Phenotypic performance
Obvious difference was presented between P1 and P2 (Fig. 1). Concerning the SD-related traits, P1 presented significant shorter MIL, more SNMI and denser SDMI (averagely 64.9 cm, 132.4 siliques, and 2.1 siliques/cm) than P2 (averagely 75.4 cm, 79.8 siliques, and 1.1 siliques/cm respectively) (Table 1). All the three traits in the F2 and the DH population presented continuous distributions, suggesting the quantitative trait nature of SDMI, SNMI and MIL. The correlation coefficient between SDMI and SNMI was 0.59 ~ 0.66 (p < 0.01) in different environments, being higher than that between SDMI and MIL (-0.25~-0.36, p < 0.01), implying that silique number contributes more to silique density trait. Broad sense heritability (H2 B) of 89%, 89% and 85% was detected for SDMI, SNMI and MIL, nevertheless, great variations (CV values) were detected in unsegregated populations (P1, P2 and F1) for all the three traits particular SNMI and MIL. Moreover, significant environment effects and genotype×environment effects were detected from the DH population (Online Resource 2). These data suggested that SDMI, SNMI and MIL are all complex quantitative traits which are easily influenced by environment.
Table 1
SDMI, SNMI and MIL among different populations under multiple environments
Population
|
SDMI (silique/cm)
|
SNMI
|
MIL (cm)
|
Range
|
Mean ± SD
|
CV(%)
|
Range
|
Mean ± SD
|
CV(%)
|
Range
|
Mean ± SD
|
CV(%)
|
Environment A
|
|
|
|
|
|
|
|
P1
|
2.1–2.9
|
2.2 ± 0.2
|
12.6
|
103–178
|
131.8 ± 23.5
|
17.8
|
50–88
|
60.3 ± 12.5
|
20.8
|
P2
|
0.8–1.4
|
1.1 ± 0.1
|
13.0
|
45–114
|
84.8 ± 17.0
|
20.1
|
53–101
|
79.5 ± 12.5
|
15.7
|
F1
|
1.3–2.2
|
1.6 ± 0.2
|
12.9
|
86–181
|
128.0 ± 26.4
|
20.6
|
57–112
|
81.8 ± 14.6
|
17.8
|
F2
|
0.7-2.0
|
1.4 ± 0.2
|
14.8
|
41–127
|
92.2 ± 17.8
|
19.3
|
31–102
|
68.1 ± 11.1
|
16.3
|
BC1P1
|
1.1–2.7
|
1.6 ± 0.3
|
17.3
|
54–162
|
105.3 ± 23.9
|
22.7
|
33–88
|
68.4 ± 14.0
|
20.4
|
BC1P2
|
0.8–1.7
|
1.3 ± 0.2
|
17.8
|
62–144
|
93.0 ± 17.5
|
18.8
|
45–110
|
73.3 ± 12.1
|
16.5
|
DH
|
0.6–1.9
|
1.3 ± 0.3
|
19.8
|
37–155
|
90.4 ± 20.9
|
23.1
|
37–95
|
68.3 ± 10.7
|
15.7
|
Environment B
|
|
|
|
|
|
|
P1
|
1.9–2.5
|
2.0 ± 0.2
|
11.3
|
88–189
|
132.9 ± 26.5
|
20.0
|
44–100
|
69.6 ± 14.7
|
21.1
|
P2
|
0.9–1.3
|
1.1 ± 0.1
|
10.5
|
53–90
|
74.8 ± 9.0
|
12.1
|
51–84
|
71.3 ± 8.7
|
12.3
|
F1
|
1.3–1.9
|
1.6 ± 0.2
|
11.0
|
79–192
|
137.1 ± 25.6
|
18.7
|
69–110
|
83.9 ± 10.9
|
13.0
|
F2
|
1.1–2.7
|
1.8 ± 0.3
|
17.2
|
67–197
|
115.4 ± 26.7
|
23.1
|
37–100
|
62.9 ± 12.7
|
20.2
|
BC1P1
|
1.3–2.8
|
2.0 ± 0.3
|
16.1
|
72–220
|
126.3 ± 29.3
|
23.2
|
36–90
|
64.6 ± 12.3
|
19.0
|
BC1P2
|
0.9–2.3
|
1.4 ± 0.3
|
18.2
|
45–144
|
101.1 ± 21.7
|
21.4
|
40–100
|
74.6 ± 16.3
|
21.9
|
DH
|
0.8-2.0
|
1.4 ± 0.3
|
18.5
|
32–148
|
86.5 ± 19.1
|
22.1
|
26–84
|
62.7 ± 10.5
|
16.8
|
Environment C
|
|
|
|
|
|
|
|
|
DH
|
0.7-2.0
|
1.4 ± 0.3
|
20.1
|
50–155
|
92.4 ± 20.4
|
22.1
|
43–91
|
68.6 ± 10.9
|
15.8
|
Genetic model of SDMI
Genetic inheritance models for SDMI were calculated based on the phenotypic data of P1, P2, F1, F2, BC1P1 and BC1P2 populations. A mixed major gene plus polygene inheritance model was used for the goodness-of-fit test of each model. The PG-ADI and MX1-AD-ADI models produced the highest likelihood and the lowest AIC values under two environments (Online Resource 3), and presented no significant parameter in the goodness-of-fit test (Online Resource 4), being in conformity with the optimal model conditions. This result further proved that the SDMI is a complex trait which is possibly controlled by multi minor genes with additive-dominant-epistatic effects (PG-ADI) or by one additive-dominance major gene plus additive-dominance-epitastic minor genes (MX1-AD-ADI).
RAD-seq data and genetic map
An average of 0.99 G clean data (over 10× of rapeseed genome) was yielded for each sequenced line, with GC content of 41.2% and Q30 > 90.4%. In range, 94.6%~99.3% of the clean reads were aligned to the reference genome of rapeseed (Online Resource 5). In total, 2,521 SNPs and 2,657 Indels were detected from the DH population, among which 2,388 polymorphic loci were subjected to the genetic map construction. In final, 2,028 loci were allocated into 19 linkage groups (LGs), with the fewest on LG A10 (33 loci) and the most number on LG C03 (291 loci) (Online Resource 6). The lengths of LGs ranged from 120.68 to 179.13 cM, summed in a total length of 2,801.63 cM, with an average distance of 1.38 cM between neighboring loci.
QTLs for SDMI, SNMI and MIL
A total of eight, 14 and three QTLs were identified for SDMI, SNMI and MIL under three environments, respectively, individually explaining 7.9–32.7% of the phenotypic variation (Table 2, Online Resource 7). The QTLs for SDMI located on LG A5, C3 and C6, explaining 9.14–17.67% of the phenotypic variation. No overlap was detected for QTLs for MIL with other two traits, whereas, overlaps were detected between QTLs for SDMI and SNMI. For example, three overlapped QTLs for SDMI identified from three environments (qSDMIC6.1, qSDMIC6.2 and qSDMIC6.3b) were found from the region 62.5–75.4 cM on LG C06 (16.1–27.3 Mb on chromosome C06), individually explaining 11.3–17.7% of the total phenotypic variation; this region also located three QTLs for SNMI (qSNMIC6.1b, qSNMIC6.2a and qSNMIC6.3c), explaining 8.6–13.1% of the total variation (Table 2). It was interesting that one adjacent region on LG C06, i.e., 55.7–62.6 cM (11.6–16.1 Mb), was also identified as the confidence interval of one QTL for SDMI (qSDMIC6.3a) and two QTLs for SNMI (qSNMIC6.1a and qSNMIC6.3b). The data suggest that the QTL regions on LG C06 possibly contain important genetic components affecting both SDMI and SNMI.
Table 2
QTLs detected for SDMI, SNMI and MIL under three environments
QTL
|
Environment
|
Linkage
group
|
Peak position (cM)
|
LOD
value
|
Confidence interval (cM)
|
Physical position (Mb)
|
Additive effect
|
R2
(%)
|
SDMI
|
|
|
|
|
|
|
|
|
qSDMIC6.1
|
A
|
C06
|
62.61
|
6.22
|
62.5–67.8
|
16.10–16.90
|
0.19
|
17.67
|
qSDMIC3.2a
|
B
|
C03
|
88.91
|
4.08
|
85.1–89.9
|
58.40-60.31
|
0.31
|
11.58
|
qSDMIC3.2b
|
B
|
C03
|
104.61
|
4.01
|
104.2-108.6
|
42.39–47.09
|
-0.31
|
10.84
|
qSDMIC6.2
|
B
|
C06
|
70.81
|
4.18
|
62.6–75.4
|
16.10-27.26
|
0.25
|
11.26
|
qSDMIA5.3
|
C
|
A05
|
84.11
|
3.38
|
77.4–95
|
12.39–19.72
|
0.18
|
9.14
|
qSDMIC6.3a
|
C
|
C06
|
58.71
|
3.82
|
57.2–62.5
|
13.10–16.10
|
0.13
|
11.50
|
qSDMIC6.3b
|
C
|
C06
|
68.21
|
4.18
|
62.5–73.6
|
16.10-27.22
|
0.12
|
11.55
|
qSDMIC6.3c
|
C
|
C06
|
93.51
|
3.87
|
88-111.1
|
34.77–44.85
|
0.12
|
10.70
|
SNMI
|
|
|
|
|
|
|
|
|
qSNMIA5.1
|
A
|
A05
|
22.61
|
3.53
|
13.4–33.5
|
27.28–30.77
|
-8.58
|
9.50
|
qSNMIC5.1b
|
A
|
C05
|
106.91
|
3.04
|
101.3-119.6
|
6.54–14.98
|
-6.82
|
7.88
|
qSNMIC6.1a
|
A
|
C06
|
57.71
|
3.21
|
56.7–62.6
|
13.10–16.90
|
10.18
|
8.39
|
qSNMIC6.1b
|
A
|
C06
|
68.21
|
3.28
|
62.6–75.3
|
16.90-27.26
|
9.60
|
8.57
|
qSNMIA10.2
|
B
|
A10
|
90.51
|
4.58
|
78.2-103.3
|
3.75–9.26
|
32.83
|
32.72
|
qSNMIC6.2a
|
B
|
C06
|
70.81
|
4.94
|
63-74.8
|
16.10-27.26
|
16.26
|
12.18
|
qSNMIC6.2b
|
B
|
C06
|
80.41
|
4.63
|
74.8–85.4
|
27.22–34.77
|
15.65
|
11.59
|
qSNMIC8.2
|
B
|
C08
|
58.71
|
4.53
|
48.7–63.1
|
23.34–29.26
|
14.59
|
11.19
|
qSNMIA6.3
|
C
|
A06
|
53.71
|
6.46
|
50.7–59.7
|
9.68–14.38
|
11.51
|
17.59
|
qSNMIC6.3a
|
C
|
C06
|
49.61
|
5.10
|
40.6–51.1
|
9.07–10.79
|
8.89
|
14.33
|
qSNMIC6.3b
|
C
|
C06
|
57.71
|
6.04
|
55.7–60.7
|
11.55–13.50
|
10.25
|
16.46
|
qSNMIC6.3c
|
C
|
C06
|
67.61
|
4.52
|
61.5–67.9
|
13.50–16.90
|
8.72
|
13.05
|
qSNMIC7.3a
|
C
|
C07
|
77.81
|
3.43
|
76.1–78.8
|
24.08–27.83
|
-6.07
|
9.57
|
qSNMIC7.3b
|
C
|
C07
|
86.31
|
5.26
|
81.6–87.7
|
16.30-20.01
|
-7.43
|
14.09
|
MIL
|
|
|
|
|
|
|
|
|
qMILA8.1
|
A
|
A08
|
15.01
|
5.98
|
6.8–19.3
|
9.61–17.73
|
-7.49
|
17.37
|
qMILA9.1a
|
A
|
A09
|
80.41
|
3.75
|
73.1–87.3
|
21.34–34.89
|
5.44
|
11.27
|
qMILC2.2
|
B
|
C02
|
138.41
|
3.75
|
133.6-145.5
|
20.44–27.93
|
10.06
|
11.86
|
In order to confirm and narrow the important genetic region for SDMI on C06, two bulks with extreme phenotypic performance were constructed from the DH population and subjected to deep genomic resequencing. Over 23.1 Gb data and 3.1 million SNPs/Indels were generated from each pool. Among these, 4996 SNPs/Indels on chromosome C06 were polymorphic between two pools. After calculating Δ(SNP-index) of each locus and visualizing the Δ(SNP-index) trends by sliding window method, a peak with higher value than the threshold 0.697 (p < 0.05) was identified on chromosome C06. The significant interval started from 25.93Mb and ended at 26.13 Mb (0.15 Mb), being located within one of the QTL regions detected before (16.1–27.3 Mb) (Fig. 2), locating with 18 genes in total.
Expression of genes in QTL region
In order to identify the candidate genes, RNA-seq was carried out between the low and high SDMI bulks, yielding more than 19 million clean reads for each sample. Over 87% of these were aligned to the reference genome of B. napus and a total of 9,469 genes exhibited more than two-fold transcriptional differences between two samples. Of these, only two genes from the aforementioned 0.15 Mb interval on chromosome C06 showed significant differential expression levels, including one auxin-related gene involved in organ size (BnARGOS) and one gene encoding methylenetetrahydrofolate reductase 1 (BnMTHFR1) (Online Resource 8). Both genes exhibited 2.2-fold higher expression in the bulk with high SDMI. As revealed by qRT-PCR, BnARGOS showed significant higher expression level in five DH lines with high SDMI as compared with the level in four DH lines with low SDMI, whereas BnMTHFR1 seemed have no obvious association with the phenotypic variance (Fig. 3).