Identification of novel loci for salt stress at the seed germination stage in soybean

Abstract

Salt tolerance in soybean [Glycine max (L.) Merr.] at the seed germination stage is a critical determinant of stable stand establishment in saline soil. This study examined one population of 184 recombinant inbred lines (RILs, F_7:11) derived from a cross between Kefeng1 and Nannong1138-2 and one natural population consisting of 196 soybean landraces. A total of 11 quantitative trait loci (QTLs) and 22 simple sequence repeat (SSR) loci associated with three salt tolerance indices were detected by linkage and association mapping. The SSR marker Sat_162 was found to be closely linked to the co-localized QTLs at a site 792,811 bp from the gene Glyma08g12400.1, which was verified in response to salt stress at the germination stage. Five SSR markers, Satt201, BE475343, CSSR306, Satt664 and Satt567, were co-associated with two of the salt tolerance indices, and two SSR markers, Satt156 and Satt636, were co-associated with all three salt tolerance indices. Furthermore, elite alleles and their carrier materials were identified by analyzing alleles at the loci associated with these salt tolerance indices. These results may be beneficial for the future breeding of soybean salt tolerance at the germination stage using marker-assisted selection and molecular pyramiding breeding.

Introduction

Soil salinity is a key abiotic stress and adversely affects crop productivity and quality (Hamwieh and Xu 2008, Hamwieh et al. 2011, Wang et al. 2011). Due to excessive or poorly managed irrigation and fertilization (Flowers and Yeo 1995), areas of soil salinity are continuously increasing: approximately 20% of agricultural land worldwide is currently affected by salt (Flowers and Yeo 1995). According to the salinity tolerance (maximum salinity without yield loss), soybean (Glycine max (L.) Merr.) is classified as a moderately salt-tolerant crop (Ashraf and Wu 1994, Phang et al. 2008). However, salinity stress inhibits seed germination, and the soybean yield is reduced when soil salinity exceeds 5 dS/m (Ashraf and Wu 1994, Phang et al. 2008). Papiernik et al. (2005) reported that soybean production was reduced by 40% with increasing salinity stress (7 dS/m). Shao et al. (2009) found that the soybean yield were deceased by 52.5% and 61.0%, respectively, when they were irrigated with 14–15 and 18–20 dS/m saline water in a field condition. The development of salt-tolerant soybean cultivars is an efficient way to utilize salinized land and maintain sustainable soybean production under conditions of high salinity (Hamwieh et al. 2011). Nonetheless, the distribution of salt is uneven in fields (Hamwieh et al. 2011), and salt tolerance is a complex trait that is affected by many genetic and non-genetic factors (Ashraf and Foolad 2013, DeRose-Wilson and Gaut 2011, Long et al. 2013, Pathan et al. 2007, Wang et al. 2011). Therefore, elucidating the genetic architecture of salt tolerance and implementing molecular marker-assisted techniques will advance the breeding of salt-tolerant soybeans (Pathan et al. 2007).

Quantitative traits are typically dissected by linkage analysis and association mapping. The former detects quantitative trait loci (QTLs) using mapping methods for experimental populations derived from bi-parental crosses, whereas the latter is based on linkage disequilibrium (LD) and identifies genotype-phenotype correlations in natural populations. Compared with linkage analysis, association mapping requires less research time to establish an associated population, provides higher mapping resolution and facilitates gene discovery (Hu et al. 2013, Long et al. 2013). Regardless, association mapping and linkage analysis are complimentary to each other in terms of providing prior knowledge, cross-validation, and statistical power (Yu and Buckler 2006). In recent years, salt tolerance has been analyzed in soybean mainly through linkage mapping yet rarely through association mapping. Two co-dominant polymerase chain reaction (PCR) markers closely linked to salt tolerance have been reported by Guo et al. (2000), and a number of major quantitative trait loci (QTLs) for salt (including alkaline salt) tolerance have been detected in previous studies (Chen et al. 2008, Ha et al. 2013, Hamwieh and Xu 2008, Hamwieh et al. 2011, Lee et al. 2004, Pathan et al. 2007, Tuyen et al. 2010, 2013). For example, Lee et al. (2004) detected a major salt tolerance QTL near the Sat_091 simple sequence repeat (SSR) of the N linkage group (LG) in the F_2:5 population of a cross between cultivars S-100 (salt tolerant) and Tokyo (salt sensitive). In this case, the QTL and the Ncl locus initially reported by Abel (1969) were likely involved. According to pedigree tracking, the SSR markers Sat_091 and Satt237, which flank the major tolerance QTL, had been consistently associated with the salt tolerance of the descendants for over 60 years (Lee et al. 2004). This QTL was also confirmed in a different mapping population by different researchers (Guan et al. 2014, Ha et al. 2013, Hamwieh and Xu 2008, Hamwieh et al. 2011), and the region includes the cloned causal gene underlying this salt tolerance locus (Guan et al. 2014, Qi et al. 2014). A significant QTL for controlling alkaline salt tolerance was identified in LG D2, explaining 50.2% of the short tandem repeat (STR) variation in the F₆ population derived from the cross between Jackson and JWS156-1 (Tuyen et al. 2010). The QTL was then fine-mapped using a 3.30-cM interval between GM17-11.6 and Satt447, with a physical map length of approximately 771 kb (Tuyen et al. 2013). Utilizing F_7:11 recombinant inbred lines (RILs; 184) derived from a cross between Kefeng No.1 and Nannong1138-2, Chen et al. (2008) detected a major salt tolerance QTL in the G LG during seedling growth in both field and greenhouse experiments. In addition, Qiu et al. (2011) mapped 23 QTLs related to salt tolerance at the germination stage in soybean backcross introgression lines constructed with Hong-feng11 as the recurrent parent and Harosoy as the donor parent. Association analysis of salt tolerance has also been reported in soybean (Kan et al. 2015, Zhang et al. 2014). Epistatic association mapping for salt tolerance was performed using 257 soybean cultivars with 135 SSR markers, and 83 QTL-by-environment (QE) interactions for the salt tolerance index were detected, with several soybean salt tolerance genes found in close proximity to frequently observed QTLs (Zhang et al. 2014). Association mapping of soybean seed germination under salt stress was also conducted in one natural population consisting of 191 soybean landraces with 1142 single-nucleotide polymorphisms (SNPs), with eight SNPs and 13 suggestive SNPs being associated with salt tolerance indices. In addition, nine possible candidate genes were obtained by blast analysis. Five genes, Glyma08g12400.1, Glyma08g09730.1, Glyma18g47140.1, Glyma09g00460.1, and Glyma09g00490.3, have been verified in response to salt stress at the germination stage (Kan et al. 2015).

The above studies largely focused on salt tolerance at the soybean seedling stage but rarely at the germination stage. However, seed germination is critical for ensuring new seedlings and enhancing yield during the plant growth cycle (Wang et al. 2011). It is established that soybean germplasms have different degrees and mechanisms of salt tolerance at different developmental stages (Pathan et al. 2007, Phang et al. 2008). Although QTLs for salt stress at the germination stage have been identified by linkage mapping (Qiu et al. 2011) and association analysis (Kan et al. 2015, Zhang et al. 2014), little is known about the genetic mechanisms of salt tolerance at the germination stage using QTL and association mapping strategies (Kan et al. 2015, Qiu et al. 2011, Zhang et al. 2014). Moreover, the genetic mechanism of salt tolerance at the seed germination stage is very important for improving the salt tolerance of soybean. Thus, the objectives of this study were to determine the genetic architecture of soybean salt tolerance at the seed germination stage via linkage and association analyses and to identify major loci contributing to salt tolerance at the seed germination stage for the genetic improvement of soybean salt tolerance.

Materials and Methods

Plant materials and salt tolerance evaluation

The linkage analysis population included 184 RILs (F_7:11). The population (designated as NJRIKY) was derived from a cross between Kefeng1 and Nannong1138-2 and was developed by single-seed descent at the National Center for Soybean Improvement of China (Fu et al. 2006, Hu et al. 2013). An association mapping panel including 196 accessions was used in this experiment. The germplasm number, name and origin of all accessions are listed in Supplemental Table 1. The association mapping panel seeds were from four environments: the Jiangpu Experimental Station of the Nanjing Agricultural University (32.12° N 118.37° E), Nanjing, China, in 2012 (E1) and 2013 (E2); the Experimental Farm of the Jiangsu Yanjiang Institute of Agricultural Sciences (31.58° N 120.53° E), Nantong, China, in 2012 (E3); and the Experimental Farm of the Agricultural College of Yangzhou University (32.23° N 119.25° E), Yangzhou, China, in 2012 (E4) (Hao et al. 2012). The RILs population seeds were from two environments: the Jiangpu Experimental Station of the Nanjing Agricultural University (32.12° N 118.37° E), Nanjing, China, in 2012 (E1) and 2013 (E2).

The soybean seeds were germinated under a no-salt condition (0 mM NaCl solution) or salt stress (150 mM NaCl solution); the concentration of 150 mM NaCl used for the salt treatment was previously verified (Kan et al. 2015). The RILs population seeds from two environments and the association mapping panel seeds from four environments were germinated with three replications for both the salt-treated and control groups. For germination, the soybean seeds were first dried in an oven for three days and adjusted to a moisture content of ~10%. Forty healthy dried seeds of uniform size, normal seed coats, no breakage or disease spots were selected, weighed and placed in sterilized Petri dishes (9 cm diameter) on two sheets of filter paper. A solution of 15 mL (0 or 150 mM NaCl) was added, and the seeds were incubated in the dark in the same temperature-controlled germination chamber at 25°C ± 1°C. After 24 h, the imbibed seeds were weighed, and we calculated the number of germinated seeds; germination was considered to have occurred when the root length and the shoot length reached one seed length and one-half of the seed length, respectively (Wang et al. 2011). Subsequently, we rinsed the seeds, placed them in new dishes with new filter paper, and added 5 mL of solution (0 or 150 mM NaCl). In the next six days, we calculated the number of germinated seeds; the seeds were then rinsed, and 5 mL of solution (0 or 150 mM NaCl) was added to the dishes every 24 h until the end of the experiment.

The following three germination related traits were evaluated: 1) the imbibition rate (IR) = (W2 – W1)/W1, where W2 represents the seed weight after imbibition for 24 h and W1 represents the dry seed weight before imbibition (Wang et al. 2011); 2) the germination index (GI) = ∑Gt/Dt, where Gt represents the number of germinated seeds at N days and Dt represents N days of germinated seeds; and 3) the germination rate (GR) % = (the number of germinated seeds/seed number for the test) × 100. Salt tolerance (ST) was defined as the ratio of germination-related traits (IR, GI and GR) under salt conditions to that under no-salt conditions (Long et al. 2013).

Phenotypic data analysis

Descriptive statistics and correlation analysis were performed based on the mean values of all phenotypic data from the RILs population across two environments and the natural population across four environments. Analysis of variance (ANOVA) for all traits was performed using the PROC GLM model, and Pearson’s correlations between traits were calculated using the PROC CORR model with the software SAS 9.0 (SAS Institute 1999).

QTL mapping

The mean ST value of three replications for each RIL and a genetic linkage map (Fu et al. 2006) were used for the QTL analysis. In total, 224 polymorphic markers, including 221 SSR markers, three EST-SSRs and one R gene (resistance to soybean mosaic virus), were applied to this RILs population. The constructed linkage map covered 2,625.9 cM of the soybean genome and converged into 24 LGs, with an average distance of 11.8 cM between markers. The linkage mapping was performed with composite interval mapping (CIM) Model 6 using QTL Cartographer V 2.5_011 (http://statgen.ncsu.edu/qtlcart/). The control marker number and window size were set at 5 and 10 cM, respectively. The forward and backward regression method was selected, and empirical thresholds were computed using permutation test analyses (1000 permutations, overall error level 5%) (Churchill and Doerge 1994). QTLs considered to be significant were those having LOD (logarithm of odds) peaks that exceed the genome-wide thresholds; suggestive QTLs were those having LOD peaks that did not exceed the genome-wide threshold but were higher than a threshold of 2.0 (King et al. 2013). Confidence intervals were set as the map interval corresponding to a 1-LOD decline on either side of the LOD peak.

Genotyping and association analysis

For each accession of the association mapping panel, genomic DNA was extracted from young leaves using the CTAB (cetyltrimethylammonium bromide) method described by Doyle and Doyle (1990), with slight modifications. We incubated the sample at 65°C and extracted twice with chloroform-isoamyl alcohol (24:1). In total, 205 SSR markers covering 20 soybean chromosomes were selected from published genetic maps (Hwang et al. 2009, Song et al. 2004). Certain SSRs selected (Hwang et al. 2009) were not available on the public USDA map (Song et al. 2004), and these positions on the public map were calculated based on the two SSR markers closest to the marker. The 10-μL PCR reaction mixture consisted of 20 ng total DNA, 0.4 μM forward and reverse primers, 200 μM each dNTP, 1 × PCR buffer (10 mM Tris-HCl, pH 8.3, 50 mM KCl), 2 mM MgCl₂ and 0.5 U Taq DNA polymerase. PCR was performed with an initial denaturation at 94°C for 3 min, followed by 35 cycles of 94°C for 30 s, 54°C for 60 s and 72°C for 1 min, with a final extension at 72°C for 8 min. The reactions were performed using an MJ Research PTC 225 DNA engine thermal cycler (Bio-Rad, USA). The PCR products were separated on a 8% non-denaturing polyacrylamide gel with a 29:1 ratio of acrylamide: bisacrylamide and then silver-stained as described by Santos et al. (1993). The stained bands were analyzed based on their migration distance relative to the pBR322 DNA Marker (Tiangen Biotech Co., Ltd., Beijing, China) using Quantity One software Version 4.4.0 (Bio-Rad Laboratories).

The genetic structure of the panel was investigated using a Bayesian model based on STRUCTURE analysis and distance-based principal component analysis. The Bayesian model-based program STRUCTURE 2.2 (Pritchard et al. 2000) was employed to infer the population structure using a burn-in of 100,000, a run length of 100,000, and a model allowing for admixture and correlated allele frequencies. The number of groups (K) was set from two to ten, with seven independent runs each, and the most likely number of clusters was then determined using the Delta K method described by Evanno et al. (2005). The correct estimation of K from the Kinship matrix was calculated with SPAGeDi 1.2 software (Hardy and Vekemans 2002), which was used to estimate a kinship matrix for each pair of accessions; according to Yu et al. (2006), negative kinship values were set to zero. The level of LD between the pairs of SSR markers was calculated using TASSEL V2.1 software (http://www.maizegenetics.net/); D’ and r² were used to measure LD for each pair of loci, and the significance (P-value) of each SSR pair was determined using 100,000 permutations (Fig. 1).

Fig. 1

Scatterplot of r² and D′ of SSR pairs versus the syntenic marker genetic distance (cM) in the entire population.

In this study, the mixed-model association mapping approach was used to identify the marker-trait associations Q (structure) + K (Kinship), where Q describes a fixed effect and K a co-variable. The association analysis was conducted using TASSEL V2.1 software (Bradbury et al. 2007). To identify significant associations, P-values were compared to the Bonferroni threshold (−Log (0.05/205) = 3.61), and SSRs significantly associated with salt tolerance were identified according to this threshold (P ≤ 0.05/205 = 2.44 × 10⁻⁴ or −Log P ≥ 3.61) (Hu et al. 2013). In addition, SSRs of interest that were significantly (2 < −Log P < 3.61) associated with salt tolerance indices were defined as suggestive SSRs (King et al. 2013).

Mining of elite alleles and carrier materials

The phenotypic effects of the alleles associated with the observed phenotypes were estimated by comparing the average phenotypic value between accessions with the specific allele and were based on the average value of all accessions. The equations were as follows: a_i = ∑X_ij/n_i – ∑X/n, $AN=\frac{a_i}{\sum N_k/n_k}\times 100\%$, where a_i represents the phenotypic effect of i allele, X_ij represents the phenotypic value of j material with i allele, n_i represents the number of materials with i allele, ∑X/n represents the mean of the phenotypic value of all materials, and AN represents the ratio of an allele effect of the locus and a null allele effect of the locus (Breseghello and Sorrells 2006, Wen et al. 2008). A result of ai > 0 corresponds to a positive allele, whereas ai < 0 corresponds to a negative allele. The phenotypic allele effect was estimated through a comparison between the average phenotypic value over accessions with the specific allele and that of accessions with the “null allele”; the average positive (or negative) allele effect of a locus was then calculated over the estimated phenotypic effects of all positive (or negative) alleles (Breseghello and Sorrells 2006, Wen et al. 2008).

Results

Phenotypic variation and correlation analysis

The means, standard deviations and ranges of three germination-related traits (IR, GI and GR) and three salt tolerance indices (ST-IR, ST-GI and ST-GR) of the two parental lines, 184 RILs and 196 soybean accessions are shown in Table 1. Overall, higher values were observed for the three germination-related traits under no-salt conditions compared to under salt conditions. The mean IR value of Kefeng1 was lower than that of Nannong1138-2, and the mean GI and GR values and the mean ST-IR, ST-GI and ST-GR of Kefeng1 were higher than those of Nannong1138-2. There were statistically significant differences for GI and GR under salt stress as well as ST-GI and ST-GR between the two parents (t test, P = 1.85 × 10⁻², P = 2.99 × 10⁻³, P = 2.66 × 10⁻², P = 2.42 × 10⁻², respectively). In contrast, no differences regarding IR, GI, and GR under no-salt conditions, IR under salt stress, and ST-IR were observed. The germination traits of the two populations under the salt-stressed conditions had larger variations than without salt stress. ANOVA indicated significant (P < 0.001) genetic variation for the three germination traits among the 184 lines, 196 accessions and treatments. Furthermore, the mean ST-IR, ST-GI and ST-GR varied from 0.73 to 1.11, from 0.14 to 1.14, and from 0.20 to 1.07, respectively, in the RILs population and from 0.83 to 1.08, from 0.16 to 0.70, and from 0.31 to 1.43, respectively, in the natural population; significant (P < 0.001) genetic variation for these indices in the two populations was found. In addition, there were significant (P < 0.001 and P < 0.01) genetic variations for the three germination-related traits and the salt tolerance indices in the two populations across the two and four environments, respectively. In contrast, there was no difference for IR and ST-IR in the RILs population across two environments. The phenotypic segregation of ST-IR, ST-GI and ST-GR in the two populations approximately fit a normal continuous distribution, indicating that these three salt tolerance indices are controlled by multiple-genes (Fig. 2).

Table 1 Descriptive statistics and ANOVA of three germination-related traits under 0 mM NaCl (C) or 150 mM NaCl (S) conditions and three salt tolerance indices based on the means of the traits in the parents, 184 recombinant inbred lines (RILs) and 196 soybean accessions

Trait	Treatment	Parents		RILs					196 soybean accessions
		Kefeng No. 1	Nannong1138-2	Mean ± SDa	Range	Gb	Tc	Ed	Mean ± SDa	Range	Gb	Tc	Ed
IRe	C	130.63	135.62	136.40 ± 0.07	107.59–155.96	***	***	ns	134.68 ± 6.04	105.06–152.42	***	***	***
	S	125.94	128.16	127.05 ± 0.07	85.35–144.69				124.39 ± 4.70	100.84–136.09
ST-IRf	S/C	0.97	0.94	0.93 ± 0.04	0.73–1.11	***		ns	0.92 ± 0.03	0.83–1.08	***		***
GIg	C	36.25	35.72	26.00 ± 11.10	2.72–52.00	***	***	***	35.94 ± 9.38	7.03–55.75	***	***	***
	S	19.93	12.26	12.01 ± 0.52	0.47–33.76				16.68 ± 6.63	2.32–36.09
ST-GIh	S/C	0.58	0.36	0.46 ± 0.14	0.14–1.14	***		***	0.45 ± 0.10	0.16–0.70	***		***
GRi	C	75.42	69.17	55.68 ± 0.18	7.92–88.75	***	***	***	68.63 ± 13.22	15.83–94.58	***	***	***
	S	60.63	44.17	39.54 ± 0.18	1.67–85.00				51.39 ± 15.26	10.21–86.88
ST-GRj	S/C	0.82	0.64	0.70 ± 0.17	0.20–1.07	***		**	0.74 ± 0.14	0.31–1.43	***		***

***  Significant at P < 0.001;

**  Significant at P < 0.01; ns, not significant.

^a  SD standard deviation.

^b  Genotype.

^c  Treatment.

^d  Environment.

^e  Imbibition rate.

^f  Salt tolerance (ST) was defined as the ratio of the imbibition rate under salt conditions to that under no salt conditions.

^g  Germination index.

^h  Salt tolerance (ST) was defined as the ratio of the germination index under salt conditions to that under no salt conditions.

ⁱ  Germination rate.

^j  Salt tolerance (ST) was defined as the ratio of the germination rate under salt conditions to that under no salt conditions.

Fig. 2

Frequency distributions of three salt tolerance indices (ST-IR, ST-GI and ST-GR) in recombinant inbred lines (RILs) (A) and natural population (B) based on the means of the traits from two environments and four environments. ST-IR, salt tolerance was defined as the ratio of the imbibition rate under salt conditions to that under no salt conditions; ST-GI, salt tolerance was defined as the ratio of the germination index under salt conditions to that under no salt conditions; ST-GR, salt tolerance was defined as the ratio of the germination rate under salt conditions to that under no salt conditions. The white histogram represents the frequency distributions of ST-IR; the black histogram represents the frequency distributions of ST-GI; the patterned histogram represents the frequency distributions of ST-GR.

Pearson’s correlations between the three germination-related traits and the salt tolerance indices were analyzed based on the means of two mapping populations (Table 2). For the two populations, ST-IR was significantly negatively correlated with ST-GR (P < 0.001), whereas ST-GI and ST-GR were significantly positively correlated (P < 0.001). However, there was no correlation between ST-IR and ST-GI in the RILs population; conversely, a significant negative correlation was found between ST-IR and ST-GI in the natural population (P < 0.05). Furthermore, ST-IR was significantly negatively correlated with GI and GR under both salt conditions (P < 0.001), and ST-GI and ST-GR were strongly associated with GI and GR under control and salt conditions, except in the RILs population between ST-GI and GI under control conditions. The correlations between three salt indices (ST-IR, ST-GI, and ST-GR) and IR under control conditions were similar in the two populations. However, the correlations between the three salt indices (ST-IR, ST-GI, and ST-GR) and IR under salt conditions were different in the two mapping populations. Additionally, there were significant correlations between the three germination traits but no correlations between IR under salt stress conditions or GI and GR under control or salt stress conditions in the two populations.

Table 2 Phenotypic correlations between three germination-related traits under 0 mM NaCl (C) or 150 mM NaCl (S) conditions and salt tolerance indices based on the means of the traits in 184 recombinant inbred lines (RILs) and 196 soybean accessions

Trait	Treatment	IRa		ST-IRb	GIc		ST-GId	GRe		ST-GRf
		C	S	S/C	C	S	S/C	C	S	S/C
IRa	C		0.87***	−0.17*	0.16*	0.17*	0.07ns	0.18*	0.16*	0.05ns
	S	0.74***		0.12ns	−0.12ns	−0.06ns	0.03ns	−0.05ns	−0.07ns	−0.04ns
ST-IRb	S/C	−0.22**	0.48***		−0.42***	−0.36***	−0.14*	−0.35***	−0.38***	−0.27***
GIc	C	0.34***	0.004ns	−0.42***		0.86***	0.23**	0.91***	0.88***	0.38***
	S	0.22**	−0.10ns	−0.42***	0.87***		0.60***	0.77***	0.93***	0.58***
ST-GId	S/C	−0.10ns	−0.18*	−0.12ns	0.09ns	0.51***		0.17*	0.51***	0.80***
GRe	C	0.23**	−0.06ns	−0.38***	0.95***	0.79***	0.02ns		0.84***	0.22**
	S	0.20**	−0.12ns	−0.43***	0.90***	0.94***	0.36***	0.86***		0.65***
ST-GRf	S/C	0.03ns	−0.19*	−0.30***	0.39***	0.67***	0.71***	0.27***	0.69***

*,** and ***  Significant at P < 0.05, P < 0.01 and P < 0.001; ns, not significant. The correlation coefficients of the natural population and RILs are listed at the top right and bottom left of the Table, respectively.

^a  Imbibition rate.

^b  Salt tolerance (ST) was defined as the ratio of the imbibition rate under salt conditions to that under no salt conditions.

^c  Germination index.

^d  Salt tolerance (ST) was defined as the ratio of the germination index under salt conditions to that under no salt conditions.

^e  Germination rate.

^f  Salt tolerance (ST) was defined as the ratio of the germination rate under salt conditions to that under no salt conditions.

QTLs for three salt tolerance indices

The salt tolerance indices data from two environments and the average data across two environments of the RILs population were used for QTL detection. A total of 11 putative QTLs, including three for ST-IR, six for ST-GI, and two for ST-GR, were mapped to six chromosomes (2, 7, 8, 10, 17 and 18) for the three salt tolerance indices, clustering on chromosome 8 (Table 3). These 11 QTLs, with LOD values ranging from 2.12 to 7.19, explained 4.49–25.94 % of the phenotypic variation. The positive alleles of five QTLs were from Kefeng1, and six QTLs were from Nannong1138-2.

Table 3 QTLs for salt tolerance indices based on the mean traits of three replications in 184 recombinant inbred lines (RILs)

Environment	Trait	QTLs	Chromosome	Marker	Position	LOD	Confidence interval	Additive effect	R2 (%)	Reference
E1	ST-IRa	qST-IR-8	8	Sat_162	25.8	7.19 (2.76)	21.4–33.0	+	25.94	Kan et al. 2015, Zhang et al. 2014
		qST-IR-18	18	satt217	77.1	2.61* (2.76)	66.7–95.0	+	5.28	Cho et al. 2002, Zhang et al. 2014
	ST-GIb	qST-GI-2-1	2	satt428	146.1	2.42* (2.95)	125.1–160.1	+	17.80
		qST-GI-2-2	2	sat_351	291.5	3.37 (2.95)	287.5–291.8	+	7.60
	ST-GRc	qST-GR-8	8	Sat_162	24.8	5.17 (3.01)	19.8–32.4	−	15.62	Kan et al. 2015, Zhang et al. 2014
		qST-GR-7	7	satt245	48.4	2.85* (3.01)	36.2–59.4	+	6.10	Chen et al. 2008, Zhang et al. 2014
E2	ST-IRa	qST-IR-8	8	Sat_162	22.8	5.11 (2.78)	9.0–31.8	+	15.87	Kan et al. 2015, Zhang et al. 2014
		qST-IR-18	18	satt130	84.0	2.78 (2.78)	72.1–99.0	+	7.80	Cho et al. 2002, Zhang et al. 2014
	ST-GIb	qST-GI-17	17	sat_292	96.5	2.25* (3.05)	61.7–108.51	−	5.41
		qST-GI-10-1	10	sat_231	20.0	2.35* (3.05)	4.0–47.9	−	9.98
	ST-GRc	qST-GR-8	8	Sat_162	24.8	5.15 (2.95)	6.0–31.8	−	18.45	Kan et al. 2015, Zhang et al. 2014
		qST-GR-7	7	satt245	48.4	3.21 (2.95)	37.2–57.4	+	7.85	Chen et al. 2008, Zhang et al. 2014
Combined	ST-IRa	qST-IR-8	8	Sat_162	25.8	5.90 (2.77)	17.4–32.0	+	20.94	Kan et al. 2015, Zhang et al. 2014
		qST-IR-2	2	Satt266	168.9	2.30* (2.77)	150.4–191.6	−	4.53	Chen et al. 2008
	ST-GIb	qST-GI-8	8	Sat_294	169.9	2.12* (2.87)	158.7–188.6	−	4.49
		qST-GI-10-2	10	BE801128	222.9	3.19 (2.87)	220.8–233.6	−	9.25
	ST-GRc	qST-GR-8	8	Sat_162	23.8	3.60 (3.00)	9.6–36.2	−	9.57	Kan et al. 2015, Zhang et al. 2014

The genome-wide threshold is shown in parentheses;

*  suggestive QTLs, QTLs with LOD peaks that did not exceed the genome-wide threshold but were higher than a threshold of 2.0.

+, the positive additive effect is from Kefeng No. 1; –, the negative effect is from Nannong1138-2.

^a  Salt tolerance (ST) was defined as the ratio of the imbibition rate under salt conditions to that under no salt conditions.

^b  Salt tolerance (ST) was defined as the ratio of the germination index under salt conditions to that under no salt conditions.

^c  Salt tolerance (ST) was defined as the ratio of the germination rate under salt conditions to that under no salt conditions.

For the three salt tolerance indices, six QTLs, including two for ST-IR, two for ST-GI, and two for ST-GR, were detected in E1. Three QTLs, qST-IR-8, qST-GI-2-2 and qST-GR-8, were significant QTLs, and the other three QTLs, qST-IR-18, qST-GI-2-1 and qST-GR-7, were suggestive QTLs. In E2, six QTLs, including two for ST-IR, two for ST-GI, and two for ST-GR, were detected in the NJRIKY population. Four QTLs, qST-IR-8, qST-IR-18, qST-GR-8 and qST-GR-7, were significant, and the other two QTLs, qST-GI-17 and qST-GI-10-1, were suggestive. Based on the average data for the RILs population across two environments, a total of five QTLs were identified, including two for ST-IR, two for ST-GI, and one for ST-GR, as shown in Table 3. Three QTLs, qST-IR-8, qST-GI-10-2 and qST-GR-8, were significant, and the other two QTLs, qST-IR-2 and qST-GI-8, were suggestive.

In addition, qST-IR-18, qST-GR-7, qST-IR-8 and qST-GR-8 were detected in both E1 and E2. Moreover, qST-IR-8 and qST-GR-8 were both detected across two environments. QTLs clustering was also found; two QTLs, qST-IR-8 and qST-GR-8, were co-localized, with both being closely linked to the same marker: Sat_162.

Genetic diversity and population structure

The set of 205 SSR markers with a genome-wide distribution detected a total of 2100 alleles among the 196 soybean accessions. The average number of alleles per locus was 10.20 ± 0.54, ranging from 2 to 47. The polymorphic information content (PIC) ranged from 0.03 to 0.95, with an average of 0.65 ± 0.01, and the genetic diversity ranged from 0.03 to 0.95, with an average of 0.68 ± 0.01 (Table 4).

Table 4 Summary of allele number, gene diversity and PIC of the 205 SSR markers in the 196 cultivated soybean accessions

Index	Means ± SD	Max	Min
Allele number	10.20 ± 0.54	47.00	2.00
Gene diversity	0.68 ± 0.01	0.95	0.03
PIC	0.65 ± 0.01	0.95	0.03

PIC: Polymorphism information content.

SD: Standard deviation.

Seventy SSR markers across the 20 chromosomes were selected, and model-based STRUCTURE analysis revealed three subpopulations (Fig. 3A). The first group, referred to as POP1, contained 47 accessions; the second group, POP2, included 61 accessions, and the third group, POP3, contained 88 accessions (Fig. 3B).

Fig. 3

Population structure of 196 cultivated soybean accessions. (A) The true K of the 196 cultivated soybeans according to Evanno et al. (2005). (B) The distribution of each line in the three subpopulations. Each individual is represented by a single vertical line that is broken into four colored segments, the lengths of these segments are proportional to each of the four clusters; the color is proportional in a single vertical line, indicating the proportion of the line that belongs to a subpopulation.

Genetic relatedness and linkage disequilibrium

Relative kinship estimates based on the 70 SSR markers showed that 54.93% of the pairwise kinship estimates were equal to zero, suggesting no relatedness between these pairs of lines. Of the pairwise kinship, 35.74% ranged from zero to 0.1, suggesting weak similarity; only 8.53% showed various degrees of relatedness, with relative kinship ranging from 0.1 to 0.50. For the remaining 0.50% of the entire coeffient matrix data set, relative kinship was higher than 0.50. This pattern of genetic relatedness revealed few lines with strong similarity, with most accessions being weakly related in this complex soybean association mapping panel (Fig. 4). A K matrix was constructed for association analysis.

Fig. 4

The distribution of the pair-wise relative kinship coefficients for the 196 cultivated soybean accessions.

The extent of LD was detected among the 17,010 pairs of SSR loci for all accessions. Overall, 18.88% of those markers were in LD (based on P < 0.05). However, LD for pairs of markers from the same chromosome (within the same linkage group) was significantly higher (31.08%) than 18.88% (Fig. 1).

SSRs associated with the three salt tolerance indices

Based on the Bonferroni threshold (P ≤ 0.05/205 = 2.44 × 10⁻⁴ or −Log P ≥ 3.61), five SSR markers significantly associated with the three salt tolerance indices were identified (Table 5). Five SSR markers, including four SSR markers for ST-IR (Satt287, Sat_286, BE475343 and Satt156) and one SSR for ST-GR (Satt342), were found on chromosomes 16, 19, 2 and 1. However, no SSRs for ST-GI were detected in this study.

Table 5 Markers with significant association signals (P ≤ 2.44 × 10⁻⁴ or −log P ≥ 3.61) for salt tolerance indices based on the means of the traits in 196 soybean accessions

Trait	SSR	Chromosome	Position	−Log P	APAE	ANAE	Reference
ST-IRa	Satt277*	6	107.59	2.19	0 (0)	0 (0)	Cho et al. 2002, Kan et al. 2015, Zhang et al. 2014
	BE475343	2	30.74	6.20	+0.09 (+8.92)	−0.03 (−3.47)
	Satt287	16	15.69	3.77	+0.04 (+4.12)	−0.15 (−15.55)
	Sat_286	19	87.42	5.90	+0.04 (+3.92)	−0.02 (−1.85)
	Satt156	19	56.14	5.41	0 (0)	0 (0)	Lee et al. 2004
	Satt201*	7	13.56	2.25	0 (0)	0 (0)
	CSSR306*	7	8	2.55	+0.01 (+1.25)	0 (0)
	Satt636*	7	5	3.36	+0.02 (+1.62)	−0.01 (−0.75)
ST-GIb	Satt664*	19	92.66	3.43	0 (0)	0 (0)
	Satt156*	19	56.14	3.33	0 (0)	0 (0)	Lee et al. 2004
	Satt20*1	7	13.56	2.28	0 (0)	0 (0)
	Satt567*	7	33.47	2.51	0 (0)	0 (0)
	Satt636*	7	5	2.70	+0.11 (+30.26)	−0.03 (−8.33)
ST-GRc	Satt363*	6	98.07	2.62	+0.05 (+6.61)	−0.06 (−12.05)	Cho et al. 2002, Kan et al. 2015, Zhang et al. 2014
	Satt342	1	48.14	4.07	+0.07 (+9.53)	−0.05 (−6.34)	Yue et al. 2001
	BE475343*	2	30.74	2.46	+0.09 (+9.17)	−0.03 (−3.57)
	Satt664*	19	92.66	3.43	0 (0)	0 (0)
	Satt156*	19	56.14	3.17	0 (0)	0 (0)	Lee et al. 2004
	CSSR306*	7	8	3.34	0 (0)	−0.08 (−9.65)
	Satt567*	7	33.47	2.51	0 (0)	0 (0)
	Satt636*	7	5	3.26	+0.07 (+10.97)	−0.09 (−13.26)
	Sat_091*	3	79.51	3.00	+0.09 (+11.27)	−0.11 (−14.43)	Guan et al. 2014, Ha et al. 2013, Hamwieh and Xu 2008, Hamwieh et al. 2011, Lee et al. 2004

SSR, simple sequence repeat; LG, linkage group; APAE, the average positive allele effect; ANAE, the average negative allele effect.

SSR* was defined as a suggestive SSR significantly (2 < −log P < 3.61) associated the salt tolerance indices.

^a  Salt tolerance (ST) was defined as the ratio of the imbibition rate under salt conditions to that under no salt conditions.

^b  Salt tolerance (ST) was defined as the ratio of the germination index under salt conditions to that under no salt conditions.

^c  Salt tolerance (ST) was defined as the ratio of the germination rate under salt conditions to that under no salt conditions.

In addition, suggestive SSRs of salt tolerance were defined as SSRs significantly (2 < −Log P < 3.61) associated with salt tolerance indices (Table 3). As a result, four suggestive SSRs for ST-IR, five for ST-GI and eight for ST-GR were detected. Additional results were as follows: Satt201 was co-associated with ST-IR and ST-GI; BE475343 and CSSR306 were co-associated with ST-IR and ST-GR; Satt664 and Satt567 were co-associated with ST-GI and ST-GR; and Satt156 and Satt636 were co-associated with ST-IR, ST-GI and ST-GR.

Mining of elite alleles and carrier materials

On the basis of 22 SSR loci associated with the three salt tolerance indices, alleles of loci associated with the traits were further analyzed. As 11 SSR loci had no “null allele”, the phenotypic allele effects of these loci were not estimated. Elite alleles and their carrier materials of 11 SSRs are shown in Supplemental Tables 2, 3. Among the SSRs associated with ST-IR, BE475343 had the highest average positive effects (8.92%), whereas Satt287 had the highest average negative effects (15.55%). CSSR306 had only positive effects on ST-IR. Among the alleles associated with ST-IR, Sat_286-A133 and Sat_286-A152 had the most positive and negative phenotypic effects on ST-IR (19.10% and 6.65%, respectively), and their carrier materials were S134 and S12, respectively. For ST-GI, the proportion of the average positive allele effect of Satt636 was 30.26%, and the proportion of the average negative allele effect of Satt636 was 8.33%. In addition, Satt636-A185 and Satt636-A162 had the most positive and negative phenotypic effects on ST-GI (59.72% and 15.28%, respectively), and their carrier materials were S37 and S40, respectively. Among the SSRs associated with ST-GR, Sat_091 had both the highest average positive effects (11.27%) and the highest average negative effects (15.55%). Among the alleles associated with ST-GR, Satt636-A185 and BE475343-A250 had the most positive and negative phenotypic effects on ST-GR (31.85% and 58.51%, respectively), and their carrier materials were S15 and S134, respectively.

Discussion

Genetic diversity within cultivated soybean germplasms is of great value for ongoing soybean improvement efforts and association mapping. A variety of factors affect the amount and distribution of genetic diversity in soybean, including the extent of the domestication bottleneck, the breeding process and the amount of gene exchange between cultivated soybeans and their wild relatives. The genetic diversity of the soybean natural population has been reported in several previous studies, with the total number of alleles ranging from 826 to 1936 and the average alleles per locus ranging from 5.0 to 19.7 in different studies (Li et al. 2008, 2010, Wang et al. 2010, Wen et al. 2009). In this study, a total of 2,100 alleles and an average alleles per locus of 10.24 were identified in 196 accessions using 205 SSR markers. Such results inconsistent with previous studies may be due to a difference in the samples and markers used. Nonetheless, a population with abundant genetic diversity was used for association mapping in this study, and population structure can cause some allele frequencies to differ significantly between subpopulations, leading to Type I error in association mapping, especially in autogamous species, such as soybean (Ersoz et al. 2009). In an effort to reduce the false positives caused by population structure, in the present study, the structure was controlled using 70 SSR markers across the 20 chromosomes, and association analysis was performed using a mixed-model (Q+K) by controlling for both population structure and cryptic familial relatedness, as described by Yu et al. (2006).

Seed germination is one of the plant growth stages that is highly susceptible to salt and is severely inhibited as salinity increases (Fredj et al. 2013). Inhibition of seed germination by salt stress can be due to toxic ion effects, decreases in water uptake, and reduced in nutrient mobilization efficiency (Long et al. 2013, Wang et al. 2011). Moreover, the three salt indices (ST-IR, ST-GI, and ST-GR) represent available traits that describe the mechanisms of soybean salt tolerance during seed germination based on correlations between the traits (Kan et al. 2015). We found a total of 11 QTLs and 22 SSR loci associated with the three salt tolerance indices using QTL mapping and association mapping, respectively. One SSR marker was closely linked to the co-localized QTLs, and seven SSR markers were co-associated with two or three of the salt tolerance indices. Because of the relatively high correlation between ST-GI and ST-GR, the number of co-localized or co-associated SSRs was the largest; therefore, the genetic mechanisms between ST-GI and ST-GR could be similar (Kan et al. 2015). Because no correlation was found between ST-IR and ST-GI in the RILs population, even though significant negative correlation were found between ST-IR and ST-GI in the natural population (P < 0.05), no co-localized SSRs and a fewer number of co-associated SSR markers were detected in the two populations. These results could dissect the genetic basis of soybean salt tolerance at the germination stage and be beneficial for developing marker-assisted selection in the future breeding of soybean salt tolerance.

Based on the genetic map, the Williams 82 physical map and the soybean whole-genome sequence in SoyBase (http://www.soybase.org), we compared the results for the three salt tolerance indices by QTL mapping with that by association mapping, and analyzed the results with previously reported QTLs. However, the loci for salt tolerance indices identified by QTL mapping were not consistent with those identified by association analysis. The possible reasons are as follows. In association analysis, a locus must have an effect in multiple lines to be detected, whereas a single locus may exhibit a major effect in a bi-parental cross population if other factors are not segregating (Hu et al. 2013, Wang et al. 2008). Therefore, it is possible that loci detected in a bi-parental population may go undetected using association mapping procedures (Wang et al. 2008). In addition, association analysis has limited power for detecting loci if alleles are rare in a natural population and subpopulation-specific; in contrast, QTL mapping can detect loci in a population constructed with lines belonging to different subpopulations (Famoso et al. 2011, Hu et al. 2013). In the present study, Kefeng1 and Nannong1138-2, the parental lines of the RILs population, are also two lines of the natural population classified as pop1 from southern China and pop2 from northern China, respectively. Finally, different molecular markers were used in the analysis of the two population. For example, there were 14 SSR markers on chromosome 8 for QTL mapping and 6 SSR markers on chromosome 8 for association analysis in this study. Thus, lower marker density was the possible reason for no common genomic regions for these traits identified by linkage and association mapping. Alhough the loci detected by linkage mapping and association analysis had no cross-validation in our study, the results of the two methods were also complimentary to each other based on the number of loci identified, providing a more robust understanding of the genetic architecture of salt tolerance than any single method (Hu et al. 2013). Among the association mapping results, Satt277, which was associated with ST-IR, and Satt363, which was associated with ST-GR, have previously been mapped (Cho et al. 2002, Zhang et al. 2014) to a location approximately 500 kb from BARC-025705-05001 (Kan et al. 2015). The SSR marker Satt363 was found to be associated with ST-GR, located in the same region as the QTL for resistance to multiple races of Heterodera glycines (Yue et al. 2001). The marker Satt664, co-associated with ST-GI and ST-GR, is approximately 1.88 cM from Satt513 in LG L, as reported by Qiu et al. (2011). The co-associated SSR marker Satt156 was found to be located in the same genomic region reported by Lee et al. (2004). Moreover, the SSR marker Sat_091, which was associated ST-GR, was previously identified to be closely linked to salt tolerance (Guan et al. 2014, Ha et al. 2013, Hamwieh and Xu 2008, Hamwieh et al. 2011, Lee et al. 2004), and the causal gene underlying this salt tolerance locus has been cloned (Guan et al. 2014, Qi et al. 2014). Regarding the QTL mapping results, the localizations of five SSR markers (Sat_162, satt217, satt130, satt245 and Satt266) closely linked to the QTLs (qST-IR-8, qST-GR-8, qST-IR-18, qST-GR-7 and qST-IR-2) have been reported in previous studies (Chen et al. 2008, Cho et al. 2002, Kan et al. 2015, Zhang et al. 2014). The SSR marker Sat_162 was closely linked to the co-localized QTLs qST-IR-8 and qST-GR-8, located in the same genomic regions reported by Kan et al. (2015) and Zhang et al. (2014) using association analysis. The candidate salt tolerance gene Glyma08g12400.1, localized to a site 550,651 bp from BARC-041663-08059, has been shown to respond to salt stress at the germination stage (Kan et al. 2015). Glyma08g12400.1 is homologous to the SOS6 gene in Arabidopsis, which encodes the cellulose synthase-like protein AtCSLD5. Moreover, SOS6 plays a critical role in osmotic stress tolerance and is likely involved in the regulation of reactive oxygen species (ROS) under stress (Zhu et al. 2010). The physical distance between the candidate gene Glyma08g12400.1 and the SSR marker Sat_162 was found to be 792,811 bp. These SSR markers located in previously reported QTLs reconfirmed the previous findings; these results might also indicate important targets for the identification of salt tolerance genes. Moreover, these findings demonstrated that association mapping and linkage analysis are valid and provide further genetic information that is complementary for the improvement of breeding procedures. In addition, some of the loci reported here are newly identified. These novel SSRs will be important for furthering our understanding of the genetic basis of salt tolerance in soybean at the germination stage.

Elite alleles and their carrier materials were obtained using an analysis of the alleles at the loci associated with the three salt tolerance indices. As the phenotypic effects of positive (and negative) alleles were different from each other, the average positive (and negative) effects of a locus were also different from each other, indicating the potential of genetic recombination for breeding purposes (Wen et al. 2008). For Satt636, the phenotypic effects on ST-IR of positive (and negative) alleles varied from 0.13 to 7.06 (−0.13 ~ −1.62); the effects on ST-GI of positive (and negative) alleles varied from 17.50 to 59.72 (−1.39 ~ −15.28) and those on ST-GR of positive (and negative) alleles from 2.08–31.85 (−2.60 ~ −23.92). The average positive (and negative) effects for ST-IR, ST-GI, and ST-GR were 1.62 (−0.75), 30.26 (−8.33) and 10.97 (−13.26), respectively. In addition, the same marker locus could be associated with multiple traits, with the associated alleles performing distinctly in direction and size, and the covariation of the same allele in two related traits might be the genetic basis of their phenotypic correlation (Wen et al. 2008). These results may provide new salt tolerance gene accessions for salt tolerance pyramiding breeding in soybean and may significantly promote salt tolerance breeding in this important crop.

In conclusion, 11 QTLs and 22 SSR loci associated with three salt tolerance indices were obtained using linkage and association mapping. In addition, we identified one SSR marker closely linked to the co-localized QTLs and seven SSR markers co-associated with two or three salt tolerance indices, and elite alleles and their carrier materials were obtained using an analysis of alleles at the loci associated with the three salt tolerance indices. These results could lay the foundation for understanding the genetic basis of soybean salt tolerance at the germination stage and improve salt tolerance using marker-assisted selection (MAS) and molecular pyramiding breeding in soybean.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (31301341), the Fundamental Research Funds for the Central Universities (KJQN201421), and Jiangsu Collaborative Innovation Center for Modern Crop Production (JCIC-MCP).

Literature Cited

•  Abel,  G.H. (1969) Inheritance of the capacity for chloride inclusion and chloride exclusion by soybeans. Crop Sci. 9: 697–698.
•  Ashraf,  M. and  L.  Wu (1994) Breeding for salinity tolerance in plants. Crit. Rev. Plant Sci. 13: 17–42.
•  Ashraf,  M. and  M.R.  Foolad (2013) Crop breeding for salt tolerance in the era of molecular markers and marker-assisted selection. Plant Breed. 132: 10–20.
•  Bradbury,  P.,  Z.  Zhang,  D.  Kroon,  T.  Casstevens,  Y.  Ramdoss and  E.S.  Buckler (2007) TASSEL: software for association mapping of complex traits in diverse samples. Bioinformatics 23: 2633–2635.
•  Breseghello,  F. and  M.E.  Sorrells (2006) Association mapping of kernel size and milling quality in wheat (Triticum aestivum L.) cultivars. Genetics 172: 1165–1177.
•  Chen,  H.T.,  S.Y.  Cui,  S.X.  Fu,  J.Y.  Gai and  D.Y.  Yu (2008) Identification of quantitative trait loci associated with salt tolerance during seedling growth in soybean (Glycine max L.). Aust. J. Agric. Res. 59: 1086–1091.
•  Cho,  Y.,  V.N.  Njiti,  X.  Chen,  K.  Triwatayakorn,  M.A.  Kassem,  K.  Meksem,  D.A.  Lightfoot and  A.J.  Wood (2002) Quantitative trait loci associated with foliar trigonelline accumulation in Glycine Max L. J. Biomed. Biotechnol. 2: 151–157.
•  Churchill,  G.A. and  R.W.  Doerge (1994) Empirical threshold values for quantitative trait mapping. Genetics 138: 963–971.
•  DeRose-Wilson,  L. and  B.S.  Gaut (2011) Mapping salinity tolerance during Arabidopsis thaliana germination and seedling growth. PLoS ONE 6: e22832.
•  Doyle,  J.J. and  J.L.  Doyle (1990) Isolation of plant DNA from fresh tissue. Focus 12: 13–15.
•  Ersoz,  E.S.,  J.  Yu and  E.S.  Buckler (2009) Applications of linkage disequilibrium and association mapping in maize. In:  Kriz,  A.L. and  B.A.  Larkins (eds.) Molecular genetic approaches to maize improvement, Springer-Verlag, New York, pp. 173–195.
•  Evanno,  G.,  S.  Regnaut and  J.  Goudet (2005) Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Mol. Ecol. 14: 2611–2620.
•  Famoso,  A.N.,  K.  Zhao,  R.T.  Clark,  C.W.  Tung,  M.H.  Wright,  C.  Bustamante,  L.V.  Kochian and  S.R.  McCouch (2011) Genetic architecture of aluminum tolerance in rice (Oryza sativa) determined through genome-wide association analysis and QTL mapping. PLoS Genet. 7: e1002221.
•  Flowers,  T.J. and  A.R.  Yeo (1995) Breeding for salinity resistance in crop plants. Where next? Aust. J. Plant Physiol. 22: 875–884.
•  Fredj,  M.B.,  K.  Zhani,  C.  Hannachi and  T.  Mehwachi (2013) Effect of NaCl priming on seed germination of four coriander cultivars (Coriandrum sativum). Eurasia J. BioSci. 7: 11–29.
•  Fu,  S.X.,  Y.  Zhan,  H.J.  Zhi,  J.Y.  Gai and  D.Y.  Yu (2006) Mapping of SMV resistance gene Rsc-7 by SSR markers in soybean. Genetica 128: 63–69.
•  Guan,  R.X.,  Y.  Qu,  Y.  Guo,  L.L.  Yu,  Y.  Liu,  J.H.  Jiang,  J.G.  Chen,  Y.L.  Ren,  G.Y.  Liu,  L.  Tian et al. (2014) Salinity tolerance in soybean is modulated by natural variation in GmSALT3. Plant J. 80: 937–950.
•  Guo,  B.,  L.J.  Qiu,  G.H.  Shao,  R.Z.  Chang,  L.J.  Liu,  Z.Y.  Xu,  X.H.  Li and  J.Y.  Sun (2000) Tagging salt tolerant gene using PCR markers in soybean. Sci. Agric. Sin. 33: 10–16.
•  Ha,  B.K.,  T.D.  Vuong,  V.  Velusamy,  H.T.  Nguyen,  J.G.  Shannon and  J.D.  Lee (2013) Genetic mapping of quantitative trait loci conditioning salt tolerance in wild soybean (Glycine soja) PI 483463. Euphytica 193: 79–88.
•  Hamwieh,  A. and  D.H.  Xu (2008) Conserved salt tolerance quantitative trait locus (QTL) in wild and cultivated soybeans. Breed. Sci. 58: 355–359.
•  Hamwieh,  A.,  D.D.  Tuyen,  H.  Cong,  E.R.  Benitez,  R.  Takahashi and  D.H.  Xu (2011) Identification and validation of a major QTL for salt tolerance in soybean. Euphytica 179: 451–459.
•  Hao,  D.R.,  H.  Cheng,  Z.T.  Yin,  S.Y.  Cui,  D.  Zhang,  H.  Wang and  D.Y.  Yu (2012) Identification of single nucleotide polymorphisms and haplotypes associated with yield and yield components in soybean (Glycine max) landraces across multiple environments. Theor. Appl. Genet. 124: 447–458.
•  Hardy,  O.J. and  X.  Vekemans (2002) SPAGeDi: a versatile computer program to analyses partial genetic structure at the individual or population levels. Mol. Ecol. Notes 2: 618–620.
•  Hu,  Z.B.,  H.R.  Zhang,  G.Z.  Kan,  D.Y.  Ma,  D.  Zhang,  G.X.  Shi,  D.L.  Hong,  G.Z.  Zhang and  D.Y.  Yu (2013) Determination of the genetic architecture of seed size and shape via linkage and association analysis in soybean (Glycine max L. Merr.). Genetica 141: 247–254.
•  Hwang,  T.Y.,  T.  Sayama,  M.  Takahashi,  Y.  Takada,  Y.  Nakamoto,  H.  Funatsuki,  H.  Hisano,  S.  Sasamoto,  S.  Sato,  S.  Tabata et al. (2009) High-density integrated linkage map based on SSR markers in soybean. DNA Res. 16: 213–225.
•  Kan,  G.Z.,  W.  Zhang,  W.M.  Yang,  D.Y.  Ma,  D.  Zhang,  D.R.  Hao,  Z.B.  Hu and  D.Y.  Yu (2015) Association mapping of soybean seed germination under salt stress. Mol. Genet. Genomics 290: 2147–2162.
•  King,  K.E.,  G.A.  Peiffer,  M.  Reddy,  N.  Lauter,  S.F.  Lin,  S.  Cianzio and  R.C.  Shoemaker (2013) Mapping of iron and zinc quantitative trait loci in soybean for association to iron deficiency chlorosis resistance. J. Plant Nutr. 36: 2132–2153.
•  Lee,  G.J.,  H.R.  Boerma,  M.R.  Villagarcia,  X.  Zhou,  T.E.  Carter Jr.,  Z.  Li and  M.O.  Gibbs (2004) A major QTL conditioning salt tolerance in S-100 soybean and descendent cultivars. Theor. Appl. Genet. 109: 1610–1619.
•  Li,  Y.H.,  R.X.  Guan,  Z.X.  Liu,  Y.S.  Ma,  L.X.  Wang,  L.H.  Li,  F.Y.  Lin,  W.J.  Luan,  P.Y.  Chen,  Z.  Yan et al. (2008) Genetic structure and diversity of cultivated soybean [Glycine max (L.) Merr.] landraces in China. Theor. Appl. Genet. 117: 857–871.
•  Li,  Y.H.,  W.  Li,  C.  Zhang,  L.  Yang,  R.Z.  Chang,  B.S.  Gaut and  L.J.  Qiu (2010) Genetic diversity in domesticated soybean (Glycine max) and its wild progenitor (Glycine soja) for simple sequence repeat and single-nucleotide polymorphism loci. New Phytol. 188: 242–253.
•  Long,  N.V.,  O.  Dolstra,  M.  Malosetti,  B.  Kilian,  A.  Graner,  R.G.F.  Visser and  C.G.  van der Linden (2013) Association mapping of salt tolerance in barley (Hordeum vulgare L.). Theor. Appl. Genet. 126: 2335–2351.
•  Papiernik,  S.K.,  C.M.  Grieve,  S.M.  Lesch and  S.R.  Yates (2005) Effects of salinity, imazethapyr, and chlorimuron application on soybean growth and yield. Commun. Soil Sci. Plant Anal. 36: 951–967.
•  Pathan,  M.S.,  J.D.  Lee,  J.G.  Shannon and  H.T.  Nguyen (2007) Recent advances in breeding for drought and salt stress tolerance in soybean. In:  Jenks,  M.A.,  P.M.  Hasegawa and  S.M.  Jain (eds.) Advances in molecular-breeding toward drought and salt tolerant crops, Springer, Dordrecht (the Netherlands), pp. 739–773.
•  Phang,  T.H.,  G.H.  Shao and  H.M.  Lam (2008) Salt tolerance in soybean. J. Integr. Plant Biol. 50: 1196–1212.
•  Pritchard,  J.K.,  M.  Stephens and  P.  Donnelly (2000) Inference of population structure using multilocus genotype data. Genetics 155: 945–959.
•  Qi,  X.P.,  M.W.  Li,  M.  Xie,  X.  Liu,  M.  Ni,  G.H.  Shao,  C.  Song,  A.K.Y.  Yim,  Y.  Tao,  F.L.  Wong et al. (2014) Identification of a novel salt tolerance gene in wild soybean by whole-genome sequencing. Nat. Commun. 5: 1–11.
•  Qiu,  P.C.,  W.B.  Zhang,  C.Y.  Liu,  H.W.  Jiang,  C.D.  Li,  H.M.  Fan,  Q.L.  Zeng,  G.H.  Hu and  Q.S.  Cheng (2011) QTL identification of salt tolerance in germination stage of soybean. Legume Genomics Genet. 2: 20–27.
•  Santos,  F.R.,  S.D.  Pena and  J.T.  Epplen (1993) Genetic and population study of a Y-linked tetranucleotide repeat DNA polymorphism with a simple non-isotopic technique. Hum. Genet. 90: 655–656.
•  Shao,  G.H.,  H.M.  Lam,  C.H.  Peng,  Z.C.  Zheng and  F.L.  Wang (2009) Salinity. In Lam, H.M., R.Z. Chang, G.H. Shao and Z.T. Liu (eds.) Research on tolerance to stresses in Chinese soybean, China Agriculture Press, Beijing, pp. 68–69.
•  Song,  Q.J.,  L.F.  Marek,  R.C.  Shoemaker,  K.G.  Lark,  V.C.  Concibido,  X.  Delannay,  J.E.  Specht and  P.B.  Cregan (2004) A new integrated genetic linkage map of the soybean. Theor. Appl. Genet. 109: 122–128.
•  Tuyen,  D.D.,  S.K.  Lal and  D.H.  Xu (2010) Identification of a major QTL allele from wild soybean (Glycine soja Sieb. & Zucc.) for increasing alkaline salt tolerance in soybean. Theor. Appl. Genet. 121: 229–236.
•  Tuyen,  D.D.,  H.M.  Zhang and  D.H.  Xu (2013) Validation and high-resolution mapping of a major quantitative trait locus for alkaline salt tolerance in soybean using residual heterozygous line. Mol. Breed. 31: 79–86.
•  Wang,  J.,  P.E.  McClean,  R.  Lee,  R.  Jay Goos and  T.  Helms (2008) Association mapping of iron deficiency chlorosis loci in soybean (Glycine max L. Merr.) advanced breeding lines. Theor. Appl. Genet. 116: 777–787.
•  Wang,  M.,  R.  Li,  W.  Yang and  W.  Du (2010) Assessing the genetic diversity of cultivars and wild soybeans using SSR markers. Afr. J. Biotechnol. 9: 4857–4866.
•  Wang,  Z.F.,  J.F.  Wang,  Y.M.  Bao,  Y.Y.  Wu and  H.S.  Zhang (2011) Quantitative trait loci controlling rice seed germination under salt stress. Euphytica 178: 297–307.
•  Wen,  Z.X.,  T.J.  Zhao,  Y.Z.  Zheng,  S.H.  Liu,  C.E.  Wang,  F.  Wang and  J.Y.  Gai (2008) Association analysis of agronomic and quality traits with SSR markers in Glycine max and Glycine soja in China: II. exploration of elite alleles. Sci. Agric. Sin. 34: 1339–1349.
•  Wen,  Z.X.,  Y.L.  Ding,  T.J.  Zhao and  J.Y.  Gai (2009) Genetic diversity and peculiarity of annual wild soybean (G. soja Sieb. et Zucc.) from various eco-regions in China. Theor. Appl. Genet. 119: 371–381.
•  Yu,  J. and  E.  Buckler (2006) Genetic association mapping and genome organization of maize. Curr. Opin. Biotechnol. 17: 155–160.
•  Yu,  J.,  G.  Pressoir,  W.H.  Briggs,  Bi.I.  Vroh,  M.  Yamasaki,  J.F.  Doebley,  M.D.  McMullen,  B.S.  Gaut,  D.M.  Nielsen,  J.B.  Holland et al. (2006) A unified mixed-model method for association mapping that accounts for multiple levels of relatedness. Nat. Genet. 38: 203–208.
•  Yue,  P.,  D.A.  Sleper and  P.R.  Arelli (2001) Mapping resistance to multiple races of Heterodera glycines in soybean PI 89772. Crop Sci. 41: 1589–1595.
•  Zhang,  W.J.,  Y.  Niu,  S.H.  Bu,  M.  Li,  J.Y.  Feng,  J.  Zhang,  S.X.  Yang,  M.M.  Odinga,  S.P.  Wei,  X.F.  Liu et al. (2014) Epistatic association mapping for alkaline and salinity tolerance traits in the soybean germination stage. PLoS ONE 9: e84750.
•  Zhu,  J.,  B.H.  Lee,  M.  Dellinger,  X.  Cui,  C.  Zhang,  S.  Wu,  E.A.  Nothnagel and  J.K.  Zhu (2010) A cellulose synthase-like protein is required for osmotic stress tolerance in Arabidopsis. Plant J. 63: 128–140.