Mol Breeding (2015) 35:6 DOI 10.1007/s11032-015-0200-1
Native Fusarium head blight resistance from winter wheat cultivars ‘Lyman,’ ‘Overland,’ ‘Ernie,’ and ‘Freedom’ mapped and pyramided onto ‘Wesley’-Fhb1 backgrounds Jonathan T. Eckard • Jose L. Gonzalez-Hernandez • Melanie Caffe • William Berzonsky • William W. Bockus G. Francois Marais • P. Stephen Baenziger
•
Received: 28 May 2014 / Accepted: 14 October 2014 / Published online: 15 January 2015 Ó Springer Science+Business Media Dordrecht 2015
Abstract Pyramiding QTL from multiple sources for FHB resistance presents an opportunity to enhance the FHB resistance of elite wheat germplasm. Mapping FHB resistance QTL directly in wheat breeding populations would eliminate the need for purpose-built mapping populations and thus accelerate markerassisted pyramiding efforts. In this study, we demonstrate that multiple QTL for FHB resistance can be mapped directly in early-generation breeding populations by application of identical-by-descent (IBD)based linkage mapping. IBD-based linkage analysis was conducted using 565 segregating F1 progeny
J. T. Eckard J. L. Gonzalez-Hernandez (&) M. Caffe W. Berzonsky Department of Plant Sciences, South Dakota State University, Brookings, SD, USA e-mail:
[email protected] Present Address: W. Berzonsky Bayer Crop Sciences, Lincoln, NE, USA W. W. Bockus Department of Plant Pathology, Kansas State University, Manhattan, KS, USA G. F. Marais Department of Plant Sciences, North Dakota State University, Fargo, ND, USA P. S. Baenziger Department of Agronomy and Horticulture, University of Nebrasaka-Lincoln, Lincoln, NE, USA
derived from 28 four-way crosses among Fhb1 donor lines and multiple native sources of resistance including Lyman, Overland, Ernie, and Freedom. A total of 15 QTL for FHB resistance were mapped on chromosomes 1A, 1B, 2A, 3A, 3B, 4A, 4B, 4D, 5A, 6A, 6D, and 7D, including known loci Fhb1, Fhb5, and Rht-B1. QTL conferring native resistance in the cultivars Lyman and Overland are mapped for the first time in this study, including a QTL on chromosome 1AS (Qfhb.sdsu-1A) explaining between 4.5 and 9.9 % of the phenotypic variance in all evaluations. Marker haplotypes for these QTL regions can be used to conduct marker-assisted selection and fixation of resistance alleles in subsequent generations of these breeding populations. Keywords Winter wheat Fusarium head blight QTL mapping Native resistance
Introduction Fusarium head blight (FHB), caused predominantly by Fusarium graminearum Schwabe, is a disease of considerable economic and health concern in wheat growing regions of the world due to yield and quality loss as well as mycotoxin accumulation in infected grain (Bai and Shaner 2004). The development of resistant wheat varieties through conventional breeding programs is a primary focus of ongoing efforts aimed at minimizing the impact of the disease. Host resistance to FHB is a complex trait conditioned by oligogenic to
123
6 Page 2 of 16
polygenic inheritance, and quantitative trait loci (QTL) have been identified on nearly every wheat chromosome (Liu et al. 2009; Lo¨ffler et al. 2009; Buerstmayr et al. 2009). Several components of FHB resistance have been defined (Mesterha´zy et al. 2005), including resistance to initial infection (type I) and resistance to spread of infection throughout the spike (type II). A major effect QTL on chromosome 3B identified from resistant Chinese landraces, designated as Fhb1, has been extensively exploited in spring wheat breeding programs to develop resistant cultivars. A USDA markerassisted backcrossing project has recently introgressed Fhb1 into several elite hard winter wheat (HWW) and soft winter wheat (SWW) cultivars with the intent of enhancing FHB resistance of adapted winter wheat germplasm (Cai et al. 2013; Jin et al. 2013). Screening of adapted germplasm in the USA has also identified several cultivars with native FHB resistance. U.S. SWW cultivars ‘Ernie’ and ‘Freedom’ have been shown to have moderate levels of FHB resistance that is unrelated to Fhb1 (Jin et al. 2013). Previous mapping studies have identified quantitative trait loci (QTL) for FHB resistance on chromosomes 3B, 4B, and 5A from ‘Ernie’ (Liu et al. 2007; Abate et al. 2008) and a QTL on chromosome 2A from ‘Freedom’ (Gupta et al. 2001). U.S. HWW cultivars ‘Lyman’ and ‘Overland’ have also been shown to have moderate levels of FHB resistance (Jin et al. 2013), although no putative QTL have been reported for these cultivars to date. It has been demonstrated that pyramiding other resistance QTL with Fhb1 provides enhanced resistance to FHB (Miedaner et al. 2006; Shi et al. 2008; Burlakoti et al. 2009). Therefore, pyramiding these native sources of FHB resistance with Fhb1 should present an opportunity to further enhance FHB resistance of adapted wheat germplasm. Efficient pyramiding of native FHB resistance QTL with Fhb1 will require the use of genetic markers linked to the QTL to facilitate marker-assisted selection. While there is certainly no limitation in the ability to map resistance QTL, the ability to transition between QTL mapping and marker-assisted selection has been sparse (Beavis 1998; Bernardo 2008). This is largely because QTL mapping in plants has been confined to experimental populations derived from a single cross between two divergent (e.g., resistant and susceptible) inbred lines. While these experimental populations provide a simplistic and powerful setting for mapping QTL, they have a number of serious limitations hindering marker-
123
Mol Breeding (2015) 35:6
assisted selection efforts. First, QTL effects estimated from experimental mapping populations represent only the fixed effects of allele substitution between the parental genotypes and thus cannot be inferred for other genetic backgrounds (Xie et al. 1998). Different mapping populations segregate for different QTL, and the actual relevance and effects of the QTL detected in these mapping populations depend on the QTL alleles, allelic frequencies, and epistatic interactions present in the population targeted for selection. Second, the linkage phase between marker and QTL alleles in an experimental mapping population cannot be inferred for other genetic backgrounds. Unless there is strong linkage disequilibrium between the marker and QTL in a breeder’s germplasm, the breeder has no relevant information on which marker alleles to select in order to improve resistance in their populations. Finally, the development and evaluation of large purpose-built mapping populations are time consuming and expensive, which means that resources are detracted from breeding efforts aimed at cultivar development (Crepieux et al. 2005). To avoid these limitations associated with the use of experimental populations, QTL for FHB resistance could be mapped directly in early-generation breeding populations developed from multiple sources of resistance. Mapping resistance QTL in early generations would enable marker-assisted selection and pyramiding in subsequent generations of the same breeding populations. However, QTL mapping in breeding populations is complicated by multi-allelic inheritance, inconsistent linkage phases between marker and QTL alleles, and complex pedigree structures. Fortunately, statistical methods for linkage analysis in pedigreed populations have been developed to address these complications and can be adopted for the purpose of QTL mapping in plant breeding populations (Jannink et al. 2001; Sneller et al. 2009; Wu¨rschum 2012). Linkage analysis methods for pedigreed populations trace the inheritance of parental alleles by estimating the proportion of alleles that are identical-by-descent (IBD) between related individuals at marker loci. Crepieux et al. (2005) demonstrated the use of an IBD-based variance components method to map known major genes for quality traits in inbred lines derived from wheat breeding populations. It has been demonstrated that Fhb1 can be mapped in earlygeneration wheat breeding populations using IBDbased methods for linkage analysis (Rosyara et al.
Mol Breeding (2015) 35:6
2009). These studies indicate that IBD-based linkage analysis has the potential to map major effect loci in plant breeding populations. However, it remains to be determined if multiple smaller effect resistance QTL segregating within an Fhb1 background can be mapped using IBD-based linkage analysis. If possible, then IBD-based linkage analysis would allow for simultaneous mapping and marker-assisted pyramiding of FHB resistance QTL in breeding populations. To assess the possibility of this approach, we applied IBD-based linkage mapping in winter wheat breeding populations derived from four-way crosses among Fhb1 donor lines and native sources of resistance. Our objectives were to 1) verify that multiple FHB resistance QTL can be mapped in early-generation breeding populations, 2) map QTL associated with FHB resistance in adapted winter wheat cultivars, and 3) develop early-generation germplasm combining Fhb1 with native sources of FHB resistance that can be used in breeding programs.
Materials and methods Population development Crosses were made among 10 winter wheat parental lines (i.e., founders) to develop 28 segregating four-way F1 populations. Two backcross-derived lines carrying the Fhb1 resistance allele in a ‘Wesley’ background (WesleyFhb1-BC06 and Wesley-Fhb1-BC56) and an experimental line AL-107-6106 (Alsen/NE00403//NE02583-107) were used as donors of Fhb1 in each cross. Founders providing native sources of resistance were the HWW varieties ‘Lyman’ (KS93U134/Arapahoe) and ‘Overland’ (Millennium sib//Seward/Archer) and the SWW varieties ‘Ernie’ (Pike/MO9965) and ‘Freedom’ (GR876/OH217). The remaining founders, which provided desirable agronomic traits, were NE06545 (KS92-946-B-I5-1/Alliance), NI08708 (CO980829/Wesley), and ‘McGill’ (NE92458/Ike). A total of 565 four-way F1 plants were derived from the 28 four-way crosses, with an average of 20 four-way F1 plants per cross (Table 1). Each four-way F1 plant was selfed to derive F2 seed for progeny testing. Phenotypic evaluations Seed of founders and four-way F1 plants were vernalized for 8 weeks and then transplanted as individual
Page 3 of 16 6
plants in 4 9 4 inch pots in a greenhouse. The plants were laid out in a completely randomized design. An aggressive Fusarium graminarum isolate, Fg4, was used for FHB inoculations in the greenhouse. The fungus was cultured on strength PDA (12 g potato dextrose, 15 g agar, 1 L dH20) with 0.2 % lactic acid. Individual spikes were spray inoculated with a conidial spore suspension containing 25,000 cfu/mL at 50 % anthesis and covered with polyethylene bags for 48 h. The temperature in the greenhouse was maintained between 21 and 26 °C. A total of 838 spikes were inoculated on 452 of the four-way F1 plants, with an average of approximately two spikes inoculated per plant (Table 2). Inoculated spikes were evaluated at 7, 14, and 21 days after inoculation (dai) for FHB severity (i.e., proportion of symptomatic spikelets). FHB severity was measured by counting the number of spikelets with disease symptoms out of the total number of spikelets on each inoculated spike. F2 seed was used to establish selfed-progeny tests in the field to estimate the genetic value of each four-way F1 plant for FHB resistance. A total of 542 four-way F1 plants yielded sufficient seed for progeny testing. The progeny tests were planted in disease nurseries at two sites, located in Volga SD and Manhattan KS (Table 2). Within each site, F2 lines were planted as single row plots, where each plot was established with a minimum of 20 seed. Of the 542 F2 lines, 339 were replicated over both sites, whereas 203 lines were evaluated at only a single site due to limited seed availability (Table 2). A mixture of Fusarium graminearum isolates (Fg1, Fg4, Fg5, Fg6, Fg30, Fg41, Fg56, Fg62, Fg63, Fg64, and Fg70) cultured on PDA were used for field inoculations. F2 lines were spawn inoculated with infested corn kernels spread on the soil surface about 1 month prior to heading, and heads were mist irrigated beginning at heading at all sites to provide constant disease pressure. Additionally, direct spray inoculation was conducted at 50 % anthesis for each plot at the Volga site using a conidial spore suspension containing 100,000 spores/mL. The extent of FHB infection in each field plot was assessed at 21 days after flowering. FHB severity was scored on 20 heads per plot using a 10-point visual scale described by Stack and Mcmullen (1998), where each score corresponds to a percent of the spike infected. Spatial analysis was conducted using the ‘fields’ package in R (Furrer et al. 2013). A thin-plate spline regression was fit to FHB severity data based on the
123
6 Page 4 of 16
Mol Breeding (2015) 35:6
Table 1 Summary of breeding populations used for pyramiding FHB resistance. Founders indicated in bold were the Fhb1 donors in each cross Population
Pedigree
4-way F1 plants
F1 plants phenotyped
F2 lines phenotyped
SNP genotyped
01
WesFHB-BC56/NE06545//Ernie/Overland
20
15
20
03
Ernie/WesFHB-BC06//Ernie/NE06545
26
19
21
No
05
Ernie/WesFHB-BC06//Lyman/AL-107-6106
22
20
17
Yes
06
Ernie/WesFHB-BC56//Ernie/Lyman
40
34
40
Yes
09
Ernie/WesFHB-BC56//NI08708/Lyman
40
35
39
Yes
10
Ernie/Lyman//Ernie/WesFHB-BC06
12
10
11
No
14
Ernie/Overland//Freedom/WesFHB-BC56
5
4
4
No
16
Ernie/Overland//Overland/WesFHB-BC56
24
22
25
Yes
17
Ernie/Overland//NI08708/WesFHB-BC06
33
25
30
Yes
20
Ernie/NE06545//McGill/WesFHB-BC56
28
26
29
No
23 26
Ernie/McGill//Lyman/WesFHB-BC06 Freedom/WesFHB-BC06//Ernie/Overland
12 12
10 8
10 10
Yes Yes
27
Freedom/WesFHB-BC06//Lyman/AL-107-6106
28
Freedom/WesFHB-BC06//Overland/WesFHB-BC56
Yes
7
2
7
Yes
11
6
9
Yes
30
Freedom/WesFHB-BC56//Ernie/NE06545
4
3
4
No
35
Freedom/Ernie//Overland/WesFHB-BC56
34
23
34
Yes
36
Freedom/Ernie//NI08708/WesFHB-BC06
29
22
28
No
40
Freedom/Overland//Lyman/AL-107-6106
8
2
10
Yes
41
Freedom/NI08708//WesFHB-BC56/NE06545
7
3
7
No
45
AL-107-6106/Overland//Lyman/WesFHB-BC06
11
4
10
Yes
48
AL-107-6106/Overland//NI08708/Lyman
14
6
16
Yes
54
Lyman/WesFHB-BC56//Ernie/Lyman
37
32
36
Yes
57
Lyman/WesFHB-BC56//NI08708/Lyman
31
22
29
No
64
Overland/WesFHB-BC56//Ernie/Lyman
44
33
43
Yes
65
Overland/WesFHB-BC56//Ernie/NE06545
5
5
5
No
67 71
Overland/McGill//Lyman/WesFHB-BC06 NI08708/WesFHB-BC06//Ernie/NE06545
12 8
7 6
11 7
Yes No Yes
76
NI08708/Lyman//Overland/WesFHB-BC56 Total population
row and column positions estimate spatial trends. Predicted values from the thin-plate spline regression were assumed as environmental indices for each row– column position in the greenhouse and field. Best linear unbiased prediction Mixed models were used to provide best linear unbiased predictions (BLUPs) of FHB resistance for each four-way F1 plant from greenhouse and selfedprogeny evaluations. These BLUPs are adjusted for fixed effects and regress the phenotypes toward the mean as a function of the repeatability of the data. The
123
29
23
30
565
427
542
later property provides a natural framework for handling unbalanced data. FHB severity observed from greenhouse evaluations was fit to the following binomial mixed model: exp Xij b þ ui þ eij E Yij ¼ Pij ¼ ð1Þ 1 þ exp Xij b þ ui þ eij where Pij is the proportion of infected spikelets on the jth spike of the ith plant, Yij is a binomial random variable (the number of symptomatic spikelets out of the total number of spikelets) with mean Pij and variance Pij(1 - Pij), b is a vector of fixed effects, Xij is a vector of covariates relating to the fixed effects for
Mol Breeding (2015) 35:6
Page 5 of 16 6
Table 2 Summary statistics for phenotypic evaluations of four-way F1 plants for FHB resistance Evaluation
Entries
Spikes/entry
Distribution of entry means (severity) Min
Q1
Mean
Q3
Entry mean heritability
Max
Greenhouse 7 DAI (F1 plants)
419
1–6
0.00
0.08
0.19
0.25
1.00
0.33
Greenhouse 14 DAI (F1 plants)
452
1–6
0.00
0.37
0.53
0.67
1.00
0.32
Greenhouse 21 DAI (F1 plants)
451
1–6
0.08
0.60
0.73
0.91
1.00
0.41
Field progeny test (F2 lines)
531
20–40
0.05
0.28
0.35
0.41
0.90
0.46
the jth spike of the ith plant, ui is the random effect of the ith plant, and eij is the random effect of the jth spike of the ith plant. Normality of the conditional distribu tion was assumed, such that ui N 0; r2u and eij N 0; r2e . The inclusion of an individual level random effect, eij, accounted for overdispersion (extra-binomial variance). The fixed effects included the overall mean, deviations for each inoculation/ flowing date, and environmental indices predicted from spatial analysis to account for environmental covariance. The ‘lme4’ package in R (Bates et al. 2013) was used to fit the model and estimate the BLUPs for each four-way F1 plant (ui). Note that the BLUPs are estimated on a logit scale, given the implicit logit link function in Model 1. Using variance components estimated from Model 1, the heritability of four-way F1 plant means (entry mean heritability) was estimated as: H2 ¼
r2u r2u þ r2e s þ 3:29=ns
ð2Þ
where s is the number of spikes observed per plant, n is the number spikelets per spike (approximately 15 on average), and 3.29 is standard deviation of the standard logistic latent variable. The same general form of Model 1 was used to analyze FHB severity data observed from selfedprogeny evaluations. In this case, Pij is the proportion of spikelets infected in the jth plot of the line derived from the ith four-way F1 plant, ui is the random effect of the ith four-way F1 parent, and eij is the random plot effect. To derive the binomial response, each evaluated spike was assumed to be comprised of 15 spikelets. The fixed effects included overall mean and site effects. The model was fit as described above to estimate BLUPs for each four-way F1 plant-based
severity on the F2 lines. Entry mean heritability for selfed-progeny evaluations was estimated using the same form of Model 2, with s defined as the number of replications, and n is defined as the number of spikelets per plot. Genotyping DNA was extracted from healthy leaf tissue collected from plants in the greenhouse. Tissue samples from founder lines were pooled over multiple plants. Precipitated nucleic acid was pelletized by centrifugation and washed with 70 % molecular grade ethanol and suspended in 10 mM Tris buffer (pH 8.0) containing 40 lg/mL of RNase A. Founder lines and all 565 four-way F1 plants were genotyped at 26 polymorphic simple sequence repeat (SSR) marker loci. SSR markers were selected to target Fhb1 and resistance loci reported previously in literature from ‘Freedom’ and ‘Ernie.’ PCR was conducted at the USDA-ARS Hard Winter Wheat Genetics Research Unit, Manhattan, KS, using the method described by Zhang et al. (2012). PCR products were separated using an ABI Prism 3730 Genetic Analyzer, and allele calls were made from fluorescence peaks using GeneMarker version 1.6 (SoftGenetics, LLC). Of the 28 populations, 10 populations segregated only for Fhb1 and resistance loci derived from ‘Ernie’ and/or ‘Freedom.’ These 10 populations were genotyped only for SSR markers, since positions of resistance loci segregating in these populations were known a priori. The remaining 18 populations were derived from Lyman and/or Overland and thus potentially segregated for novel resistance loci (Table 1). Therefore, these 18 populations were genotyped for 9,000 SNP (single-nucleotide polymorphism) marker loci to facilitate genome-wide QTL mapping. SNP
123
6 Page 6 of 16
genotyping was conducted using an Infinium 9,000 SNP iSelect Beadchip assay developed for wheat (Cavanagh et al. 2013). The assay was performed using the Illumina BeadStation and iScan instruments at the USDA-ARS Biosciences Research Laboratory, Fargo, ND. GenomeStudio version 2011.1 (Illumina) was used for cluster analysis and SNP genotype calling. The minimum ‘GenTrain’ score (a measure of the reliability of SNP calling based on cluster distribution) was reduced to 0.05 in GenomeStudio to facilitate delineation of compressed but unambiguous SNP clusters. Genotype clusters were then visually assessed for each SNP and manually revised to improve genotype calling. Mendelian inheritance errors for both SSR and SNP loci were detected using the ‘prepare’ function of CRIMAP version 2.504 (Green et al. 1990). For those cases where genotyping errors could not be rectified, the genotypic data were replaced with missing values. A framework linkage map consisting of 3,785 loci covering the entire wheat genome was jointly estimated over all breeding populations using CRI-MAP as detailed by Eckard et al. (2014). QTL linkage mapping Identity-by-descent (IBD)-based linkage analysis was conducted using the software package S.A.G.E version 6.3 (S.A.G.E. 2012). A detailed overview of the steps involved in conducting linkage analysis in S.A.G.E is provided by Morris and Stein (2012). Maximum-likelihood allele frequencies for each locus were estimated using the ‘FREQ’ program in S.A.G.E. Null alleles at SSR marker loci were specified as recessive alleles, thus allowing the use of partial genotypes resulting from the segregation of null alleles in the linkage analysis. Multipoint IBD allele sharing probabilities were estimated using the ‘GENIBD’ program of S.A.G.E. For the purpose of IBD estimation, each four-way cross was assumed to be an unrelated pedigree. ‘Trait modelfree’ linkage analysis was performed using the ‘RELPAL’ program in S.A.G.E. The ‘RELPAL’ program implements a mixed (two-level) model extension of Haseman-Elston regression that is described in detail by Wang and Elston (2005). Since the phenotypes were already adjusted for fixed effects (including the mean) during estimation of the BLUPs, only the random effects (second level) of
123
Mol Breeding (2015) 35:6
the Haseman-Elston regression were applicable. Therefore, for a given chromosome region, the variance–covariance structure of kth pedigree, Vk, was defined as, Vk ¼ COV yik ; yjk ¼
r2g þ r2p þ r2e p^ij r2g þ /ij r2p
if i ¼ j if i ¼ 6 j ð3Þ
where yik and yjk are the BLUPs for four-way F1 plants i and j in pedigree k, r2g is the additive genetic variance due to a QTL at the tested chromosomal region, r2p is the remaining polygenic genetic variance, r2e is the residual variance, p^ij is the estimated proportion of alleles shared IBD at the chromosomal region, and /ij is the kinship coefficient. A one-sided score test was used to test for significance of r^2g (Verbeke and Molenberghs 2003, S.A.G.E. 2012). Empirical P values were computed for r^2g by permutation (S.A.G.E. 2012). The log10 (P value) was plotted across the genome to graphically display linkage results. The proportion of the variation for BLUPs explained by a QTL (R2) was calculated as r^2g =r^2y , where r^2y is the variance of the BLUPs for the trait. For each chromosome, haplotype analysis was conducted using the ‘chrompic’ function in CRI-MAP to reconstruct the most likely fully typed and phase known haplotypes for each four-way F1 plant. Using these data, the transmission of founder haplotypes to each four-way F1 plant within the QTL region was estimated. The effects of the founder alleles at the QTL were then estimated as, yi ¼ l þ xi a þ ei where yi is the BLUP for ith four-way F1 plant, l is the grand mean (equals 0 for BLUPs), x is a design matrix of inheritance probabilities, xi is a vector of QTL inheritance probabilities from each founder to the ith plant, a is a vector of additive effects to be estimated for founder QTL alleles (sum equal to 2), and ei is the random residual effect. For nonrecombinant gametes, the inheritance probabilities were equal to 0 or 1. For recombinant gametes, the inheritance probability for a founder allele was estimated by the proportion of the QTL region comprised by the founder genotype. A negative additive effect with a P value \ 0.10 was classified as a resistance allele at the QTL.
Mol Breeding (2015) 35:6
Page 7 of 16 6
Results Summary of phenotypic data Mean FHB severity in the greenhouse was 0.19, 0.53, and 0.73 at 7, 14, and 21 days after inoculation, respectively (Table 2). Mean FHB severity in the selfed-progeny field evaluations was 0.35, with individual site means ranging from 0.31 to 0.42. Considerable phenotypic variation was observed among fourway F1 plants in the greenhouse, and F2 lines in the field, with the range of entry mean severity approaching 0.0 and 1.0 at all evaluation periods (Table 2). Spatial analysis detected significant environmental covariance in the greenhouse (P value \ 0.0001). Thin-plate spline regression explained 10, 11, and 6 % of the variation for FHB severity in the greenhouse at 7, 14, and 21 days after inoculation, respectively. Date of inoculation effects was also highly significant (P value \ 0.0001). Therefore, both the environmental index (spatial covariate) and date of inoculation effects were kept as fixed effects for estimation of BLUPs for greenhouse severity. After spatial adjustment, entry mean heritability for greenhouse severity ranged from 0.32 to 0.41. No significant spatial covariance was observed in the field for the sefledprogeny evaluations. The entry mean heritability for selfed-progeny evaluations was 0.46. The distributions BLUPs for the four-way F1 parents were approximately normally distributed for all phenotypic evaluations, satisfying assumptions of QTL linkage analysis.
BLUPs based on greenhouse evaluations at 7 days after inoculation were only moderately correlated with greenhouse evaluations at 14 and 21 days after inoculation (Table 3). However, evaluations at 14 and 21 days after inoculation were highly correlated (r = 0.78). AUDPC had strong correlations with BLUPs obtained from all measurement periods (Table 3). BLUPs from selfed-progeny evaluations were weakly correlated with greenhouse evaluations, with correlations of 0.07, 0.16, and 0.20 with greenhouse evaluations at 7, 14, and 21 days after inoculation, respectively (Table 3). Correlations between BLUPs and four-way F1 plant means for FHB severity ranged from 0.68 to 0.94 for different evaluations (Table 3), indicating substantial adjustment of the phenotypic data for fixed effects and imbalance in the data. QTL linkage mapping A highly significant QTL was detected on the short arm of chromosome 1A across all greenhouse evaluations and field-based selfed-progeny evaluations of the fourway F1 plants (Fig. 1). This QTL, designated as Qfhb.sdsu-1A, explained between 4.5 % and 8.1 % of the variance in greenhouse BLUPs and explained 9.9 % of the variance in selfed-progeny BLUPs (Table 4). Peak significance for linkage to Qfhb.sdsu-1A was between Xiwa7021 (75.0 cM from p-terminus) and Xiwa459 (75.9 cM from p-terminus) for all greenhouse evaluations. The 1.5 LOD support interval for Qfhb.sdsu-1A from the selfed-progeny evaluations also
Table 3 Correlations among BLUPs (above diagonal), and between phenotypic means and BLUPs (diagonal) for evaluations at 7, 14, and 21 days after inoculation (DAI), area under the disease progression curve (AUDPC) and selfed-progeny evaluations (PROG)
7 DAI
7 DAI
14 DAI
21 DAI
AUDPC
PROG
0.684
0.502
0.360
0.629
0.066
(\0.001)
(\0.001)
(\0.001)
(\0.001)
(0.215)
14 DAI
–
0.775
0.787
0.939
0.161
(\0.001)
(\0.001)
(\0.001)
(0.002)
21 DAI
–
–
0.829
0.849
0.200
(\0.001)
(\0.001)
(\0.001)
–
0.798
0.183
(\0.001)
(\0.001)
AUDPC PROG
– –
– –
–
–
0.938 (\0.001)
P values for the correlations are shown in parentheses
123
6 Page 8 of 16
Mol Breeding (2015) 35:6
Fig. 1 Whole genome scans for linkage to FHB resistance QTL. Vertical lines designate breaks between chromosomes, and horizontal lines represent the empirical P value threshold of 0.001 (-log10 P value = 2). Arrows indicate locations of significant QTL peaks
spanned Xiwa7021, suggesting that this SNP is tightly linked with Qfhb.sdsu-1A. The 15.8 cM region between Xiwa2922 and Xiwa7956 contained the peak for Qfhb.sdsu-1A across all phenotypic evaluations. Inheritance of the region between Xiwa2922 and Xiwa7956 from either Lyman or Overland resulted in a significant reduction in FHB severity (Table 4).
123
Another QTL detected in both greenhouse and selfed-progeny evaluations mapped to the short arm of chromosome 3B (Fig. 1). The 1.5 LOD support interval indicated that this QTL is located within a 14.8 cM region between Xbarc133 and Xiwa2493 (Table 4). This QTL explained 10.4 % of the variance in greenhouse BLUPs at 21 dai and explained 10.0 %
Mol Breeding (2015) 35:6
Page 9 of 16 6
Table 4 Summary of QTL detected by IBD-based linkage analysis, including the region of peak significance, P values for the test of linkage, percent of variance in the BLUPs explained by the QTL region (R2), and putative sources of resistance alleles R2 (%)
Source of resistance
Trait locus
Evaluation
1.5 LOD support interval
Peak position
P value (-log10)
Qfhb.sdsu-1Aa
7 DAI
Xiwa764-Xiwa7956 (45.1 cM)
Xiwa459
7.0
8.1
14 DAI
Xiwa4577-Xiwa7956 (12.0 cM)
Xiwa7021
7.0
4.5
Lyman*
21 DAI
Xiwa2922-Xiwa459 (9.9 cM)
Xiwa7021
7.0
7.4
Lyman*; Overland*
AUDPC
Xiwa2922-Xiwa7956 (15.8 cM)
Xiwa459
7.0
6.5
Lyman**
Selfed-progeny
Xiwa7796-Xiwa7021 (26.0 cM)
Xiwa7151
4.5
9.9
Lyman*; Overland**
Qfhb.sdsu-1B.1a
21 DAI
Xiwa415-Xiwa8543 (33.4 cM)
Xiwa7422
2.5
8.6
Lyman*; Overland*
Qfhb.sdsu-1B.2a
Selfed-progeny
Xiwa6063-Xiwa3120 (9.9 cM)
Xiwa2861
3.6
12.5
Lyman**; WesleyFhb1-BC56**; Overland*
7 DAI
Xiwa6745-Xiwa1512 (10.2 cM)
Xbarc212
2.7
2.5
14 DAI
Xiwa6745-Xiwa1512 (10.2 cM)
Xbarc212
2.2
5.8
Qfhb.sdsu-2A
Overland*
Qfhb.sdsu-3A
7 DAI
Xiwa2153-Xiwa3198 (8.1 cM)
Xiwa2617
3.8
6.5
Fhb1
21 DAI
Xbarc133-Xiwa2493 (14.8 cM)
Xiwa2908
2.1
10.4
Wesley-Fhb1-BC56***
AUDPC
Xbarc133-Xiwa2493 (14.8 cM)
Xgwm493
2.0
6.6
Wesley-Fhb1-BC56***
Selfed-progeny
Xbarc133-Xiwa2493 (14.8 cM)
Xiwa2493
3.1
10.0
Wesley-Fhb1-BC56***
7 DAI
Xiwa4199-Xiwa4261 (21.9 cM)
Xiwa482
4.3
6.1
14 DAI
Xiwa5349-Xiwa1170 (21.8 cM)
Xiwa482
2.5
4.5
Lyman**
AUDPC
Xiwa5349-Xiwa4867 (6.6 cM)
Xiwa482
3.6
5.5
Lyman**
Selfed-progeny
Xiwa5200-Xiwa3584 (30.1 cM)
Xiwa8432
2.2
8.6
Wesley-Fhb1-BC56**; AL-107-6106*
Evaluation
1 LOD support interval
P value (-log10)
R2 (%)
Qfhb.sdsu-4A.1a
Trait locus
Peak position
Qfhb.sdsu-4A.2
Selfed-progeny
Xiwa7394-Xiwa1900 (17.6 cM)
Xiwa3161
3.6
6.6
Qfhb.sdsu-4B
Selfed-progeny
Xwmc710-Xgwm258 (27.9 cM)
Xiwa2313
3.4
6.5
Qfhb.sdsu-4D
7 DAI
Xiwa752-Xiwa161 (29.9 cM)
Xiwa1633
4.1
4.6
Source of resistancea
Wesley-Fhb1BC56***
Qfhb.sdsu-5A
14 DAI
Xiwa1253-Xiwa8154 (24.2 cM)
Xiwa1546
2.3
5.0
Overland*
Qfhb.sdsu-6A.1a
AUDPC 14 DAI
Xiwa1253-Xiwa3263 (23.1 cM) Xiwa2812-Xiwa1282 (32.9 cM)
Xiwa3349 Xiwa3527
3.0 5.7
7.5 5.3
Overland* Overland**
21 DAI
Xiwa2812-Xiwa1282 (32.9 cM)
Xiwa2235
3.2
6.5
Overland***; Lyman***
AUDPC
Xiwa1423-Xiwa5401 (28.4 cM)
Xiwa3527
6.0
7.5
Overland**
Qfhb.sdsu-6A.2
Selfed-progeny
Xiwa272-Xiwa7913 (29.6 cM)
Xiwa7913
3.7
9.9
Wesley-Fhb1BC56***
Qfhb.sdsu-6Da
Selfed-progeny
Xiwa2338-Xiwa984 (4.9 cM)
Xiwa6625
4.2
12.1
Wesley-Fhb1BC56***; Overland*
Qfhb.sdsu-7D
7 DAI
Xiwa2545-Xiwa5391 (14.4 cM)
Xiwa1247
5.8
3.9
14 DAI
Xiwa6320-Xiwa1537 (34.4 cM)
Xiwa1247
4.1
3.4
21 DAI
Xiwa2545-Xiwa6822 (21.5 cM)
Xiwa1247
5.6
5.3
Wesley-Fhb1-BC56*; AL-107-6106*
123
6 Page 10 of 16
Mol Breeding (2015) 35:6
Table 4 continued Trait locus
Evaluation
1 LOD support interval
Peak position
P value (-log10)
AUDPC
Xiwa2545-Xiwa1537 (33.4 cM)
Xiwa1247
4.2
R2 (%)
3.4
Source of resistancea Wesley-Fhb1-BC56*; AL-107-6106*
* P value \ 0.05; ** P value \ 0.01; *** P value \ 0.001 a
Potentially novel QTL based on previous literature
of the variance in the selfed-progeny BLUPs. Peak significance for linkage to the QTL on chromosome 3B in the greenhouse was between Xiwa2908 (12.0 cM from p-terminus) and Xgwm493 (16.0 cM from p-terminus). Peak significance for the QTL shifted to Xiwa2493 (20.6 cM from p-terminus) based on selfed-progeny evaluations. Inheritance of the 14.8 cM support region between Xbarc133-Xiwa2493 from Wesley-Fhb1-BC56 resulted in a highly significant reduction in FHB severity (p \ 0.001). Given that this region is known to harbor Fhb1, and the source of resistance was an Fhb1 donor, this QTL was designated as Fhb1. The only other QTL detected in both greenhouse and selfed-progeny evaluations was located on chromosome 4A. This QTL, designated as Qfhb.sdsu-4A, explained between 4.5 % and 6.1 % of the variation in greenhouse BLUPs and explained 8.6 % of the variance in selfedprogeny BLUPs. Peak significance for linkage to Qfhb.sdsu-4A was found at Xiwa482 (114.4 cM from p-terminus) for greenhouse evaluations and at Xiwa8432 (100.2 cM from p-terminus) for selfed-progeny evaluations. Inheritance of this region from Lyman resulted in highly significant reduction in FHB severity in the greenhouse. Conversely, Wesley-Fhb1-BC56 conferred highly significant resistance in the selfed-progeny evaluation. The LOD profile for the QTL was bimodal for selfed-progeny evaluations (Fig. 1). Given the different sources of resistance at this locus in the greenhouse and selfed-progeny evaluations, and bimodal LOD profiles, it is unclear whether Qfhb.sdsu-4A represents a single locus or two tightly linked loci. The remaining QTL detected in this study were specific to either greenhouse or selfed-progeny evaluations (Table 4). Overall, QTL were identified on chromosomes 1A, 3B, 4A, 5A, 6A and 7D for greenhouse evaluations based on AUDPC (Table 4). In addition to Qfhb.sdsu-1A, Fhb1 and Qfhb.sdsu-4A, the QTL on chromosomes 6A (Qfhb.sdsu-6A.1) and 7D (Qfhb.sdsu7D) were detected across multiple evaluations periods in
123
the greenhouse (Table 4). Cumulatively, the QTL detected from greenhouse evaluations explained 37 % of the phenotypic variance in the greenhouse BLUPs. Given that the variation among the BLUPs represents the total genotypic variance, there was a considerable amount of ‘missing heritability’ remaining after accounting for the QTL detected in this study. For selfedprogeny evaluations, QTL were detected on chromosomes 1A, 1B, 3B, 4A, 4B, 6A, and 6D, explaining a total of 69.5 % of the phenotypic variation. Therefore, a larger portion of the total genetic variance was explained by the QTL detected in the selfed-progeny evaluations. Effects of pyramiding QTL Considering only those QTL significant in greenhouse evaluations, the number of FHB resistance alleles inherited by the four-way F1 plants ranged from 0 to 14, with an average of four resistance alleles per plant (Fig. 2). FHB severity consistently decreased with the number of inherited resistance alleles. Regression of the number of resistance alleles inherited by each plant against the AUDPC in the greenhouse resulted in a strong negative linear fit, with an R2 of 0.928 (Fig. 2). For QTL significant in selfed-progeny evaluations, the number of FHB resistance alleles inherited by the fourway F1 plants ranged from 0 to 11, with an average of four resistance alleles per plant (Fig. 2). Regression of the number of resistance alleles inherited by each fourway F1 plant against the field FHB severity resulted in a significant negative linear trend, with an R2 of 0.776 (Fig. 2). Therefore, resistance alleles in the four-way F1 plants had cumulative additive effects on reducing FHB severity in the selfed-progeny.
Discussion IBD-based linkage analysis applied to 28 earlygeneration breeding populations established with
Mol Breeding (2015) 35:6
Page 11 of 16 6
(a)
80
y = -0.0924x + 0.4496 R² = 0.9284
0.4 0.2
60
0
50
-0.2 40 -0.4 30
-0.6
20
Greenhouse BLUPs
Number of four-way F 1 plants
70
0.6
-0.8
10
-1 -1.2
0 0
1
2
3
4
5
6
7
8
9 10 11 12 13 14
Number of resistance alleles 0.06
(b)
y = -0.0068x + 0.0352 R² = 0.7763
60
0.04
50
0.02
40
0
30
-0.02
20
-0.04
10
-0.06
Selfed-progeny BLUPs
Number of four-way F 1 plants
70
-0.08
0 0
1
2
3
4
5
6
7
8
9
10
11
Number of resistance alleles
Fig. 2 The number of FHB resistance alleles carried by fourway F1 plants and regressions of number of resistance alleles against BLUPs for a AUDPC from greenhouse evaluations and b severity from selfed-progeny evaluation
multiple sources of resistance enabled mapping and validation of 15 QTL located on chromosomes 1A, 1B, 2A, 3A, 3B, 4A, 4B, 4D, 5A, 6A, 6D, and 7D. These loci were also simultaneously validated in this study given that they were detected over multiple populations and genetic backgrounds. Qfhb.sdsu-1A, Qfhb.sdsu-4A.1, and Fhb1, which were efficacious not only across multiple genetic backgrounds, but also across greenhouse and selfed-progeny evaluations, should be relevant for marker-assisted selection across winter wheat breeding populations. Previous studies have focused on validating IBDbased linkage analysis in plant breeding populations by mapping a small number of major effect loci or genes through targeted genotyping. For example, Rosyara et al. (2009) mapped Fhb1 to its known position using IBD-based linkage analysis by genotyping 11 SSR markers on chromosome 3B. Crepieux et al. (2005) mapped Glu-1 genes for high molecular
weight glutenin subunits and the Ha gene for kernel hardness to known positions using IBD-based linkage analysis by genotyping 65 SSR markers on homeologous groups 1 and 5. These previous studies have shown that IBD-based linkage analysis can reliably map one or more major effect loci in plant breeding populations. However, the current study represents the first application of IBD-based linkage analysis for a genome-wide QTL scan in plant breeding populations without a priori knowledge of QTL positions. The ability to map such a large number of QTL for FHB severity in this study demonstrates the power of IBDbased linkage analysis combined with dense genotyping to detect multiple segregating loci for a complex trait in plant breeding populations. Detection of this number of QTL would have required numerous experimental bi-parental mapping populations, since each bi-parental population would have only reflected a small portion of the total genetic architecture of FHB resistance in the breeding program. This study also provides the first mapping results for native resistance alleles from the North American cultivars Lyman and Overland. Lyman was found to confer resistance alleles at the QTL on chromosomes 1A, 1B, 3A, 4A, and 6A, while Overland was found to confer resistance at the QTL on chromosomes 1A, 1B, 2A, 5A, and 6A. Thus, several of the loci associated with resistance from the cultivars Lyman and Overland overlap, including Qfhb.sdsu-1A, Qfhb.sdsu-1B, and Qfhb.sdsu-6A.1. It is possible that Lyman and Overland have the same alleles or functionally similar alleles at these loci, since they were derived from public breeding efforts utilizing the same indigenous germplasm pool. The QTL detected on the short arm of chromosome 1A (Qfhb.sdsu-1A) with resistance derived from Lyman and Overland is of particular interest due to its large and consistent effect in both greenhouse and selfed-progeny evaluations. Interestingly, Lyman is known to harbor the 1AL.1RS wheat-rye translocation derived from Amigo (Berzonsky 2013, personal communication). The 1RS chromosome arm from rye has been associated with FHB resistance in European winter wheat cultivars carrying the 1BL.1RS translocation (Schmolke et al. 2005; Holzapfel et al. 2008; Liu et al. 2009; Buerstmayr et al. 2009). Since both the 1AL.1RS and 1BL.1RS translocation chromosomes carry a similar portion of the 1RS chromosome arm, it could be hypothesized that the 1AL.1RS translocation chromosome carried by Lyman
123
6 Page 12 of 16
harbors a resistance locus in the 1RS region. However, there is no indication that the 1AL.1RS translocation is harbored by Overland. Confirmation of whether Qfhb.sdsu-1A resides on 1RS or simply linked to this region requires further research. Holzapfel et al. (2008) mapped a QTL for FHB severity on 1AS in two of four bi-parental crosses among European winter wheat cultivars. Liu et al. (2013) mapped a repeatable QTL on 1AS for FHB index in the cross between US soft winter wheat Becker and Massey. A QTL on 1AS was also detected for FHB incidence and severity in a biparental cross of durum wheat (Singh et al. 2008). Although not detected in all phenotypic evaluations, Qfhb.sdsu-1B.1, Qfhb.sdsu-4A.1, and Qfhb.sdsu-6A.1 may also represent a novel sources of native resistance to FHB from Lyman and Overland that can be exploited in breeding programs. The QTL on the short arm of chromosome 5 (Qfhb.sdsu-5A) was approximately centered on the SSR locus Xbarc56, which is tightly linked to Xbarc180 (* 2.6 cM). Xbarc180 is highly diagnostic for Qfhs.ifa-5A (Fhb5) (Anderson 2007), a QTL for FHB resistance that has been consistently detected in mapping studies derived from Asian, North American, South American, and European germplasm (Buerstmayr et al. 2009). Thus it is clear that Qfhb.sdsu-5A is colocalized with Fhb5. Liu et al. (2007) detected a resistant allele from Ernie at Fhb5 in a bi-parental mapping population. However, a significant allelic effect was not observed for Ernie at Qfhb.sdsu-5A in this study. Instead, a significant reduction in FHB severity was observed for progeny inheriting the allele at Qfhb.sdsu-5A from Overland, suggesting that Overland may carry a resistance allele at Fhb5. The QTL detected on the short arm of chromosome 3B (3BS) was located between SSR loci Xbarc133 and Xbarc147, which is the region that is well known to harbor Fhb1. Furthermore, resistance at the 3BS region was conferred by Wesley-Fhb1-BC56, which was a donor for Fhb1 in this study. Inheritance of this region from Wesley-Fhb1-BC06 and AL-107-6106 also reduced FHB severity, but the effect was not statistically significant due to an insufficient number of progeny derived from these lines. Therefore, the 3BS locus detected in this study is clearly associated with segregation for Fhb1. The loci Qfhb.sdsu-1B.2, Qfhb.sdsu4A.1, Qfhb.sdsu-4B, Qfhb.sdsu-6A.2, Qfhb.sdsu-6D, and Qfhb.sdsu-7D also had resistance alleles derived from either Wesley-Fhb1-BC56 or AL-107-6106. Non-
123
Mol Breeding (2015) 35:6
Fhb1 resistance derived from ‘Sumai 3’ has been reported on chromosomes 4A (Yang et al. 2005a), 6A (Anderson et al. 2001), and 7D (Sneller et al. 2001), near the loci detected in this study. Furthermore, the region on 4B has been reported for sources of Asian resistance including Wuhan1 and Wangshuibai (Buerstmayr et al. 2009) and Chokwang (Yang et al. 2005b). These results may suggest that the Fhb1 donor lines used in this study harbor Sumai-3 derived resistance alleles at several loci in addition to Fhb1. Semi-dwarfing loci Rht-B1 and Rht-D1 have well documented effects on FHB resistance, with dwarfing alleles generally conferring susceptibility either through pleiotropy or linkage to other loci effecting FHB resistance (Miedaner and Voss 2008; Srinivasachary et al. 2009). Thus, FHB resistance QTL are often reported to be colocalized with Rht-B1 and RhtD1. In this study, Qfhb.sdsu-4B was located in a 22.5 cM region between SSR loci Xwmc710 and Xgwm495. Based on the SSR consensus map of Somers et al. (2004), Rht-B1 falls within this interval, tightly linked to Xwmc710. This evidence strongly suggests that Rht-B1 is the causal gene underlying Qfhb.sdsu-4B. However, the recurrent parent for Wesley-Fhb1-BC56 and Wesley-Fhb1-BC06, which significantly increased FHB resistance at Qfhb.sdsu4B, is a semi-dwarf cultivar with the Rht-B1b dwarfing allele (Guedira et al. 2010). Ernie also harbors the RhtB1b allele, but conferred significant susceptibility at Qfhb.sdsu-4B, suggesting that the effect associated with Rht-B1 was not due to a pleiotropic effect. Consistent with these results, Srinivasachary et al. (2009) reported that Rht-B1b did not increase susceptibility to FHB, suggesting that the effects of dwarfing loci on FHB were due to linkage rather than pleiotropy. Due to poor marker coverage on chromosome 4D, it was not possible to determine the degree to which Qfhb.sdsu-4D was colocalized with Rht-D1. Peak significance for Qfhb.sdsu-2A was at Xbarc212, which corresponds closely to the position of a QTL on chromosome 2AS reported by Gupta et al. (2001) with resistance inherited from Freedom. However, Freedom was not found to confer significant resistance at Qfhb.sdsu-2A in the current study. It should be noted, however, that the power to detect resistance alleles inherited from a parental source in this study was dependent on the number of progeny derived from that parent line. Given that Freedom was only used to derive 117 four-way F1 plants, it is not surprising the allelic
Mol Breeding (2015) 35:6
effect from Freedom at this locus was not significant. This limitation of power to detect allelic effects is demonstrated by analysis of allelic effects of Fhb1. The Fhb1 donor Wesley-Fhb1-BC56 from which 369 progenies were derived was found to confer significant resistance and Fhb1, whereas significant effects were not observed for Fhb1 donors Wesley-Fhb1-BC06 and AL-107-6106 from which 184 and 62 progenies were derived, respectively. It should be noted that the experiments used to evaluate FHB resistance in this study were unbalanced. Unbalanced data are intrinsic to early-generation breeding populations, since seed amounts and populations sizes are generally limited in earlygeneration breeding nurseries. Analysis of these data requires the application of statistical methods that can simultaneously adjust for fixed affects and account for the differences in the precision of estimated entry means. Mixed linear models, as used in this study, provide a natural framework for handling the unbalanced phenotypic data derived from early-generation breeding populations. BLUPs estimated from fitting the mixed linear models are regressed back to the overall mean to a degree based on the repeatability of each entry mean, a property which maximizes the correlation between the predicted and actual genetic values for entries (Piepho et al. 2008). Moderate correlations between entry means and BLUPs for the four-way F1 plants indicate that significant adjustment of the phenotypic data was required to account for fixed effects and the unbalanced nature of the data. The moderate heritabilities for FHB severity in this experiment, ranging from 0.32 to 0.46, underscore the complex polygenic nature of FHB resistance. Given these heritability estimates, the weak correlations among greenhouse and selfed-progeny evaluations are expected. The theoretical maximum phenotypic correlation among independent evaluations, assuming a genetic correlation of 1, would be the product of the square root of the heritabilities between evaluations, which ranges from 0.38 to 0.43 between the greenhouse and selfed-progeny evaluations. The differential expression of QTL observed between the greenhouse and selfed-progeny evaluations indicates that the genetic correlation is clearly less than one, thus resulting in weak phenotypic correlations. Focus of further breeding efforts should be given to those loci that have repeatable effects across evaluations, including Qfhb.sdsu-1A, Qfhb.sdsu-4A.1, and Fhb1.
Page 13 of 16 6
Cumulative effects of pyramided FHB resistance alleles were primarily additive in both greenhouse and selfed-progeny evaluations. Therefore, marker-assisted selection for those genotypes carrying the largest number of resistance alleles should serve to enhance resistance to FHB. Marker haplotypes of the founder lines conferring resistance at the QTL in this study can directly be utilized in the breeding populations developed here to select for and fix FHB resistance alleles at multiple loci in multiple genetic backgrounds. Therefore, application of linkage analysis to early-generation breeding populations enabled a transition between QTL mapping and marker-assisted selection efforts as early as the F2 generation, without any extraneous population development. We therefore assert that the approach demonstrated in this study can enable the direct transition between QTL mapping and molecular breeding efforts. Using a conventional approach, several generations would have been required to develop a separate mapping population for each source of resistance, followed by QTL validation in separate populations, all prior to commencing breeding efforts to pyramid the QTL. However, it should be noted that the desirability of a particular marker haplotype varies among individual breeding populations. This results from the fact that the marker haplotypes are not always unique among founder lines and thus provide identicalby-state information rather than identical-by-descent information. Therefore, marker-assisted selection must be population specific and based on the founders of each population when the haplotypes are not unique among the founder lines. While IBD-based linkage enabled the detection of a large number of QTL for FHB resistance in plant breeding populations, there are aspects of the methods that need further refinement. First, application of IBDbased linkage analysis to plant breeding populations in this study and in previous studies (Rosyara et al. 2009; Ali et al. 2013) has required the assumption of independence among crosses. In reality, plant breeding populations tend to be highly interrelated by common founders, and ignoring the relationships among populations vastly reduces the number of sib-pair observations used for linkage analysis and thus reduces the power to map QTL. The assumption of independence among crosses is necessary simply because available software implementations of IBD-based linkage analysis of complex pedigrees have been developed for analysis of human and animal populations and thus require a sex designation
123
6 Page 14 of 16
for each individual in the pedigree. This is problematic as parental lines in a plant breeding populations can be used as both a male and female in different crosses, which makes the reconstruction of the complete set of relationships among populations impossible within the framework of these existing programs. Furthermore, incorporating the relationships among crosses would essentially result in the formation of a single complex pedigree. Joint analysis of this complex pedigree would require estimating the necessary pairwise IBD sharing probability among all 565 four-way F1 plants based on multipoint information for each of the 3,807 loci. Using Markov-chain Monte Carlo (MCMC) sampling methods to approximate multipoint IBD sharing probabilities given such a large dataset would likely have required months of computational time, even with unlimited computational resources provided. Methods for multithreading IBD analysis and simplifying MCMC sampling algorithms for phase known founder lines might alleviate some of this computational burden. The second limiting aspect is the ability to estimate the allelic effects at the QTL to enable marker-assisted selection. Since the number of alleles at the QTL is unknown in general pedigrees, IBD-based linkage analysis estimates the variance attributable to a QTL rather than the actual allelic effects. The haplotype analysis employed in this study tests for the unique effect of the allele inherited from each founder. Therefore, this approach only has power to detect allelic effects if a founder is represented by a large number of progeny across the breeding populations. If a large number of founder lines are used relative to the number of progeny, then an identical-bystate haplotype sharing approach such as that described by Jansen et al. (2003) may provide better estimates of allelic effects. Therefore, as noted by Jannink et al. (2001), the primary hindrance to the application of IBDbased linkage analysis in plant breeding populations is still the lack of available algorithms and software suited to the particular attributes of plant populations and mating systems. Future research therefore needs to be focused on the development of these resources to improve the power of QTL mapping in plant breeding populations.
Conclusions Application of IBD-based linkage analysis to 28 fourway F1 winter wheat breeding populations segregating
123
Mol Breeding (2015) 35:6
for Fhb1 and native sources of resistance enabled mapping of 15 FHB resistance QTL. Several of these QTL corresponded to known loci contributing to FHB resistance, including Fhb1, Fhb5, Rht-B1, and Rht-D1, validating the reliability of the IBD-based linkage analysis for detected multiple QTL for FHB resistance in plant breeding populations. A QTL on chromosome 1AS (Qfhb.sdsu-1A) with resistance conferred from Lyman and Overland was detected in all greenhouse and selfed-progeny evaluations, explaining between 4.5 to 9.9 % of the phenotypic variance. Qfhb.sdsu-1A, along with less repeatable QTL on chromosomes 1B, 4A and 6A, appear to be novel QTL for native FHB resistance that can be exploited in North American wheat breeding programs. Germplasm developed in this study should be useful for the purpose of developing breeding lines for FHB resistance, and QTL mapping results presented here are directly applicable to marker-assisted selection for this material. This study demonstrates the utility of IBD-based linkage analysis for integrating QTL mapping and plant breeding efforts. However, further work is needed in the development of software implementations of IBD-based linkage analysis that are adapted to the particular attributes of plant populations, including hermaphroditic mating systems, large interrelated populations, and systems of inbreeding before these methods are routinely employed to support molecular plant breeding efforts.
References Abate ZA, Liu S, McKendry AL (2008) Quantitative trait loci associated with deoxynivalenol content and kernel quality in the soft red winter wheat ‘‘Ernie’’. Crop Sci 48(4):1408–1418 Ali MB, Ibrahim AMH, Malla S, Rudd J, Hays DB (2013) Family-based QTL mapping of heat stress tolerance in primitive tetraploid wheat (Triticum turgidum L.). Euphytica 192(2):189–203 Anderson JA (2007) Marker-assisted selection for fusarium head blight resistance in wheat. Int J Food Microbiol 119:51–53 Anderson JA, Stack RW, Liu S, Waldron BL, Fjeld AD, Coyne C, Moreno-Sevilla B, Fetch JM, Song QJ, Cregan PB, Frohberg RC (2001) DNA markers for Fusarium head blight resistance QTLs in two wheat populations. Theor Appl Genet 102(8):1164–1168 Bai G, Shaner G (2004) Management and resistance in wheat and barley to Fusarium head blight. Annu Rev Phytopathol 42:135–161 Bates D, Maechler M, Bolker B, Walker S (2013) lme4: linear mixed-effects models using Eigen and S4. R package version 1.0-4. http://CRAN.R-project.org/package=lme4
Mol Breeding (2015) 35:6 Beavis WD (1998) QTL analyses: power, precision, and accuracy. In: Paterson AH (ed) Molecular dissection of complex traits. CRC Press, Boca Raton, pp 145–162 Bernardo R (2008) Molecular markers and selection for complex traits in plants: learning from the last 20 years. Crop Sci 48(5):1649–1664 Buerstmayr H, Ban T, Anderson JA (2009) QTL mapping and marker-assisted selection for Fusarium head blight resistance in wheat: a review. Plant Breed 128(1):1–26 Burlakoti RR, Mergoum M, Kianian SF, Adhikari TB (2009) Combining different resistance components enhances resistance to Fusarium head blight in spring wheat. Euphytica 172(2):197–205 Cai J, Bai G, Jin F, St Amand P, Bockus W, Baenziger S (2013) Effects of Fhb1 on resistance to wheat FHB in different hard winter wheats. In: Canty S, Clark A, Salat Y, Van Sanford D (eds) Proceedings of the 2013 national Fusarium head blight forum. Milwaukee, WI, p 12 Cavanagh CR, Chao S, Wang S, Huang BE, Stephen S, Kiani S, Forrest K, Saintenac C, Brown-Guedira GL, Akhunova A, See D, Bai G, Pumphrey M, Tomar L, Wong D, Kong S, Reynolds M, Lopez da Silva M, Bockelman H, Talber L, Anderson J, Dreisigacker S, Baenziger S, Carter A, Korzun V, Morrell PL, Dubcovsky J, Morell M, Sorrells ME, Hayden MJ, Akhunov E (2013) Genome-wide comparative diversity uncovers multiple targets of selection for improvement in hexaploid wheat landraces and cultivars. Proc Natl Acad Sci USA 110(20):8057–8062 Crepieux S, Lebreton C, Flament P, Charmet G (2005) Application of a new IBD-based QTL mapping method to common wheat breeding population: analysis of kernel hardness and dough strength. Theor Appl Genet 111(7): 1409–1419 Eckard JT, Gonzalez-Hernandez JL, Chao S, St Amand P, Bai G (2014) Construction of dense linkage maps ‘‘on the fly’’ using early generation wheat breeding populations. Mol Breed. doi:10.1007/s11032-014-0116-1 Furrer R, Nychka D, Sain S (2013) Fields: tools for spatial data. R package version 6.8. http://CRAN.R-project.org/package=fields Green P, Falls K, Crooks S (1990) Documentation for CRIMAP, version 2.4: www.animalgenome.org Guedira M, Brown-Guedira G, Van Sanford D, Sneller C, Souza E, Marshall D (2010) Distribution of Rht genes in modern and historic winter wheat cultivars from the Eastern and Central USA. Crop Sci 50(5):1811–1822 Gupta A, Lipps PE, Campbell KG, Sneller CH (2001) Identification of QTL associated with resistance to FHB in Ning 7840 and Freedom. In: 2001 National Fusarium head blight forum proceedings (p 175) Holzapfel J, Mohler V, Haeberle J, Schweizer G, Miedaner T, Voss H-H, Korzun V, Hartl L (2008) Genome distribution of QTL for Fusarium head blight resistance in European wheat germplasm. In: Appels R, Eastwood R, Lagudah E, Langridge P, Mackay M, McIntyre L, and Sharp P (eds) Proceedings of 11th international wheat genetic symposium, Brisbane, QLD, Australia, 24–29 August 2008. Sydney University Press, Sydney Jannink JL, Bink MCAM, Jansen RC (2001) Using complex plant pedigrees to map valuable genes. Trends Plant Sci 6(8):337–342
Page 15 of 16 6 Jansen RC, Jannink JL, Beavis WD (2003) Mapping quantitative trait loci in plant breeding populations: use of parental haplotype sharing. Crop Sci 43(3):829–834 Jin F, Zhang D, Bockus W, Baenziger PS, Carver B, Bai G (2013) Fusarium head blight resistance in U.S. winter wheat cultivars and elite breeding lines. Crop Sci 53(5):2006–2013 Liu S, Abate ZA, Lu H, Musket T, Davis GL, McKendry AL (2007) QTL associated with Fusarium head blight resistance in the soft red winter wheat Ernie. Theor Appl Genet 115(3):417–427 Liu S, Hall MD, Griffey CA, McKendry AL (2009) Meta-analysis of QTL associated with Fusarium head blight resistance in wheat. Crop Sci 49(6):1955–1968 Liu S, Griffey CA, Hall MD, McKendry AL, Chen J, Brooks WS, Brown-Guedira G, Van Sanford D, Schmale DG (2013) Molecular characterization of field resistance to Fusarium head blight in two US soft red winter wheat cultivars. Theor Appl Genet 126(10):2485–2498 Lo¨ffler M, Scho¨n CC, Miedaner T (2009) Revealing the genetic architecture of FHB resistance in hexaploid wheat (Triticum aestivum L.) by QTL meta-analysis. Mol Breed 23(3):473–488 ´ , Barto´k T, Ka´szonyi G, Varga M, To´th B, Varga J Mesterha´zy A (2005) Common resistance to different Fusarium spp. causing Fusarium head blight in wheat. Eur J Plant Pathol 112(3):267–281 Miedaner T, Voss H-H (2008) Effect of dwarfing genes on Fusarium head blight resistance in two sets of near-isogenic lines of wheat and check cultivars. Crop Sci 48(6):2115–2122 Miedaner T, Wilde F, Steiner B, Buerstmayr H, Korzun V, Ebmeyer E (2006) Stacking quantitative trait loci (QTL) for Fusarium head blight resistance from non-adapted sources in an European elite spring wheat background and assessing their effects on deoxynivalenol (DON) content and disease severity. Theor Appl Genet 112(3):562–569 Morris N, Stein C (2012) Model-free linkage analysis of a quantitative trait. Stat Hum Genet Methods Protoc 850:301–316 Piepho HP, Mohring J, Melchinger AE, Buchse A (2008) BLUP for phenotypic selection in plant breeding and variety testing. Euphytica 161:209–228 Rosyara UR, Gonzalez-Hernandez JL, Glover KD, Gedye KR, Stein JM (2009) Family-based mapping of quantitative trait loci in plant breeding populations with resistance to Fusarium head blight in wheat as an illustration. Theor Appl Genet 118(8):1617–1631 S.A.G.E. (2012) Statistical analysis of genetic epidemiology. Release 6.3: http://darwin.cwru.edu Schmolke M, Zimmermann G, Buerstmayr H, Schweizer G, Miedaner T, Korzun V, Ebmeyer E, Hartl L (2005) Molecular mapping of Fusarium head blight resistance in the winter wheat population Dream/Lynx. Theor Appl Genet 111:747–756 Shi JR, Xu DH, Yang HY, Lu QX, Ban T (2008) DNA marker analysis for pyramided of Fusarium head blight (FHB) resistance QTLs from different germplasm. Genetica 133(1):77–84 Singh AK, Knox RE, Clarke FR, Clarke JM, Somers DJ, Fedak G, Singh A, DePauw R (2008) Fusarium head blight QTL
123
6 Page 16 of 16 mapping in durum wheat and Triticum carthlicum sources of resistance. In: Appels R, Eastwood R, Lagudah E, Langridge P, Mackay M, McIntyre L, Sharp P (eds) Proceedings of 11th international wheat genetic symposium, Brisbane, QLD, Australia, 24–29 August 2008. Sydney, Sydney University Press Sneller C, Gupta A, Lipps P, Campbell K (2001) Mapping QTL for fusarium head blight resistance from wheat genotypes Freedom and Ning 7840. In: Fedak G, Choo A (eds) Canadian workshop on Fusarium head blight, Ottawa, ON, 3–5 November 2001. AAFC Publisher, Ottawa, ON Sneller CH, Mather DE, Crepieux S (2009) Analytical approaches and population types for finding and utilizing QTL in complex plant populations. Crop Sci 49(2):363–380 Somers DJ, Isaac P, Edwards K (2004) A high-density microsatellite consensus map for bread wheat (Triticum aestivum L.). Theor Appl Genet 109(6):1105–1114 Srinivasachary Gosman N, Steed A, Hollins TW, Bayles R, Jennings P, Nicholson P (2009) Semi-dwarfing Rht-B1 and Rht-D1 loci of wheat differ significantly in their influence on resistance to Fusarium head blight. Theor Appl Genet 118(4):695–702 Stack RW, Mcmullen MP (1998) A visual scale to estimates severity of Fusarium head blight in wheat. Fargo, ND
123
Mol Breeding (2015) 35:6 Verbeke G, Molenberghs G (2003) The use of score tests for inference on variance components. Biometrics 59(2): 254–262 Wang T, Elston R (2005) Two-level Haseman–Elston regression for general pedigree data analysis. Genet Epidemiol 22:12–22 Wu¨rschum T (2012) Mapping QTL for agronomic traits in breeding populations. Theor Appl Genet 125(2):201–210 Xie C, Gessler DD, Xu S (1998) Combining different line crosses for mapping quantitative trait loci using the identical by descent-based variance component method. Genetics 149(2):1139–1146 Yang Z, Gilbert J, Fedak G, Somers D (2005a) Genetic characterization of QTL associated with resistance to Fusarium head blight in a doubled-haploid spring wheat population. Genome 196:187–196 Yang J, Bai G, Shaner GE (2005b) Novel quantitative trait loci (QTL) for Fusarium head blight resistance in wheat cultivar Chokwang. Theor Appl Genet 111(8):1571–1579 Zhang X, Bai G, Bockus W, Ji X, Pan H (2012) Quantitative trait loci for Fusarium head blight resistance in U.S. hard winter wheat cultivar Heyne. Crop Sci 52(3):1187–1194
