Introduction

The progenitor of cultivated soybeans [Glycine max (L.) Merrill] (Fukuda 1933; Hymowitz 1970), wild soybean (Glycine soja Sieb. & Zucc.) as an genetic resource for soybean breeding, is widely distributed in eastern Asia. This species is an herbaceous, annual and self-pollenating plant with a twining or procumbent stems, and commonly grows in fields such as watersides, roadsides, sparse woodlands, brushes, tussocks, abandoned farmlands, ditches, sunny hillsides, and around villages. Over the past 30 years a great number of wild soybean populations are unfortunately extinct or have been significantly reduced in China. Now this species has been placed in the protected plant catalogue because of its great significance to breeding and its serious loss of distribution habitats in China.

Many studies of genetic diversity related to the conservation of wild soybean have been conducted. Choi et al. (1999) reported a geographic pattern of SSR variation of wild soybean along five main Korean rivers. Tozuka et al. (1998) demonstrated the clinal distribution of mitochondrial DNA-RFLP in Japanese wild soybean. Wang and Takahata (2007) compared the geographic distribution of SSR variation between China and Japan. Chinese wild soybean has been studied for geographic diversity in morphology (Dong et al. 2001) and for geographical variation in chloroplast and mitochondrial DNA (Shimamoto et al. 1998; Xu et al. 2002). Isozyme analysis was carried out for a few natural populations in Korean, Japan (Kiang et al. 1992; Fujita et al. 1997) and China (Li et al. 1995).

The conservation of genetic diversity in wild soybean is actualized usually through both in situ and ex situ approaches. Effective conservation of the genetic diversity of natural populations is essential, which is involved in the cost, worthiness and validity of the conservation. Jin et al. (2003a) has suggested a sampling strategy for a single natural population of wild soybean through an ISSR analysis in a natural population case. However, another important sampling strategy is worthy of consideration, that is that how all the subpopulations in a region should be treated in in situ or ex situ conservation. It is important to understand the degree of genetic diversity, and its distribution patterns, and subpopulation genetic structures under various habitats or ecogeographical environments. In China, facing the fact that natural populations of wild soybean have been reducing in number and area, how to effectively protect wild soybean resource has become pressing. Our objective of this work were to assess the subpopulation genetic diversity in different habitats all over a region and understand how the subpopulations in a geographical or ecological region should be dealt with when conserving high genetic diversity with respect to sampling strategy for situ or ex situ conservation within a region.

Wild soybean is sensitive to environmental selection (Wang and Li 2012), and different ecogeographical locations and habitats would result in population genetic differentiation and emergence or elimination of specific alleles among populations in vast areas. To avoid overlarge ecogeographical effect because of vast regional areas, we selected an area-suitable ecotope, Beijing region as the studying simulation region.

Materials and methods

Sampling

We collected geographical site subpopulations of wild soybean in Beijing region (Table 1; Fig. 1). The sampling sites were selected considering representative geographic limits, most common habitats and practical distribution status. These sites were about 13–100 km apart (beeline distance) and separated in four areas: north mountain area (sites 1, 2 and 3 were distributed along the same river valley), east hill area (sites 4 and 5), central plain area (sites 6, 7, 8 and 9) and western mountain area (site 10, located in the eastern edge of this area, near the central plain area). Site 1 grew on a semiarid sandy soil close to a lake. Site 2 was a newly generated founder population of a small area with homogeneous morpha on a waterside. Sites 3, 5 and 8 had fertile habitat conditions. Site 4 was situated on an arid sandy soil, presenting dwarfish forms of plants with small leaves. Sites 6 and 7 were along an arroyo of a river. Site 9 grew in a semi-moist riverbank. Site 10 lay to roadsides on the foot of a mountain. Seeds of total 294 individual plants were gathered from 28 to 30 individuals at least 5–15 m apart from each other in each site based on Sjögren and Wyöni (1994)’s suggestion on plant sampling and Jin et al. (2003a)’s sampling strategy within a wild soybean natural population for conservation of genetic diversity. Twenty-eight Beijing cultivated soybeans were used for identification of the origin of the big seed individuals that appeared in a subpopulation.

Table 1 Location and habitats of the sampled site subpopulations of wild soybean in Beijing
Full size table

Survey of subpopulation phenotypic characters

To identify morphological characters of the sampled plants, the seed samples from 294 individual plants were grown as 294 individual families by ten seeds each family in a condition-uniform experimental field and measured for main morphological traits. Each family line was harvested (6–9 plants survived in families) and measured for the mean values of quantitative traits (plant height, aboveground dry matter weight, yield per plant and 100-seed weight), and the population mean values were calculated by the arithmetic average of all family lines in subpopulations (Table 3).

SSR marker analysis

Total DNA was extracted from 20-d old leaves of the 294 family seedlings. Thirty-six public SSRs over 20 genetic linkage groups (Cregan et al. 1999) were applied to Polymerase chain reaction (PCR) amplification for these samples (Table 2). These markers were selected for their stability, as described in two reports on SSR assessment of genetic diversity in Chinese soybeans (Xie et al. 2003; Wang et al. 2003). PCR amplification was carried out in a volume of 20 μl reaction mixture containing 100 ng of template DNA, 10 pmol of each primer, 100 μM of each of the four dNTPs, 1 U Taq polymerase, and 2 μL of 10 × PCR buffer. Cycling consisted of 30 s denaturation at 94 °C, 30 s annealing at 47 °C, 30 s extension at 72 °C for 30 cycles. The amplified products were run on a 6 % gel (SDS-PAGE) and the bands were developed with silver staining.

Table 2 Estimate of number of alleles (Na), genetic differentiation for SSR loci in the subpopulations of wild soybean in Beijing
Full size table

Data analysis

Each pair of primers as a locus and each variant (bands) as an allele were recorded. POPGENE1.32 (Yeh et al. 2000) was used to estimate the observed total number of alleles (Na), Shannon–Weaver index (I, Shannon and Weaver 1949), and Nei’s gene diversity statistics (H S : intra-subpopulation genetic diversity; D ST : inter-subpopulation genetic diversity; G ST : amount of inter-subpopulation gene differentiation, Nei 1973) (Table 2). A clustering analysis for site subpopulations was performed in software PowerMarker version 3.25 (Liu and Muse 2005) using UPGMA with the software NEIGHBOR of PHYLIP version 3.5c (Felsenstein 1993) based on Nei’s genetic distance (Nei and Takezaki 1983). An AMOVA analysis was used to assess within-population genetic variation and population genetic parameter estimates (Table 3) using PowerMarker 3.25 (Liu and Muse 2005). Arlequin 3.1 (Excoffier et al. 2005) was used to calculate overall variation of subpopulations. A principal component analysis (PCA) based SSR variation was used to identify genetic differentiation among the subpopulations using NTSYScp version 2.1 (Rohlf 2000). A significance test was made for the population mean values of four quantitative traits (100-seed weight, aboveground dry matter per plant, yield per plant, and plant height) to determine the subpopulation morphological differentiation. Shannon–Weaver index was also used to estimate the population phenotypic diversity. For this estimate, the morphological quantitative traits (mean measured values of field-cultivated plants for each single-sampled individual) were each divided into several graded-samples and their frequencies were calculated. Plant height (m): seven grades of every 0.50 m interval from ≤1.00 to >3.50 m; aboveground dry matter weight (g): ten grades of every 10.00 g interval from ≤10.00 to >90.00 g; yield per plant: ten grades of every 5.00 g interval from ≤5.00 to >45.00 g and 100-seed weight: six grades of every 0.50 g interval from ≤1.00 to >3.00 g. A PCA of morphology was used to identify the natural big-seeded plants (G. gracilis) and the typical wild plants in a mixed subpopulation using the software NTSYScp version 2.1. Six morphological characters [100-seed weight (g), plant dry weight (g), plant height (cm), single-plant yield (g), leaf length (mm), and leaf width (mm)] were used for the PCA.

Table 3 Genetic parameter estimates and phenotypic performances of the subpopulations
Full size table

Results

SSR-polymorphic variation

In total, 454 alleles were detected at the 36 loci, ranging from 2 (satt197 and satt345) to 19 alleles (satt530) in the samples, with an average of 12.61 alleles for loci, 45.40 for sites and 1.26 for each locus per site, respectively (Table 2). The numbers of alleles at loci among site subpopulations varied from 1 to 8 bands (Table 2). Some loci appeared polymorphic at some sites but monomorphic at other sites. Locus satt197 was monomorphic at nine sites. At sampling sites, G ST values indicated that the degrees of gene differentiation occurred at different extents between loci with a mean value of 0.559 (range from 0.247 at locus Satt197 to 0.870 at locus Satt334) for loci in sites (Table 2).

Genetic diversity in relation to subpopulation establishment history

The genetic parameters in the sites were estimated (Table 3). The values indicated that all parameters drastically varied between the subpopulations. Subpopulation polymorphic loci (P) averaged a 72 % value, ranging from 13.89 to 97.22 %. The mean number of alleles per locus was 2.85, ranging from 1.14 to 4.39 in sites. The relationship between the genetic diversity and the history of the subpopulation establishment was revealed in the case of site 2. This site grew alone on a small area of no more than about 400 m2 at a crook of a montane rivulet without serious environmental stresses and bore similar morpha among the individuals and had the lowest genetic diversity (Na = 41, P = 13.89 %, A p  = 1.139, h = 0.019, and I = 0.034) (Tables 2, 3). The AMOVA analysis showed the lowest intrapopulation variation in the site 2 subpopulation (Table 3). This site subpopulation was inferred to be a newly formed founder population generated by migration of the upriver lesser seeds during a short period so as to contain such low level of genetic diversity.

Genetic diversity in relation to ecological habitats

The AMOVA analysis indicated that higher genetic variation existed among subpopulations (=57.46 %, d.f. = 9; P = 0.000001) and within subpopulation (=42.54 %, d.f. = 576; P = 0.000001). The levels of subpopulation genetic diversity were basically related to the environmental conditions of habitats, as shown in Fig. 2, the founder and drought subpopulations at sites 2 and 4 with low levels of genetic diversity were congregated in narrow spaces, respectively; while subpopulations at sites 3, 5, and 8 with high levels of genetic diversity were scattered in broad spaces. Obviously, good habitat conditions would generate high levels of genetic diversity in wild soybean populations. For example, site 4 subpopulation at a drought habitat had the second lowest total number of alleles (Na = 44) (Table 2), percentage of polymorphic loci (P = 25), number of alleles per locus (A p  = 1.306), gene diversity (h = 0.043) and Shannon Index (I = 0.076) (Table 3), while other subpopulations, particularly those at sites 3, 5 and 8 located in moist and fertile soil habitats, had higher values (Na = 140–158, P = 91.67–97.22, A p  = 3.861–4.389, h = 0.554–0.608, and I = 1.006–1.145) (Tables 2, 3) and three river system subpopulations had medium–high genetic diversity (Na = 69–97, P = 52.78–86.11, A p  = 1.917–2.722, h = 0.207–0.384, and I = 0.342–0.640) (Tables 2, 3). The site 4 subpopulation survived a drought selection, making the genetic background simplified.

Geographic regionality of subpopulation clustering

Genetically geographical structures of the site subpopulations were found to be related to habitats and formative history. The dendrogram obtained from the 36 SSR markers for the ten sites (Fig. 3) showed a geographic and ecological or environmental stress-depending clustering of site subpopulations and contained four divergent groups, i.e. respectively belonging to north-east mountain-hill areas (sites 1, 2, 3 and 5), central plain (sites 7, 8 and 9), west mountain area (site 10), and a drought habitat (site 4). Sites 1, 2 and 3 clustered into one wide group (about 63 km between sites 1 and 2 and about 64 km between sites 2 and 3 along the river). Although sites 6 and 7 were separated by 17.5 km along an arroyo, unexpectedly only site 6 deviated from its geographical area of the central plain and fell into a distant geographic area of clustering (Fig. 3).

Unique alleles in different habitat subpopulations

All subpopulations were found to contain alleles unique to the entire regional population (Table 4). However, the most distinct characterization for the subpopulations was that there were fewer unique alleles and fewer loci containing unique alleles in the founder population (site 2), the drought habitat (site 4) and the river system subpopulations (sites 6, 7 and 9) than in other subpopulations (Table 4). The founder population had smaller number of alleles than the drought-stress subpopulation.

Table 4 Distribution of unique allele among the subpopulations in Beijing region
Full size table

Phenotypic diversity in subpopulations

Significance test of subpopulation phenotypic characters showed that morphologically, nine subpopulations one another had significant differences in one to three quantitative traits (Table 5), and only the site 4 drought subpopulation had significant differences with all other subpopulations in the four quantitative traits measured, suggesting that the ecological environments of the habitats could impact their respective subpopulation morpha (Table 5). Shannon–Weaver indexes (I) in the ten subpopulations also showed that the drought subpopulation at site 4 had the lowest phenotypic diversity (Table 3) and the smallest phenotypic values in seed size (1.20 g), aboveground dry matter (17.82 g), yield per plant (6.03 g), and plant height (106 cm). Most water system subpopulations (sites 1, 3, 6, and 7) excepting the semiarid site 9 had greater seeds (1.93–2.09 g) and high Shannon indexes (I) in morphology (I = 1.270–1.553). Site 9 subpopulation had relatively lower phenotypic diversity (I = 0.913), this might be the cause of its growing in a semiarid sandy soil so as to bear adaptive convergence of morphology although it was located on a riverbank. The subpopulations (sites 3, 5 and 8) with good habitat conditions had high phenotypic indexes (I = 1.328–1.553), particularly site 3 of the best habitat conditions gave the highest I value. While the drought subpopulation (site 4) and the founder population (site 2) showed two lowest I values, 0.795 and 1.093, respectively (Table 3). The phenotypic diversity and genetic diversity had similarly fluctuated patterns in Shannon index (I) (Fig. 4), with a significant value of r = 0.719.

Table 5 Significance test for the average values of four morphological traits among the subpopulations
Full size table

Occurrence of big-seeded plants in site 1 subpopulation

Twenty-three or 63.89 % loci and 9 subpopulations were observed to have natural crossing (Tables 2, 3). The heterozygosity (H o) was detected to be mean 0.005 and range 0.000–0.017 in the sites (Table 3). A big-seeded type of over 3.0 g 100-seed weight is called semi-wild soybean (G. gracilis by Skvortzow 1927). Plants with 3.38–4.47 g in 100-seed weight were found in the site 1, they grew closely together within about 6 m2. These big-seeded plants were 2.08–3.25 fold seed mass of the typical G. soja plants and also very different from the typical G. soja type in other morphological traits at significance levels of differences (P < 0.01) (Table 6), were absolutely different from the typical wild soybean plants in morphology (Fig. 5). The genetic differentiation between the big-seeded and typical G. soja plants were characterized by the phylogenetic tree performed with the SSR loci (Fig. 6).

Table 6 Morphological characteristics for the big-seeded type (G. gracilis) and typical wild type (G. soja) plants collected in site 1 subpopulation under experimental field conditions (in Beijing)
Full size table

The SSR data showed that the big-seeded plants had so many rarely unique alleles or special genotypes at 10 loci (i.e. allele M at satt300, C at satt197, B and C at satt556, G at satt386, J at satt185, J at satt268, A at satt334, K at satt352, I at satt239, B and C at satt530), which did not appear in all the typical G. soja plants (Table 7). These unique alleles were inferred to be introgressive alleles from cultivated soybeans around the site because usually, only crossing between wild and cultivated soybeans can cause big-seeded offspring as a result of the gene effect of seed size that is involved in many genes with minor effect in the subgenus Soja (Liu et al. 2007).

Table 7 Genotypes at 36 SSR loci detected in 22 typical wild type plants (G. soja) and 8 big-seeded type plants (G. gracilis) within site 1 subpopulation
Full size table

We also observed that heterozygosity at satt346 (heterozygote EA) in the big-seeded type, which was caused by the segregation of the offspring between an AA (or EE) wild soybean plant and an EE (or AA) cultivated soybean. Loci satt556 and satt530 had two special genotypes BB and CC at their respective locus in the big-seed plants. There were two possible explanations for each locus case: one, the cultivated soybean donor plant originally was a BC heterozygote and it crossed with a typical G. soja plant to segregate BB and CC big-seeded plants; second, a cultivated donor with BB or CC genotype hybridized a typical G. soja plant with a genotype (CC or BB) but the CC and BB genotypes of typical G. soja were not collected in the samples from this site 1 in this study. The gene flow took place from cultivated into wild soybean, thereby generating the hybrid plants between wild and cultivated soybeans. So many unique alleles in these big-seeded plants could not come from mutations synchronously occurring in these big-seeded plants. In other words, the typical G. soja plants within this site were not the donors of these unique alleles in the big-seeded plants, which must be transferred from nearby soybean(s) around the site.

Discussion

Ecogeographic regionality of wild soybean subpopulations

Wild soybean had a variable level of genetic diversity in geographic sites (Table 3). In this study, a mean value (G st = 0.559) of gene differentiation. Isozyme research has indicated that riverbank populations of wild soybean would have very low genetic differentiation in Japan (Kiang et al. 1992; Fujita et al. 1997), which appears to be due to the “water system effect” in the along-river system populations. The clustering of site subpopulations studied here suggested that there was a genetic similarity for geographical ecotopes in wild soybean (Fig. 3); the subpopulations were distinctly clustered as corresponding geographical groups (Fig. 1), the north-east mountain-hill area, the central plain area, and the west mountain area, with the exception of drought site 4 and riverside site 6. Sometimes genetic differentiation between subpopulations may be rapid, just as shown in Fig. 3, where sites 6 and 7 were isolated in a distance of 17.5 km apart along a wadi but showed larger genetic differences between them.

Fig. 1
figure 1

Sampling sites of wild soybean subpopulations studied in Beijing region. Sites 1, 2 and 3 were distributed along a river valley in the northern mountain area. Sites 4 and 5 were located in the eastern foothill. Sites 6, 7, 8 and 9 were on the central plain; but sites 6 and 7 were distributed along a wadi. Site 10 lay on the eastern edge of the western mountainous, near the central plain

Full size image
Fig. 2
figure 2

Principal component analysis for ten subpopulations of wild soybean in Beijing region based SSR markers. Site 2 founder and site 4 drought subpopulations with low levels of genetic diversity were distinctly congregated in narrow spaces, respectively. Eight big-seeded plants in site 1 were distinguished from other plants. Subpopulations at sites 3, 5, and 8 with high levels of genetic diversity were scattered in broad spaces

Full size image
Fig. 3
figure 3

UPGMA dendrograms based Nei’s on (1983) genetic distance estimated by using 36 SSR markers indicating relationship between ten site subpopulations. The numbers on clades are genetic distance

Full size image

Correlation between ecological environments or habitats and genetic diversity

Autogamous plants commonly have lower within-population genetic variation and higher among-population genetic variation (Loveless and Hamrick 1984). In the Beijing regional natural populations, the within-population genetic diversity (H S ) was 0.35 and the among-population diversity (D ST ) was 0.45. Nevertheless, large geographical-divided regions have very high intra-region genetic diversity and very low inter-region genetic variation (Wang and Takahata 2007; We et al. 2009) owing to the adjoining effect between regions and the large genetic heterogeneity within vast regional regions.

Our results indicated that the genetic diversity was related to ecological environments, habitats. A significant accordance between genetic and morphological variations in the subpopulations was observed, as demonstrated by the Fig 4, where similarly-fluctuated patterns of the I values between the phenotypic and genetic diversities were illustrated across the subpopulations, with a significant value of r = 0.719. Fertile habitats showed higher genetic diversity; such as site 3 as an extreme case of best environmental conditions correspondingly had the highest genetic and the phenotypic diversities. Commonly along-river system habitat subpopulations would have relatively lower genetic differences (Tables 3, 4).

Fig. 4
figure 4

Distribution patterns of genetic and phenotypic Shannon–Weaver indexes across ten subpopulations of wild soybean sampled in Beijing region. The phenotypic and genetic diversities had similarly fluctuated patterns

Full size image
Fig. 5
figure 5

Principal component analysis for the site 1 subpopulation of wild soybean. Eight big-seeded natural hybrid plants were identified from the other 22 typical wild soybean of this site in six morphological traits (Table 6). Two first CP explain over 85 % of variability. Black circle typical wild soybean plants (nos. 1–22); Black Triangle big-seeded hybrid plants (nos. 23–30)

Full size image
Fig. 6
figure 6

An UPGMA tree obtained by software NTSYSpc version 2.1 using 36 SSR markers. Eight big seed lines nos. 23–30 were clustered into a group distinctly differentiated from other wild soybean lines within site 1 subpopulation

Full size image

Impacts of drought and founder bottlenecks on genetic structure variation

Evolutionary approaches such as the population founder effect and natural selection can affect the level of genetic variation within soybean species (Hyten et al. 2006). In wild barley, selection has played a role in molding the geographic structure of genetic diversity at some loci and loci with very little diversity and homogenous geographic distributions may have experienced a selective sweeping (Morrell et al. 2003). The impact of the two genetic bottlenecks on genetic structure and diversity was observed in the wild soybean subpopulations studied here. The two subpopulations, founder population at site 2 and arid sandy soil population at site 4, showed low genetic variation (Table 3). Our genetic diversity analysis and geographical UPGMA clustering suggested that the bottleneck impact of drought stress appeared be inferior to the founding effect on subpopulation genetic diversity but superior on genetically geographic grouping (Table 3; Fig. 3). Although site 2 founder population also had far fewer polymorphic loci and low genetic diversity, it still showed regionality within the north-east mountain-foothill ecotope (Fig. 3).

Geographical and genetic distances

Genetic distance reflects the identity of genetic variation or the interpopulation correlations within a region or ecotope. Kuroda et al. (2006) found clinal spatial structure of genetic variation by SSRs, with a positive correlation within a range of about 200 km in Japan. Fujita et al. (1997) analysed enzymatic genetic structures of four populations of wild soybean along a river reach of 50 km and found a significant correlation between genetic and geographical distances. Our results indicated that geographical beeline distance and genetic distance had a positive but nonsignificant correlation (r = 0.307, P > 0.05) within a limit of about 100 km. Seeds can spread a long distance in water or by animals, human or other natural factors (Cain et al. 2000). Seed dispersal distances of 0.2–12.4 km were observed in Japanese wild soybean (Kuroda et al. 2006). However, no long-distance seed dispersal was detected in this study. As a self-pollinated wild soybean, near subpopulations not always imply having close genetic distance and similar genetic variation, for example sites 6 and 7 along a wadi in the center plain area, although they were more similar to each other than to site 5 of the east hill area in numbers of alleles and unique alleles, genetic parameters and phenotypic diversity (Tables 3, 4), site 6 had genetic distance closer to site 5 (Fig. 3). This could be attributed to the similar genetic structure between sites 5 and 6, involved in the more similar categories and frequencies of alleles at the loci, showing genetic drift of natural populations in genetic structure.

The origin of G. gracilis type wild soybean

Big-seeded wild soybean was thought as a species G. gracilis (Skvortzow 1927). Hermann (1962) and Hymowitz (1970) suggested that G. gracilis should be a variant of G. max and originated from introgressive hybridization between wild and soybean (Hymowitz 1970). There have been some inferential studies based on morphology (Broich and Palmer 1980) and DNA variations (Sisson et al. 1978; Shoemaker et al. 1986; Close et al. 1989; Abe et al. 1999). In this study, allochthonous genes from cultivated soybeans in these big-seeded type plants, showing them to be hybrids between wild and cultivated soybeans. In this way, this study provided indirect population evidence for the hybridization origin of semi-wild soybean G. gracilis (Table 6).

Regional ex situ conservation strategy

Wild soybean has been decreasing in total distribution area, density, and number of populations and how to conserve this genetic resource is a pressing issue in China. Several studies using one single populations for sampling strategy of wild soybean in populations proposed that sampling should be conducted at intervals of at least 10 m (Jin et al. 2003b) and contain 30–45 individual plants (Jin et al. 2003a; Zhao et al. 2006) across the entire population to maximize the amount of genetic diversity.

However, the strategy of selecting subpopulations across a geographical or ecological or special area to conduct collecting samples of seeds is important for maximizing level of the genetic diversity in ex-situ conservation. Hu and Wang (1985) propounded setting up a sampling site and sampling 100 individual plants at every >10 km or several 10 km in a region, while Li et al. (1995) suggested collecting as many individual plants as possible in an area, and Fu et al. (2002) deemed it appropriate to collect samples from as many subpopulations as possible within areas and an appropriate number of plants in each population. Our results showed higher genetic variation between subpopulation (57.46 %) and within subpopulation (42.54 %) based AMOVA analysis, and that all individual subpopulations contained unique alleles, even the founder and the drought-stressed subpopulations (Table 4). So, theoretically all subpopulations should be sampled and as many individual plants as possible should be collected in consideration of such relatively high within-subpopulation genetic variation if conserving as high genetic diversity as possible in ex situ conservation. However, we think that founder subpopulations and environmental stress habitat subpopulations could be moderately reduced in sampling size to save conserving cost because of their low genetic diversity. In addition, it could be considered that number of sampling populations could be reduced along a river system considering seed dispersal between the upper and lower reaches.

Considering biosafety of the wild soybean gene pool

Despite no evidence that transgenic soybeans escape their genes into natural wild soybean populations in the fields, currently several artificial designed field experiments confirmed that transgenes could escape from transgenic soybeans to wild soybean in field conditions (Nakayama and Yamaguchi 2002; Chen et al. 2004; Mizuguti et al. 2009, 2010; Liu et al. 2012). In addition to this study, natural hybrids between wild and cultivated soybeans have been also found in eastern Asia (Kuroda et al. 2010; Wang et al. 2010). Visible direct evidence has been provided in our recent investigations of pollen flow from cultivated soybeans into wild soybean species (Wang and Li 2011, 2013). These documents suggest that transgenes can introgress into wild soybean like traditional soybean genes, which might lead to “genetic pollution” to this species and thereby incurr unknown ecological environments. The release of genetically modified (GM) soybeans appears to be only a matter of time in China. Therefore, some measures of spatial distance isolation should be considered to isolate wild soybean populations from GM soybeans in China.