Mapping additive and epistatic QTLs for forage quality and yield in soybean [Glycine max (L.) Merri.] in two environments

Abstract

Soybean plants have high protein content and can be used as a supplementary source of high-protein feed. To map quantitative trait loci (QTL) for the content of crude protein (CP), neutral detergent fibre (NDF), acid detergent fibre (ADF) and dry weight of plant (DWP) in R2 stage of soybean, two recombinant inbred lines, RIL3613 and RIL6013, containing 134 and 156 RILs, derived from the cross of Dongnong L13 × Heihe 36 and Dongnong L13 × Henong 60, were planted for two consecutive years. Based on a simple sequence repeat (SSR) linkage map, QTLs of CP, NDF, ADF and DWP were mapped by interval mapping (IM) and inclusive composite interval mapping method (ICIM) using additive effect, epistatic effect and environmental interaction model. The variance components of genotype, environment and genotype × environment (G × E) interaction for quality and yield traits in the two RIL populations were significant under multiple environmental conditions. Eighteen additive effect QTLs on 10 of 20 soybean chromosomes explained 7.02%–15.67% and 2.13%–11.42% of the phenotypic variation in RIL3613 and RIL6013, respectively. Three epistatic QTL pairs related to CP and six ones for DWP were identified. Eight additive effect QTLs for CP, eight ones for NDF, three ones for ADF, and another three ones for DWP were identified by genotype × environment interaction analysis. One epistatic QTL for CP, 11 epistatic QTL pairs for ADF and 38 ones for NDF were identified. These results can provide better understanding of the genetic basis of soybean feed quality and yield.

Keywords:

• Soybean
• crude protein content
• neutral detergent fibre content
• acid detergent fibre content
• dry weight of plant
• QTL

Introduction

Soybean [Glycine max (L.)] is not only an important source of edible protein and oil, but also has rich nutritional value. Like most legumes, soybean feeds tend to be high in protein and low in fibre content, and are considered to have similar nutritional value compared to alfalfa [1]. Soybean as a feed crop has a long history of cultivation. Soybean was introduced to North America in 1765 and was grown primarily as a forage crop for the next 155 years. Starting from the 1920s, soybean has been grown as a seed crop mainly for seed protein and oil. With the growing demand for soybean oil and meal, in 1941, the acreage sown to seed soybeans in the United States exceeded that sown to forage soybeans for the first time [2]. The traditional soybean breeding target is mainly the seed yield and quality, which is reflected in the seed protein and oil content and other quantitative traits [3]. However, the progress made in the last two decades in improving the feed quality and biomass of soybeans indicates a growing interest in breeding forage soybeans [1, 4–6].

In the study of biomass per plant, shoot dry weight is one of the acceptable indicators. This method is often used to estimate plant yield, but it is also an accurate measure of plant biomass [7]. With the increasing demand for high quality feed, it is imperative to combine the dry weight of plant (DWP) per plant with crude protein content (CP), neutral degraded fibre content (NDF) and acid degraded fibre content (ADF) as a comprehensive evaluation index to evaluate the genetic variation of soybean feed quality and yield for the improvement of soybean varieties. Significant genetic variation in feed quality traits had been detected in soybean [8–10]. With the objective to improve the feed quality and yield related traits in soybean efficiently, several studies on mapping quantitative trait locus (QTL) for feed quality and yield had been conducted based on a variety of genetic backgrounds, environments and statistical methods. Asekova et al. [1] used a set of 188 recombinant inbred lines (RILs) derived from a cross between wild soybean (PI 483463) and cultivated soybean (Hutcheson) to locate QTL controlling feed quality traits. Six CP QTLs, six NDF QTLs and two ADF QTLs were identified. These QTLs were distributed on 8 out of 20 soybean chromosomes, and some QTLs for CF and ADF on L chromosome were identified stably in multiple environments. The results indicated that the genomic region labelled BARC-029975-06765 may carry a gene that has a significant effect on the expression of feed quality traits. Santos et al. [11] used soybean varieties Bossier (high biological nitrogen fixation capacity) and Embrapa 20 (medium biological nitrogen fixation capacity) to obtain a RIL, and identified 2 DWP QTLs from it, explaining 15.4% of the total variation of DWP [12]. Zhang et al. [13] found eight QTLs controlling DWP using 152 RILs derived from a cross between Nannong94-156 and Bogao. A coincident QTL for SDW flanked by S506-Satt534 on linkage group B2-1 was detected across years at low phosphorus condition, which explained 41.4% of the phenotypic variation in combination. Up to date, 7 CP QTLs, 8 NDF QTLs, 2 ADF QTLs and 47 DWP QTLs have been integrated on the public genetic linkage map (http://www.soybase.org/).

The genetic effects include additive effects, epistasis and interactions of QTLs with the environment. Epistasis effects refer to additive × additive interaction components for RIL population, which may play an important role in the improvement of quantitative traits [14]. Under the assumption that there is no epistasis, QTL mapping genetic models will lead to biased estimation of QTL parameters, therefore, models containing the epistasis are proposed [12]. In quantitative genetics, additive × environment (AE) and epistasis × environment (AAE) are two major genetic components that contribute significantly to phenotypic variation of complex traits [15, 16]. With advances in molecular techniques and statistical methods in recent decades, many studies on soybean have identified many pairs of epistatic QTLs [17–23]. However, the studies on mapping QTL for forage quality and yield traits have mainly focused on the additive effect and neglected AA and AAE [1, 11, 13].

With the objective to provides theoretical basis and to search technical support for forage soybean breeding, mapping QTLs for CP, NDF, ADF and DWP was conducted based on two soybean RIL populations derived from the crosses of Dongnong L13 × Henong 60 and Dongnong L13 × Heihe 36 in this study.

Materials and methods

Plant materials and field design

Two populations, RIL3613 and RIL6013, containing 134 and 156 RIL_2:8, respectively, were obtained from crosses Dongnong L13 × Heihe 36 and Dongnong L13 × Henong 60. Field experiments were conducted in two environments (Baojia, 45°29′46. 6″N 126°26′30. 7″E, Shuangcheng District, Harbin, 2017; Animal husbandry science base, 45°30′54. 1″N 127°02′14. 5″E, Acheng District, Harbin, 2018). Both parents and recombinant inbred lines were planted in the field. The field experiment was carried out in a randomized block design and repeated 3 times. Each plot is 5 m long, the row spacing is 0.7 m, and the plant spacing is 0.06 m. The field management was similar as the general field cultivation. Three soybean plants from each plot of each RIL were collected at the R2 growth stage (full-bloom stage) [24] for forage quality and dry weight measurements from individual plants.

Measurement of forage traits

Each harvested plant is placed in a separate paper bag and placed in a 60 °C forced-air oven (YS-120203N, Vision Scientific Co.,106 Ltd) until constant weight. Dried samples were ground with a cyclone mill (Hood Mixer, AC 220 Super grinder) and passed through a 1-mm sieve for traits of forage quality and yield determination. CP was determined by Kjeldahl nitrogen determination method (Kjeltec 8400, FOSS), NDF and ADF were determined by Van Soest method (Fibertec 8000, FOSS).

Statistical analysis

For forage quality and yield traits between parents and inbred lines Microsoft Excel 2016 were used to perform statistical analysis on the mean value, SD, CV and other parameters of parental samples and RILs samples, and Graph Prism 8 was used for plotting. The SAS program (Version 9.3;SAS Institute, 2013) was applied to conduct analysis on of variance and estimate generalized heritability (h²). The model of analysis of variance (ANOVA) included environment, genotype, genotype × environment interaction effect, and the following formula is used to estimate heritability.

For single environments:

For the multi-environment average values:

_{Where h2 is broad-sense heritabilit,} ${\sigma }_G^2$_{is the variance of genotype,}${\sigma }_{\epsilon }^2$_{is the variance of error,}${\sigma }_{GE}^2$ _{indicates the variance of genotype by environment effect, r is the number of replications and e is the number of environments in the study.}

QTL mapping

The SSR linkage map has been constructed in previous studies [25]. The total lengths of the SSR linkage maps for RIL3613 and RIL6013 were 2849.54 cM and 1886.8 cM, respectively, and the mean interval lengths were 21.92 cM and 16.13 cM, respectively. QTL mapping was performed for the average of quality and yield traits of each line in each environment by interval mapping (IM) and inclusive composite interval mapping (ICIM) [26, 27]. Using the software QTL IciMapping v4.2, firstly, the ICIM-ADD and ICIM-EPI algorithms in BIP were used to map the additive-effect and epistatic-effect QTLs in single environment and the average of two environments, and then MET model was used to analyze the additive and epistatic effect QTLs in two environments jointly. Recombination frequencies between linked loci were transformed into distances (cM) using Kosambi’s mapping function [28]. The existence of QTL was determined by setting the minimum LOD threshold to 2.5. The naming of QTL follows the QTL nomenclature of McCouch et al. [29].

Results

Phenotypic analysis

CP, NDF, ADF and DWP of soybean were measured in two RIL populations and three parents for two environments. Among RIL3613 and RIL6013 populations, four traits showed significant variation among individuals (Table 1), and the values of each trait were distributed normally (Figures 1 and 2). Compared with the phenotype of the parents (Table 2), transgressive inheritance existed in this four traits. Furthermore, the skewness and kurtosis values of most of the traits of RIL3613 and RIL6013 were less than 1, indicating that the four traits of RIL3613 and RIL6013 conform to the quantitative genetic model [30]. Therefore, RIL3613 and RIL6013 are suitable for QTLs localization. The variance of genotype, environment and gene–environment interaction of the four traits was significant under two environments (p < 0.0001) (Table 2). The generalized heritabilities of soybean forage quality and yield traits under single environment was significantly higher than those under two combined environments. These results indicated that the genetic basis of feed quality and yield traits of soybean were different in various environments and specific QTL would be detected in multiple environments.

Figure 1. Histogram of RIL3613 population forage quality and dry weight of plant in two environments.

Figure 2. Histogram of RIL6013 population forage quality and dry weight of plant in two environments.

Table 1. Joint analysis of inheritance in multiple and single environments.

Trait	Joint environment				Environment1		Environment2
	MSE	MSG	MSGE	h2	MSG	h2	MSG	h2
RIL3613
CP	27.04**	18.35**	8.94**	0.51	15.60**	0.86	11.74**	0.82
NDF	2996.63**	101.15**	0.72	0.84	51.15**	0.89	50.71**	0.86
ADF	2364.78**	53.50**	0.63	0.74	29.78**	0.82	25.68**	0.76
DWP	0.0085**	0.0021**	0.0012	0.47	0.0015**	0.84	0.0029**	0.76
RIL6013
CP	1046.53**	19.87**	0.03	0.82	12.94**	0.92	8.20**	0.86
NDF	3151.19**	91.48**	0.33	0.70	58.18**	0.89	40.43**	0.73
ADF	2486.18**	50.66**	0.37	0.81	38.61**	0.81	17.75**	0.83
DWP	0.0034**	0.0041	0.0056	0.58	0.0014**	0.65	0.0059**	0.67

Note:

** Significant at p < 0.0001.

Table 2. Evaluation of the forage quality and dry weight of plant of the RIL populations and their parents in different environments.

Environment.	Parent			RIL3613						RIL6013
	Heihe36	DongnongL13	Henong60	Mean	SD	Range	Skew.	Kur.	CV (%)	Mean	SD	Range	Skew.	Kurt.	CV (%)
CP (%)
SC17	17.53	20.09	15.66	18.19	1.84	14.09–24.33	0.38	0.34	10.10	19.07	2.20	12.21–26.74	0.20	0.76	11.53
AC18	22.46	22.33	23.31	21.45	1.81	17.37–26.47	0.29	−0.03	8.44	22.60	1.63	19.86–26.54	0.25	−0.94	7.23
AVRG	20.00	21.21	19.49	19.82	1.83	15.83–25.40	0.34	0.16	9.27	20.84	1.92	15.94–25.64	0.23	−0.09	9.38
NDF (%)
SC17	39.39	41.16	39.35	43.46	4.12	33.61–54.14	0.15	−0.19	9.47	42.04	4.39	28.91–53.07	0.10	0.13	10.43
AC18	38.80	41.50	32.72	39.45	4.11	29.75–50.28	0.18	−0.17	10.36	37.72	3.48	25.05–47.24	−0.21	0.32	9.24
AVRG	39.10	41.33	36.04	41.46	4.12	35.97–49.65	0.17	−0.18	9.92	39.88	3.90	26.74–49.56	−0.06	0.23	9.84
ADF (%)
SC17	21.16	26.17	22.02	25.53	3.15	17.13–33.36	0.10	−0.35	12.35	25.27	3.54	16.43–37.72	0.55	0.63	14.01
AC18	25.39	25.03	19.15	22.03	2.91	15.86–29.1	0.34	−0.31	13.23	21.58	2.38	16.89–27.36	0.11	−0.71	11.05
AVRG	23.28	25.60	20.59	23.78	3.03	19.39–28.88	0.22	−0.33	12.79	23.43	2.96	17.26–34.62	0.33	−0.04	12.53
DWP (kg)
SC17	0.017	0.016	0.024	0.015	0.003	0.060–0.028	0.27	0.86	20.02	0.018	0.003	0.009–0.032	0.76	0.86	16.67
AC18	0.035	0.024	0.023	0.027	0.004	0.019–0.045	0.68	0.93	14.81	0.027	0.007	0.015–0.090	0.97	0.91	25.93
AVRG	0.026	0.020	0.024	0.021	0.004	0.040–0.037	0.48	0.89	17.42	0.023	0.005	0.012–0.061	0.87	0.89	21.30

Additive effect QTLs

Three and eleven QTLs controlling forage quality and yield traits were identified in RIL3613 and RIL6013 populations growing under three environments (2017,2018, AVE), which were located on 10 of 20 soybean chromosomes (A1, A2, B1, D1A, D1b, F, J, K, M, N). In RIL3613 population, three CP additive effect QTLs were identified in three linkage groups (D1A, K and N), with PVE values of 7.02%–15.67% (Table 3 and Figure 3). The PVE value of qCP-Gm16-2 was greater than 10%. qCP-Gm07-1, qCP-Gm16-2 and qCP-Gm19-3 all had positive additive effects, indicating that the alleles from the female parent Dongnong L13 could increase CP.

Figure 3. Additive and epstatic QTL identified in RIL3613 (above) and RIL6013 (bottom) populations.

Green colour represents protein content QTL; Red colour represents ADF QTL; Mazarine colour represents NDF QTL; Bluish colour represents DWP QTL

Table 3. Additive QTLs in RIL3613 and RIL6013.

QTL	Population	Methods	Environment	Trait	Chromosome	Interval	Physical region	LOD^A	PVE(%)^B	ADD^C	LeftCI	RightCI
qCP-Gm07-1	RIL3613	IM	2018	CP	Gm07	Satt536-Satt655	Gm07:?-17727925	3.08	7.02	0.60	170.5	194.5
qCP-Gm16-2	RIL3613	ICIM	2018	CP	Gm16	Sat_350-Satt414	Gm16:?-8910056	4.28	15.67	0.92	101.5	123.5
qCP-Gm19-2	RIL3613	ICIM	AVE	CP	Gm19	Satt313-Satt373	Gm19:34589359-49191398	3.00	8.50	0.66	150.5	172.5
qADF-Gm01-1	RIL6013	ICIM	2017	ADF	Gm01	Sat_413-Sat_160	Gm01:354691-53236911	2.87	5.51	−2.79	51.5	63.5
	RIL6013	ICIM	AVE	ADF	Gm01	Sat_413-Sat_160	Gm01:354691-53236911	7.42	2.38	−3.98	54.5	62.5
qADF-Gm02-1	RIL6013	ICIM	AVE	ADF	Gm02	sat_096-sat_289	Gm02:?-50122158	3.44	2.20	−3.43	31.5	42.5
qADF-Gm08-2	RIL6013	ICIM	AVE	ADF	Gm08	satt119-sat_406	Gm08:?-3987718	3.06	2.34	−3.93	27.5	32.5
qADF-Gm11-1	RIL6013	ICIM	AVE	ADF	Gm11	satt197-sat_123	Gm11:8879480-36660539	2.63	2.13	−4.39	42.5	46.5
qADF-Gm19-2	RIL6013	ICIM	AVE	ADF	Gm19	sat_195-satt448	Gm19:21374872-41916849	4.31	2.19	−4.32	47.5	52.5
qCP-Gm02-1	RIL6013	ICIM	2018	CP	Gm02	satt546-staga002	Gm02:43775564-49358842	3.19	10.90	1.00	47.5	57.5
qCP-Gm18-1	RIL6013	IM	2018	CP	Gm18	satt688-Sat_372	Gm18:3264343-7811432	2.65	7.61	0.68	89.5	103.5
qNDF-Gm01-1	RIL6013	ICIM	AVE	NDF	Gm01	Sat_413-Sat_160	Gm01:354691-53236911	2.51	6.11	−2.90	49.5	64.5
qNDF-Gm02-1	RIL6013	IM	AVE	NDF	Gm02	satt546-staga002	Gm02:43775564-49358842	5.32	5.89	−4.08	51.5	58.5
	RIL6013	ICIM	2017	NDF	Gm02	staga002-sat_289	Gm02:49358807-50122158	3.51	10.35	−2.98	46.5	49.5
	RIL6013	ICIM	AVE	NDF	Gm02	sat_096-sat_289	Gm02:?-50122158	4.83	11.42	−3.91	33.5	45.5
qNDF-Gm03-1	RIL6013	IM	2017	NDF	Gm03	Satt584-Satt660	Gm03:9758110-39627584	2.54	7.67	−2.47	21.5	28.5
	RIL6013	ICIM	AVE	NDF	Gm03	Satt584-Satt660	Gm03:9758110-39627584	2.65	3.54	−2.61	21.5	28.5
qNDF-Gm15-2	RIL6013	IM	AVE	NDF	Gm15	satt685-satt231	Gm15:48273606-50497797	2.94	4.37	−3.15	12.5	22.5

A: LOD, log of odd.

B: PVE means phenotypic variation explanation ration.

C: ADD means additive effects.

Two CP, five ADF and four NDF additive effect QTLs were identified on eight chromosomes (linkage groups A1, A2, B1, D1b, F, J, M, N) in tRIL6013 population, with the PVE values were 7.61%–10.90%, 2.13%–5.51% and 3.54%–11.42%, respectively (Table 3). The PVE values of qCP-Gm02-1 and qNDF-Gm02-1 were both greater than 10%. The 2 QTLs of CP (qCP-Gm02-1 and qCP-Gm18-1) had a positive additive effect, which means that the alleles derived from Dongnong L13 enhanced CP, while the five ADF QTLs (qADF-Gm01-1, qADF-Gm02-1, qADF-Gm08-2, qADF-Gm11-1 and qADF-Gm19-2) and four NDF QTLs (qNDF-Gm01-1, qNDF-Gm02-1, QNDF-Gm03-1 and qNDF-Gm15-2) had a negative additive effect, indicating that the alleles of Henong 60 increase ADF or NDF.

In RIL6013 population, two QTLs with multiple effects simultaneously controlled NDF and ADF, and one QTL with multiple effects simultaneously controlled CP and NDF. Meanwhile, two QTL SSR intervals (sat_096-sat_289, Sat_413-Sat_160) simultaneously controlled NDF and ADF and one QTL SSR interval (satt546-staga002) simultaneously controlled CP and NDF. The additive effect of QTLs controlling CP was positive, while the additive effect of QTLs controlling NDF and ADF were negative, which means that the use of these QTLs may simultaneously increase CP and reduce ADF and NDF.

Epistatic effect QTLs

Nine epistatic effect QTL pairs for either CP or DWP were identified in two RIL populations under two environments (Table 4 and Figure 3). Among them, one epistatic effect QTL pair for CP showed PVE value of 11.587% in RIL3613 population, and two epistatic effect QTL pairs for CP showed PVE value of 6.921% and 7.286% in RIL6013 population, respectively. qCP-Gm10-1∼qCP-Gm19-2 showed positive epistatic effects, whereas qCP-Gm14-1∼qCP-Gm16-3 and qCP-Gm11-2∼qCP-Gm16-1 showed negative epistatic effects. Similarly, five and one epistatic effect QTL pairs for DWP with PVE values of 1.692%∼3.032% and 3.463% were identified in RIL3613 and RIL6013, respectively. qDWP-Gm05-1∼qDWP-Gm15-1 and qDWP-Gm17-1∼qDWP-Gm19-1 showed a positive epistatic effect, and the remaining 6 pairs showed a negative epistatic effect.

Table 4. Epistatic QTL in RIL3613 and RIL6013.

TraitName	Population	Analysis method	QTL_1	Marker Interval	Physical regions	QTL_2	Marker Interval	Physical regions	LOD	PVE(%)	Add1	Add2	AddbyAdd
CP	RIL3613	ICIM	qCP-Gm10-1	Satt358-Sat_303	Gm10:1022833-2697889	qCP-Gm19-2	Satt313_Satt373	Gm19:34589359-49191398	5.385	11.587	0.266	0.352	0.839
CP	RIL6013	IM	qCP-Gm14-1	satt577-Satt070	Gm14:668802-17406274	qCP-Gm16-3	Sat_255_sat_394	Gm16:?-?	6.597	7.286	0.292	0.087	−1.388
CP	RIL6013	IM	qCP-Gm11-2	sat_128-Sat_095	Gm11:9980934-29344307	qCP-Gm16-1	sat_228_Sct_065	Gm16:3049971-10499035	5.019	6.921	0.022	0.158	−1.360
DWP	RIL3613	ICIM	qDWP-Gm05-1	SYNOD26A-Sat_171	Gm05:?-34163967	qDWP-Gm15-1	Satt483_Satt231	Gm15:34902061-50497797	5.431	3.032	0.002	0.003	0.004
DWP	RIL3613	ICIM	qDWP-Gm02-1	Satt698-Satt271	Gm02:8738184-51321546	qDWP-Gm15-1	Satt483_Satt231	Gm15:34902061-50497797	6.100	2.663	−0.003	0.003	−0.004
DWP	RIL3613	ICIM	qDWP-Gm02-3	Sat_096-Satt546	Gm02:?-43775623	qDWP-Gm02-2	AI856415_Satt041	Gm02:11234544-42619574	5.241	2.622	0.003	−0.004	−0.003
DWP	RIL3613	ICIM	qDWP-Gm02-1	Satt698-Satt271	Gm02:8738184-51321546	qDWP-Gm03-1	Satt675_Satt080	Gm03:6923174-33419014	5.141	2.122	−0.004	0.003	−0.004
DWP	RIL3613	ICIM	qDWP-Gm03-1	Satt675-Satt080	Gm03:6923174-33419014	qDWP-Gm10-1	Satt358_Sat_291	Gm10:1022833-12546650	5.864	1.692	0.004	−0.004	−0.005
DWP	RIL6013	ICIM	qDWP-Gm17-1	sat_194-sat_001	Gm17:20438248-36745724	qDWP-Gm19-2	sat_405_sat_195	Gm19:12437468-21374929	5.131	3.463	−0.003	−0.002	0.003

Additive × environment interaction QTL

The phenotypic data obtained in the two environments were used for QTL mapping of CP, ADF, NDF and DWP based on the multi-environment QTL mapping model, and a total of 22 additive × environment QTLs were detected (Table 5). Eight CP QTLs were located on six linkage groups, A2, D1A, F, K, M and N. Three ADF QTLs were mapped and distributed on three linkage groups, C2, I and M. Eight NDF QTLs were located on six linkage groups, A1, A2, B1, C1, C2 and M. Three DWP QTLs were located and distributed on three linkage groups, A1, A2 and F.

Table 5. Additive × environment QTLs in RIL3613 and RIL6013.

QTL	Populatiom	Method	Trait	Chromosome	Interval	Physical region	LOD	PVE	PVE(A)	PVE(AbyE)	Add	AbyE_01	AbyE_02
qADF-Gm06-1	RIL6013	ICIM	ADF	Gm06	sat_246-satt277	Gm06:14444925-17174009	3.205	4.973	4.512	0.461	0.805	0.257	−0.257
qADF-Gm14-1	RIL6013	ICIM	ADF	Gm14	satt168-sat_009	Gm14:8210513-?	2.578	2.424	2.314	0.111	−0.616	0.135	−0.135
qADF-Gm18-1	RIL3613	ICIM	ADF	Gm18	AZ254740-Satt570	Gm18:2390346-3162756	3.369	6.077	5.933	0.144	0.793	0.123	−0.123
qCP-Gm02-1	RIL6013	ICIM	CP	Gm02	satt546-staga002	Gm02:43775564-49358842	5.116	6.126	6.096	0.030	0.912	−0.064	0.064
qCP-Gm07-1	RIL3613	IM	CP	Gm07	Satt536-Satt655	Gm07:?-17727925	3.271	6.568	4.936	1.633	0.360	−0.207	0.207
qCP-Gm11-1	RIL6013	IM	CP	Gm11	BE806308-sat_272	Gm11:199131-2710583	3.190	4.021	3.783	0.237	0.778	0.195	−0.195
qCP-Gm16-2	RIL3613	ICIM	CP	Gm16	Sat_350-Satt414	Gm16:?-8910056	4.046	8.566	2.205	6.362	0.229	−0.389	0.389
	RIL3613	IM	CP	Gm16	Sat_350-Satt414	Gm16:?-8910056	4.046	6.943	1.787	5.157	0.229	−0.389	0.389
qCP-Gm18-1	RIL6013	IM	CP	Gm18	satt688-Sat_372	Gm18:3264343-7811432	3.056	2.567	2.436	0.131	0.420	−0.098	0.098
qCP-Gm18-2	RIL6013	IM	CP	Gm18	satt352-satt564	Gm18:?-47617860	3.300	3.327	3.325	0.003	0.695	0.020	−0.020
qCP-Gm19-1	RIL3613	ICIM	CP	Gm19	Satt182-Sat_134	Gm19:2014557-15995288	2.520	5.918	5.529	0.389	0.350	−0.093	0.093
	RIL3613	IM	CP	Gm19	Satt182-Sat_134	Gm19:2014557-15995288	2.948	5.989	5.388	0.601	0.384	−	0.128
qCP-Gm19-2	RIL3613	ICIM	CP	Gm19	Satt313-Satt373	Gm19:34589359-49191398	3.377	7.945	7.944	0.001	0.410	0.004	−0.004
	RIL3613	IM	CP	Gm19	Satt313-Satt373	Gm19:34589359-49191398	2.903	5.652	5.646	0.006	0.384	0.013	−0.013
qDWP-Gm01-1	RIL6013	ICIM	DWP	Gm01	satt482-Sat_160	Gm01:7666341-53236911	2.649	27.640	5.391	22.250	−0.639	1.298	−1.298
qDWP-Gm02-4	RIL6013	ICIM	DWP	Gm02	staga002-sat_289	Gm02:49358807-50122158	4.377	45.405	10.136	35.269	−0.883	1.646	−1.646
qDWP-Gm11-1	RIL6013	ICIM	DWP	Gm11	BE806308-sat_123	Gm11:199131-36660539	2.861	19.594	0.449	19.145	−0.176	1.148	−1.148
qNDF-Gm01-1	RIL6013	ICIM	NDF	Gm01	Sat_413-Sat_160	Gm01:354691-53236911	2.766	5.295	5.061	0.234	−1.369	−0.294	0.294
qNDF-Gm02-1	RIL6013	ICIM	NDF	Gm02	staga002-sat_289	Gm02:49358807-50122158	4.275	7.576	6.389	1.187	−1.909	−0.823	0.823
qNDF-Gm03-1	RIL6013	ICIM	NDF	Gm03	Satt584-Satt660	Gm03:9758110-39627584	2.574	4.935	4.502	0.434	−1.304	−0.405	0.405
qNDF-Gm05-1	RIL3613	ICIM	NDF	Gm05	Satt625-Satt717	Gm05:18349654-33736447	2.825	1.602	1.596	0.006	−0.909	0.055	−0.055
qNDF-Gm05-2	RIL3613	ICIM	NDF	Gm05	Sat_171-SYNOD26A	Gm05:34163916-?	3.018	7.829	7.828	0.001	2.009	−0.001	0.001
qNDF-Gm06-1	RIL3613	ICIM	NDF	Gm06	Satt681-Sat_336	Gm06:488842-?	3.274	2.702	2.701	0.001	1.235	0.024	−0.024
qNDF-Gm18-1	RIL3613	ICIM	NDF	Gm18	Sat_210-Satt688	Gm18:1621238-3264375	3.440	2.766	2.765	0.001	1.296	0.026	−0.026
qNDF-Gm18-2	RIL3613	ICIM	NDF	Gm18	Satt610-Satt570	Gm18:2665085-3162756	2.577	1.849	1.848	0.001	1.117	0.022	−0.022

The additive effects of qCP-Gm02-1, qCP-Gm07-1, qCP-Gm11-1, qCP-Gm16-2, qCP-Gm18-1, qCP-Gm18-2, qCP-Gm19-1 and qCP-Gm19-2 ranged from 0.229 to 0.912, which suggested the alleles that increased CP were from the female parent Dongnong L13. The additive effect contribution rate for phenotype variation (ACPV) of qCP-Gm16-2 was less than that of the additive-by-environment interaction effect contribution rate for phenotype variation (AECPV), and AECPV of the remaining 7 QTLs were less than ACPV. The additive effects of qADF-Gm06-1 and qADF-Gm18-1 were both positive; that is, the alleles that increase ADF came from the female parent Dongnong L13. The additive effect of qADF-Gm14-1 was negative; that is, the alleles that increased ADF were from the male parent Henong 60, and ACPV of the above three QTLs was greater than AECPV.

The additive effects of qNDF-Gm01-1, qNDF-Gm02-1, qNDF-Gm03-1 and qNDF-Gm05-1 were all negative; that is, the alleles that increased NDF were derived from the male parent Henong 60 and Heihe 36. The additive effects of qNDF-Gm05-2, qNDF-Gm06-1, qNDF-Gm18-1 and qNDF-Gm18-2 were all positive, indicating that the alleles that increased NDF came from the female parent Dongnong L13. ACPV of these QTLs were greater than their AECPV.

The additive effects of qDWP-Gm01-1, qDWP-Gm02-4 and qDWP-Gm11-1 were all negative; that is, the alleles that increased DWP were from the male parent Hinong 60. AECPV was much greater than ACPV.

Epistatistic × environment interaction QTL

In the two RIL populations, 50 epistatic (additive × additive)-by-environment interaction effect (AAE) QTL pairs (Table 6) were located. Among them, 31 epistatic effects QTL pairs were positive, and 19 epistatic effects showed negative effects.

Table 6. Epistatic × environment QTL in RIL3613 and RIL6013.

Trait	Population	Analysis method	QTL_1	Chromosome1	Interval 1	Physical regions 1	QTL_2	Chromosome2	Interval 2	Physical regions 2	LOD	LOD(AA)	LOD(AAbyE)	PVE	PVE(AA)	PVE(AAbyE)	Add1	Add2	AddbyAdd	A1byE_01	A1byE_02	A2byE_01	A2byE_02	AAbyE_01	AAbyE_02
ADF	RIL3613	ICIM	qADF-Gm03-1	Gm03	Satt125-Satt080	Gm03:19368145-33419014	qADF-Gm12-1	Gm12	Satt314-Satt293	Gm12:28026665-36045909	5.112	4.971	0.141	15.875	15.799	0.076	−0.320	0.228	1.215	−0.102	0.102	0.076	−0.076	0.088	−0.088
ADF	RIL3613	ICIM	qADF-Gm05-1	Gm05	Sat_171-SYNOD26A	Gm05:34163916-?	qADF-Gm13-1	Gm13	Satt334-Satt510	Gm13:28415974-30590395	6.082	6.073	0.010	10.227	10.213	0.015	0.315	−0.913	1.305	−0.166	0.166	0.059	−0.059	0.043	−0.043
ADF	RIL3613	ICIM	qADF-Gm13-1	Gm13	SOYHSP176-Satt334	Gm13:27847477-28416021	qADF-Gm16-1	Gm16	Sat_350-Satt414	Gm16:?-8234168	5.143	5.079	0.064	16.027	15.969	0.058	−0.472	0.161	1.308	0.039	−0.039	−0.009	0.009	0.079	−0.079
ADF	RIL3613	ICIM	qADF-Gm03-1	Gm03	Satt125-Satt080	Gm03:19368145-33419014	qADF-Gm16-1	Gm16	Satt596-Satt183	Gm16:14155157-25103068	5.458	5.460	0.000	14.480	14.478	0.002	−0.345	−0.579	−1.155	−0.076	0.076	−0.037	0.037	0.008	−0.008
ADF	RIL3613	ICIM	qADF-Gm01-2	Gm01	Sat_346-Satt515	Gm01:31964313-35653134	qADF-Gm17-1	Gm17	Sat_333-Sct_192	Gm17:1753070-2841460	5.208	5.174	0.035	17.373	17.322	0.050	−0.583	−0.797	1.625	0.112	−0.112	−0.082	0.082	0.083	−0.083
ADF	RIL3613	ICIM	qADF-Gm11-2	Gm11	Satt197-Satt359	Gm11:8879539-36868269	qADF-Gm17-3	Gm17	Sat_326-Sat_022	Gm17:39561478-44682665	6.360	6.333	0.027	12.195	12.175	0.020	0.641	0.386	−1.118	0.159	−0.159	0.134	−0.134	−0.048	0.048
ADF	RIL3613	ICIM	qADF-Gm17-2	Gm17	Sat_194-Sat_001	Gm17:20438248-36745724	qADF-Gm18-2	Gm18	Satt610-Satt612	Gm18:2665085-56453831	5.138	5.084	0.054	11.140	11.122	0.018	−0.496	0.072	1.225	−0.013	0.013	0.011	−0.011	0.051	−0.051
ADF	RIL3613	ICIM	qADF-Gm08-1	Gm08	Satt538-Satt378	Gm08:46545825-46760244	qADF-Gm18-3	Gm18	Satt503-Satt288	Gm18:53955525-55407084	5.178	5.032	0.147	16.619	16.524	0.095	0.100	−0.210	1.367	−0.008	0.008	−0.047	0.047	0.102	−0.102
ADF	RIL3613	ICIM	qADF-Gm16-2	Gm16	Sct_193-Sat_224	Gm16:?-8234168	qADF-Gm19-1	Gm19	Satt182-Satt497	Gm19:2014557-33697967	5.929	5.784	0.145	22.371	22.300	0.071	0.100	−0.064	−1.511	0.070	−0.070	0.078	−0.078	−0.081	0.081
ADF	RIL3613	ICIM	qADF-Gm18-2	Gm18	Satt610-Satt612	Gm18:2665085-56453831	qADF-Gm19-2	Gm19	Satt448-Satt373	Gm19:41916805-49191398	5.233	5.122	0.111	11.150	11.139	0.011	0.599	−0.256	1.237	0.160	−0.160	−0.071	0.071	0.083	−0.083
ADF	RIL3613	ICIM	qADF-Gm16-1	Gm16	Satt414-Satt596	Gm16:8909747-14155412	qADF-Gm20-1	Gm20	Satt239-Satt354	Gm20:24129710-38123544	5.924	5.851	0.073	19.676	19.617	0.059	0.380	−0.063	−1.584	−0.003	0.003	−0.063	0.063	−0.090	0.090
CP	RIL3613	ICIM	qCP-Gm02-2	Gm02	Satt271-Sat_069	Gm02:46353731-46353789	qCP-Gm19-1	Gm19	Satt182-Satt497	Gm19:2014557-33697967	5.196	5.096	0.101	7.102	7.096	0.006	−0.041	−0.131	0.538	−0.201	0.201	0.117	−0.117	0.058	−0.058
NDF	RIL3613	ICIM	qNDF-Gm01-2	Gm01	Sat_346-Satt515	Gm01:31964313-35653134	qNDF-Gm10-1	Gm10	Satt243-Sat_307	Gm10:46088332-46827471	5.897	5.897	0.000	41.255	41.254	0.001	−1.360	0.997	−3.498	0.019	−0.019	−0.001	0.001	−0.006	0.006
NDF	RIL3613	ICIM	qNDF-Gm05-1	Gm05	Satt625-Satt155	Gm05:18349654-29157354	qNDF-Gm10-1	Gm10	Satt243-Sat_307	Gm10:46088332-46827471	7.251	7.251	0.000	57.351	57.351	0.000	−1.912	0.632	−3.480	0.009	−0.009	−0.002	0.002	−0.008	0.008
NDF	RIL3613	ICIM	qNDF-Gm02-2	Gm02	Sat_069-Sat_183	Gm02:46353731-47404775	qNDF-Gm10-1	Gm10	Satt243-Sat_307	Gm10:46088332-46827471	5.406	5.403	0.003	60.096	60.095	0.001	−1.899	0.534	−3.734	0.042	−0.042	0.002	−0.002	−0.015	0.015
NDF	RIL3613	ICIM	qNDF-Gm07-2	Gm07	Satt697-Sat_389	Gm07:?-160140	qNDF-Gm10-1	Gm10	Satt243-Sat_307	Gm10:46088332-46827471	7.235	7.232	0.003	57.391	57.391	0.000	2.056	0.627	3.690	0.026	−0.026	−0.021	0.021	−0.010	0.010
NDF	RIL3613	ICIM	qNDF-Gm06-2	Gm06	Satt281-Satt376	Gm06:6524074-15482935	qNDF-Gm10-1	Gm10	Satt243-Sat_307	Gm10:46088332-46827471	5.361	5.361	0.000	30.473	30.472	0.001	−1.850	0.741	−3.333	−0.003	0.003	−0.013	0.013	−0.001	0.001
NDF	RIL3613	ICIM	qNDF-Gm01-2	Gm01	Sat_346-Satt515	Gm01:31964313-35653134	qNDF-Gm11-1	Gm11	Satt197-Satt359	Gm11:8879480-36868313	5.208	5.207	0.001	12.829	12.827	0.002	−0.584	−0.591	1.793	0.012	−0.012	0.021	−0.021	0.019	−0.019
NDF	RIL3613	ICIM	qNDF-Gm07-2	Gm07	Satt567-Satt697	Gm07:4510477-?	qNDF-Gm12-1	Gm12	Satt314-Satt293	Gm12:28026665-36045909	5.297	5.297	0.000	31.365	31.365	0.000	−0.538	0.170	2.316	0.032	−0.032	−0.030	0.030	0.004	−0.004
NDF	RIL3613	ICIM	qNDF-Gm04-1	Gm04	Satt565-SOYGPATR	Gm04:511133-525164	qNDF-Gm13-3	Gm13	Satt114-GMRUBP	Gm13:27718778-?	5.127	5.123	0.004	17.360	17.360	0.001	−0.637	−0.942	2.253	0.019	−0.019	−0.014	0.014	0.040	−0.040
NDF	RIL3613	ICIM	qNDF-Gm07-2	Gm07	Satt567-Satt697	Gm07:4510477-?	qNDF-Gm13-1	Gm13	Sat_039-Satt425	Gm13:1909333-22874057	5.691	5.691	0.000	40.996	40.992	0.004	−0.917	−0.362	2.657	0.033	−0.033	−0.009	0.009	−0.006	0.006
NDF	RIL3613	ICIM	qNDF-Gm10-1	Gm10	Sat_307-Satt243	Gm10:46088332-46088382	qNDF-Gm14-1	Gm14	Satt474-Sct_094	Gm14:33076661-37718098	5.389	5.389	0.000	37.916	37.916	0.000	0.641	−1.097	−3.241	−0.007	0.007	0.006	−0.006	−0.002	0.002
NDF	RIL3613	ICIM	qNDF-Gm10-1	Gm10	Sat_307-Satt243	Gm10:46088332-46088382	qNDF-Gm15-1	Gm15	Satt720-Satt651	Gm15:4135853-6805232	6.014	6.015	0.000	50.080	50.080	0.000	0.438	−1.834	−3.075	−0.016	0.016	−0.031	0.031	−0.003	0.003
NDF	RIL3613	ICIM	qNDF-Gm01-2	Gm01	Satt515-Sat_346	Gm01:31964264-31964313	qNDF-Gm15-2	Gm15	Satt231-Satt483	Gm15:34902061-50497797	5.117	5.117	0.000	29.368	29.368	0.001	−0.330	−1.032	2.849	0.015	−0.015	0.008	−0.008	0.009	−0.009
NDF	RIL3613	ICIM	qNDF-Gm13-1	Gm13	Satt425-Sat_039	Gm13:1909333-1909366	qNDF-Gm16-1	Gm16	Sct_193-Sat_224	Gm16:?-8234168	6.913	6.913	0.000	47.064	47.064	0.000	−0.782	−1.054	2.940	−0.002	0.002	0.006	−0.006	0.003	−0.003
NDF	RIL3613	ICIM	qNDF-Gm12-1	Gm12	Satt293-Satt434	Gm12:36045835-38453642	qNDF-Gm16-2	Gm16	Satt596-Satt183	Gm16:14155157-25103068	6.120	6.119	0.001	25.035	25.035	0.000	0.055	−0.541	−2.227	0.016	−0.016	0.011	−0.011	0.008	−0.008
NDF	RIL3613	ICIM	qNDF-Gm07-2	Gm07	Satt697-Sat_389	Gm07:?-160140	qNDF-Gm16-2	Gm16	Satt596-Satt183	Gm16:14155157-25103068	6.644	6.644	0.000	42.492	42.492	0.000	0.464	−0.857	−2.721	0.034	−0.034	0.013	−0.013	0.000	0.000
NDF	RIL3613	ICIM	qNDF-Gm10-1	Gm10	Sat_307-Satt243	Gm10:46088332-46088382	qNDF-Gm17-1	Gm17	Sat_333-Sct_192	Gm17:1753070-2841460	6.665	6.664	0.001	38.570	38.570	0.000	0.564	1.875	2.805	−0.010	0.010	0.012	−0.012	−0.004	0.004
NDF	RIL3613	ICIM	qNDF-Gm13-1	Gm13	Satt425-Sat_039	Gm13:1909333-1909366	qNDF-Gm17-1	Gm17	Sct_192-Sat_284	Gm17:2841403-6551348	5.317	5.314	0.003	46.141	46.141	0.000	−0.782	0.908	−2.931	0.001	−0.001	0.004	−0.004	0.010	−0.010
NDF	RIL3613	ICIM	qNDF-Gm07-2	Gm07	Satt567-Satt697	Gm07:4510477-?	qNDF-Gm17-3	Gm17	Sat_194-Sat_001	Gm17:20438248-36745724	5.119	5.118	0.001	19.874	19.871	0.003	−0.616	0.142	1.935	0.025	−0.025	0.014	−0.014	0.015	−0.015
NDF	RIL3613	ICIM	qNDF-Gm08-1	Gm08	Satt177-Sat_215	Gm08:?-9211939	qNDF-Gm17-3	Gm17	Sat_001-Sat_326	Gm17:36745649-39561533	5.177	5.177	0.000	18.943	18.941	0.002	−0.532	−1.133	2.561	0.004	−0.004	0.018	−0.018	0.003	−0.003
NDF	RIL3613	ICIM	qNDF-Gm17-3	Gm17	Satt669-Sat_194	Gm17:?-20438335	qNDF-Gm17-3	Gm17	Sat_001-Sat_326	Gm17:36745649-39561533	5.287	5.287	0.000	27.682	27.680	0.002	0.933	0.166	−2.339	−0.005	0.005	0.019	−0.019	−0.013	0.013
NDF	RIL3613	ICIM	qNDF-Gm16-1	Gm16	Sct_193-Sat_224	Gm16:?-8234168	qNDF-Gm17-3	Gm17	Sat_001-Sat_326	Gm17:36745649-39561533	5.059	5.059	0.000	31.175	31.175	0.000	−0.563	−0.521	2.541	−0.004	0.004	0.013	−0.013	0.000	0.000
NDF	RIL3613	ICIM	qNDF-Gm11-1	Gm11	Satt359-Satt197	Gm11:8879480-8879539	qNDF-Gm17-3	Gm17	Sat_326-Sat_022	Gm17:39561478-44682665	5.139	5.136	0.003	15.159	15.156	0.004	0.779	0.616	−1.711	0.027	−0.027	0.039	−0.039	0.021	−0.021
NDF	RIL3613	ICIM	qNDF-Gm10-1	Gm10	Sat_307-Satt243	Gm10:46088332-46088382	qNDF-Gm18-2	Gm18	Satt610-Satt570	Gm18:2665085-3162756	5.284	5.284	0.000	24.128	24.128	0.000	1.282	−1.478	−3.409	−0.001	0.001	0.020	−0.020	−0.007	0.007
NDF	RIL3613	ICIM	qNDF-Gm06-3	Gm06	Satt316-Satt202	Gm06:47485158-47820561	qNDF-Gm18-3	Gm18	Satt501-Sat_203	Gm18:15845994-52210860	5.407	5.406	0.001	22.048	22.045	0.003	−0.526	0.343	2.132	0.035	−0.035	−0.028	0.028	0.012	−0.012
NDF	RIL3613	ICIM	qNDF-Gm13-4	Gm13	Satt334-Satt510	Gm13:28415974-30590395	qNDF-Gm18-3	Gm18	Satt501-Sat_203	Gm18:15845994-52210860	5.259	5.258	0.000	13.490	13.490	0.000	−0.207	0.034	1.569	0.035	−0.035	−0.066	0.066	−0.006	0.006
NDF	RIL3613	ICIM	qNDF-Gm17-1	Gm17	Sct_192-Sat_284	Gm17:2841403-6551348	qNDF-Gm19-1	Gm19	Sat_134-Satt182	Gm19:15995288-2014557	5.128	5.128	0.000	34.001	34.001	0.001	−0.276	0.391	2.493	0.004	−0.004	−0.051	0.051	0.003	−0.003
NDF	RIL3613	ICIM	qNDF-Gm15-1	Gm15	Satt720-Satt651	Gm15:4135853-6805232	qNDF-Gm19-1	Gm19	Sat_134-Sat_191	Gm19:15995215-33892139	5.273	5.274	0.000	23.509	23.509	0.001	−0.479	−0.447	2.010	−0.041	0.041	−0.005	0.005	0.008	−0.008
NDF	RIL3613	ICIM	qNDF-Gm17-2	Gm17	Sat_292-Satt669	Gm17:12771389-?	qNDF-Gm20-1	Gm20	Satt239-Satt354	Gm20:24129710-38123544	5.043	5.041	0.002	15.967	15.966	0.001	−0.672	−1.004	2.082	0.010	−0.010	−0.001	0.001	−0.020	0.020
NDF	RIL3613	ICIM	qNDF-Gm15-2	Gm15	Satt231-Satt483	Gm15:34902061-34902099	qNDF-Gm20-1	Gm20	Satt239-Satt354	Gm20:24129710-38123544	5.623	5.623	0.000	30.947	30.945	0.001	−1.077	−0.463	2.681	0.013	−0.013	0.002	−0.002	0.009	−0.009
NDF	RIL3613	ICIM	qNDF-Gm02-2	Gm02	Sat_069-Sat_183	Gm02:46353731-47404775	qNDF-Gm20-1	Gm20	Satt239-Satt354	Gm20:24129710-38123544	6.861	6.861	0.000	31.805	31.805	0.000	−0.971	−0.300	2.778	0.050	−0.050	0.011	−0.011	−0.001	0.001
NDF	RIL3613	ICIM	qNDF-Gm07-1	Gm07	Sat_389-Satt245	Gm07:160085-9320867	qNDF-Gm20-1	Gm20	Satt239-Satt354	Gm20:24129710-38123544	5.266	5.266	0.001	23.824	23.824	0.000	−1.319	−0.627	2.426	0.019	−0.019	0.010	−0.010	0.013	−0.013
NDF	RIL3613	ICIM	qNDF-Gm12-1	Gm12	Satt314-Satt293	Gm12:28026665-28026706	qNDF-Gm20-1	Gm20	Satt239-Satt354	Gm20:24129710-38123544	7.854	7.855	0.000	38.800	38.800	0.000	−1.237	−0.713	3.016	−0.018	0.018	0.014	−0.014	0.005	−0.005
NDF	RIL3613	ICIM	qNDF-Gm18-2	Gm18	Satt610-Satt570	Gm18:2665085-3162756	qNDF-Gm20-1	Gm20	Satt239-Satt354	Gm20:24129710-38123544	7.377	7.376	0.001	25.750	25.749	0.001	−0.938	−1.588	2.918	0.017	−0.017	0.005	−0.005	0.022	−0.022
NDF	RIL3613	ICIM	qNDF-Gm16-1	Gm16	Sct_193-Sat_224	Gm16:?-8234168	qNDF-Gm20-1	Gm20	Satt239-Satt367	Gm20:24129775-2622475	5.234	5.229	0.005	15.329	15.324	0.005	−1.481	0.495	1.848	0.033	−0.033	0.020	−0.020	−0.033	0.033
NDF	RIL3613	ICIM	qNDF-Gm10-1	Gm10	Sat_307-Satt243	Gm10:46088332-46088382	qNDF-Gm20-1	Gm20	Satt367-Satt270	Gm20:2622475-34223223	7.969	3.976	3.993	40.156	40.156	0.001	0.628	−1.687	−3.071	−0.001	0.001	0.042	−0.042	−0.015	0.015
NDF	RIL3613	ICIM	qNDF-Gm13-4	Gm13	Satt334-Satt510	Gm13:28415974-30590395	qNDF-Gm20-2	Gm20	Satt270-Sat_170	Gm20:34223176-38657738	5.947	5.941	0.007	26.471	26.469	0.001	−0.174	0.471	2.175	0.014	−0.014	0.028	−0.028	−0.028	0.028
NDF	RIL6013	ICIM	qNDF-Gm01-3	Gm01	AZ302047-satt402	Gm01:41867763-41867832	qNDF-Gm13-2	Gm13	satt030-satt343	Gm13:8722687-11494511	5.197	4.914	0.284	6.690	6.514	0.175	0.767	−0.249	−1.278	0.052	−0.052	−0.158	0.158	−0.235	0.235
NDF	RIL6013	IM	qNDF-Gm06-3	Gm06	satt202-satt371	Gm06:47820517-49159668	qNDF-Gm15-2	Gm15	satt685-satt231	Gm15:48273606-50497797	5.881	5.317	0.564	7.416	7.129	0.287	0.417	0.080	−1.359	−0.082	0.082	−0.023	0.023	−0.321	0.321

One epistatic effect QTL pair for CP was identified, qCP-Gm02-2∼qCP-Gm19-1. The epistatic (additive × additive) contribution rate for phenotype variation (AACPV) was greater than the epistatic × environment interaction contribution rate for phenotype variation (AAECPV), indicating that CP was strongly controlled by the epistatic effects.

A total of 11 epistatic-effect QTL pairs for ADF were identified, and the PVE of AA was 10.213%∼22.3%, while PVE of AAE was between 0.002% and 0.551%. For all of the 11 epistatic effect QTL pairs for ADF, the PVE of epistatic QTLs was greater than that of AAE interaction, indicating that it was strongly controlled by the epistatic effects. Seven pairwise interaction sites (qADF-Gm03-1∼qADF-Gm12-1, qADF-Gm05-1∼qADF-Gm13-1, qADF-Gm13-1∼qADF-Gm16-1, qADF-Gm01-2∼qADF-Gm17-1, qADF-Gm17-2∼qADF-Gm18-2, qADF-Gm08-1∼qADF-Gm18-3 and qADF-Gm18-2∼qADF-Gm19-2) showed positive epistatic effects, while the remaining four pairwise interaction sites showed negative epistatic effects.

Thirty-eight epistatic effect QTL pairs for NDF were identified, and the PVE of AA was 6.514%∼60.095%. The PVE range of AAE is between 0%∼0.287%. For all NDF epistatic-effect QTL pairs, the PVE for the epistatic QTLs was greater than the PVE for the AAE interaction; that is, it was deeply impacted by the epistatic effects. Among them, 15 pairwise interaction sites showed positive epistatic effects, while the remaining 23 pairwise interaction sites showed negative epistatic effects.

Discussion

Associated populations can improve the efficiency of QTL mapping

In previous studies, a recombinant inbred line derived from two parents was often used for QTL mapping. The number of polymorphism markers between parents may be limited, leading to low marker density, fewer detected intervals and low efficiency of QTL detection. To overcome this limitation, plant breeders had adopted a multi-population improvement strategy, which has been used in rice, arabidopsis, maize and soybean [31–36]. In this study, we used two RIL populations with the same female parent (Dongnong L13) to locate QTLs for forage quality and yield traits (Tables 3 and 5). Ten additive effect QTLs were identified in the RIL3613 population, whereas 18 additive effect QTLs were identified in the RIL6013 population, and the intervals of these QTLs were not duplicated, indicating that the genetic mechanisms of CP, ADF, NDF and DWP in the two populations were different, and the detection efficiency could be improved by using two populations.

Multiple QTL statistical methods can improve the detection efficiency

Various QTL mapping methods have been used in research, among which IM and ICIM have been widely used in QTL detection. The main difference between these two statistical methods is the different statistical models used. IM does not control the background error in the analysis of inferred QTL, while ICIM method uses adjacent molecular markers to control the background error. Therefore, the results of QTLS mapping will differ. In this study, two different statistical methods were used to identify QTLs. A total of 100 and 15 QTLs were identified by ICIM and IM, respectively. Among them, 5 QTLs (qCP-Gm19-2, qCP-Gm16-2, qCP-Gm19-1, qNDF-Gm02-1 and qNDF-Gm03-1) were identified by both statistical methods, whereas the other QTLs were identified by a single method. Similarly, Ning et al. (2018) used single marker analysis (SMA), composite interval mapping (CIM) and inclusive composite interval mapping (ICIM) methods to identify QTLs of vertical distribution of pod number, and detected different QTLs with different methods [25]. Therefore, it can be seen that multiple methods can verify the accuracy of QTL detection, and can also increase the detection efficiency of QTLs.

Multi-environment experiment could improve the detection efficiency of QTL for forage quality and yield

The quality of soybean forage is easily affected by environmental conditions[1]. This study found that in addition to forage quality, DWP was also easily affected by environmental conditions (Table 1). In order to overcome the influence of the environment on the results of QTL mapping, identify the stable expression of QTLs in different environments, and accurately estimate the location and function of QTLs, it is necessary to conduct analysis in multiple environments over many years. There are two methods to analyze the results of multi-environment experiments. One is to analyze the results of a single environment alone. The other is to combine the data collected from multiple environments [25, 37–40]. Compared with the single environmental analysis method, the statistical method combining multiple environments can accurately estimate the error, improve the detection ability of QTLs, and estimate the location and effect of QTLs more accurately [41]. In this study, both single environment (BIP function) and multiple environment (MET function) methods were used to map QTLs for three forage quality traits and one yield trait. Fourteen additive effect QTLs and nine epistatic effect QTL pairs were identified in the single-environment genetic effect model (BIP) and 22 additive effect QTLs and 50 epistatic effect QTL pairs were identified in the multi-environment QTLs × environment model (MET), of which only eight additive effect QTLs were detected by both methods. These results indicate that the combination of the two methods can improve the efficiency of QTL localization.

Quantitative traits are controlled by both genetic factors and environmental influences. The sensitivity of a certain QTL effect to the environment can be evaluated by comparing the phenotypic contribution rate of QTL main effect and the phenotypic contribution rate of genotype × environmental interaction effect. In this study, the additive phenotypic contribution rates of 18 additive effect QTLs were higher than that of additive × environmental interaction effects, and the phenotypic contribution rates of 50 epistatic effect QTL pairs were higher than that of epistatic × environmental interaction effects (Tables 5 and 6). These QTLs showed stability in multiple environments, indicating that their effects could be expressed stably in multiple environments, which is of great significance for breeding with multiple environment stability. On the contrary, the phenotypic contribution rate of additive × environment interaction was higher than that of additive effect in four QTLs, and there were no epistatic QTL pairs in which the phenotypic contribution rate of epistatic × environment interaction was higher than that of epistatic effect in this study. In addition, 11 QTLs were even only identified in a single environment (Table 3). The QTLs that are only suitable for a single environment, are environment-specific QTLs, and have practical significance for breeding varieties in specific environments.

The epistasis effect QTL is an important genetic basis for forage quality and yield traits

The epistasis of QTL × QTL and its interaction with the environment play an important role in the genetic research of quantitative traits, and it is an important component of quantitative trait inheritance [42]. It is of great significance to study the epistasis effect of complex quantitative traits for genetic breeding. Asekova et al. [1] only studied the additive effect of QTLs, but did not study the epistatic effect of QTLs. Laurie et al. [43] divided epistatic effect QTL pairs into three types: significantly additive QTL × significantly additive QTL, significantly additive QTL × non-significantly additive QTL, and non-significantly additive QTL × non-significantly additive QTL according to the two QTLs interacting with each other. In this study, we mapped a total of 59 epistatic effect QTL pairs by the epistatic effect model, including 4 CP QTL pairs, 11 ADF QTL pairs, 38 NDF QTL pairs and 6 DWP QTL pairs. A total of 64 QTLs were identified. Among them, 6 QTLs were identified as significant additive QTLs by the additive effect model, whereas the other 58 QTLs were non-significant additive QTLs. Therefore, through epistatic effect analysis, QTL locus that could not be identified by additive effect model could be found, providing genetic basis for further exploration of forage quality and yield traits. Therefore, QTL analysis of additive effect and epistatic effect should be carried out simultaneously when analyzing the genetic basis of forage quality and yield traits, which can improve the efficiency of molecular assisted selection and genetic analysis.

Breeding value of co-localization with QTLs

Some QTLs are localized to the same genomic region, and this pleiotropy is common in most QTLs [44]. Most QTLs are aggregated, which is probably the result of close association of multiple genes. It is an important problem in genetic breeding to analyze whether a single gene affects multiple traits simultaneously. If pleiotropy is caused by the tight linkage of multiple genes, then the beneficial genes can be exploited by breaking the linkage. However, if a single effect controls multiple traits, it may not be possible to break the link between them because it carries some negative effects at the same time, thus limiting its effective use. In this study, some QTLs controlling multiple traits were identified. For example, in the Gm01 linkage group, qADF-Gm01-1, qNDF-Gm01-1 and qDWP-Gm01-1 share a genomic interval (satt482-Sat_160). In the Gm02 linkage group, the gene interval staga002-sat_289 affects CP, ADF, NDF and DWP simultaneously. In the Gm11 linkage group, the gene interval satt197-sat272 affects CP, ADF and DWP simultaneously. In the Gm18 linkage group, qADF-Gm18-1 and qNDF-Gm18-2 overlap in position and coexist in the intergenic region (Satt610-Satt570). The QTLs controlling ADF and NDF in satt482-Sat160, staga002-sat_289 and satt197-sat272 are all negative additive effects, and CP are all positive additive effects. The alleles of the female parent Dongnong L13 promoted simultaneous reduction of ADF and NDF and increased CP. Fibre is an important forage component that affects animal digestibility and energy utilization. An increase in fibre content will reduce the digestibility of the forage, ultimately leading to a decrease in energy content. Fibre content is usually determined by measuring NDF and ADF [45, 46]. With the increase of ADF, the digestibility of forage decreased [47]. Cell wall digestibility of grasses and legumes is often limited by the content and composition of lignin components. Since digestibility is negatively correlated with cell wall lignin and fibre content, reducing fibre content can help develop feed crops with higher digestibility [48]. Since the protein content corresponds to the digestible portion of dry matter in hay [49], increasing the crude protein content should improve the digestibility of hay. Therefore, using the gene regions of QTLs above can better achieve the purpose of improving the nutritional quality of forage soybean.

Comparison with previous studies

Comparing the QTLs identified in this study with the results of previous studies, it was found that the genetic regions of qDWP-Gm01-1 and qDWP-Gm02-4 overlap with the genetic interval of QTLs previously discovered by Zhang et al. [50]. The genetic interval of qDWP-Gm11-1 overlaps with the genetic interval of QTLs mapped by Zhang et al. [50] and Liang et al. [20]. The genetic interval of qCP-Gm19-3 overlaps with the QTL previously discovered by Asekova et al. [1]. The comparison with previous studies shows the authenticity of these QTLs. The other QTLs in this study are all reported for the first time and need further study in the future.

Conclusions

Eight CPs, eight ADFs, nine NDF and three DWP additive effect QTLs were identified in two soybean populations. Four epistatic effect QTL pairs for CP, 12 for ADF, 38 for NDF and 8 for DWP were also identified. Four QTLs were consistent with previous studies, and the rest were newly discovered QTLs. These results proved theoretical and practical significance for future maker assisted selection.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability

The data that support the findings reported in this study are available from the corresponding author upon reasonable request.

Additional information

Funding

This research was financially supported by Hundredthousand and Million Project of Heilongjiang Province for Engineering and Technology Science (2019ZX16B01) to Hailong Ning.