Identication of additive and epistatic effects QTLs for seed oil content in soybean based on an integrating map of two RIL populations

In (AE an eight-environment conjoint analysis based two with an integrating was conducted. An new high-density integrated genetic map containing 2212 SNP markers and covering 5718.01 cM with an average distance of 2.61 cM were constructed by the combination of two linkage maps of two associated recombinant inbred line (A-RIL) populations. A total of 64 additive effect and additive × environment interaction (AE) QTL were identied on 19 chromosomes by both ICIM and IM methods, and the proportion of phenotypic variations explained (PVE) range of QTL related to oil content was 1.29–10.75%, of which 19 QTLs had overlapping marker intervals, and qOil-5-1 was identied simultaneously in both RIL populations. Compared with previous SSR positioning results, it is found 8 SNP sites within the QTL physical interval located in the SSR sites. Among them, 4 QTLs were new found. Twelve pairs of epistatic QTLs (additive × additive, AA) and QTL interactions with environments (AAE) for oil content were identied by the ICIM method, of which 3 QTLs were new found, and 2 additive effect QTLs, qOil-9-2 and qOil-15-1, linked to the other two QTLs to produce epistatic effects. A total of 5 potential candidate genes were identied based on genetic ontology and annotated information showing the relationship with seed oil content and/or fatty acid biosynthesis and metabolism.

(PVE) range of QTL related to oil content was 1.29-10.75%, of which 19 QTLs had overlapping marker intervals, and qOil-5-1 was identi ed simultaneously in both RIL populations. Compared with previous SSR positioning results, it is found 8 SNP sites within the QTL physical interval located in the SSR sites. Among them, 4 QTLs were new found. Twelve pairs of epistatic QTLs (additive × additive, AA) and QTL interactions with environments (AAE) for oil content were identi ed by the ICIM method, of which 3 QTLs were new found, and 2 additive effect QTLs, qOil-9-2 and qOil-15-1, linked to the other two QTLs to produce epistatic effects. A total of 5 potential candidate genes were identi ed based on genetic ontology and annotated information showing the relationship with seed oil content and/or fatty acid biosynthesis and metabolism.

Conclusion
These QTLs with different effects provide the good basis for molecular-assisted breeding of soybean oil content-related traits and further ne mapping of related genes. Background Soybean [Glycine max (L.) Merr.] has become one of the most primary crops in the world, high quantity oil (average 20%) that are widely used for human consumption and industrial purpose, such as cooking oils and biodiesel [1,2]. Thus, development of increasing the soybean oil content is great signi cance for quality improvement.
However, oil content is quantitative traits not only controlled by multiple genes having small or large effects but also in uenced by environments [3,4]. Using molecular markers for indirect selection of important agronomic traits, and molecular marker-assisted selection (MAS) could improve the e ciency of traditional plant breeding [5,6]. In last three decade years, numerous quantitative trait loci (QTL) for oil content in soybean seed have been reported [5,[7][8][9][10][11][12][13], about 327 oil QTLs listed in the Soybase Database (http://www.soybase.org). And a few soybean oil content QTLs have been identi ed in multi-environment and multi-genetic backgrounds [5,7,14]. However, the resolution and accuracy with which QTL mapping can identify causal genetic changes is limited by the low total number of recombination events present in biparental mapping populations [2,15]. QTL for oil content detected by different populations derived from the common parent have been less reported.
Furthermore, previous mapping studies carried out for oil trait in soybean were mainly based on the identi cation of additive effect QTLs, while less efforts have been made on the study of complex genetic effects such as epistasis and environment effects. Epistatic effects and QTL × environments interactions effect are all-important genetic effects [16,17]. Epistasis, interaction between one pair of loci located in the same or different chromosomes, as an important genetic basis of complex traits [18]. Some studies had shown that epistasis signi cantly affected the expression of genes and genetic variation underlying soybean [17,[19][20][21][22][23][24]. Hou et al.(2014) [20] mapped oil content QTLs using simple repeat sequence (SSR) markers derived from Charleston and Dongnong594, which detected 4 pairs oil epistatic effect QTLs, and Qi et al. (2017) [12] identi ed epistatic interactions for seed oil content under multi-environments based on a high-density single-nucleotide polymorphisms (SNP) using the same population. The key for a better resolution of QTL architecture was the use of a genetic linkage map with relatively high marker coverage, and can predict functional genes.
In the previous results, we used SSR map to analysis the additive and epistasis (additive × additive) QTLs based on multi-environment data of two associated recombinant inbred lines (RIL) populations with common parents (RIL3613 and RIL6013) [25]. In this study, SNP mapping was used to re-identify the QTLs of oil content with the same population, the aim is to improve the detection e ciency, shorten the localization interval and candidate gene identi cation.  [15]. Finally, a total of 120 and 139 lines from RIL3613 and RIL6013 were used in this research, respectively.  Supplementary Table S1. All plant materials were grown in a randomized complete block design with three replications (3 m in length, 0.70 m in apart and the seeds of an individual line were sown at 0.06 cm intervals). The eld experiment was managed identically to the local soybean production.

Measurement of oil content
Seed phenotypic measurement was determined by ten mature plants randomly selecting in the middle row of per plot. Seed oil content (dry seeds, with water content of about 10%) was determined with three times by Infratec 1241 Grain Analyzer (FOSS, Sweden).

Variation analysis and heritability of phenotypic data
The signi cance of the difference in oil content between the parents of each population was determined by the t test, and the signi cance of the genotype difference between RILs and the environments was determined by ANOVA. The frequency distribution was analyzed by Microsoft Excel 2007.
For the multi-environment average value, the formula is as follows: Where h 2 is broad sense heritability, is the variance of genotype, is the variance of error, indicates variance of genotype by environment effect, r is the number of replications and e is the number of environments in the study, , and were estimated using a mixed method implemented by Proc Mixed in SAS 9.2 (SAS Institute, Cary, NC, United States).

SNP genotyping and construction of genetic linkage map
By the SoySNP660K Beadchip (Beijing Boao Biotechnology Co. Ltd), three parents and 259 RIL individuals were genotyped and 600010 SNP markers were obtained. And 2212 SNP markers were obtained using the Bin function of QTL IciMapping4.0 software (www.isbreeding.net) to analyze the obtaining individual SNP data for redundant markers, removing the markers without polymorphism and with a missing rate > 10%, the partially separated markers and selecting a marker at each 100K interval (Supplementary Figure S1). Genetic maps of each population were constructed by using the MAP function of QTL IciMapping 4.0 software, divided the linkage groups according to the anchor information of markers, and arranged by nnTwoOpt method. Using CMP to integrate the corresponding chromosomes of the two populations into one map, and nally obtained the genetic map of the A-RIL populations containing 20 linkage groups.  Table 1). The SNP map of soybean covered 5718.01 cM with an average marker density of 2.61 cM ( Table 1). The genetic length of each chromosome ranged from 170.98 cM on Chr_18/LG G to 587.11 cM on Chr_13/LG F. On average, each linkage group is covered by 110.60 SNP markers ( Figure S1).

Phenotypic variation
Phenotypic data of seed oil content of parents and RILs in two RIL populations across 8 environments have been shown in Table 2. The phenotypic data of parents indicate a wide range of variation, with the mean values of 17.28%-21.50% and 17.28%-21.40% in the RIL3613 and RIL6013 populations, respectively. The kurtosis were recorded >1 in two environments, Harbin in 2015 (E3) and Shuangcheng in 2017 (E8) of RIL3613, in other environments in two populations, kurtosis and skewness were recorded <1. The phenotypic performance of the RIL populations was continuously distributed and relatively consistent, suggesting that the segregations of oil t a normal distribution model, characterized by transgressive inheritance, and soybean seed oil is controlled by polygenes and are suitable for QTL mapping ( Figure 2 and Table 2). The ANOVA results of RILs showed that genotype, environment, genotype × environment were signi cantly variation (P=0.01). The heritability (h 2 ) of oil content was 28.1% and 43.6% in RIL3613 and RIL6013 populations, respectively, indicating that the environment also affected the expression of oil content traits (  We further detected a common QTL simultaneously across two genetic backgrounds and methods ( Figure 2 and Table 4). The qOil-5-1 (marker interval Gm05_4062384-Gm05_5158657) was located on chromosome 5, the LOD value ranged from 2.66 to 4.28, and the PVE by additive QTLs and PVE by environments interactions ranged from 2.09% to 2.72% and 1.77% to 2.71%, respectively. The qOil-5-1 showed the positive additive effect in RIL6013 population, however, it showed the negative additive effect in RIL3613 population based on the different genetic backgrounds.
We found 19 regions overlapped QTLs (including qOil-5-1) by comparing the consequence of the different genetic backgrounds, which were mapped on chromosomes 2, 5, 9, 13, 15, 17 and 19, and 10 of them were identi ed in populations RIL3613 and RIL6013 ( Figure 2 and Table 4). PVE of additive QTLs and PVE of QTL by environments interactions ranged from 1.57% to 9.33% and 0.86% to 7.63%, respectively. The additive effect direction of QTL appeared negative and positive across different genetic backgrounds and methods.
Epistasis effect QTLs and epistasis QTLs interaction with environment for oil content A total of 12 pairs of epistatic QTLs (additive × additive, AA) and epistasis QTL interaction with environment (AAE) were identi ed, covering 9 linkage groups (chrs. 1, 3, 9, 10, 11, 12, 15, 18 and 19) across two RIL populations for the oil content by ICIM method in multiple environments conjoint analysis ( Figure 3 and Table 5). And all the PVE by AA ranged from 1.61% to 3.31%. The 1 pairs QTLs showed negative epistatic effect in population RIL3613, and all the QTLs showed positive epistatic effect in population RIL6013.
Several pairs epistatic effect and AAE QTLs mapped on the same chromosomes or with other chromosomes, and one QTL also could interact with multiple QTLs.

Comparison with the QTL detected from SSR and SNP map
In present research, we using ICIM and IM of MET module with the populations RIL3613 and RIL6013 to identify QTLs associated with oil content of soybean. A total of 64 additive effect QTLs for oil content were  Table S3). Compared with the SSR genetic map [25], the SNP genetic map can shorten the interval for searching candidate genes and improve the detection power.

The additive QTLs variation
Eight SNP sites within the QTL physical interval located in the SSR map and nineteen additive common QTLs were detected by two methods and in multiple backgrounds ( Figure 2, Table 4 and Supplementary  Figure S2). Among them, the same QTL that was located simultaneously in two RIL populations, qOil-5-1 (marker interval, Gm05_4062384-Gm05_5158657) showed similar position the consensus QTL detected by Wang et al (2012) [49] in cross SD02-4-59 × A02-381100. This further con rmed the accuracy of QTLs identi cation in this study. The rest four QTLs (qOil-13-1, qOil-13-4, qOil-13-5 and qOil-17-2) are new found.

Genetic basement of epistasis QTL and interactions with environments
Epistasis, or the interaction of a pair of loci may play an important role in the genetic of complex quantitative traits [18,52]. If epistasis is ignored, many individual loci could not be detected, which will weaken the detection power of QTL [53]. In present study, PVE of A and AE of 64 additive QTL ranged from 1.29% to 10.75% and 0.86% to 10.66%, respectively, while those of AA and AAE effect of 12 pairs of epistatic QTL ranged from 1.61% to 3.31% (Figure 3 and Table 5). By the comparison, AA and AAE effect were less than A and AE effect, this is similar to the study of Teng et al. (2017) [22]. Epistasis may also play an essential role in trait improvement even if epistatic variance components are low [17,54]. In this research, two individual QTL of two pairs of epistasis effects QTLs were constructed by the intervals from the same chromosome 9 and 15, respectively, including Gm09_19759328-Gm09_19958039 interacted with Gm09_683799-Gm09_1859079, Gm15_8908864-Gm15_9850704 interacted with Gm15_2753022-Gm15_29807975, and other pairs epistasis effects QTLs were mapped on different chromosomes. One QTL also could interact with multiple QTLs. In the RIL6013 population, six QTLs interacted with more than one QTL, Gm01_41350513-Gm01_42850670, Gm09_585590-Gm09_1201285, Gm09_683799-Gm09_1859079, Gm11_18740411-Gm11_34453671, Gm12_7885195-Gm12_20168509, Gm15_8908864-Gm15_9850704 interacted with each other or other QTLs. Twelve pairs of epistatic QTLs included 14 sites, among which 11 sites contained, overlapped or crossed regions with the previous QTLs (Supplementary Figure S3) [9, 10, 29, 37-39, 47-49, 51, 55-59]. It also further illustrates the stability of these locus, and the results were similar to Yang et al. (2013) [53]. The other three sites (Gm01_4091303-Gm01_5085864, Gm01_41350513-Gm01_42850670 and Gm18_47550661-Gm18_48024037) have not been reported and can be considered as the newly discovered QTLs in this study. Boer et al. (2002) [60] believed that the QTL interactions with environments could be regarded as a speci c expression of QTL caused by year, location, temperature and other factors. Therefore, multi-environment joint analysis can be used to identify the stability of QTL and estimate the AAE effect. In this study, the PVE value by AAE effects ranged from 0.09 to 1.26% (Figure 3 and Table 5), only a pair AAE effects QTL (Gm01_4091303-Gm01_5085864~Gm19_41753854-Gm19_42163159), the PVE value of AAE effects was less than the PVE value of AA effects in the RIL3613 population, meaning that the environment played an important role, while the impact of environment were weak in remaining 11 pairs QTLs. The results showed that AA and AAE effects were important genetic effects in QTL localization [12,61].

Relationship between additive effect QTLs and epistatic QTLs
In this study, the additive effect QTL also had epistasis effects [53,62,63], which should be carefully analyzed and evaluated in MAS to improve the seed oil content in soybean. Two additive effect QTLs qOil-9-2 (marker interval, Gm09_585590-Gm09_1201285) and qOil-15-1 (marker interval, Gm15_8908864-Gm15_9850704) showed interactions with other two QTLs (Table 4 and Table 5), such as Gm09_585590-Gm09_1201285 interacted with two QTLs Gm01_41350513-Gm01_42850670 and Gm11_18740411-Gm11_34453671, Gm15_8908864-Gm15_9850704 interacted with Gm01_41350513-Gm01_42850670, Gm09_19759328-Gm09_19958039, Gm11_18740411-Gm11_34453671, and Gm12_20168509-Gm12_7885195 interacted with Gm15_2753022-Gm15_29807975. The results showed that, should not only consider the effect of one site but also the interaction of multiple sites, not only consider the main effect of QTL, but also the additive effect, AA effects and QTL interactions with environments in the application of MAS breeding.

Conclusion
In the A-RILs populations, we identi ed 8 SNP sites within the QTL physical interval located in the SSR map and 19 common additive QTLs, qOil-5-1 simultaneously existed two RILs populations, of which 4 QTLs were new found. Moreover, a total of 12 pairs of epistatic QTLs (AA) and epistasis QTL interactions with environments (AAE) were identi ed, covering 9 of the 20 linkage groups, of which 3 QTL were new found.
Around these QTLs, 5 potential candidate genes were identi ed.

Availability of data and materials
All the relevant data are all availed.

Competing Interests
The authors declare that the research was conducted in the absence of any commercial or nancial relationships that could be construed as a potential con ict of interest.

Funding
The authors gratefully acknowledge the nancial support for this study provided by grants from the Key Science and Technology Project in Heilongjiang Province (SC2019ZX16B0039).   Additive QTLs of oil content in RIL3613 (red bars) and RIL6013 (blue bars) Chromosomes genetic groups were arranged tandemly as a circle. The green bar is the QTL simultaneously in two RIL populations. The elongation is the common QTLs for two RIL populations.

Figure 3
Epistatic interactions of oil content QTLs in RIL3613 (red lines) and RIL6013 (green lines) populations.
Chromosomes genetic groups were arranged tandemly as a circle.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.