JOURNAL OF

GENETICS AND GENOMICS J. Genet. Genomics 37 (2010) 545−556 www.jgenetgenomics.org

Identification of genomic regions determining flower and pod numbers development in soybean (Glycine max L.) Dan Zhang, Hao Cheng, Hui Wang, Hengyou Zhang, Chunying Liu, Deyue Yu * National Center for Soybean Improvement, National Key Laboratory of Crop Genetics and Germplasm Enhancement, Nanjing Agricultural University, Nanjing 210095, China Received for publication 15 March 2010; revised 11 May 2010; accepted 10 June 2010

Abstract Flower and pod numbers per plant are important agronomic traits underlying soybean yield. So far quantitative trait loci (QTL) detected for flower and pod-related traits have mainly focused on the final stage, and might therefore have ignored genetic effects expressed during a specific developmental stage. Here, dynamic expressions of QTL for flower and pod numbers were identified using 152 recombinant inbred lines (RILs) and a linkage map of 306 markers. Wide genetic variation was found among RILs; 17 unconditional and 18 conditional QTL were detected for the two traits at different developmental stages over two years. Some QTL were detected only at one stage and others across two or more stages, indicating that soybean flower and pod numbers development may be governed by time-dependent gene expression. Three main QTL (qfn-Chr18-2, qfn-Chr20-1, and qfn-Chr19) were detected for flower number, and two main QTL (qpn-Chr11 and qpn-Chr20) were detected for pod number. The phenotypic variation explained by them ranged from 6.1% to 34.7%. The markers linked to these QTL could be used in marker-assisted selection for increasing soybean flower and pod numbers, with the ultimate aim of increasing soybean yield. Comparison of the QTL regions for flower and pod numbers traits with the related genes reported previously showed that seven and four related genes were located in the QTL regions of qfn-Chr11 and qfn-Chr19, respectively. These results provide a basis for fine mapping and cloning of flower and pod development-related genes. Keywords: conditional QTL; unconditional QTL; developmental quantitative genetics; flower number; pod number; soybean

Introduction Flower and pod numbers are the most important agronomic traits for seed production in soybean (Glycine max L.). Flower and pod numbers per plant determine the effective flower and pod numbers, which are key components of soybean yield. Pod number per plant is an especially important factor in determining soybean yield (Board and Tan, 1995). Flower and pod numbers are typi* Corresponding author. Tel & Fax: +86-25-8439 6410. E-mail address: [email protected] DOI: 10.1016/S1673-8527(09)60074-6

cal developmental quantitative characters reflecting expression changes over time. The discovery of molecular QTL and QTL-assisted selection for these quantitative straits would greatly aid the breeding procedure, because traditional soybean breeding for quantitative traits is slow and difficult. Marker-assisted selection could potentially improve the selection of yield traits that have low heritability by using markers with high heritability (Sun et al., 2006). Understanding the developmental genetic mechanism of flower and pod numbers will help to elucidate the mechanism of soybean yield, and is important for the development of high-yield cultivars.

546

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

In the past decades, many QTL for agronomic traits have been detected at individual development stages in soybean. These include morphological traits (Keim et al., 1990; Mansur et al., 1993; Lee et al., 1996; Zhang et al., 2004), reproductive traits (Keim et al., 1990; Mansur et al., 1993), and traits for seed quality (Qiu et al., 1999), disease resistance (Concibido et al., 1994; Webb et al., 1995; Yuan et al., 2002), and yield (Chung et al., 2003; Wang et al., 2004). Moreover, several studies have focused on mapping of QTL associated with soybean pod number-related traits (Zhang et al., 2004). However, these traits were measured only after harvest, and the detected QTL were based mainly on phenotypic values at the final stage of plant development. Thus, these studies ignored the contribution to flower and pod numbers of loci underlying the expression of distinct genes at different developmental stages. In addition, these QTL cannot account for net genetic effects in a specific time interval of plant development, which are the essential components of quantitative traits (Yan et al., 1998). According to the theory of developmental genetics, genes are expressed selectively at different growth stages (Zhu, 1995). This means that different QTL might have different expression dynamics during different developmental stages. To thoroughly understand the genetic functions of QTL, we should therefore not only determine their effects at a given time or stage, but also their expression dynamics throughout all growth stages in different environments (Liu et al., 2008). The risk of the traditional approach is that the loci mapped are composites of clustered or linked loci underlying different stages of pod number development. In recent years, several studies have mapped QTL and estimated their effects on soybean developmental traits at different stages (Sun et al., 2006; Li et al., 2007; Teng et al., 2008). However, there have been no reports on QTL analysis of the dynamic development of total flower and pod numbers in soybean. This might have been for two reasons: flower and pod numbers have low heritability (Wang and Wang, 1992), and counting the numbers of flowers and pods is difficult. Previous study suggested that the heritability of flower and pod numbers was up to 70%, and the correlation coefficient between the yield and number of pods per plant was significant (r = 0.508) (Shaban, 2005). The broad-sense heritability of pod number was estimated between 22% and 32% in F2 and F3 generations (Yang, 1975). Thus, the number of seeds produced by a soybean population is an important determinant of yield

(Brevedan et al., 1978). The aim of this study was to identify unconditional and conditional QTL that determine flower and pod numbers development in soybean. These genomic regions could underlie the developmental increase in flower and pod numbers, and provide a basis for fine mapping and cloning of the genes responsible for flower and pod numbers, with the ultimate aim of increasing soybean yield.

Materials and methods Plant materials and genetic linkage map The recombinant inbred line (RIL) population used in this study, comprising 152 F8:10 lines, was derived from a cross between Bogao and Nannong94-156. The RIL population was developed via single-seed descent at the National Center for Soybean Improvement (Nanjing Agricultural University, Nanjing, China). A genetic map comprising 306 markers has been constructed in our laboratory (Zhang et al., 2009).

Plant growth and phenotypic investigation Two independent trials of the flowering and pod-forming phases of soybean were carried out in 2006 and 2007, respectively, in Nanjing, China. The sowing dates were June 20, 2006 and June 15, 2007. The 154 genotypes (two parents plus 152 RILs) were grown in a completely randomized design. Plants were sown in a 10 L pot with 4 kg dry soil, with two plants sown in each pot. There were two replicates, and each replicate contained six plants. The soil had 26.51 mg/kg available P, 0.26 g/kg of total nitrogen, 72.4 mg/kg of available K, and 12.8 g/kg of organic matter. During the experimental periods, the temperature in the greenhouse was around 25°C at night and 30°C−38°C during the day. The plants were irrigated with tap water. In this study, the parent Nannong94-156 flowered and set pods at about 41 and 86 days after sowing, respectively; the parent Bogao flowered and set pods at about 39 and 98 days, respectively. Flowers and pods were counted as described by Fu and Chen (2002) and Wang et al. (2006). The number of fresh flowers was counted between 6 AM and 9 AM every 2 days beginning at R1 (beginning of flowering; investigation starting dates were July 20, 2006

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

and July 18, 2007) until flowering ceased. The number of young pods longer than 1 cm was counted every 4 days from R3 (beginning of pod elongation; investigation starting dates were August 10, 2006 and August 7, 2007) until maturity. We measured the flower number 11 times in 2006 and 13 times in 2007. We measured pod numbers nine times in 2006 and 11 times in 2007. We used t1, t2, …, t13 to denote the corresponding stages.

547

yj(t|t–1) = µ0(t|t–1) + µ*( t|t–1)X*j +∑µi(t|t–1)Xij + εj(t|t–1), where yj(t|t–1) is the conditional phenotypic value of the jth individual; µ0(t|t–1) is the conditional population mean; µ*( t|t–1) is the conditional QTL effect; X*j is the coefficient for conditional QTL effect; µi(t|t–1) is the conditional effect for the ith marker; Xij is the coefficient for the ith marker effect; and εj(t|t–1) is the conditional residual error of the jth individual.

Data analysis and QTL mapping The composite interval mapping (CIM) program of WinQTLCart version 2.5 (Wang et al., 2005) was used to detect unconditional and conditional QTL for the two traits. The empirical thresholds were computed using the permutation test (1,000 permutations, overall error level 5%) for CIM (Churchill and Doerge, 1994). A LOD value corresponding to an experiment-wide threshold of P = 0.05 was used to declare a QTL as significant. The estimate of the QTL position was the point of maximum LOD score in the region under consideration. The genetic effect of an unconditional QTL was the net accumulation of several gene sets from the start of plant growth to the time point t. Conditional phenotypic values at time t were calculated by subtracting the phenotypic means measured at time t-1 from the mean at time t (Zhu, 1995). The derived genetic effect reflected the changes accumulating in the 2 or 4 days prior to the measurement, rather than the net genetic effect of the unconditional QTL. In this study, 152 genotypes were tested over 2 years. The unconditional QTL were assessed based on the phenotypic values at time t[y(t) ] (Zhu, 1995), in which the genetic effect was the accumulation of the individual gene effect from the initial time of plant growth to a time point t: yj(t) = µ0(t) + µ*(t)X*j + µi(t)Xij + εj(t), where yj(t) is the phenotypic value of the jth individual measured at time t; µ0(t) is the population mean at time t; µ*(t) is the cumulative QTL effect at time t; X*j is the coefficient for the QTL effect; µi(t) is the cumulative effect for the ith marker effect at time t; Xij is the coefficient for the ith marker effect; and εj(t) is the residual error of the jth individual at time t. The conditional phenotypic means [y(t|t-1)] were obtained by a mixed model approach in which the genetic effect was contributed by the specific developmental stage between time t-1 to time t (Zhu, 1995).

Results Phenotypic variation Phenotypic values for flower and pod numbers measured at different growth stages are shown in Table 1. The mean flower and pod numbers of the high value parent, Nannong94-156, were higher than those of Bogao, and the difference between the two parents for the two traits was significant at all stages of measurement. For flower number, the average was from 11.7 at t1 to 138.0 at t12 in 2006, and from 3.0 at t1 to 126.7 at t13 in 2007. Pod number of the parents and RIL population increased consistently across all stages of measurement. The average pod number of the RIL population increased from 3.2 to 61.1 in 2006 and 3.6 to 57.6 in 2007. Transgressive segregation for flower and pod numbers was prominent at all stages of measurement. In Table 1, both skew and kurtosis values of the two traits were less than 1.0 at most stages of measurement, suggesting that the segregation of the two traits fitted normal distribution.

QTL analysis for flower number Ten unconditional QTL underlying flower number at different developmental stages were identified and mapped onto eight chromosomes (2, 4, 9, 11, 15, 18, 19, and 20) based on two years’ data (Table 2). Of these QTL, three QTL were detected only in 2006 and one was detected only in 2007. Six were detected in both 2006 and 2007 (qfn-Chr4, qfn-Chr11, qfn-Chr15, qfn-Chr18-2, qfn-Chr20-1, and qfn-Chr19, referred to as common QTL). For example, qfn-Chr18-2 was detected on chromosome 18 in four consecutive stages in 2006, from t9 to t12, and in four consecutive stages in 2007, from t4 to t7. This QTL explained 18.1% of the largest phenotypic variation. Another common QTL (qfn-Chr19), which explained 15.4% of the

548

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

largest phenotypic variation, was detected on chromosome 19 in 10 consecutive stages in 2006, from t1 to t10, and in six consecutive stages in 2007, from t2 to t7. This QTL was an allele from the high value parent Nannong94-156 and greatly increased the accumulation of flower numbers on plants; however, it was not detected at the final stage. We noticed that none of the QTL mentioned could be detected at the same time at any of the growth stages, suggesting that none of the QTL expressed their effects during all growth stages in soybean. Therefore, conventional QTL mapping done only at the final stage will miss some QTL. Ten conditional QTL for flower number were identified at different developmental stages in 2006 and 2007, and mapped onto nine chromosomes (2, 3, 4, 11, 14, 15, 18, 19, and 20). Nine common conditional QTL, qfn-Chr2, qfn-Chr3, qfn-Chr4, qfn-Chr11, qfn-Chr14, qfn-Chr18-2,

qfn-Chr20-1, qfn-Chr20-2, and qfn-Chr19, were identified in both experiments. Among them, qfn-Chr18-2 on chromosome 18 was detected in four consecutive stages in 2006, from t4 to t7, and in six consecutive stages in 2007, from t7 to t13. qfn-Chr2 was detected at two developmental stages (t1 and t11) in 2006 and four QTL, qfn-Chr14, qfn-Chr2, qfn-Chr15, and qfn-Chr3, were detected at two consecutive developmental stages in 2007. This might indicate that the genes were expressed at different stages in each year. In addition, 10 QTL detected by the unconditional method, except for qfn-Chr18-1 and qfn-Chr9, were also detected by the conditional method, which indicates that these QTL were expressed mainly at one or more specific stages. The fact that qfn-Chr18-1 and qfn-Chr9 were not detected by the conditional method might reflect their minute effects at the specific period. In contrast, qfn-Chr14

Table 1 Statistical analyses of soybean flower and pod numbers for two parents and recombinant inbred line (RIL) population at different growth stages Parent mean

RIL population

Traits

Stage 2006

2007

2006

2007

2006

2007

2006

2007

2006

2007

Flower number

t1

004.6

01.1

001.7

000.4

011.7 ± 9.67

002.9 ± 2.58

01.0

00.9

00.2

−0.3

t2

017.7

03.1

011.5

003.1

034.4 ± 21.29

010.1± 7.24

00.6

01.0

−0.4

00.8

t3

032.8

06.3

022.9

014.5

051.4 ± 25.12

016.1 ± 10.46

00.3

00.7

−0.4

−0.2

t4

047.6

12.2

059.6

027.0

067.5 ± 26.5

026.3 ± 18.04

00.0

00.5

00.2

−0.8

t5

055.4

30.9

086.5

059.9

079.5 ± 27.27

039.5 ± 23.30

−0.3

00.4

01.0

−0.8

t6

061.5

41.9

107.8

078.7

089.6 ± 26.33

054.1 ± 26.95

00.0

00.2

00.4

−0.6

t7

067.5

49.7

127.3

095.9

100.9 ± 27.79

065.0 ± 29.35

00.0

00.1

−0.2

−0.5

t8

074.4

59.0

142.7

116.2

110.9 ± 30.16

079.8 ± 33.27

00.0

00.0

−0.1

−0.1

t9

086.8

70.3

156.8

128.0

120.9 ± 34

092.5 ± 36.61

00.0

−0.1

−0.1

00.3

t10

095.9

72.8

166.1

140.5

128.4 ± 38.63

103.3 ± 39.30

00.1

−0.2

−0.1

00.6

t11

098.4

80.3

174.4

148.7

133.9 ± 42.87

110.9 ± 41.50

00.3

−0.2

00.0

00.7

t12

102.0

92.8

176.5

156.9

137.9 ± 46.44

120.9 ± 44.87

00.3

−0.1

−0.2

01.0

t13

0

/

96.5

0

/

158.5

0

/

126.6 ± 46.68

00

/

−0.1

0

/

01.0

t1

000

00

000

000

003.2 ± 2.29

003.6 ± 3.49

01.0

01.2

−0.3

−0.4

t2

001.6

02.2

010.4

000.0

007.8 ± 6.67

007.3 ± 9.63

00.8

01.1

00.3

−0.1

t3

002.7

16.6

035.2

005.6

013.9 ± 11.68

015.3 ± 12.53

00.9

00.9

01.3

−0.4

t4

020.6

24.7

062.0

011.1

027.1 ± 15.35

020.2 ± 16.16

00.3

00.6

−0.1

−0.8

Pod number

BG

Mean (± SD)

NN94-156

Skew

Kurtosis

t5

026.8

42.3

080.0

028.6

037.0 ± 14.34

025.8 ± 23.20

00.2

00.4

00.3

−1.0

t6

043.4

52.9

094.0

064.0

045.4 ± 12.98

032.3 ± 24.57

00.5

00.2

00.9

−1.0

t7

060.9

54.5

100.4

081.5

053.9 ± 18.11

038.7 ± 22.93

00.1

00.0

−0.4

−0.9

t8

062.0

56.4

100.4

088.3

059.8 ± 21.74

045.2 ± 19.36

00.1

00.1

−0.7

00.6

t9

060.8

57.0

100.4

091.5

0061.1 ± 21.82

052.4 ± 18.16

00.1

00.3

−0.7

00.1

t10

00

/

57.0

0

/

091.5

0

/

056.9 ± 17.10

00

/

00.3

0

/

00.5

t11

00

/

57.0

0

/

091.5

0

/

059.6 ± 17.12

00

/

00.3

0

/

00.4

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

549

Table 2 Unconditional and conditional soybean QTL for flower number at different stages of measurement 2006 QTL

Marker interval

Stage

T LOD

qfn-Chr2

BE475343-Satt095

Satt080-Satt009

Add.

R%

LOD 2.7

t1

3.0

−2.5

6.6

t2

3.6

−4.9

10.2

t4

3.5

−6.5

7.5

Sct_046-Sat_207

2

Add.

R%

−2.4

7.9

3.7

1.6

8.9

4.1

−1.9

12.1

3.2

−2.5

8.2

LOD

Add.

t|t-1 2

R%

qfn-Chr11

Satt273-Satt710

Satt197-Satt251

qfn-Chr15

Sat_264-Sat_177 Sat_124-Satt598

qfn-Chr18-1 Satt352-Satt303

qfn-Chr18-2 Satt138-Satt564

5.5

1.3

15.9

3.1

1.5

9.3

−2.4

10.3

−1.7

10.3

3.1

−2.2

9.9

t12

3.5

−3.1

11.9

t13

5.6

−2.4

16.5

t8

2.8

−8.3

8.7

t13

2.8

−10.7

6.3

t9

3.1

−9.3

7.7

t10

4.0

−10.9

9.6

t11

4.3

−11.8

10.2

t12

5.0

−13.6

11.7

t13

5.1

−14.2

11.9

5.6

1.8

21.1

t1

3.1

3.1

9.7

t9

qfn-Chr14

R2%

3.2

t9 qfn-Chr9

Add.

LOD

3.8

t10 t11

qfn-Chr4

tt

t|t-1 2

t11 qfn-Chr3

2007

3.2

3.0

9.7

4.4

−3.2

13.9

t3

4.0

−1.8

9.9

4.0

−1.8

9.9

t4

5.8

−2.5

13.4

5.8

−2.5

13.4

2.7

−0.9

6.5

t11

3.6

2.0

13.5

t12

2.8

2.3

6.7

4.3

2.2

15.1

5.0

2.6

14.5

t2

3.4

−7.5

12.1

4.0

3.5

10.6

t3

4.0

−8.8

12.7

3.6

4.4

9.7

t4

3.1

−8.2

11.1

t5

3.5

−9.1

13.9

t6

3.2

−9.4

14.6

t6

2.9

6.3

10.9

t7

3.2

7.7

9.1

t8

3.3

8.3

8.8

t4

3.0

2.6

9.8

t5

7.0

2.8

18.0

3.8

3.8

8.6

t6

7.1

2.6

18.6

3.8

5.7

9.0

t7

4.8

2.1

14.0

2.7

6.1

6.4

2.7

1.9

6.4

t9

4.0

10.0

9.7

t10

5.0

12.9

18.1

6.0

2.7

13.0

t11

4.3

13.7

10.7

5.6

2.1

15.7

t12

3.7

13.7

9.0

4.6

2.9

10.7

5.7

2.0

12.4

t13

(to be continued on the next page)

550

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

Table 2 (Continued) 2006 QTL

Marker interval

Stage

t LOD

qfn-Chr19

Satt166-Dt1

2007 t

t|t-1

Add.

2

R%

LOD 2.7

2

Add.

R%

2.4

6.1

2.7

2.6

7.1

3.1

5.9

7.9

2.8

2.5

6.4

t3

3.5

7.0

8.4

4.1

4.1

15.3

t4

4.0

7.2

8.8

3.3

5.3

10.7

t5

3.0

6.5

7.4

2.9

6.3

6.8

t6

4.7

8.3

11.8

3.0

7.0

9.7

t7

5.3

9.5

13.3

2.8

9.5

8.5

t8

5.0

10.6

13.9

t9

3.9

10.8

11.3

t10

3.1

11.2

9.0

3.1

−2.1

7.6

t1

4.1

−3.4

11.0

3.2

−2.4

7.4

4.0

−3.4

11.2

Sat_421-Satt330

2.7

9.0

6.5

2.8

9.5

6.5

3.0

−1.0

8.6

t2

4.8

−7.9

13.8

6.3

−4.2

17.6

t3

5.0

−9.0

13.9

5.8

−5.2

14.1

t4

3.5

−7.5

9.4

3.8

−6.2

9.5

t5

4.9

−8.6

11.9

t6

3.8

−8.7

9.6

3.9

−2.5

13.6

t7

3.1

−2.4

8.7

3.0

−8.4

7.5

t8

3.2

−2.3

9.5

3.3

−9.5

8.3

t9

2.7

−2.3

7.3

t10

3.9

−2.7

12.0

t13 qfn-Chr20-2

R%

t2

t12 Sat_155-Sat_420

Add.

t1

t11 qfn-Chr20-1

LOD

t|t-1 2

t2

3.0

−5.5

6.6

t3

3.8

−7.7

10.0

t8

3.3

2.9

15.2

t9

3.9

3.0

13.0

t11

3.0

1.8

10.5

R2%

LOD

Add.

2.8

0.8

6.4

7.1

4.1

25.6

3.5

2.2

14.5

4.4

2.3

9.3

Add., additive effect; t, unconditional QTL; t|t – 1, conditional QTL.

and qfn-Chr3, which were undetected by the unconditional method, were detected by the conditional method during stages t2-t3 and t10-t11. This might be explained by the fact that the net effects of QTL in t2-t3 or t10-t11 stages were partially counteracted by the cumulative effects before the t2 or t10 stages. As shown in Table 2, the main effect QTL, qfn-Chr18-2, qfn-Chr19, and qfn-Chr20-1, were stably detected in association with flower number accumulation across the two years. Phenotypic values of 10 typical inbred lines with high or low flower number at different

developmental stages are shown in Fig. 1. The relationship between phenotypic values and associated molecular markers indicated that the accuracy of markers Satt138, Satt564, Satt166, Dt1, Sat_155, and Sat_420 (linked to QTL qfn-Chr18-2, qfn-Chr19 and qfn-Chr20-1) for selecting flower number in soybean were 100%, 90%, 100%, 80%, 100%, and 100%, respectively. The favorable alleles for qfn-Chr18-2 and qfn-Chr19 were all derived from the high value parent Nannong94-156 and the allele for qfn-Chr20-1 was derived from the low value parent Bogao.

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

551

Fig. 1. The major soybean QTL (by LOD score and additive effect) associated with flower number, the markers used for selection, and their accuracy of application in the breeding program. t1, t2, …, t13 denote the stages of measurement. denotes DNA band of Nannong94-156 type, denotes DNA band of Bogao type. Percent denotes the accuracy of the markers for selection in typical inbred lines.

QTL underlying pod number Eight unconditional QTL underlying pod number at different developmental stages were detected in 2006 and 2007 (Table 3). These loci were located on seven chromosomes (2, 3, 5, 8, 11, 15, 18, and 20). Of these, four QTL (qpn-Chr11, qpn-Chr15, qpn-Chr18, and qpn-Chr20) were detected in both years. The main effect common QTL, qpn-Chr11, was detected on chromosome 11 in seven stages in 2006, and in eight consecutive stages in 2007. This QTL explained more than 30% of phenotypic variation in two experiments, which greatly increased the accumulation of pods. Another main effect common QTL, qpn-Chr20, was detected on chromosome 20 in seven consecutive stages in 2006, from t3 to t9, and in six consecutive stages in 2007, from t2 to t7. This QTL explained more than 10% of phenotypic variation at different stages and years. However, these QTL (qpn-Chr2, qpn-Chr3, qpn-Chr5, qpn-Chr8, qpn-Chr15 and qpn-Chr18) were not detected at the final stage (Table 3). Only two unconditional QTL, qpn-Chr11 and qpn-Chr20, were detected at the final measuring stage (t9 in 2006 and t12 in 2007). This result also indicated that conventional QTL mapping done only at the final stage will miss some QTL. Seven conditional QTL underlying pod number were detected and mapped onto seven chromosomes (3, 5, 11, 16, 18, 19 and 20) at different developmental stages in

2006 and 2007. Among them, qpn-Chr11, qpn-Chr18, and qpn-Chr20 were detected in both years. qpn-Chr11 was expressed in two continuous developmental stages in 2006 (t6 and t7). This QTL accounted for 14.8% of the phenotypic variation at t6; however, at t7, this QTL explained 19.3% of the phenotypic variation. This indicated that the net genetic effects were different at every specific stage and the relevant genes were expressed continuously through two measuring stages, rather than being the cumulative result of genetic effects for the two specific growth stages for unconditional QTL. In addition, only one conditional QTL, qpn-Chr16, was detected at the final measuring stage t9 in 2006, and was mapped onto chromosome 16. This QTL accounted for 12.4% of the phenotypic variation. The main effect QTL, qpn-Chr11 and qpn-Chr20, were stably detected in association with pod accumulation across the two years (Table 3). Phenotypic values of ten typical inbred lines with high or low pod number at different developmental stages are shown in Fig. 2. The relationship between phenotypic values and associated molecular markers indicated that the accuracy of markers Satt197, Satt251, Sat_155 and Sat_420 (linked to QTL qpn-Chr11 and qpn-Chr20) for selecting pod number in soybean were 100%, 100%, 80%, and 90%, respectively. The favorable alleles for QTL qpn-Chr11 were all derived from the high value parent Nannong94-156.

552

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

Table 3 Unconditional and conditional soybean QTL for pod number at different stages of measurement 2006 QTL

Marker interval

Stage

T LOD

Add.

R%

BE475343-Satt095

t1

3.4

−1.4

8.3

qpn-Chr3

Satt234-Sat_239

t4

3.9

−5.6

12.1

qpn-Chr8 qpn-Chr11

Satt276-Sat_368 Sat_215-Sat_409 Satt197-Satt251

t

t|t-1 2

qpn-Chr2

qpn-Chr5

2007

t5

3.1

−4.4

8.4

t6

2.8

−3.8

8.0

t1

3.0

−1.4

12.4

t2

7.3

−3.7

20.4

LOD

2

Add.

R%

3.5

1.9

13.9

3.4

−1.4

14.9

Satt231-Sat_381

Add.

R%

t4

3.9

−6.0

14.4

t5

3.0

−5.1

11.9

t1

2.6

1.4

11.0

2.9

1.7

10.1

t2

3.4

2.6

10.0

2.9

3.1

10.6

5.9

8.1

20.0

t4

3.7

5.5

12.2

7.8

11.9

30.9

t5

3.3

5.5

14.1

7.2

12.8

28.9

t6

5.8

11.3

26.9

4.6

3.3

14.8

8.3

15.0

34.7

t7

4.3

7.9

19.0

5.0

3.8

19.3

t8

5.9

11.3

31.0

t9

5.8

11.3

26.9

t3

3.5

−3.2

7.9

t4

4.3

−4.8

9.0

t5

3.6

−4.3

8.2

t6

3.6

−3.9

8.5

t3

qpn-Chr15

LOD

t|t-1 2

6.5

13.2

31.8

4.5

10.0

27.0

t10

3.2

−4.8

8.2

t11

2.8

−4.4

7.2

5.0

−3.3

11.9

t2

5.1

−6.5

12.2

t3

4.4

−6.5

8.9

qpn-Chr16

Satt596-Satt622

t9

qpn-Chr18

Satt138-Satt564

t1

2.8

−1.2

8.0

4.7

−0.4

12.4

4.2

−1.5

12.2

t4

5.9

−1.9

15.5

t8

2.7

−2.3

9.5

t10 qpn-Chr19

Satt373-Satt495

t10

qpn-Chr20

Sat_155-Sat_420

t2

3.5

−3.4

12.5

t3

4.4

−4.1

12.6

6.0

−7.5

16.2

t4

5.5

−6.1

13.8

4.3

−7.3

10.5

t5

5.0

−5.6

13.0

3.2

−7.2

8.8

t6

3.4

−4.4

10.3

3.6

−8.2

10.9

t7

3.0

−5.5

8.5

2.7

−6.3

6.9

t8

4.0

−7.3

10.6

3.6

−7.7

15.5

t9

4.0

−7.3

10.6

3.5

−7.0

15.2

t10

2.7

−5.6

11.6

t11

2.9

−5.4

11.0

t12

2.9

−6.2

14.4

Add., additive effect; t, unconditional QTL; t|t – 1, conditional QTL.

Add.

R2%

4.4

3.7

18.4

4.5

1.7

12.2

2.7

1.0

6.8

LOD

3.1

1.4

12.6

4.7

−1.8

11.9

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

553

Fig. 2. The major soybean QTL (by LOD score and additive effect) associated with pod number, the markers used for selection, and their accuracy of application in the breeding program. t1, t2, …, t11 denote the corresponding stages of measurement. denotes DNA band of Nannong94-156 type, denotes DNA band of Bogao type. Percent denotes the accuracy of the markers for selection in typical inbred lines.

Discussion Genetic basis determining flower and pod numbers development The theory of developmental genetics considers that QTL might have expression dynamics during trait development, even though they might have the same final effects (Zhu, 1995; Atchley and Zhu, 1997). Many complex traits develop through the actions of genes that might behave differently during growth periods and across environments (Atchley and Zhu, 1997). Although molecular marker techniques have provided powerful tools to dissect these complex traits, it is still difficult to directly handle different developmental stages across various environments for QTL mapping (Yan et al., 1998; Wu et al., 1999). In previous studies, many soybean QTL were determined at a single developmental stage, usually the mature stage, in which most of the genetic information might remain undetected. In this study, the number of QTL related to soybean flower and pod numbers and their genetic effects varied at different stages of measurement. More significant QTL were detected across all developmental stages than at any one specific stage. Most QTL controlling flower and pod numbers were not expressed, or their

genetic effects were not significant, at the final stage. For example, some major QTL such as qpn-Chr11, qfn-Chr19 and qfn-Chr20-1, which significantly affected flower and pod numbers, were only detected at the early stages of measurement, and were undetected at the final stage. This indicates that a parallel between gene expression and dynamic QTL expression exists, and that some major QTL of soybean may have been neglected by previous studies that only surveyed the mature stage. Similar conclusions have been reported by Yan et al. (1998) for the rice tiller trait, by Sun et al. (2006) for soybean plant height and pod number of the main stem, and by Li et al. (2007) for soybean protein and oil content. Our results also revealed that gene expression exhibited multiple patterns; some gene effects were maintained for long periods of time, and others disappeared quickly. The dynamic expression process of flower and pod numbers implied that QTL affecting these traits might be governed by time-dependent gene expression. Consistent with our results, some QTL developed for rice tiller number consistently existed for more than five measuring stages (Yan et al., 1998). qfn-Chr11 and qfn-Chr4, on the other hand, were detected only at one stage, indicating specific gene expression at a specific stage of measurement. We also detected some conditional QTL as unconditional QTL. For example, unconditional QTL qfn-Chr18-2,

554

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

which significantly affected flower number, was also detected continuously by the conditional mapping method. However, other conditional QTL were not found by traditional unconditional analysis. The unconditional QTL explained the cumulative gene actions from initial time to time t; therefore, the variation of cumulative gene effects might be diminished if genes with opposite genetic effects were expressed at the same or nearby locations. This might explain why many QTL were not detected by the unconditional mapping method, especially at the mature stage in this study. To breed high-yield cultivars, a new approach would be to identify the major QTL that facilitated high yield at different developmental stages before maturity, rather than using the QTL detected at the mature stage, which will reduce the influence of opposite genetic effects. In addition, we found some conditional QTL at various stages, for example, qfn-Chr2 appeared at t1, t2, and t4. This indicates that some genes might repeat their expression at different times. QTL for flower and pod numbers detected on chromosomes 11, 18, and 19 mostly possessed positive additive effects, whereas QTL detected on other chromosomes mostly possessed negative additive effects. In addition, the total net additive genetic effects of QTL were parallel to the apparent peaks and troughs in the numbers of flowers and pods. This indicates that flower development and pod

number is affected by the increase of some genetic effects, together with the decrease of other genetic effects. A better understanding of the dynamic expression mechanism of QTL will therefore be helpful for artificial regulation of QTL expression, with the eventual aim of effective use of QTL.

QTL for flower and pod numbers and related genes In this study, two major QTL for pod number on chromosomes 11 and 20, as well as QTL qpn-Chr3, qpn-Chr16 and qpn-Chr19, were detected for pod number of the main stem by Sun et al. (2006). In addition, Zhang et al. (2004) also found a QTL on chromosome 16 associated with formation of pod per node. On the basis of the genetic position of these QTL and the available sequence annotation in Soybase (http://www. soybase.org/), the relationship between the QTL detected for these two traits and related genes reported in other plants can be visualized by comparative genomics and bioinformatics. These QTL regions include six homologs of JMJC, five homologs of SEC14, three homologs of WD40, two homologs of PEBP-PKIP, and one homolog of an F-box gene (Hepworth et al., 1999; Mo et al., 2006; Lin et al., 2007; Sitaraman et al., 2008; Sun et al., 2008; Yu et al., 2008). As shown in Table 4, these related genes are

Table 4 Flower and pod numbers-related genes located in the regions of five main soybean QTL QTL region

Related genes

Function

Homologs

Reference

qfn-Chr2

DQ865290

Modulated flower development

PEBP-PKIP

Lin et al., 2007

DQ865291

Modulated flower development

PEBP-PKIP

Lin et al., 2007

NM_128829

Flower development

WD40

Sitaraman et al., 2008

qfn-Chr3

NM_128829

Flower development

WD40

Sitaraman et al., 2008

qfpn-Chr11

NP_001065492

Flower development regulatory

JMJC

Sun et al., 2008

NP_680116

Relative of early flower 6

JMJC

Yu et al., 2008

Ak176420

Predominately transcribed in flowers

SEC14

Mo et al., 2007

AY057587

Predominately transcribed in flowers

SEC14

Mo et al., 2007

AY050419

Predominately transcribed in flowers

SEC14

Mo et al., 2007

BT000834

Predominately transcribed in flowers

SEC14

Mo et al., 2007

Np_565514

Flower development

SEC14

Mo et al., 2007

NP_001065492

Flower development regulatory

JMJC

Sun et al., 2008

NP_680116

Relative of early flower 6

JMJC

Yu et al., 2008

NM_102834

Required for floral meristem

F-box

Hepworth et al., 1999

NM_128829

Flower development

WD40

Sitaraman et al., 2008

NP_001065492

Flower development regulatory

JMJC

Sun et al., 2008

NP_680116

Relative of early flower 6

JMJC

Yu et al., 2008

qfn-Chr19

qpn-Chr20-1

The accession number was determined by searching the NCBI database using the BLAST algorithm in the NCBI database (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

distributed throughout the whole soybean genome, as well as in some hotspots. Seven related genes located at the interval of qfn-Chr11 comprise five SEC14-homologous genes and two JMJC-homologous genes. The predicted function of qfn-Chr11 is the regulation of flower development and control of early flowering; qfn-Chr19 is also predicted to regulate flower development, as well as being required for floral meristem development. The predicted function of qfn-Chr2 is the control of flower development and modulation of flowering time. This suggests that the mechanism of action of these QTL in flower development might be to stimulate cooperation between different flower development regulatory factors. In conclusion, analysis of QTL in a segregating population can identify markers linked to QTL that could be used in marker-assisted selection for increasing soybean flower and pod numbers, with the ultimate aim of increasing soybean yield. In addition, QTL analysis for flower and pod numbers could also lead to the prediction of candidate genes determining flower and pod development in the detected QTL regions.

Acknowledgments This work was supported by the National Basic Research Program of China (Nos. 2010CB125906 and 2009CB118400), the National High-Tech Research Program of China (Nos. 2006AA10Z1C1 and 2008AA10Z153), the National Natural Science Foundation of China (No. 30771362), and the 111 Program from the Ministry of Education (No. B07030).

References Atchley, W.R., and Zhu, J. (1997). Developmental quantitative genetics, conditional epigenetic variability and growth in mice. Genetics 147: 765−776. Board, J.E., and Tan, Q. (1995). Assimilatory capacity effects on soybean yield components and pod number. Crop Sci. 35: 846−851. Brevedan, R., Egli, D., and Leggett, J. (1978). Influence of N nutrition on flower and pod abortion and yield of soybeans. Agron. J. 70: 81−84. Chung, J., Babka, H., Graef, G., Staswick, P., Lee, D., Cregan, P., Shomeaker, R., and Specht, J. (2003). The seed protein, oil, and yield QTL on soybean linkage group I. Crop Sci. 43: 1053−1067. Churchill, G.A., and Doerge, R.W. (1994). Empirical threshold values for quantitative triat mapping. Genetics 138: 963−971.

555

Concibido, V.C., Denny, R.L., Boutin, S.R., Hautea, R., Orf, J.H., and Young, N.D. (1994). DNA marker analysis of loci underlying resistance to soybean cyst nematode (Heterodera glycines Ichinohe). Crop Sci. 34: 240−246. Fu, Y.H., and Chen, Y.J. (2002). The relationship between flower pod abscission rate and growth stage of varieties. Crop J. 5: 12−13. Hepworth, S., Klenz, J., and Haughn, G. (2006). UFO in the Arabidopsis inflorescence apex is required for floral-meristem identity and bract suppression. Planta 223: 769−778. Keim, P., Diers, B.W., Olson, T.C., and Shoemaker, R.C. (1990). RFLP mapping in soybean: association between marker loci and variation in quantitative traits. Genetics 126: 735−742. Lee, S.H., Bailey, M.A., Mian, M.A.R., Shipe, E.R., Ashley, D.A., Parrott, W.A., Hussey, R.S., and Boerma, H.R. (1996). Identification of quantitative trait loci for plant height, lodging, and maturity in a soybean population segregating for growth habit. Theor. Appl. Genet. 92: 516−523. Li, W., Sun, D., Du, Y., Chen, Q., Zhang, Z., Qiu, L., and Sun, G. (2007). Quantitative trait loci underlying the development of seed composition in soybean (Glycine max L. Merr.). Genome 50: 1067−1077. Lin, M., Belanger, H., Lee, Y., Varkonyi-Gasic, E., Taoka, K., Miura, E., Xoconostle-Cazares, B., Gendler, K., Jorgensen, R., and Phinney, B. (2007). FLOWERING LOCUS T protein may act as the long-distance florigenic signal in the cucurbits. Plant Cell 19: 1488−1506. Liu, G., Zeng, R., Zhu, H., Zhang, Z., Ding, X., Zhao, F., Li, W., and Zhang, G. (2008). Dynamic expression of nine QTL for tiller number detected with single segment substitution lines in rice. Theor. Appl. Genet. 147: 1432−2242. Mansur, L.M., Lark, K.G., Kross, H., and Oliveira, A. (1993). Interval mapping of quantitative trait loci for reproductive, morphological, and seed traits of soybean (Glycine max L. Merr.). Theor. Appl. Genet. 86: 907−913. Mo, P., Zhu, Y., Liu, X., Zhang, A., Yan, C., and Wang, D. (2007). Identification of two phosphatidylinositol phosphatidylcholine transfer protein genes that are predominately transcribed in the flowers of Arabidopsis thaliana. Plant Physiol. 164: 478−486. Qiu, B.X., Arelli, P.R., and Sleper, D.A. (1999). RFLP markers associated with soybean cyst nematode resistance and seed composition in a ‘Peking’ × ‘Essex’ population. Theor. Appl. Genet. 98: 356−364. Shaban, N. (2005). Analysis of the correlation and regression coefficients of the interaction between yield and some parameters of snap beans plants. Trakia Journal of Sciences 3: 27−31. Sitaraman, J., Bui, M., and Liu, Z. (2008). LEUNIG_HOMOLOG and LEUNIG perform partially redundant functions during Arabidopsis embryo and floral development. Plant Physiol. 147: 672−681. Song, Q.J., Marek, L.F., Shoemaker, R.C., Lark, K.G., Concibido, V.C., Delannay, X., Specht, J.E., and Cregan, P.B. (2004). A new integrated genetic linkage map of the soybean. Theor. Appl. Genet. 109: 122−128. Sun, D., Li, W., Zhang, Z., Chen, Q., Ning, H., Qiu, L., and Sun, G. (2006). Quantitative trait loci analysis for the developmental behavior of soybean (Glycinemax L. Merr.). Theor. Appl. Genet. 112: 665−673. Sun, Q., and Zhou, D. (2008). Rice jmjC domain-containing gene JMJ706 encodes H3K9 demethylase required for floral organ devel-

556

Dan Zhang et al. / Journal of Genetics and Genomics 37 (2010) 545−556

opment. Proc. Natl. Acad. Sci. USA 105: 13679−13684. Teng, W., Han, Y., Du, Y., Sun, D., Zhang, Z., Qiu, L., Sun, G., and Li, W. (2008). QTL analyses of seed weight during the development of soybean (Glycine max L. Merr.). Heredity 102: 372−380. Wang, D., Graef, G., Procopiuk, A., and Diers, B. (2004). Identification of putative QTL that underlie yield in interspecific soybean backcross populations. Theor. Appl. Genet. 108: 458−467. Wang, L., and Wang, J. (1992). Genetics and Breeding in Soybean, 1st edn. (Beijing, China: Science Press), pp. 60−65 (in Chinese). Wang, S., Basten, C., and Zeng, Z. (2005). Windows QTL Cartographer 2.5. (Department of Statistics. North Carolina State University, Raleigh, NC). Wang, S.Q., Gao, J.L., Liu, K.L., and Zhang, S. (2006). Research on bean pod initiation and abscission in soja bean. Inner Mongolia Agricultural Science and Technology 2: 10 (in Chinese with an English abstract). Webb, D., Baltazar, B., Rao-Arelli, A., Schupp, J., Clayton, K., Keim, P., and Beavis, W. (1995). Genetic mapping of soybean cyst nematode race-3 resistance loci in the soybean PI 437.654. Theor. Appl. Genet. 91: 574−581. Wu, W., Li, W., Tang, D., Lu, H., and Worland, A. (1999). Time-related mapping of quantitative trait loci underlying tiller number in rice. Genetics 151: 297−303. Yan, J., Zhu, J., He, C., Benmoussa, M., and Wu, P. (1998). Quantitative trait loci analysis for the developmental behavior of tiller number

in rice (Oryza sativa L.). Theor. Appl. Genet. 97: 267−274. Yang, Q. (1975). Analysis on genetic variation in early generation of the cross materials for important agronomic traits. Acta Genetica Sinica 2: 225−230 (in Chinese with an English abstract). Yu, X., Li, L., Guo, M., Chory, J., and Yin, Y. (2008). Modulation of brassinosteroid-regulated gene expression by jumonji domain containing proteins ELF6 and REF6 in Arabidopsis. Proc. Natl. Acad. Sci. USA 105: 7618−7623. Yuan, J., Njiti, V., Meksem, K., Iqbal, M., Triwitayakorn, K., Kassem, M., Davis, G., Schmidt, M., and Lightfoot, D. (2002). Quantitative trait loci in two soybean recombinant inbred line populations segregating for yield and disease resistance. Crop Sci. 42: 271−277. Zhang, D., Cheng, H., Geng, L., Kan, G., Cui, S., Gai, J., and Yu, D. (2009). Detection of quantitative trait loci for phosphorus deficiency tolerance at soybean seedling stage. Euphytica 167: 313−322. Zhang, W.K., Wang, Y.J., Luo, G.Z., Zhang, J.S., He, C.Y., Wu, X.L., Gai, J.Y., and Chen, S.Y. (2004). QTL mapping of ten agronomic traits on the soybean (Glycine max L. Merr.) genetic map and their association with EST markers. Theor. Appl. Genet. 108: 1131−1139. Zhu, J. (1995). Analysis of conditional genetic effects and variance components in developmental genetics. Genetics 141: 1633−1639. Zhu, J. (1999). Mixed model approaches of mapping genes for complex quantitative traits. Journal of Zhejiang University (Natural Sciences Edition) 33: 327−335 (in Chinese with an English abstract).