Skip to main content

Systematic trait dissection in oilseed rape provides a comprehensive view, further insight, and exact roadmap for yield determination

Abstract

Background

Yield is the most important and complex trait that is influenced by numerous relevant traits with very complicated interrelations. While there are a large number of studies on the phenotypic relationship and genetic basis of yield traits, systematic studies with further dissection focusing on yield are limited. Therefore, there is still lack of a comprehensive and in-depth understanding of the determination of yield.

Results

In this study, yield was systematically dissected at the phenotypic, genetic to molecular levels in oilseed rape (Brassica napus L.). The analysis of correlation, network, and principal component for 21 traits in BnaZN-RIL population showed that yield was determined by a complex trait network with key contributors. The analysis of the constructed high-density single nucleotide polymorphism (SNP) linkage map revealed the concentrated distribution of distorted and heterozygous markers, likely due to selection on genes controlling the growth period and yield heterosis. A total of 134 consensus quantitative trait loci (QTL) were identified for 21 traits, of which all were incorporated into an interconnecting QTL network with dozens of hub-QTL. Four representative hub-QTL were further dissected to the target or candidate genes that governed the causal relationships between the relevant traits.

Conclusions

The highly consistent results at the phenotypic, genetic, and molecular dissecting demonstrated that yield was determined by a multilayer composite network that involved numerous traits and genes showing complex up/down-stream and positive/negative regulation. This provides a systematic view, further insight, and exact roadmap for yield determination, which represents a significant advance toward the understanding and dissection of complex traits.

Background

Yield, which usually refers to biomass or seed yield, is the most important trait of crops, such as wheat, rice, maize, soybean, peanut, cotton, and oilseed rape [1]. For example, crop straw can be used to generate heat and electricity [2]. The grains of cereal crops (including wheat, rice, maize, etc.) are the main source of starch (the first nutrient for human), and can also be used to produce bioethanol [3]. The seeds of rapeseed, soybean, and other oil crops can be supplied for producing edible oil and biodiesel [4]. With the rapid increase in the global population, there is an urgent need to increase crop yield to meet the human demand for food and energy [5]. However, yield is also the most complex trait, as it is a composite outcome of numerous contributing traits, as well as their interactions [6]. Specifically, yield is directly determined by its multiple components with a trade-off effect between them, e.g., seed number and size [7]. This means that a change in one component for yield often causes a change in other components in an opposite direction. The trade-off among yield components is generally explained by the competition among sinks (negative feedback) due to limited resources [8]. In addition, yield is indirectly affected by numerous yield-related traits in either a positive or negative direction through undetermined mechanisms [9]. These may include growth period (e.g., flowering and maturity time), plant architecture (e.g., plant height and branch number), and resistance to biotic (e.g., disease, pest, and weed) and abiotic (e.g., drought/water logging and hot/cold) stress. Therefore, characterizing the complex relationships between yield and its components or related traits is the key to understanding what and how yield is determined. This is important not only for the evolution and physiology of plants [10, 11], but also for crop genetics and breeding [12].

Previous studies have revealed the phenotypic correlation between yield and its components or related traits in various crops, such as rice [13], wheat [14], maize [15], soybean [16], peanut [17], and rapeseed [1]. Recently, a large number of studies have identified the underlying QTL for yield, and genome-wide QTL co-localization between yield and its components or related traits was found in crops, such as rice [18, 19], wheat [20], maize [15, 21, 22], soybean [16, 22, 23], cotton [24, 25], and rapeseed [9, 26]. However, the underlying genetic basis for phenotypic correlation and QTL clustering between yield traits is basically unclear. Theoretically, the phenotypic correlation and QTL colocalization between traits mechanistically result from either genetic linkage or pleiotropy [27]. Genetic linkage means that the genes for different traits are physically close to each other (Fig. 1A). Pleiotropy refers to the effect of a locus on two or more traits (Fig. 1B). In addition, pleiotropy may be due to physiological interactions among traits in which one trait acts at “upstream” of another (Fig. 1C). Although a few of these QTL clusters for yield traits have been further dissected to specific loci/genes [19, 28, 29], the exact relationship between the relevant traits has not been characterized.

Fig. 1
figure 1

Several putative models underlying the co-localization of QTL for two traits. The first and second models (A and B) are tightly linked and generally called pleiotropy, which are easily understood. The third (C) is physiological interaction: one gene can indirectly affect a trait by regulating another trait

In summary, although a large number of studies have reported the phenotypic relationship and genetic basis of yield traits, systematic research with further dissection focusing on yield is rarely performed. Therefore, there is still a lack of a comprehensive and in-depth understanding of the determination of yield. Oliseed rape is an important crop widely planted around the world for multiple purposes, including oil, fodder, and food [30]. With the successful application of catalysts in the production of biodiesel from vegetable oil [31], rapeseed also serves as a main biodiesel resource in Europe [32]. With the increasing demand for edible vegetable oil and biofuel, it is urgent to improve the seed yield of oilseed rape [33]. In the current study, taking oilseed rape as an example, yield was systematically dissected at the phenotypic, genetic, and molecular levels, with an emphasis on its complex relationship with other contributing traits. In particular, four representative hub-QTL clusters were dissected to specific loci/genes, and the causal trait relationship was revealed for the first time. The phenotypic, genetic, and molecular dissecting results were highly consistent, which provided comprehensive and further insights into yield determination.

Results

Analysis of phenotypic relationships revealed an integrated trait network with key factors to determine yield

In the BnaZN-RIL population and its two parents that were planted in six environments, 21 yield-related traits were investigated. The two parents Zhongshuang11 and No.73290 showed significant differences for 19 out of the 21 investigated traits in at least one environment (Additional file 8: Table S1). The phenotypic values of the BnaZN-RIL population showed normal or near-normal distribution for all 21 traits across the six environments (Additional file 1: Figure S1). Analysis of variance showed that genotype, environment, and their interaction had significant effects on all 21 investigated traits (Additional file 9: Table S2).

Among the 210 trait pairs, 198 (94.3%) showed a significant correlation in at least one environment and 158 (75.2%) were significantly correlated in multiple environments (Additional file 2: Figure S2; Additional file 10: Table S3). To obtain a general picture of the trait relationship, a trait network map was constructed using the correlation that was significant in more than half of the investigated environments (Fig. 2A), which displayed several obvious characteristics. First, all 21 investigated traits were woven into a complex network of interconnections, and none was independent. Second, seed yield was located at the center of this network, followed by yield components, whereas other yield-related traits were on the periphery. Third, seed yield showed a higher correlation with yield components than with other yield-related traits. In addition, yield-related traits usually showed a higher correlation with yield components than with yield itself, suggesting their indirect relationship with seed yield.

Fig. 2
figure 2

Trait relationship revealed by trait network and PCA analysis of 21 traits investigated in the BnaZN-RIL population. A: Trait network constructed by the 149 trait pairs showing significant correlation in at least half of the investigated environments. The seed yield and its components and related traits are shown in different colors. The traits are treated as nodes, which are linked with liens of different types (solid: + ; dotted: -) and widths (distinguishing the degree). BD: Loading of PC1 vs PC2, PC1 vs PC3, and PC2 vs PC3. The different traits are represented by dots of different colors as demonstrated in the legends. Circle A, B, and C represent the different groups

Although the results of principal component analysis (PCA) in Wuhan and Zhengzhou showed a slight difference, high consistency was found between different years at the same location (Additional file 11: Table S4). Therefore, the best linear unbiased prediction (BLUP) value for six environments was subjected to PCA (Table 1; Fig. 2B–D). The first principal component 1 (PC1) accounted for 24.2% of the trait variance. Among the seven traits (pod number of branch raceme (PNb), pod number of whole plant (PNw), seed yield of branch raceme (SYb), primary branch number (PBN), seed yield of whole plant (SYw), main inflorescence length (MIL), pod number of main raceme (PNm)) with negative loading, PNb and PNw showed a high value, which suggested that seed yield was primarily determined by pod number. Of the other 14 traits with positive loading, five growth period traits (budding time (BuT), flowering time (FlT), flowering end time (FET), bolting time (BoT), maturity time (MaT)) had the highest value. The second principal component (PC2) accounted for 16.3% of the trait variance. Among the eight traits with negative loading, the seed yield of the main raceme (SYm), seed number per pod (SNPP), and seed oil content (OIL) showed the highest values, which indicated that seed yield was secondarily determined by seed number per pod. Of the other 13 traits with positive loading, PRO, FET, PNw, and PNb had a high value, which was in accordance with the significant negative correlation between seed number per pod and pod number as well as between oil content and protein content (Additional file 2: Figure S2; Additional file 10: Table S3). The third principal component (PC3) accounted for 15.0% of the total variance. Of the six traits with negative loading, thousand-seed weight (TSW), MIL, and PRO showed a high value, which was in accordance with a positive correlation between seed weight and protein content in the current study (Additional file 2: Figure S2; Additional file 10: Table S3) and previous research [34]. Of the other 15 traits with positive loading, SYw, SYb, PBN, and SNPP exhibited a high value, which was in accordance with the negative correlation between seed weight and seed number per pod, branch number, and seed yield (Additional file 2: Figure S2; Additional file 10: Table S3).

Table 1 Principal component analysis of the 21 traits based on the BLUE (best linear unbiased estimation) value in six environments

Analysis of a high-density genetic map revealed a concentrated distribution of distorted segregation, residual heterozygosity, and variation in recombination frequency

To further dissect the genetic relationship between yield traits, a high-density genetic linkage map of 2207.7 cM and 1887 bins/6444 SNP markers was constructed for the BnaZN-RIL population, which covered 812.1 Mb physical distance representing 84.5% of the assembled genome of Zhongshuang11 (Fig. 3; Table 2). It should be noted that the recombination frequencies (2.92 to 5.93) of the 10 linkage groups in the A subgenome were all higher than those (1.95 to 2.75) for the 9 linkage groups in the C subgenome, with a mean of 2.72 per Mb.

Fig. 3
figure 3

Genetic linkage map constructed using the BnaZN-RIL population. The names of each of the 19 linkage groups are shown on the top and the black oval indicates the positions of the centromere. The serial numbers of mapped bins and their genetic distances are shown on the left and right sides of each linkage group, respectively. The names of bin markers with high heterozygosity and segregation distortion are highlighted using italics and color, of which red and green distinguish the skew from Zhongshuang11 and No.73290, respectively

Table 2 The summary statistics of the BnaZN-RIL linkage map

Notably, 298 bins (15.8%) displayed extremely significant segregation distortion, which tended to cluster especially at the end of linkage groups (Fig. 3; Table 2). Interestingly, dozens of markers were found to have high residual heterozygosity (ranging from 25.6 to 60.2%) in the BnaZN-RIL linkage map, most of which were concentrated at the ends of A01, A03, C01, and C02. Further single marker analysis showed that many of these markers in linkage groups A03 and C02 displayed an overdominant effect on pod number, the most important principal component of seed yield (Additional file 12: Table S5).

The constructed BnaZN-RIL genetic linkage map aligned well with the genomic map of Zhongshuang11 (Fig. 4), demonstrating its high quality. Generally, the genetic distance increased with physical distance, with a more rapid increase at both ends of the chromosomes than in the middle part, showing an S-shaped curve. There were obvious breakpoints (no markers in a large genome segment) in the alignment map, basically corresponding to the centromeric region. Interestingly, no SNP marker was located in the centromeric region, and markers flanking centromeric regions were basically monomorphic (Additional file 13: Table S6). The recombination frequencies in pericentromeric regions ranged from 0.02 (A05) to 1.04 (A02) with a mean of 0.28 per Mb, which was much lower than the corresponding mean (2.72 per Mb) calculated from the whole genome.

Fig. 4
figure 4

Alignment of the genetic map of the BnaZN-RIL population and the genomic map of Zhongshuang11. The horizontal and vertical axes show the genetic distances (cM) and physical positions (Mb), respectively. The scatter dots indicate the genetic positions of the mapped SNP markers on the genetic map of the BnaZN-RIL population and their physical positions on the pseudochromosomes (A01-A10; C01-C09) of the reference genome Zhongshuang11. The determination coefficients of the genetic and physical distances of these SNP markers are also shown on each chart

Analysis of genetic relationships revealed an integrated trait-QTL network with hub-QTL to control yield

To further dissect the genetic determination of yield, large-scale QTL mapping was conducted using the abovementioned phenotypic data and high-density linkage map of the BnaZN-RIL population. At the significance level of P = 0.05, a total of 207 QTL were identified for the 21 traits investigated in six environments, which explained 4.0–48.3% of the variance (Additional file 14: Table S7A). After the integration of overlapping identified QTL trait by trait in different environments, a total of 134 consensus QTL were obtained (Additional file 14: Table S7B). Of these, 19, 32, and 84 were for seed yield, yield component, and related traits, respectively.

Interestingly, the consensus QTL were clustered rather than randomly distributed across the genome (Fig. 5), which might explain the extensive correlation among these traits (Additional file 2: Figure S2). Then, 106 of the 134 consensus QTL were combined into 28 QTL clusters (Additional file 14: Table S7C), which might be caused by pleiotropy or tight linkage. Among the 19 consensus QTL for seed yield, 16 overlapped with the QTL for other traits (Additional file 14: Table S7C), highly in accordance with the extensive correlation between yield and other investigated traits (Additional file 2: Figure S2; Additional file 10: Table S3). Statistical analysis of the number/proportion and direction of these overlapping QTL revealed some obvious characteristics (Additional file 15: Table S8). First, the directions of these overlapping QTL between seed yield and other yield components or related traits were same rather than opposite. Second, the directions of almost all overlapping QTL of the five growth period traits were same, indicating that pleiotropy rather than tight linkage was more likely to be the underlying genetic basis.

Fig. 5
figure 5

Distribution of consensus QTL in the Zhongshuang11 genome. The 19 chromosomes are drawn as cylinders, with their names (A01-A10 and C01-C09) shown on the top. The positions of each QTL are indicated by thin horizontal lines

It should be noted that four QTL clusters (QC4, QC8, QC14, QC17) contained many consensus QTL with large effects (Additional file 14: Table S7C), which might play an important role in regulating traits. Of the 14 QTL involved in QC4 at the top of chromosome A2, eight showed reproducible and large effects, including three growth period traits (+ , 25.4–48.3%), seed yield (+ , 10.9%), PN (+ , 23.3%), PBN (+ , 13.0%), protein (−, 18.8%) and oil content (+ , 11.1%). QC8 on the lower part of chromosome A06 showed reproducible and large effects on SNPP (+ , 25.4%), PN (−, 17.7%), and pod length (PL) (+ , 8.8%), but a moderate effect on other traits including seed weight (−, 9.9%), pod density (PD) (−, 9.1%), FIT (−, 6.5%), and plant height (PH) (−, 12.4%). QC14 on the lower part of chromosome A09 showed reproducible and large effects on PL (+ , 18.8%) and seed weight (+ , 11.9%), and moderate effects on six other traits including PN (−, 8.2%), FIT (−, 4.1%), MaT (−, 13.5%), main stem length (MIL)(−, 7.0%), and primary branch height (PBH) (−, 11.0%). QC17 on the bottom of chromosome C02 had producible and large effects on growth period traits (−, 7.0%−16.4%), with a moderate effect on seed yield (−, 8.3%), PBN (−, 6.0%), and protein content (−, 6.4%).

To further link the phenotypic and genetic relationship between yield and its components or related traits, an integrated trait–QTL network was constructed (Fig. 6), which displayed several obvious features. Firstly, all 134 consensus QTL for the 21 traits were integrated into an interconnected network, none of which was independent. This indicated the extensive relationships between these traits, which might be caused by the multiple/pleiotropic roles of the underlying QTL. Secondly, there were several obvious hub-QTL that were linked with multiple traits and displayed large effects, which might play a major role in the trait–QTL network and are worthy of further study. Thirdly, most of these hub-QTL clusters had smaller effects on seed yield than their components or related traits, which indicated their pleiotropy and indirect effects on yield.

Fig. 6
figure 6

Trait–QTL network constructed for all of the investigated traits and the underlying consensus QTL. Traits and consensus QTL are treated as nodes, which are drawn using the hexagon and circle, respectively. The seed yield, yield components and yield-related traits are distinguished by the different colors. The additive-effect direction of these consensus QTL is distinguished by the lines of different endpoint types (+ : arrow; -: vertical line). The abbreviations of trait names are the same as those in Table 1, and the width of the lines indicates the R2 of the consensus QTL

Further dissection of four representative hub-QTL revealed the causal trait relationship and underlying target or candidate genes in yield determination

To further dissect the complex trait relationship within yield determination at molecular level, four representative hub-QTL were selected to construct high-generation near isogenic lines (NILs) (Additional file 3: Figure S3) for accurate evaluation of their phenotypic effects and trait relationships as well as fine-mapping and identification of underlying genes.

BnaA2.FLC indirectly affected yield through influencing three yield components via regulating the growth period

QC4 was narrowed to a 71-kb region between SNP markers seq-new-rs24859 (A02: 1,971 kb) and seq-new-rs32262 (A02: 2,043 kb). Relative to Zhongshuang11 (Table 3), the homologous NIL_QC4 showed the largest decrease in BoT (− 26.1%), followed by BuT, SYb, PNb, SYw, PBH, PNw, and PBN (from − 19.8 to − 14.2%), while SYm, PNm, FlT, PH, and PD showed a moderate reduction (from − 12.0 to − 8.9%), whereas the other eight traits showed only small change (from − 5.0 to 6.2%). These results were generally consistent with their effects on these traits in the preliminary mapping population of BnaZN-RIL (Additional file 14: Table S7C), such as the largest effect on several growth period traits. Further conditional QTL analysis using the NIL segregation population demonstrated a complex up-/down-stream and positive/negative regulation between these traits (Fig. 7A), i.e., QC4 had a large and direct effect on the growth period, which then had indirect and pleiotropic effects on PH ( +), PBN ( +), PN ( +), SNPP ( +), seed weight (−), and the final seed yield ( +). This was understandable because longer vegetative growth generally produces more leaves and biomass, therefore positively correlating with branch and pod number [35]. In the semi-winter growing area of China, oilseed rape cultivars with late maturity often encounter high-temperature ripening, leading to the decreased seed weight due to inadequate seed filling [36].

Table 3 Phenotypic evaluation of the recurrent parent Zhongshuang11 and corresponding NILs for four representative hub-QTL clusters
Fig. 7
figure 7

Further dissection of four representative hub-QTL clusters (AD: QC4, QC8, QC14, QC17). The up-/downstream and positive/negative feedback relationships between the relevant traits were linked with lines of different widths and types, which indicate the size and direction (positive: arrow; negative: vertical line) of the effect, respectively

The fine-mapped region of QC4 contained only 17 annotated genes in the reference genome of Zhongshuang11 (Table S9), only BnaA02G0035100ZS(BnaA2.FLC) was homologous to the known flowering time gene FLC (AT5G10140) in Arabidopsis. BnaA2.FLC had been previously identified to be the causal [37] or candidate [38] gene for flowering time major QTL in the same genomic region of chromosome A2. In addition, RNA sequencing of the shoot apical meristem at the initial stage of floral bud differentiation showed that the expression of BnaA2.FLC was much higher (56.8-fold) in Zhongshuang11 than in No.73290 (Additional file 17: Table S10), which was highly in accordance with the positive additive-effect direction of QC4 on growth period traits (Additional file 14: Table S7). Further sequence analysis showed that there was a 10-bp insertion in the core 40-bp motif of the promoter in No.73290 (Additional file 4: Figure S4), which might decrease its expression. These results highly supported that BnaA2.FLC was the target gene of QC4.

QC8 directly affected yield by regulating seed number per pod likely through the embryogenesis gene BnaA6.EMB93

QC8 had previously been fine-mapped to a 267-kb region between SSR markers BrSF47-10 and BrSF46-167, using the BC4F2 population and its recombinant progeny [39]. However, no recombinant was found in a small genomic fragment of 103 kb between SSR markers BrSF46-28 and BrSF46-78, even in a very large BC5F2 population of 37,976 plants [40]. Relative to Zhongshuang11 (Table 3), the homologous NIL_QC8 showed the largest decrease in SNPP (− 23.8%), followed by SYb (− 13.4%), SYw (− 12.8%), and SYm (− 11.6%), while TSW (+ 8.4%), PL (− 7.3%), PNm (+ 6.9%), PNw (+ 5.7%), and PNb (+ 5.1%) showed moderate change, and the differences of other 12 traits were not significant. These results were also highly consistent with their effects on these traits in the preliminary mapping population of BnaZN-RIL (Additional file 14: Table S7C), such as the largest effect on seed number per pod. Conditional QTL analysis using the NIL segregation population also revealed the causal relationships between these traits (Fig. 7B), where QC8 showed a direct and large effect on seed number per pod, which then had indirect and pleiotropic effects on PL ( +), PN (−), and seed weight (−). These results were highly consistent with the previous finding that the change in seed number per pod is usually accompanied by a change in fruit length in the same direction (but the reverse is not true), indicating positive feedback between seed setting and fruit growth [41]. In addition, the opposite pleiotropy between SNPP and pod number/seed weight could be explained by the competition among sink organs due to limited resources, which resulted in trade-off/negative feedback between them [39]. However, the moderate negative effects of QC8 on pod number and seed weight could not counteract its large positive effect on seed number per pod. It also had a considerable positive effect on the final seed yield.

Among the 19 genes annotated in the 103-kb region of QC8 (Additional file 16: Table S9), only BnaA06G0400200ZS’s homologue (EMB93/AT2G03050) is involved in known biological processes (such as ovule differentiation and development, fertilization and seed development) related to seed number in Arabidopsis. The EMB93 gene encodes a mitochondrial transcription termination factor that is involved in embryogenesis, whose mutation results in embryo lethality (https://www.arabidopsis.org/servlets/TairObject?id=34053&type=locus). There were 18 SNPs between the coding sequences of BnaA6.EMB93 in Zhongshuang11 and No.73290, only one of them caused amino acid variation that was not in the functional domains (Additional file 5: Figure S5). However, there was an ≈11 kb insertion in the upstream regulatory region of BnaA6.EMB93 in No.73290 but not in Zhongshuang11, which was highly consistent with its decreased expression in the ovules of different stages in NIL_QC8 (Additional file 6: Figure S6). These results suggested that BnaA6.EMB93 was the most likely candidate gene of QC18.

The cytochrome p450 gene BnaA9.CYP78A9 indirectly affected yield through influencing seed weight via regulating pod length and photosynthetic area.

QC14 was successfully delimited to a 90-kb region between SNP markers Bn-A09-p30171993 (A09: 57,344 kb) and Bn-A09-p30260475 (A09: 57,435 kb). Compared to the recurrent parent Zhongshuang11, the homologous NIL_QC14 showed the largest decrease in PL (− 26.1%), followed by SYm (− 19.6%), TSW (− 19.0%), SYw (− 17.6%), and SYb (− 16.6%), whereas the other two yield components and 14 yield-related traits exhibited no significant difference (Table 3). The effects of QC14 in high-generation NILs were highly similar to those in the preliminary mapping population of BnaZN-RIL (Additional file 14: Table S7C), where it had a stable and large effect on pod length, followed by seed weight. Conditional QTL analysis further revealed the causal relationship between these traits (Fig. 7C), where QC14 had a direct effect on pod length, thus had indirect pleiotropic effects on seed weight ( +), and final yield ( +). This is understandable as longer pod generally means a larger photosynthetic area that is able to produce more assimilates for seed filling [34], which essentially reflects the positive feedback between the source and sink.

Among the 13 genes annotated in the 90-kb region of QC14 (Additional file 16: Table S9), only BnaA09G0560100ZS’s homologue (CYP78A9/AT3G61880) is involved in regulating silique and seed development in Arabidopsis (https://www.arabidopsis.org/servlets/TairObject?id=36508&type=locus). In addition, the expression level of BnaA09G0560100ZS showed a large difference (fold-change = 24.0 and 3.3; P-value = 1.2E−13 and 8.7E−3) between Zhongshuang11 and NIL_QC14 in both pod walls and seeds (Additional file 6: Figure S6). Although the coding sequence of BnaA09G0560100ZS had no difference between Zhongshuang11 and NIL_QC14, a CACTA-like transposable element was present in its upstream regulatory region in Zhongshuang11, but absent in NIL_QC14 (Additional file 7: Figure S7). A very recent study showed that this CACTA-like transposable element in the upstream region of BnaA9.CYP78A9 acted as an enhancer to increase its expression, which was responsible for a major QTL-qSLW.A9 for silique length and seed weight in rapeseed [42]. All of the above results strongly supported that BnaA9.CYP78A9 was the target gene of QC14.

QC17 indirectly affected yield by influencing three yield components likely through the growth period gene BnaC2.MAF2

QC17 was narrowed to a 744 kb genomic region between two SNP markers Bn-scaff_16139_1-p1393867 and seq-new-rs22829. Compared to the recurrent parent Zhongshuang11 (Table 3), the homologous NIL_QC17 showed the largest increase on BoT (+ 15.5%) and PNb (+ 15.1%), followed by PNw (+ 13.0%), BuT (+ 12.8%), SYb (+ 12.6%), SYw (+ 10.2%), PBN (+ 9.0%), and PNm (+ 8.6%), and moderate changes in FlT (+ 7.2%), TSW (− 7.0%), SYm (+ 5.9%), PD (+ 5.3%), and SNPP (+ 4.7%). The effects of QC17 on these traits were similar to those in the preliminary mapping population of BnaZN-RIL (Additional file 14: Table S7), where it had stable and large effects on growth period traits. The integration with conditional QTL results demonstrated the causal relationship between traits controlled by QC17 (Fig. 7D), where it had a direct and large effect on the growth period, which then had indirect and pleiotropic effects on PBN ( +), PN ( +), SNPP ( +), seed weight (−), and the final yield ( +).

The detailed analysis of the 101 genes annotated in the 744 kb region of QC17 revealed seven genes (BnaC02G0541300ZS, BnaC02G0542400ZS, BnaC02G0542800ZS, BnaC02G0545600ZS, BnaC02T0546100ZS, BnaC02T0546200ZS, and BnaC02T0546300ZS) that are homologous to the known genes (LNK, MTL1, EIP9, MAF2, and MAF4) controlling flowering time in Arabidopsis (Additional file 16: Table S9). Notably, all seven genes showed no expression or very low expression levels except for BnaC2.MAF2 (Additional file 17: Table S10), which was the most likely candidate gene of QC17.

Conclusions

In the current study, yield was systematically dissected at the phenotypic, genetic, and molecular levels using oilseed rape as an example. At the phenotypic level, analysis of 21 traits in a representative recombinant inbred line (RIL) population showed that yield was determined by a complex trait network with key contributors. At the genetic level, large-scale mapping and analysis of QTL showed that yield was controlled by an integrated QTL network with obvious hub-QTL that regulated multiple traits with large effects. At the molecular level, four representative hub-QTL were further fine-mapped, the causal relationships between the relevant traits were revealed, and the target or candidate genes were also identified. The highly consistent results at the phenotypic, genetic, and molecular dissecting provided a systematic view and further insight into the determination of yield in crops.

Discussion

Yield was determined by a complex trait network with key contributors

In the current study, there were several important findings in dissecting the phenotypic relationship between yield and its components or related traits by using the different analysis methods. Additionally, the constructed trait network exhibited an obvious center-periphery structure, where seed yield was located at the center, followed by yield components and related traits (Fig. 2A). This suggested that yield-related traits could indirectly affect yield by influencing yield components, which was highly supported by the results of further dissection of four representative hub-QTL. Among them, QC 8 directly affected yield, whereas QC4, QC14, and QC17 indirectly influenced yield. Further principal component analysis revealed several key traits (mainly represented by three yield components) as well as their influencing factors in determining yield. For example, seed yield was mainly determined by pod number and branch number, which were largely influenced by growth period, which was also highly supported by the results of further dissection of QC4 and QC17 (Fig. 7A, D). These results were highly accordant with population-level studies showing that the high yield of oilseed rape mainly depended on more branch number and pod number, followed by more and larger seeds [35, 43, 44].

Analysis of an integrated genetic and physical map provided insight into evolution and heterosis

The in-depth analysis of the integrated genetic and physical map resulted in several novel/interesting findings, which had great significance for genetics and breeding. To our knowledge, this is the first report that has calculated the recombination frequency in centromeric regions and compared with other regions in Brassica, although the genome-wide recombination frequency has been estimated [45]. The higher recombination frequencies of chromosomes A1 to A10 than C1 to C9 (Table 2) might be attributed to the lower proportion of repetitive sequences in the A subgenome than in the C subgenome [46, 47]. The probability of segregation distortion showed a gradual downward trend from the peak marker with the most significant Chi-square value (Pχ2), which might be closely linked with the segregation distortion loci [48]. Interestingly, all three major QTL clusters of the growth period were distorted to alleles with early growth/development (Fig. 3; Additional file 14: Table S7), which might be subjected to selection during the development of the BnaZN-RIL population. This result strongly suggested that selection via a genetic hitchhiking effect had an important role in the generation of segregation distortion in this population [49]. The higher proportion of distorted markers in the current BnaZN-RIL population than in the previously reported BnaZN-F2 population [45] derived from the same parents could be largely due to more generations of selection in the RIL population than in the F2 population [50]. Unexpectedly, several regions with high (even > 50%) residual heterozygosity were found in the BnaZN-RIL population that had been self-crossed for seven generations (theoretical heterozygosity was only 1.56%). Interestingly, the markers with high heterozygosity tend to cluster at the end of linkage groups (Fig. 3), and the heterozygotes in these regions usually perform better than the corresponding homozygotes (Additional file 12: Table S5), which has great significance for evolution and heterosis [51].

Yield was controlled by an integrated trait-QTL network with obvious hub-QTL

The integrated high-quality genetic and physical map provided an ideal platform to accurately dissect the genetic relationship between yield and its components or related traits. The most significant characteristics regarding these QTL were their clustered distribution (Fig. 5) and high-proportion (> 80%) of overlap (Additional file 15: Table S8), which were highly consistent with the extensive correlation between these traits (Additional file 3: Figure S3). The extensive trait correlation and QTL overlap in the current study were further supported by an integrated trait-QTL network that involved all of the 134 consensus QTL for 21 traits (Fig. 6). This strongly indicated the complexity of the genetic improvement of composite traits such as yield [1]. Generally, the overlapping QTL clusters/hotspots of yield [18,19,20, 29, 52,53,54] usually include several other QTL for yield components and/or related traits. Therefore, the overlapping QTL cluster for yield can be caused by the pleiotropic effects of a specific gene or the combined effects of several tightly linked genes, since a detected QTL in preliminary mapping may be dissected into several sub-QTL after further fine-mapping [24, 55, 56]. It is speculated that some of the seed yield QTL could not be detected due to the cancellation of opposite effects of several tightly linked QTL for the same or different yield components or related traits. Relative to yield components or related traits with higher heritability, yield QTL were generally unrepeatable and showed smaller effects, which made it difficult to directly clone and utilize them. The integrated trait-QTL network showed obvious hub-QTL clusters that linked multiple traits and displayed large effects, which played a major role in trait relations and were likely to be the targets for cloning.

Hub-QTL dissection provided causal explanations for yield formation and determination

Through the further dissection of four representative hub-QTL clusters using high-generation NILs, the qualitative (up/down-stream and positive/negative regulation) and quantitative (effect size) relationships between the relevant yield traits were revealed (Fig. 7). To our knowledge, this should be the exact roadmap of yield determination at the single locus/gene level, which provided a further causal explanation for yield formation. First, there were two main pathways for yield formation, including direct determination by yield components such as seed number per pod (QC8) and indirect determination by yield-related traits, such as growth period (QC4 and QC17) and source size (QC14). Second, if a yield QTL was directly caused by the genes of a specific yield component, it generally had negative pleiotropy on other yield components. Highly consistent with this, none of the cloned genes for yield QTL can simultaneously improve all yield components [42, 57,58,59,60,61,62,63,64,65,66,67]. This universal rule can explain the common phenomenon for QTL clusters of yield traits in which their effects on yield are usually smaller than those on yield components. However, the size of the effect was related to the determination period/developmental order of these traits (i.e., pod number sooner than seed number per pod sooner than seed weight), the traits determined earlier usually have large influence on those determined later but not vice versa. Taking QC8 as an example, the change in seed number per pod usually causes a large change in seed weight but only a small change in pod number. In addition, the trait relationships reflected by the four major hub-QTL clusters demonstrated that the yield variation in the BnaZN-RIL population should be mainly attributable to the difference in source (QC4/QC17 and QC14 affect biomass accumulation and pod area, respectively) and sink (QC8 controls ovule fertility and seed number per pod). The results of further dissection at the molecular level were also highly consistent with those at the genetic and phenotypic levels (e.g., pleiotropy of BnaA9.CYP78A9 → QTL cluster 14 → correlation between pod length and seed weight), which provided a systematic and deep understanding of yield determination.

Experimental procedures

Population development

The BnaZN-RIL population was developed by single-seed descent from the previously reported BnaZN-F2 population that was derived from Zhongshuang11 and No.73290 [45]. For the development of the near-isogenic line, the F1 hybrid (Zhongshuang11 × No.73290) was crossed with Zhongshuang11 for 12 generations. For each generation, BCnF1 plants heterozygous at the target QTL were screened (using flanking SSR/InDel markers) for each consecutive backcross. For the final BC12F1 generation, the heterozygous plants were also screened by background selection (using Brassica 60 K Illumina® Infinium SNP array), of which those with the highest background recovery rate (99.6%, 99.7%, 99.8%, and 99.5% for QC4, QC8, QC14, and QC17, respectively) were self-crossed to produce BC12F2 seeds.

Field experiments

Both the BnaZN-RIL population and its parents were planted in six environments, including four years at Wuhan (codes W12, W13, W14, and W16) and two years at Zhengzhou (codes Z13 and Z14). To accurately evaluate the phenotypic effect of four hub-QTL clusters, Zhongshuang11 and the corresponding NILs were planted and investigated at Wuhan in 2019. The field planting followed a randomized complete block design with three (BnaZN-RIL population) or ten (NILs) replications, respectively. Each block contained three (BnaZN-RIL population) or five (NILs) rows, respectively, with 33-cm spacing and 18 plants were evenly retained after singling. At maturity, 10 (BnaZN-RIL population) or 30 (NILs) representative individuals in the middle of each block (except for two rows on the side) were harvested from each block.

Trait investigation

A total of 21 traits were investigated and divided into five types: seed yield, yield components, growth period, plant architecture and seed constituent. Pod number (PN), seed number per pod (SNPP), pod length (PL),thousand-seed weight (TSW), flowering time (FlT), and plant height (PH) were measured as described in our previous studies [40, 45, 68, 69]. The main inflorescence length (MIL), primary branch number (PBN), primary branch height (PBH), pod density (PD), bolting time (BoT), budding time (BuT), flowering end time (FET), maturity time (MaT), seed oil content (OIL), and seed protein content (PRO) were measured as described in other studies [70,71,72].

Genotypic analyses and genetic linkage map construction

The BnaZN-RIL population of 171 lines was genotyped using the Illumina Infinium 60 K SNP chip, which contains 52,157 SNP markers [73]. Of the 17,286 polymorphic markers, 5405 showed a heterozygous or absent genotype in the parents and were removed from further analysis. Finally, a total of 11,881 high-quality and polymorphic SNP markers were obtained, which were then merged into 4993 co-segregated bins. Genetic linkage mapping was carried out using JoinMap 4.0 software [74] for each of the 19 linkage groups, with the following parameters: goodness of fit was set to ≤ 5.0 with LOD scores > 1.0 and a recombination frequency < 0.4.

QTL detection and integration

The conditional phenotype value y(T1|T2) was obtained by the mixed model approach for conditional analysis of quantitative traits using QGAStation1.0 (http://ibi.zju.edu.cn/software/qga/index.htm), where T1|T2 indicated that trait 1 was conditioned by trait 2. Conditional and unconditional QTL mapping was performed using the composite interval mapping procedure [75] incorporated in Windows QTL Cartographer 2.5 software. The walk speed, number of control markers, window size and regression method were set to 1 cM, 5, 10 cM and forward regression, respectively. The experiment-wise LOD threshold was determined by permutation test [76] with 1000 repetitions.

The identified QTL of the same trait detected in different environments were integrated into consensus QTL according to the previous report [68]. These consensus QTL were further combined into QTL clusters if their confidence intervals overlapped.

Construction of trait–QTL network

With slight modification from the previous report [77], traits and QTL were treated as nodes, which were connected by edges using Cytoscape version 3.7.2 software [78]. The QTL were renamed as the chromosome name followed by their positions. If QTL of different traits were integrated into a cluster, it would represent QTL that affect multiple traits. Yield, yield components and yield-related traits are plotted in orange, blue, and green, respectively.

qRT-PCR and RNA-seq

Total RNA was isolated using the RNeasy Plant Mini Kit (Qiagen). The cDNA was synthesized using the First Strand cDNA Synthesis Kit (Takara). Using the gene-specific primers (Table S11), quantitative reverse-transcription PCR (qRT-PCR) was performed using SYBR® Select Master Mix (2X) according to the manufacturers’ recommendations. The β-Actin gene was used as an internal control to normalize transcript levels in both B. napus and A. thaliana [34]. The relative expression level was calculated using the 2–ΔΔCt method.

RNA-seq was also performed for comparison of gene expression at the transcriptome level. The fragments per kilobase per million reads (FPKM) was used to calculate gene expression levels. DEseq2 (http://bioconductor.org/packages/stats/bioc/DESeq2/) was used to perform gene differential expression analysis. The absolute values of log2 (ratio) ≥ 1 and FDR < 0.05 were chosen as thresholds to screen for DEGs.

Cloning of full-length sequence of target genes

To verify the sequence variation of target genes in the two parents, their full-length sequences (including the coding region, 5ʹ upstream, and 3ʹ downstream) were amplified from the genomic DNA of Zhongshuang11 and No.73290 by the KOD FX enzyme (cat: KFX-101). The PCR products with expected size were cloned into the pEASY®-T1 cloning vector (cat: CT101-01)and then sequenced by Beijing Tsingke Biotechnology Co., Ltd.

Statistical analysis

Variance, correlation and principal component analyses were performed using PROC GLM, CORR, and PRINCOMP procedures incorporated into SAS 9.2 software. Based on the estimated variance components, broad-sense heritability was calculated according to the method described previously [1].

Fundings

This work was supported by the Agricultural Science and Technology Innovation Program of China (CAAS-ZDRW202105), Natural Science Foundation of China (31,771,840), Agriculture Research System of MOF and MARA of China (CARS-13), Agricultural Science and Technology Innovation Project of China (CAAS-ASTIP-2013-OCRI), and Fundamental Research Funds for Central Non-Profit Institute of Crop Sciences, CAAS (Y2020YJ09).

Availability of data and materials

The original datasets of RNA-seq are available in the Sequence Read Archive at the National Center for Biotechnology Information (NCBI) under accession number SRX1715587. The datasets supporting the conclusions of this article are included in the manuscript and additional files.

References

  1. Shi JQ, Li RY, Qiu D, Jiang CC, Long Y, Morgan C, et al. Unraveling the complex trait of crop yield with quantitative trait loci mapping in Brassica napus. Genetics. 2009;182(3):851–61.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  2. Sharma A, Singh G, Arya SK. Biofuel from rice straw. J Clean Prod. 2020;277:124101.

    CAS  Article  Google Scholar 

  3. Abo BO, Gao M, Wang YL, Wu CF, Ma HZ, Wang QH. Lignocellulosic biomass for bioethanol: an overview on pretreatment, hydrolysis and fermentation processes. Rev Environ Health. 2019;34(1):57–68.

    CAS  Article  PubMed  Google Scholar 

  4. Khanal A, Shah A. Oilseeds to biodiesel and renewable jet fuel: an overview of feedstock production, logistics, and conversion. Biofuel Bioprod Bior. 2021;15(3):913–30.

    CAS  Article  Google Scholar 

  5. Bailey-Serres J, Parker JE, Ainsworth EA, Oldroyd GED, Schroeder JI. Genetic strategies for improving crop yields. Nature. 2019;575(7781):109–18.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  6. Quarrie SA, Quarrie SP, Radosevic R, Rancic D, Kaminska A, Barnes JD, et al. Dissecting a wheat QTL for yield present in a range of environments: from the QTL to candidate genes. J Exp Bot. 2006;57(11):2627–37.

    CAS  Article  PubMed  Google Scholar 

  7. Sadras VO. Evolutionary aspects of the trade-off between seed size and number in crops. Field Crop Res. 2007;100(2–3):125–38.

    Article  Google Scholar 

  8. Gambin BL, Borras L. Resource distribution and the trade-off between seed number and seed weight: a comparison across crop species. Ann Appl Biol. 2010;156(1):91–102.

    Article  Google Scholar 

  9. Fletcher RS, Mullen JL, Heiliger A, McKay JK. QTL analysis of root morphology, flowering time, and yield reveals trade-offs in response to drought in Brassica napus. J Exp Bot. 2015;66(1):245–56.

    CAS  Article  PubMed  Google Scholar 

  10. Farnsworth EJ, Ellison AM. Prey availability directly affects physiology, growth, nutrient allocation and scaling relationships among leaf traits in 10 carnivorous plant species. J Ecol. 2008;96(1):213–21.

    Google Scholar 

  11. Miyatake T, Shimizu T. Genetic correlations between life-history and behavioral traits can cause reproductive isolation. Evolution. 1999;53(1):201–8.

    Article  PubMed  Google Scholar 

  12. Musse M, Bidault K, Quellec S, Brunel B, Collewet G, Cambert M, et al. Spatial and temporal evolution of quantitative magnetic resonance imaging parameters of peach and apple fruit relationship with biophysical and metabolic traits. Plant J. 2021;105(1):62–78.

    CAS  Article  PubMed  Google Scholar 

  13. Hittalmani S, Shashidhar HE, Bagali PG, Huang N, Sidhu JS, Singh VP, et al. Molecular mapping of quantitative trait loci for plant growth, yield and yield related traits across three diverse locations in a doubled haploid rice population. Euphytica. 2002;125(2):207–14.

    CAS  Article  Google Scholar 

  14. Sukumaran S, Dreisigacker S, Lopes M, Chavez P, Reynolds MP. Genome-wide association study for grain yield and related traits in an elite spring wheat population grown in temperate irrigated environments. Theor Appl Genet. 2015;128(2):353–63.

    CAS  Article  PubMed  Google Scholar 

  15. Peng B, Li YX, Wang Y, Liu C, Liu ZZ, Tan WW, et al. QTL analysis for yield components and kernel-related traits in maize across multi-environments. Theor Appl Genet. 2011;122(7):1305–20.

    Article  PubMed  Google Scholar 

  16. Liu W, Kim M-Y, Van K, Lee Y-H, Li H, Liu X, et al. QTL identification of yield-related traits and their association with flowering and maturity in soybean. J Crop Sci Biotech. 2011;14(1):65–70.

    Article  Google Scholar 

  17. Wang ZH, Huai DX, Zhang ZH, Cheng K, Kang YP, Wan LY, et al. Development of a high-density genetic map based on specific length amplified fragment sequencing and its application in quantitative trait loci analysis for yield-related traits in cultivated peanut. Front Plant Sci. 2018;9:827.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Wang P, Zhou GL, Cui KH, Li ZK, Yu SB. Clustered QTL for source leaf size and yield traits in rice (Oryza sativa L.). Mol Breeding. 2012;29(1):99–113.

    CAS  Article  Google Scholar 

  19. Liu TM, Yu T, Xing YZ. Identification and validation of a yield-enhancing QTL cluster in rice (Oryza sativa L.). Euphytica. 2013;192(1):145–53.

    Article  Google Scholar 

  20. Sukumaran S, Reynolds MP, Sansaloni C. Genome-wide association analyses identify QTL hotspots for yield and component traits in durum wheat grown under yield potential, drought, and heat stress environments. Front Plant Sci. 2018;9:81.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Yang JW, Liu ZH, Chen Q, Qu YZ, Tang JH, Lubberstedt T, et al. Mapping of QTL for grain yield components based on a DH population in maize. Sci Rep-Uk. 2020;10(1):13218.

    CAS  Article  Google Scholar 

  22. Zhu XT, Leiser WL, Hahn V, Wurschum T. Identification of QTL for seed yield and agronomic traits in 944 soybean (Glycine max) RILs from a diallele cross of early-maturing varieties. Plant Breeding. 2021;140(2):254–66.

    CAS  Article  Google Scholar 

  23. Ott A, Trautschold B, Sandhu D. Using microsatellites to understand the physical distribution of recombination on soybean chromosomes. PLoS ONE. 2011;6(7):e22306.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  24. Chen H, Qian N, Guo WZ, Song QP, Li BC, Deng FJ, et al. Using three selected overlapping RILs to fine-map the yield component QTL on Chro.D8 in upland cotton. Euphytica. 2010;176(3):321–9.

    Article  Google Scholar 

  25. Said JI, Lin ZX, Zhang XL, Song MZ, Zhang JF. A comprehensive meta QTL analysis for fiber quality, yield, yield related and morphological traits, drought tolerance, and disease resistance in tetraploid cotton. BMC Genomics. 2013;14(7):776.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  26. Zhao WG, Wang XD, Wang H, Tian JH, Li BJ, Chen L, et al. Genome-wide identification of QTL for seed yield and yield-related traits and construction of a high-density consensus map for QTL comparison in Brassica napus. Front Plant Sci. 2016;7:17.

    PubMed  PubMed Central  Google Scholar 

  27. Wagner GP, Zhang JZ. The pleiotropic structure of the genotype-phenotype map: the evolvability of complex organisms. Nat Rev Genet. 2011;12(3):204–13.

    CAS  Article  PubMed  Google Scholar 

  28. Jeon YA, Lee HS, Kim SH, Shim KC, Kang JW, Kim HJ, et al. Natural variation in rice ascorbate peroxidase gene APX9 is associated with a yield-enhancing QTL cluster. J Exp Bot. 2021;72(12):4254–68.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  29. Xie XB, Jin FX, Song MH, Suh JP, Hwang HG, Kim YG, et al. Fine mapping of a yield-enhancing QTL cluster associated with transgressive variation in an Oryza sativa × O. rufipogon cross. Theor Appl Genet. 2008;116(5):613–22.

    Article  PubMed  Google Scholar 

  30. Chalhoub B, Denoeud F, Liu SY, Parkin IAP, Tang HB, Wang XY, et al. Early allopolyploid evolution in the post-Neolithic Brassica napus oilseed genome. Science. 2014;345(6199):950–3.

    CAS  Article  PubMed  Google Scholar 

  31. Sivasamy A, Cheah KY, Fornasiero P, Kemausuor F, Zinoviev S, Miertus S. Catalytic applications in the production of biodiesel from vegetable oils. Chemsuschem. 2009;2(4):278–300.

    CAS  Article  PubMed  Google Scholar 

  32. Kluts IN, Brinkman MLJ, de Jong SA, Junginger HM. Biomass resources: Agriculture. Adv Biochem Eng Biot. 2019;166:13–26.

    CAS  Google Scholar 

  33. Khan SU, Saeed S, Khan MHU, Fan CC, Ahmar S, Arriagada O, et al. Advances and challenges for QTL analysis and GWAS in the plant-breeding of high-yielding: a focus on rapeseed. Biomolecules. 2021;11(10):1516.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  34. Li N, Song OJ, Peng W, Zhan JP, Shi JQ, Wang XF, et al. Maternal control of seed weight in rapeseed (Brassica napus L.): the causal link between the size of pod (mother, source) and seed (offspring, sink). Plant Biotechnol J. 2019;17(4):736–49.

    CAS  Article  PubMed  Google Scholar 

  35. Li SY, Zhu YY, Varshney RK, Zhan JP, Zheng XX, Shi JQ, et al. A systematic dissection of the mechanisms underlying the natural variation of silique number in rapeseed (Brassica napus L) germplasm. Plant Biotechnol J. 2020;18(2):568–80.

    CAS  Article  PubMed  Google Scholar 

  36. Yi S, Liang Y, Dai L, Chen L, Chai Y, Li J. Effects of high temperature on post-harvest ripening-related characteristics in Brassica napus L. J Southwest Univ. 2008;30(2):48–50.

    CAS  Google Scholar 

  37. Chen L, Dong FM, Cai J, Xin Q, Fang CC, Liu L, et al. A 2.833-kb insertion in BnFLC.A2 and its homeologous exchange with BnFLC.C2 during breeding selection generated early-flowering rapeseed. Mol Plant. 2018;11(1):222–5.

    CAS  Article  PubMed  Google Scholar 

  38. Tudor EH, Jones DM, He Z, Bancroft I, Trick M, Wells R, et al. QTL-seq identifies BnaFT.A02 and BnaFLC.A02 as candidates for variation in vernalization requirement and response in winter oilseed rape (Brassica napus). Plant Biotechnol J. 2020;18(12):2466–81.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  39. Yang YH, Shi JQ, Wang XF, Liu GH, Wang HZ. Genetic architecture and mechanism of seed number per pod in rapeseed: elucidated through linkage and near-isogenic line analysis. Sci Rep-Uk. 2016;6:24124.

    CAS  Article  Google Scholar 

  40. Zhang JJ, Zhan JP, Liu QY, Shi JQ, Wang XF, Liu GH, et al. QTL mapping and integration as well as candidate genes identification for plant height in rapeseed (Brassica napus l.). Scientia Agricultura Sinica. 2017;50(17):3247–58.

    Google Scholar 

  41. Hussain Q, Shi JQ, Scheben A, Zhan JP, Wang XF, Liu GH, et al. Genetic and signalling pathways of dry fruit size: targets for genome editing-based crop improvement. Plant Biotechnol J. 2020;18(5):1124–40.

    Article  PubMed  PubMed Central  Google Scholar 

  42. Shi LL, Song JR, Guo CC, Wang B, Guan ZL, Yang P, et al. A CACTA-like transposable element in the upstream region of BnaA9.CYP78A9 acts as an enhancer to increase silique length and seed weight in rapeseed. Plant J. 2019;98(3):524–39.

    CAS  Article  PubMed  Google Scholar 

  43. Zhao YG, Cheng Y, Lu GY, Xu JS, Fu GP, Zou XL, et al. Characteristics and variation of winter rapeseed (Brassica napus L.) cultivars under high density. Chin J Oil Crop Sci. 2015;37(3):285–90.

    Google Scholar 

  44. Bai GP, Liu KZ, Tan YQ, Yin YF, Yu HQ, Wang HZ. Effect of agronomic traits on seed yield in high-yielding rapeseed populations. Crops. 2015;6:33–8.

    Google Scholar 

  45. Shi JQ, Zhan JP, Yang YH, Ye J, Huang SM, Li RY, et al. Linkage and regional association analysis reveal two new tightly-linked major-QTLs for pod number and seed number per pod in rapeseed (Brassica napus L). Sci Rep-Uk. 2015;5:4481.

    Google Scholar 

  46. Wang XW, Wang HZ, Wang J, Sun RF, Wu J, Liu SY, et al. The genome of the mesopolyploid crop species Brassica rapa. Nat Genet. 2011;43(10):1035–9.

    CAS  Article  PubMed  Google Scholar 

  47. Liu SY, Liu YM, Yang XH, Tong CB, Edwards D, Parkin IAP, et al. The Brassica oleracea genome reveals the asymmetrical evolution of polyploid genomes. Nat Commun. 2014;5:4930.

    Article  CAS  Google Scholar 

  48. Song X, Sun X, Zhang T. Segregation distortion and its effect on genetic mapping in plants. J Agric Biotechnol. 2006;14(2):286–92.

    CAS  Google Scholar 

  49. Liu HL, Cui JT, Guo YM. Progress of segregation distortion. J Plant Genet Resour. 2009;10(4):613–7.

    CAS  Google Scholar 

  50. Xu Y, Zhu L, Xiao J, Huang N, McCouch SR. Chromosomal regions associated with segregation distortion of molecular markers in F2, backcross, doubled haploid, and recombinant inbred populations in rice (Oryza sativa L). Mol Gen Genet. 1997;253(5):535–45.

    CAS  Article  PubMed  Google Scholar 

  51. McMullen MD, Kresovich S, Villeda HS, Bradbury P, Li HH, Sun Q, et al. Genetic properties of the maize nested association mapping population. Science. 2009;325(5941):737–40.

    CAS  Article  PubMed  Google Scholar 

  52. Portis E, Barchi L, Toppino L, Lanteri S, Acciarri N, Felicioni N, et al. QTL mapping in eggplant reveals clusters of yield-related loci and orthology with the tomato genome. PLoS ONE. 2014;9(2):e89499.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Rae AM, Street NR, Robinson KM, Harris N, Taylor G. Five QTL hotspots for yield in short rotation coppice bioenergy poplar: The Poplar Biomass Loci. BMC Plant Biol. 2009;9:23.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Bharadwaj C, Tripathi S, Soren KR, Thudi M, Singh RK, Sheoran S, et al. Introgression of “QTL-hotspot” region enhances drought tolerance and grain yield in three elite chickpea cultivars. Plant Genome-Us. 2021;14(1):e20076.

    CAS  Google Scholar 

  55. Graham GI, Wolff DW, Stuber CW. Characterization of a yield quantitative trait locus on chromosome five of maize by fine mapping. Crop Sci. 1997;37(5):1601–10.

    CAS  Article  Google Scholar 

  56. Dixit S, Swamy BPM, Vikram P, Ahmed HU, Cruz MTS, Amante M, et al. Fine mapping of QTLs for rice grain yield under drought reveals sub-QTLs conferring a response to variable drought severities. Theor Appl Genet. 2012;125(1):155–69. https://doi.org/10.1007/s00122-012-1823-9.

    Article  PubMed  Google Scholar 

  57. Xing YZ, Zhang QF. Genetic and molecular bases of rice yield. Ann Rev Plant Biol. 2010;61(1):421–42.

    CAS  Article  Google Scholar 

  58. Nadolska-Orczyk A, Rajchel IK, Orczyk W, Gasparis S. Major genes determining yield-related traits in wheat and barley. Theor Appl Genet. 2017;130(6):1081–98.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  59. Jia HT, Li MF, Li WY, Liu L, Jian YA, Yang ZX, et al. A serine/threonine protein kinase encoding gene KERNEL NUMBER PER ROW6 regulates maize grain yield. Nat Commun. 2020;11(1):988.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  60. Bommert P, Nagasawa NS, Jackson D. Quantitative variation in maize kernel row number is controlled by the FASCIATED EAR2 locus. Nat Genet. 2013;45(3):334–7.

    CAS  Article  PubMed  Google Scholar 

  61. Liu L, Du YF, Shen XM, Li MF, Sun W, Huang J, et al. KRN4 controls quantitative variation in maize kernel row number. PLoS Genet. 2015;11(11):e1005670.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  62. Wang J, Lin ZL, Zhang X, Liu HQ, Zhou LN, Zhong SY, et al. krn1, a major quantitative trait locus for kernel row number in maize. New Phytol. 2019;223(3):1634–46.

    CAS  Article  PubMed  Google Scholar 

  63. Nguyen CX, Paddock KJ, Zhang ZY, Stacey MG. GmKIX8-1 regulates organ size in soybean and is the causative gene for the major seed weight QTL qSw17-1. New Phytol. 2021;229(2):920–34.

    CAS  Article  PubMed  Google Scholar 

  64. Wang XB, Li YH, Zhang HW, Sun GL, Zhang WM, Qiu LJ. Evolution and association analysis of GmCYP78A10 gene with seed size/weight and pod number in soybean. Mol Biol Rep. 2015;42(2):489–96.

    CAS  Article  PubMed  Google Scholar 

  65. Jeong N, Suh SJ, Kim MH, Lee S, Moon JK, Kim HS, et al. Ln is a key regulator of leaflet shape and number of seeds per pod in soybean. Plant Cell. 2012;24(12):4807–18.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  66. Liu J, Hua W, Hu ZY, Yang HL, Zhang L, Li RJ, et al. Natural variation in ARF18 gene simultaneously affects seed weight and silique length in polyploid rapeseed. Proc Natl Acad Sci USA. 2015;112(37):E5123–32.

    CAS  PubMed  PubMed Central  Google Scholar 

  67. Li SP, Chen L, Zhang LW, Li X, Liu Y, Wu ZK, et al. BnaC9.SMG7b functions as a positive regulator of the number of seeds per silique in Brassica napus by regulating the formation of functional female gametophytes. Plant Physiol. 2015;169(4):2744–60.

    CAS  PubMed  PubMed Central  Google Scholar 

  68. Li N, Shi JQ, Wang XF, Liu GH, Wang HZ. A combined linkage and regional association mapping validation and fine mapping of two major pleiotropic QTLs for seed weight and silique length in rapeseed (Brassica napus L.). BMC Plant Biol. 2014;14:114.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  69. Wei YQ, Wei WL, Liu DM, Zhang JJ, Zhan JP, Shi JQ, et al. QTL mapping and candidate genes analysis for flowering time in rapeseed (Brassica napus L.). Chinese J Oil Crop Sci. 2019;41(5):679–87.

    Google Scholar 

  70. Chen W, Zhang Y, Liu XP, Chen BY, Tu JX, Fu TD. Detection of QTL for six yield-related traits in oilseed rape (Brassica napus) using DH and immortalized F2 populations. Theor Appl Genet. 2007;115(6):849–58.

    CAS  Article  PubMed  Google Scholar 

  71. Li JC, Zhao XH, Nishimura Y, Fukumoto Y. Correlation between bolting and physiological properties in Chinese cabbage (Brassica rapa L. pekinensis Group). J Jpn Soc Hortic Sci. 2010;79(3):294–300.

    Article  Google Scholar 

  72. Tetteh ET, de Koff JP, Pokharel B, Link R, Robbins C. Effect of winter canola cultivar on seed yield, oil, and protein content. Agron J. 2019;111(6):2811–20.

    CAS  Article  Google Scholar 

  73. Clarke WE, Higgins EE, Plieske J, Wieseke R, Sidebottom C, Khedikar Y, et al. A high-density SNP genotyping array for Brassica napus and its ancestral diploid species based on optimised selection of single-locus markers in the allotetraploid genome. Theor Appl Genet. 2016;129(10):1887–99.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  74. Van Ooijen JW. JoinMap4, Software for calculation of genetic linkage maps in experimental populations. Wageningen: Kyazna B V; 2006.

    Google Scholar 

  75. Zeng ZB. Precesion mapping of quantitative trait loci. Genetics. 1994;136:1457–68.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  76. Churchill GA, Doerge RW. Empirical threshold values for quantitative trait mapping. Genetics. 1994;138(3):963–71.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  77. Crowell S, Korniliev P, Falcao A, Ismail A, Gregorio G, Mezey J, et al. Genome-wide association and high-resolution phenotyping link Oryza sativa panicle traits to numerous trait-specific QTL clusters. Nat Commun. 2016;7:10527.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  78. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

The authors thank Xiya Zhang, Yujiao Liu, and Songgui Xu for their help in the field work.

Author information

Authors and Affiliations

Authors

Contributions

JS, HW, and JT designed the research; JS, HL, JY performed the research; JZ, XW, and YW collected the data; JS and HL analyzed and interpreted the data; JS, HL, and XZ wrote the manuscript. XZ, JB, GK, and LG revised the manuscript. All authors have read and agreed to the published version of the manuscript.

Corresponding author

Correspondence to Jiaqin Shi.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors consent for publication.

Competing interests

All authors state that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1

: Figure S1. Frequency of distribution for each of the 21 traits investigated in the BnaZN-RIL population planted in six environments. The horizontal and vertical axes are divided with the same spacing, which shows the phenotypic value and line number, respectively. The columns of the different heights represent the number of lines in different groups. Different environments are distinguished by the different colors.

Additional file 2

: Figure S2. Qualitative and quantitative presentation of phenotypic correlation among 21 traits investigated in the BnaZN-RIL population planted in six environments. The abbreviations of 21 traits are shown near both the horizontal and vertical axes. Below the diagonal, the significant correlations among these traits are indicated by circles of different sizes; above the diagonal, the coefficients of significant correlations are shown. The direction and degree of correlation are distinguished by the different colors demonstrated in the legend.

Additional file 3

: Figure S3. Development of high-generation near-isogenic lines for the fine-mapping and evaluation of phenotypic effects. Hybrid F1 was alternatively backcrossed with Zhongshuang11 in Xining and Wuhan. In each generation, foreground selection was performed to choose heterozygous plants (at the target QTL region), which were used for consecutive backcrosses. Finally, the heterozygous BC12F1 NILs with the highest background recovery rate were self-obsessed to obtain BC12F2 seeds, for the individual hub-QTL. The homologous BC12F2 NILs were subjected to phenotypic investigation and compared with the recurrent parent Zhongshuang11.

Additional file 4

: Figure S4. Comparison of the promoter sequence of BnaA2.FLC between Zhongshuang11 and No.73290. (A) Alignment of the 2.9 kb sequence in the upstream regulatory region of BnaA2.FLC between Zhongshuang11 and No.73290. (B) Structural comparison of the 2.9 sequence in the upstream regulatory region of BnaA2.FLC between Zhongshuang11 and No.73290. There were 12 SNPs and two InDels in the promoter region of BnaA2.FLC between Zhongshuang11 and No. 73290. Of these, the largest difference was a 10-bp InDel within the core 40-bp motif.

Additional file 5

: Figure S5. Comparison of the BnaA6.EMB93 sequence between Zhongshuang11 and No.73290. (A) Alignment of the coding sequence of BnaA6.EMB93 between Zhongshuang11 and No.73290. Dark blue and no fill background represent the consensus and different sequences, respectively. There are a total of 17 SNPs between the two parents. (B) Alignment of the protein sequence of BnaA6.EMB93 between Zhongshuang11 and No.73290. Only amino acid 228 showed difference between the two parents, which is not within the functional domain of this protein. (C) Structural analysis of the transposon inserted into the promoter region of BnaA6.EMB93 in No.73290.

Additional file 6

: Figure S6. Quantitative analysis of the expression of BnaA6.EMB93 and BnaA9.CYP78A9. (A) The relative expression level of BnaA6.EMB93 in the ovaries of Zhongshuang11 and NIL_QC8. The horizontal axis shows buds of different sizes (1-8 mm) before flowering and ovaries at different days after flowering (DAF). (B) The relative expression level of BnaA9.CYP78A9 in the ovaries of Zhongshuang11 and NIL_QC14. The horizontal axis shows pod walls at two weeks after flowering and seeds at four weeks after flowering. * P <0.05, ** P <0.01, *** P <0.001 (t-test) indicate a significant difference between Zhongshuang11 and NIL. Each data is obtained from three biological replicates.

Additional file 7

: Figure S7. Comparison of the full-length sequence of BnaA9.CYP78A9 between Zhongshuang11 and NIL_QC14. (A) The full-length genic structure of BnaA9.CYP78A9 in Zhongshuang11 and NIL_QC14. There was no difference in the coding sequence, but a 3.6-kb CACTA-like TE insertion into its upstream regulatory region in Zhongshuang11. (B) The alignment of coding sequence of BnaA9.CYP78A9 in Zhongshuang11 and NIL_QC14.

Additional file 8

: Table S1. Phenotypic variation of both parents and the BnaZN-RIL population in six environments.

Additional file 9

: Table S2. Analysis of variance and estimation of heritability for the 21 traits investigated in the BnaZN-RIL population.

Additional file 10

: Table S3. Summary statistics of the correlation among 21 traits investigated in six environments.

Additional file 11

: Table S4. Three principal components (PC1, PC2, and PC3) in the two locations and six environments.

Additional file 12

: Table S5. Single marker analyses of the highly heterozygous markers on 21 investigated traits.

Additional file 13

: Table S6. Polymorphism of SNPs within the pericentromere region of the 19 chromosomes.

Additional file 14

: Table S7. List of identified QTL (A), consensus QTL (B), and QTL clusters (C).

Additional file 15

: Table S8. Statistical analysis of consensus QTL overlapping between pair-wise combinations among the 21 investigated traits.

Additional file 16

: Table S9. The annotated genes in the fine-mapped genomic regions of four representative hub-QTL clusters.

Additional file 17

: Table S10. The DEGs identified by RNA-seq of SAM in the initial stage of floral bud differentiation between Zhongshuang11 and No.73290.

Additional file 18

: Table S11. List of primers used in this study, including fine-mapping, gene cloning, and qRT-PCR.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Liang, H., Ye, J., Wang, Y. et al. Systematic trait dissection in oilseed rape provides a comprehensive view, further insight, and exact roadmap for yield determination. Biotechnol Biofuels 15, 38 (2022). https://doi.org/10.1186/s13068-022-02134-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13068-022-02134-w

Keywords

  • Brassica napus
  • Yield determination
  • Segregation distortion
  • Residual heterozygosity
  • Trait-QTL network
  • Positive or negative pleiotropy
  • Target or candidate genes
  • Trade-off