Growth and seed traits were measured in a QTL mapping population, and the frequency distributions of all traits in the progeny showed a continuous distribution. The distribution of phenotypic values showed bi-directional transgressive segregation (Table ), revealing complex genetic bases of these traits. While seed yield in J. curcas was higher than that in J. integerrima, branch number in J. integerrima is significantly higher than that in J. curcas. The data implied that J. integerrima germplasm could be applied for hybrid breeding to improve agronomic traits, such as branch number in the fourth and tenth months, and female flower number.
Descriptive statistics on phenotype data of QTL mapping population and parents (J. curcasPZMD16,J. integerrimaS001 and F1 CI7041)
Correlation analysis among these traits was performed (Table ), and total seed weight showed a significant correlation with total branch number, female flower number and fruit number, with coefficients 0.364, 0.294 and 0.308, respectively. Therefore, these agronomic traits were suggested to be key factors for seed yields.
Correlation coefficients and significance of correlations among growth and yield traits in a QTL mapping population
The linkage map consisting of 105 DNA markers and covering 643.8
cM of the genome, converged into 11 LGs (linkage groups) corresponding to 11 chromosome pairs in jatropha. The average distance between markers was 6.6
cM. Most of the LGs were consistent with those described previously [14
QTL analyses were performed on the means of growth traits, branch number, female flower and fruit number, and seed yield (Table ; Figure ). We have detected 28 QTLs for all traits examined with LOD threshold 2.0 to 2.5 determined by permutations. Individual QTLs were detected with percentage of variation explained (PVE or R2) 3 to 21.16%, and four of them had PVE exceeding 10%.
QTLs for growth traits, seed characters
Figure 1 Summary of QTL locations detected. QTL represented by bars are shown on the left of the linkage groups, close to their corresponding markers. The lengths of the bars are proportional to the confidence intervals of the corresponding QTL in which the inner (more ...)
QTLs with positive and negative additive effects were identified, with a positive effect implying a higher value for the trait conferred by the allele from J. curcas, and negative from J. integerrima (Table ).
QTLs for growth traits
Sixteen QTLs were identified and dispersed among all the linkage groups except LGs 2 and 8. Four QTLs overlapping on the lower part of LG5, namely qH10m-5, qD4m-5, qD10m-5 and qTBN-5, were detected underlying plant height in the 10th month, stem diameter in the 4th and 10th months, and total branch number, respectively (Figure ). Additive effects of these QTLs were positive, indicating that the alleles from J. curcas increased these trait values.
Conversely, two QTLs, namely qBN4m-6 and qBN10m-6, were detected on the lower part of LG6 controlling branch number with negative additive values, indicating Jatropha integerrima allele increased branch number.
QTLs for female flower and fruit number
Six QTLs were identified and dispersed on LGs 1, 5, 6 and 7, with two QTLs, namely qFFN-6 and FruitNo-6, being located on the same region of LG6, controlling female flower number and fruit number respectively. The PVE of these two QTLs were higher than 10%, indicating their significant effects on the two important yield trait components.
QTLs for seed traits
On LGs 5 and 7, two QTLs of qTSW-5 and qTSW-7 were detected controlling total seed weight, which is one of the most economically important traits. Interestingly, QTLs underlying yield related traits were clustered at these two QTLs. At qWT-5, four QTLs underlying plant height, stem diameter, branch number and female flower number were detected. Near qTSW-7, three QTLs of qH4m-7, qTBN-7 and qFruitNo-7 were detected, controlling plant height, total branch number and fruit number respectively.
It was noteworthy that two QTL clusters were detected on LGs 5 and 7, respectively. Five QTLs were detected on the lower part of LG5 (Figure A), and four QTL clusters were detected on lower part of LG7 (Figure B).
QTL clusters on LGs 5 and 7. QTL scans of growth on linkage maps. Horizontal line indicates 5% LOD significance thresholds (2.0) based on permutation. A: LG5; B: LG7.
Favored alleles originated from two parents
Two QTL clusters were detected consisting of five and four QTLs, controlling total seed weight, plant height, stem diameter, female flower number and fruit number. The positive additive effects indicated higher values for the traits conferred by the allele from J. curcas. Meanwhile five QTLs on LG6, namely qH4m-6, qBN4m-6, qBN10m-6 qFFN-6 and qFruitNo-6, controlling plant height, branch number (in 4th and 10th months post seed germination), female flower number and fruit number respectively, were detected with negative additive effects indicating higher values conferred by J. integerrima (Table ).
Major effects of qTSW-5 and qTSW-7
A two-way analysis of variance (ANOVA) was carried out to assess genetic effects and interactions of the two QTLs of qTSW-5
controlling total seed weight. The values of different genotypes are shown in Figure . Total seed weight was significantly increased in the presence of these two QTLs. When qTSW-5
presented, total seed weight was improved from 16.66
7.26 to 42.00
g, and qTSW-7
, from 15.97
6.36 to 42.69
g (Figure A).
Figure 3 Total seed weight (g) and related traits of plants with different genotypes ofqTSW-5(AA, Aa) andqTSW-7(BB, Bb); N denotes sample number of each genotypic classes; Error bars denote SEs (Standard Errors).A: Significant major effects of the two QTLs on (more ...)
Interestingly, we found that the two QTLs for seed yield overlapped with other QTLs for other agronomic traits than seed yield itself. ANOVA showed that the QTL qTSW-5 for seed yield affected significantly plant height, stem diameter, new branch number per branch and female flower number, while qTSW-7 affected plant height, total branch number and fruit number (Figure B).
Effect of pyramiding qTSW-5 and qTSW-7
The interaction between marker effects for qTSW-5 and qTSW-7 was non-significant with a relatively low P value (0.14) (Table ), while the marker effects for qTSW-5 and qTSW-7 were non-additive (Figure ). This could be caused by the lack of power in the ANOVA due to an unequal distribution of genotypic classes (Figure ).
ANOVA of seed yield in the QTL mapping population based on genotypes of the marker loci that are most closely linked to the QTLs
Effects of pyramiding the two QTLs ofqTSW-5(genotypes of AA, Aa) andqTSW-7(BB, Bb) on seed yield. Error bars denote SEs.
Despite the non-significance of the interaction of the two QTLs, total seed weight was significantly increased in the presence of the two QTLs. Lines carrying both QTLs produced an average 61.93
g of seeds, nearly three times as much as any other genotype combinations (Figure ). Therefore, although total seed weight could be improved by introducing the two QTLs, there would be advantages to be gained by pyramiding the two QTLs.