO2) plates were applied (catalog no. 105626, Merck Millipore, Germany). For further identification of variant saponin composition, LC-PDA/MS analysis was performed using a hybrid ion-trap mass spectrometer (LTQ Orbitrap XL, Thermo Fisher Scientific Incorporated, Waltham, MA, USA) under an UFLC system (Prominence UFLC system, Shimadzu Corporation, Kyoto, Japan) with a photodiode array (PDA) detector and a C30 reverse-phase column (Develosil C30 UG-3, 2.0 × 150 mm, 3 μm; Nomura Chemical, Okayama, Japan). The saponin extracts were diluted ten times with 80% (v/v) methanol. The solvent for elution were selected for acetonitrile including 0.1% formic acid. The initial solvent concentration was adjusted to 20% (v/v) and linearly increased to 80% (v/v) in 60 min. An automatic full-scan mode over a mass of charge ratio (m/z) from 300 to 2000 was used to acquire MS data. The data of UV chromatograms and MS spectra were analyzed with Xcalibur software version 3.1 (Thermo Fisher Scientific Incorporated).

Results

Saponin composition and phenotypes in Chinese soybeans

Saponin analysis of 2959 Chinese soybean accessions distinguished 13 components in five separate saponin groups (group A: Aa, Ab, Ae, Ax, Ad, and Af; group DDMP: αg and βg; group B: Ba; group α: H-αg, I-αg and K-αg; group β: KA-αg) (Fig. 1).

Fig. 1
figure 1

Polymorphism of saponin composition detected in seed hypocotyls of Chinese soybeans by TLC analysis. There were five groups of saponin components: group-A (Aa, Ab, Ae, Ax, Af, and Ad), DDMP (αg and βg), group-α (H-αg, I-αg, and K-αg) and β (KA-αg), group B (Ba) and nine phenotypes of saponin composition: Aa, Ab, AaBc, AbBc, AaβKA, AbαH, AaαI, AbαI, and AbαK. The group-α saponins were characterized by blue colour

Full size image

Groups α and β were first found in cultivated soybeans, and the latter appeared in only one accession (sample ZDD13997 from the eastern south Zhejiang, an autumn-sowing type) and group α in 12 accessions, with separate components H-αg, I-αg and K-αg (Fig. 1). LC–MS analysis confirmed the existence of the new group-β and group-α saponins, which emerged at 22.5 min for KA-αg, 24.5 min for H-αg, 25.5 min for I-αg, and 35.9 min for K-αg (Fig. 2). These accessions with group α saponins comprised seven accessions from the northwest (ZDD10529, ZDD10728, ZDD10929 and ZDD11158 for AbαH, ZDD10554 for AbαI, ZDD11144 for AbαK, and ZDD11182 for AaαI), three accessions from the southwest (ZDD17640 and ZDD15516 for AbαH, and ZDD13221 for AbαI), and two accessions from the eastern north areas (ZDD04515 and ZDD04584 for AbαI).

Fig. 2
figure 2

HPLC chromatograms of saponin extracts from soybean hypocotyls under UV 205 nm by LC–MS analysis. The results distinctly revealed the occurrence of group-α saponin components H-αg, I-αg, and K-αg and KA-αg in soybeans (c–f). a, b were the standard saponin components of Aa and Ab types

Full size image

Saponin composition includes many components, thus far more than 100 components have been detected according to published documents (Shiraiwa et al. 1991a, c; Kudou et al. 1992, 1993; Tsukamoto et al. 1993; Krishnamurthy et al. 2014b; Takahashi et al. 2016b, c). Usually, the combination of saponin components is divided into phenotypic types according to the occurrence of saponin composition in soybean variety individuals or accessions (Tsukamoto et al. 1993). In dividing saponin composition types for Korean wild soybeans, Krishnamurthy et al. (2013) only denoted presence or absence for the group α saponins (for example, Aa + α, AaBc + α, AbBc + α). Here we adopted idiographic components of group α in addition to Krishnamurthy et al.’s phenotype grouping and identified four common types Aa, Ab, AaBc, AbBc and five new types AaβKA, AbαH, AaαI, AbαI, and AbαK (Fig. 1). The Ab + α type did not appear in Korean wild soybeans but occurred in Chinese soybeans.

Two common saponin phenotypes Aa (involved with Sg-1 a, Sg-3, sg-4, Sg-5, and sg-6 genes) and Ab (Sg-1 b, Sg-3, sg-4, Sg-5, and sg-6) were observed at a frequency of 39.2 and 60.8%, respectively, in the total accession samples. The AaBc type (Sg-1 a, Sg-3, Sg-4, Sg-5, sg-6) and AbBc type (Sg-1 b, Sg-3, Sg-4, Sg-5, sg-6), which rested with the Sg-4 dominant allele, each appeared in only one accession (ZDD17404, AaBc and ZDD17536, AbBc).

Frequencies and geographical distribution of saponin phenotypes in Chinese soybeans

The analysis showed that two types Aa and Ab were the main types of nine saponin phenotypes in Chinese cultivated soybeans, with frequencies of 39.2 and 60.3% for Aa and Ab, respectively (Table 1). Aa and Ab type are controlled by two codominant alleles of Sg-1 a and Sg-1 b, respectively, at the Sg-1 locus for Xyl and Glc at C-22 terminal sugar component (Sayama et al. 2012), and our current results demonstrated that there were regional differences in the distribution of allelic frequencies between Aa (Sg-1 a allele) and Ab (Sg-1 b allele) in Chinese soybeans. One region (Northwest) had almost the same frequencies of both Sg-1 a and Sg-1 b alleles (50.8 and 49.2%, respectively) and one region (North) showed a higher frequency (60.6%) for the Sg-1 a allele than that for Sg-1 b (39.4%); while the other six areas all showed higher and similar Sg-1 b frequencies (64.9–69.7%) than Sg-1 a allele (30.3–35.1%). By and large, the northern regions (the Yellow River valley) had a higher Sg-1 a frequency, and the Northeast and the Yangtze River valley and its southern regions had a higher Sg-1 b frequency in Chinese soybeans (Table 1; Fig. 3).

Fig. 3
figure 3

Percentage frequency distribution of the Sg-1 a allele in eight geographical regions, northeast, Northwest, North, Centre, Eastern North, Southeast, Southwest and South. The northwest and north China along Yellow River had the higher frequencies (>50%) of Sg-1 a. The ellipse denotes the area of the middle and lower reaches of the Wei River valley

Full size image

There were only two accessions from southwestern Yunnan Province that were found to carry dominant Sg-4 allele, with a very low 0.33 frequency in all the tested southwestern soybean accessions (Table 1). The sole group β saponin component KA-αg was found in one accession from Zhejiang Province in the Southeast region. Only twelve accessions have group α saponins, existing at a very low frequency of 0.41%, in Chinese cultivated soybeans, which were distributed in three regions, the Northwest, Eastern North, and Southwest. Particularly the Northwest, Gansu and Shaanxi provinces produced seven accessions and contributed a relatively higher proportion (58.33%), with more than half the number of all the group α saponin accessions (Table 1).

Allelic frequency at the Sg-1 locus among different botanically qualitative character types in Chinese soybeans

Seed-coat colour, cotyledon colour, stem erectness, pubescence colour, flower colour, leaf shape and growth habits are main botanical qualitative traits for the morphological description of soybean varieties. Some characters could be related to evolutionary degrees or ecological types of soybeans. Aa and Ab phenotypes were validated to two main saponin types in Chinese soybeans (Table 1). Here, we analyzed the allelic distribution of two main Aa and Ab phenotypes with these characters.

Cotyledon, pubescence and flower colours

The results showed that there were no obvious differences in the frequency of Sg-1 a or Sg-1 b between yellow (39.3% Sg-1 a) and green (41.8% Sg-1 a) types of cotyledon colours, among grey (40.2% Sg-1 a), brown (38.5% Sg-1 a) and no-pubescence (40.0% Sg-1 a) of pubescence colours, and between purple (39.3% Sg-1 a) and white (39.3% Sg-1 a) of flower colours. However, Sg-1 b frequency (58.2–60.7%) was higher than Sg-1 a (38.5–41.8%) in each type of the three characters (Table 2).

Seed-coat colour

Seed-coat colours presented differences in allelic frequencies among seed-coat types, i.e., black seed type showed the highest Sg-1 a frequency (58.9%), brown (38.5%), bicolour (37.5%) and yellow (37.2%) colour types had similar lower frequencies, and the green had the lowest Sg-1 a frequency (33.0%). In addition, only the black seed type had a higher Sg-1 a than Sg-1 b frequency, whereas all other types were contrariwise (Table 2).

Leaf shape

The soybean landrace accessions had four leaf shapes, and all showed higher frequencies (54.7–66.4%) for the Sg-1 b allele than for Sg-1 a (33.6–45.3%) (Table 2). There were two obvious groups in Sg-1 a frequency: rotundifolius and ovoid leaf types accumulated higher frequencies of Sg-1 a, with 45.3 and 44.0% frequencies, respectively. Elliptic and lanceolate leaf types had lower frequencies for the Sg-1 a allele, giving 33.6 and 35.0% frequencies, respectively.

Stem erectness and growth habit

There were four types of stem erectness in the accessions. Sg-1 a or Sg-1 b allelic frequency in the stem type groups depended upon the degrees of stem erectness (Table 2). Sg-1 a frequency varied from high to low with the degree of erectness: prostrate (73.6%) > semi-prostrate (61.0%) > weak-erect (46.4%) > erect (31.7%).

Stem growth habit of soybeans included three types: definite, semi-definite and indefinite growth. Similarly to stem erectness, growth habit also showed that the frequency of Sg-1 a or Sg-1 b varied with growth habit types (Table 2); the order of frequency fluctuation for Sg-1 a accumulation was: indefinite type (57.8%) > semi-indefinite (34.7%) > definite (31.2%). The indefinite type also accumulated a higher frequency of the Sg-1 a allele than of Sg-1 b.

Ecotype

The analyzed accessions belonged to three sowing-ecotypes: spring, summer, and autumn. The frequencies of Sg-1 a among the three types reduced gradually in the order spring (42.7%), summer (35.7%), and autumn (33.1%) (Table 2). Sg-1 b frequency was higher than Sg-1 a in each sowing-type.

Allelic frequency at the Sg-1 locus among different types of agronomically quantitative characters in Chinese soybeans

We investigated the degree of accumulation of the Sg-1 a or Sg-1 b allele in four important agronomically quantitative characters: 100-seed weight (g), plant height (cm), protein and oil contents (%). Some of these characters clearly exhibited differences in the frequency of Sg-1 accumulation (Table 2).

100-seed weight

In four different groups of 100-seed weight, the group ≤10.0 g had an obviously higher frequency (59.0%) of Sg-1 a than the other three groups (38.2% for 10.01–20.0 g, 26.8% for 20.01–30.0 g, and 37.0% for >30.0 g) (Table 2). Furthermore, Sg-1 a frequency was higher than Sg-1 b in the group with the smallest seeds.

Plant height

Plant height was divided into five classes from ≤50.0 to >200.0 cm. The results displayed that Sg-1 a or Sg-1 b frequency varied with plant height (Table 2). The degression of Sg-1 a frequency occurred from the tallest plant type (>200.0 cm) to the shortest one (≤50.0 cm), i.e., frequency gradually was reduced in turn from 75.0, 60.3, 56.4, 36.2 to 28.2% for the tallest plant type (>200.0 cm), high plant height (150.01–200.0 cm), medium height (100.01–150.0 cm), low plant height (50.01–100.0 cm) and the shortest plant (≤50.0 cm), respectively.

Protein and oil content

Protein and oil content are two important characters in soybeans, and are related to the degree of evolution. The highest and lowest content of protein in soybeans are the consequence of artificial selection. We observed that soybeans with a protein content of 40.01–45.0% had a higher Sg-1 a frequency (43.2%), followed by 45.01–50.0 (34.9%) protein content. The classes of soybeans with the highest and lowest content of protein had the lowest frequencies of the Sg-1 a allele, 14.3 and 29.7%, respectively. However, all the soybean protein content classes contained higher frequencies of the Sg-1 b allele than the Sg-1 a allele (Table 2).

In oil content types, the degression of Sg-1 a frequency occurred from the lowest content type (≤16.0%) to the highest content type (>22.0%), i.e., frequency gradually was reduced in turn from 44.5, 41.0, 34.9, 32.3 to 13.2% in five content class types, the lowest content (≤16.0%), low content (16.01–18.0%), medium content (18.01–20.0%), high content (20.01–22.0%) and the highest content (>22.0%), respectively.

Discussion

Saponin composition and the significance of finding group α saponins in Chinese soybeans

Soybean group A saponins Aa, Ab, Ae, Ax, Af, Ad, DDMP βg, αg and group B Ba components are common main components (Kudou et al. 1992, 1993; Shiraiwa et al. 1990, 1991a, c; Tsukamoto et al. 1993). Honda et al. (2009) identified a new group of α saponins (i.e., Sg-6 saponins) in Japanese wild soybean accessions. Subsequently the group α was confirmed to exist in Korean and Chinese wild soybeans (Krishnamurthy et al. 2014a; Takahashi et al. 2016c). Takahashi et al. (2016c) reported further research of group α saponins, where four kinds of single components H-αg, I-αg and J-αg, and K-αg were recognized definitely in Chinese wild soybeans. However, thus far, no group α saponins have been found in cultivated soybeans. In our present investigation, a total of 13 components were detected in Chinese soybeans, including two new groups (α and β) in addition to common group A (Aa, Ab, Ae, Ax, Af, Ad), DDMP (βg, αg) and group B (Ba) components. Three components of group α saponins, H-αg, I-αg and K-αg, were first identified in cultivated soybeans. Another component KA-αg was also detected, which was a new recruit to saponins and belonged to group β. This KA-αg saponin component has recently been found in a Chinese semi-wild soybean (Takahashi et al. 2016a).

Tsukamoto et al. (1993) showed six phenotypes (Aa, Ab, AcAf, AaAb, AaBc, AbBc) of saponin composition in Japanese soybeans. In view of finding the existence of new saponins (groups α and β) in Chinese soybeans, a total of nine phenotypes of saponins were identified, four of which were common phenotypes of Aa, Ab, AaBc, AbBc, and five were new types of AaβKA, AbαH, AaαI, AbαI, and AbαK.

Up to the present, group α saponins are known to exist only in wild soybeans (Honda et al. 2009; Krishnamurthy et al. 2014a). The finding that group α saponins occurred in soybeans was of a far-reaching significance; the fact that group α saponins existed in cultivated soybeans testifies to the safety of this group of α saponins to human health because these saponins have been consumed for several 1000 years in China.

Alleles for soybean saponins in east Asia

Eight loci are proposed to be involved in saponin biosynthesis, GmSGT2, GmSGT3 (Shibuya et al. 2010), Sg-1 (Shiraiwa et al.1990; Tsukamoto et al. 1993), Sg-3, Sg-4 (Tsukamoto et al. 1993; Takada et al. 2012), Sg-5 (Tsukamoto et al. 1998; Takada et al. 2013), Sg-6 (Krishnamurthy et al. 2014b), and Sg-7 (Takahashi et al. 2016a). Recessive alleles at loci Sg-3 and Sg-5 were found in Japanese and Korean wild soybeans (Tsukamoto et al. 1993, 1998; Krishnamurthy et al. 2013); a recessive sg-3 would lead to the absence of Aa, or Ab and Ax or Ad components (Tsukamoto et al. 1993), and recessive sg-5, to none of the whole group A (Tsukamoto et al. 1998). The sg-3 allele only appeared in one Japanese soybean (Tsukamoto et al. 1993) and in very few Korean wild soybeans, and recessive sg-5 allele occurred in very few Japanese and Korean wild soybeans (Tsukamoto et al. 1998; Krishnamurthy et al. 2013). The Sg-4 allele is rare in soybeans but abundant in wild soybeans; about 2% of Japanese soybeans, 67.5% of Japanese wild soybeans, and about 61.4% of Korean wild soybeans possess the Sg-4 allele (Tsukamoto et al. 1993; Krishnamurthy et al. 2014a). There were a small number of Sg-4 allele in Chinese soybeans and it existed with a very low frequency (0.067%), only two of 2959 accessions. The recessive sg-7 allele was found only in one semi-wild type wild soybean (Takahashi et al. 2016a).

The locus Sg-1 is responsible for the formation of group A saponins (Sayama et al. 2012) and there are three alleles Sg-1 a, Sg-1 b and sg-1 0 (recessive allele for no terminal sugar at C-3 position) in this locus. The Sg-1 a allelic frequency was higher in Korean and Japanese wild soybeans, with 98.4 and 85.4% frequencies, respectively; whereas cultivated soybeans had a lower Sg-1 a frequency of about 26.6% (Tsukamoto et al. 1993; Krishnamurthy et al. 2014a). The recessive sg-1 0 allele also rarely existed with a frequency of 0.3% in both Korean and Japanese wild soybeans (Tsukamoto et al. 1993; Krishnamurthy et al. 2014a). We detected two co-dominant alleles Sg-1 a and Sg-1 b in Chinese soybeans, and the Sg-1 a allele showed a frequency of 39.3%, which was slightly higher than the frequency of 26.6% in Japanese soybeans (Tsukamoto et al. 1993) but was much lower than the frequency of Sg-1 a in Glycine soja (98.4% in South Korea and 85.4% in Japan). We do not know the reasons for the accumulation of the Sg-1 b allele in cultivated soybean during the domestication from wild soybeans. A possible explanation might be that the Sg-1 b allele could be of some benefit to cultivation adaptation or human preference, and/or to some physiological function, stress resistance, or other characters of soybeans by direct or indirect artificial selection during the domestication of soybeans.

The group α saponins were genetically thought to be controlled by a single Sg-6 locus (Krishnamurthy et al. 2014b), and approximately 10% of Korean wild soybeans carried the Sg-6 allele and 90% were recessive (sg-6), i.e., without group α (Krishnamurthy et al. 2013). However, our investigation detected three different components (H-αg, I-αg, and K-αg) of group α saponins in twelve Chinese soybeans, which suggests that there could be different genes or loci controlling these respective components of group α saponins. The accumulation of no α saponins in G. max agreed with the domestication of soybeans because group α saponins also exist at a relatively lower level in G. soja (Krishnamurthy et al. 2014a).

Unexpectedly, a new group, β saponin (KA-αg component), was first found in a Chinese cultivated soybean accession. This KA-αg component was only recently found in a semi-wild soybean lacking group A saponins in China (Takahashi et al. 2016a), which was proposed to be derived from the A-αg precursor of group A by acetylation at the C-29 position. The present discovery that a KA-αg component occurred in a normal Aa type accession (ZDD13997) suggested that this ZDD13997 accession possessed the acetylating enzyme and a small part of the A-αg precursor molecules become KA-αg by acetylation at the C-29 position, and that the acetylation or acetyl of KA-αg could prevent KA-αg from attaching arabinose (Ara) to the OH group at the C-22 position. As a result, both Aa and KA-αg components were detected in soybean ZDD13997.

Accumulation of Sg-1 a allele, geographical distribution and origin of soybean

The Sg-1 a allele accumulates at high frequency in the wild soybean species (Tsukamoto et al. 1993; Krishnamurthy et al. 2014a); this peculiarity was also characterized by the evolutionarily degree-low or some particular character types in the soybeans analyzed here (Table 2). Smaller seeds (≤10.0 g/100-seed wt.), the black seed-coat, prostrate stem, indefinite growth habit, and spring-sowing type are thought to be primordial characters of soybeans, and accordingly these characters exhibited the highest accumulation level of the Sg-1 a allele. The summer and autumn-sowing types are derived from spring-sowing types, therefore, they should have relatively lower frequencies of the Sg-1 a allele. There was no frequency differentiation in the Sg-1 a locus between types of cotyledon, pubescence and flower colours.

Stem erectness and growth habit characters represented a decreased inclination of Sg-1 a allelic frequency with the degrees of character intensity, prostrate > semi-prostrate > weak-erect > erect in stem erectness and indefinite > semi-indefinite > definite in growth habit. These phenomena could be understood in consideration of the frequency decrease of Sg-1 a allele incidental to the evolution of soybeans from the wild soybean. Three quantitative characters (plant height, protein and oil contents) appeared to have some correlations with the Sg-1 locus, showing gradually decreased frequencies of the Sg-1 a allele from the biggest to the smallest classes (Table 2); we could not definitely explain the reasons for this; however, from a domestication perspective, the correlations between the Sg-1 locus and these quantitative characters accorded with the evolution of soybeans because the progenitor wild species G. soja always has higher plant height and lower oil content. The wild species has a higher average protein content of 40–45% than soybeans, and this protein content could belong to a primitive trait of soybeans, then the conscious or unconscious artificial selection towards a high or low content of protein in soybean domestication and breeding would bring about the decrease of Sg-1 a allelic frequency from the primordial to evolutionarily acquired class types according to the logic observed here that the frequency of the Sg-1 locus varied regularly with the degree of evolution (Table 2).

We observed correlations between the Sg-1 locus and the domestication or evolution of soybeans within the cultivated species, therefore, the Sg-1 locus may be used as an indicator or a marker to probe the area of origin for soybean. With regard to the area of origin of soybean in China, the clear area is unknown and there are long-standing disputes around several hypotheses, such as origination in the northeast (Fuguda 1933; Li 1994), north (Hymowitz and Newell 1981; Xu et al. 1986; Chang 1989; Li et al. 2013), south (Gai et al. 2000; Xu and Gai 2003; Zhao et al. 2004; Wen et al. 2009; Guo et al. 2010), and multiple centres in China (Lu 1978; Zhou 1998; Xu et al. 2002).

Around the area of origin of soybeans, the aforementioned mainly focus on whether the area of origin is in north (the Yellow River valley including the Huang-Huai plain) or in south China (the Yangtze River and its south). However, archaeological remains of wild and soybean seeds unearthed from over 30 sites of the Neolithic cultures in China in context dating from 7000 bc to the Longshan period (2350–1900 bc) are all located in the Yellow River valley, i.e., northern regions (Lee et al. 2007; Wu et al. 2013). Our present results obtained here exhibited higher frequencies of Sg-1 a allele in north and northwest China along the Yellow River valley. A recent report indicated a geographical area of the highest genetic diversity for Chinese soybeans on the basis of an SSR assessment, which lies 30°–35°N × 105°–110°E, including the valley of the middle and lower reaches of the Wei River and the valley of the upper reaches of the Hanjiang River (Wang et al. 2016). This area belongs to the southeastern Gansu Province and southern Shaanxi, and the middle-upper reaches of the Yellow River valley. In this investigation, group α saponins were first found in cultivated soybeans, which also implies that the group α saponin genes introgressed into the cultivars when the early domestication of soybeans started from wild soybeans. As a result of integrating the documents published for the archaeological remains of wild and cultivated soybean, the higher frequency distribution of the Sg-1 a allele observed here along the Yellow River valley and the appearance of group α soybean germplasm with a relatively higher frequency (seven of 13 group α landrace accessions) in northwestern Gansu and Shaanxi provinces, we more specifically propose that the area in the middle and lower reaches of the Wei River valley (Shaanxi and Gansu provinces), i.e., the middle-upper reaches of the Yellow River valley, is the possible area of origin of soybeans in China.