Introduction

Wheat Fusarium head blight (FHB), mainly caused by Fusarium graminearum, is an economically important wheat disease in humid and semi-humid wheat-growing areas worldwide (Parry et al. 1995; McMullen et al. 1997). The grain harvested from FHB-infected wheat spikes shows reduced test weight, poor baking quality, and low germination rate when used as seed (Bai and Shaner 2004). Grain contaminated with deoxnivalenol (DON) produced by the fungus is also a major concern for animal production and human health (Snijders 1990; Bai et al. 2001).

There is no single effective measure for complete control of wheat FHB. The use of resistant cultivars, together with proper crop management practices, is the most effective strategy for control of the disease (Bai and Shaner 1994; McMullen et al. 1997). Two types of FHB resistance have been extensively studied and well documented: type I resistance against initial infection and type II resistance against spread within the infected spike (Schroeder and Christensen 1963). So far, a few FHB resistant germplasm have been reported in wheat, such as Sumai 3 (Anderson et al. 2001), Wangshuibai (Zhou et al. 2004; Lin et al. 2006), Renan (Gervais et al. 2003), Frontana (Steiner et al. 2004), Arina (Paillard et al. 2004), Goldfield (Gilsinger et al. 2005) and so forth. Type II resistance is more stable than type I in the majority of the resistant cultivars identified to date (Bai and Shaner 2004).

Fusarium head blight resistance is a complex trait (Bai and Shaner 1994; van Ginkel et al. 1996; Ban and Suenaga 2000). Two to five major genes, plus several minor genes, have been reported from various sources of FHB resistance (Buerstmayr et al. 1997; Yao et al. 1997; Grausgruber et al. 1998, 1999; Somers et al. 2003; Zhou et al. 2004). QTLs for FHB resistance have been mapped to almost all wheat chromosomes when different mapping populations were investigated (Bai and Shaner 2004). In Sumai 3, QTLs for FHB resistance have been identified on 3BS, 5AS, 6AS, 6BS, and 3BSc, a QTL region proximal to the centromere on 3BS (Anderson et al. 2001; Buerstmayer et al. 2002; Zhou et al. 2002; Yang et al. 2003). In Wuhan 1, the QTLs for FHB resistance were mapped to 2DL and 4B (Somers et al. 2003). QTLs on other chromosomes were also reported including those on chromosomes 2A, 2B, 3A, 3B, 5D, 6D, 5B, 4A, 1B, and 7A in wheat germplasm from Europe, Brazil, and Asia (Gervais et al. 2003; Paillard et al. 2004; Steiner et al. 2004; Zhou et al. 2004). But only the QTL on 3BS from the Chinese cultivar Sumai 3 consistently showed a major effect on Type II resistance across different genetic backgrounds and environments (Bai et al. 1999; Anderson et al. 2001; Buerstmayr et al. 2002, 2003; Zhou et al. 2002, 2004; Yang et al. 2003). Other QTLs for FHB resistance exhibited a minor effect, and their expression varied significantly with genetic backgrounds and the environments where the disease was evaluated (Bai and Shaner 2004). Therefore, Sumai 3 has been extensively used as a major source of resistance to FHB in breeding programs worldwide (Rudd et al. 2001; McCartney et al. 2004). However, heavy use of narrow FHB resistance sources may increase selection pressure on the pathogens to wear away the effectiveness of the resistance genes involved. New FHB resistance germplasm are desired to broaden the genetic diversity of FHB resistant sources and improve the level of wheat resistance to FHB (Ruckenbauer et al. 2001; Gervais et al. 2003). The objectives of this study were to identify new sources of FHB resistance from the Asian wheat gene pool, and elucidate the genetic relationship among these accessions using molecular markers.

Materials and methods

Plant materials

Fifty-nine wheat accessions originated from China (38 accessions), southwestern Japan (20 accessions), and Korea (one accession) (Table 1). The majority of the Asian accessions were believed to have some degrees of FHB resistances, including the well-known FHB resistant cultivars (Sumai 3, Wangshuibai, and Ning 7840) from China. Three U.S. cultivars (Ernie, Freedom and Clark) were also included as moderately resistant and susceptible controls, respectively. Funo from Italy and Avrora from Russia were also included because Funo is a parent of Sumai 3 and Avrora is a parent of Ning 7840. Both of them have been extensively used as parents in Chinese breeding programs from the 1950s to the 1970s (Bai et al. 2003).

Table 1 Origin, pedigree and Fusarium head blight severity for 64 wheat accessions
Full size table

Evaluation of FHB resistance

Type II resistance was measured by using a syringe to inject 1,000 conidia spores of F. graminearum into a basal floret of a central spikelet of a spike at anthesis. All wheat accessions were evaluated for FHB resistance in the greenhouse of Kansas State University in 2003 as described by Bai et al. (2003). The inoculum of F. graminearum was a field isolate (GZ 3639) that originated from Kansas. This isolate has been well characterized for its high aggressiveness and DON production (Desjardins et al. 1996). After vernalization at 4°C in a growth chamber for eight weeks, six plants were transplanted into a 5′ × 5′ tora pot (Hummert Int., St. Louis, MO, USA) filled with Metro Mix 360 soil mix (Hummert Int.,), and were grown on a greenhouse bench with a 12-h daylight period. The FHB resistance evaluation experiment was repeated once, with three replications in each experiment. Four to six plants, one head per plant, were inoculated for each pot. Inoculated plants were incubated in a moist chamber for 3 days to initiate infection. Then the infected plants were moved to the original bench position and were grown at 25°C during day and 22°C during night. Infected and total spikelets in a spike were counted at the 21st day after inoculation, and the proportion of symptomatic spikelets (PSS) was calculated as a measure for disease severity.

Molecular marker analysis

DNA was isolated from seedling leaf tissue according to CTAB method (Saghai-Maroof et al. 1984). For AFLP analysis, DNA restriction digestion (with EcoRI and MseI), adapter ligation, and PCR amplification were carried out as described by Bai et al. (2003). Pre-amplification was conducted with an EcoRI primer (5′-ACTGCGTACCAATTC) and an MseI primer (5′-GATGAGTCCTGAGTAA). Selective PCR used 24 primer combinations between six IR-dye-labeled EcoRI primers with selective nucleotides of AGT, AAC, ACT, GCTG, CTCG, and CATG and five unlabeled MseI primers with selective nucleotides of CAC, CAT, CAGT, TGC, and AGTG.

Twenty-five SSR markers (Table 2) associated with FHB-resistance QTLs on 3BS (Anderson et al. 2001; Buerstmayer et al. 2002; Zhou et al. 2002), 3BSc (Somers et al. 2003), 5AS (Buerstmayer et al. 2002), 6BS (Anderson et al. 2001; Yang et al. 2003), 4B, and 2DL (Somers et al. 2003) were screened for polymorphisms among these accessions. The FHB-resistance QTLs on 4B and 2DL were originally detected from Wuhan 1, a Chinese breeding line with unknown pedigree (Somers et al. 2003), and five SSR markers linked to the two QTLs (McCartney et al. 2004) were included in this study (Table 2). The SSR markers were amplified according to the protocol described by Bai et al. (2003). For PCR detection, an M13 tail sequence was added to the 5′-end of the forward SSR primers (5′-ACGACGTTGTAAAACGAC). A PCR with 10-μl reaction volume consisted of ∼50 ng DNA, 1× PCR buffer, 0.2 mM dNTPs, 2 mM MgCl2, 1 pmol each of tailed-forward and reverse primers, as well as IR-dye-labeled M13 primer (Li-Cor, Inc., Lincoln, NE, USA). The SSR markers on 3BS were directly labeled with IR-dye in the 5′-ends without adding the M13 primer in PCR reaction mixture. To amplify SSR, a touchdown PCR profile started at 95°C for 5 min, followed by 5 cycles of 45 s at 95°C, 5 min at 68°C, and 1 min at 72°C; the annealing temperature was lowered by 2°C in each following cycle. Then 5 more cycles in which the annealing time was 2 min and the temperature was lowered 2°C in each following cycles. For the last 25 additional cycles, the annealing temperature was held constantly at 50°C, with 5 min at 72°C for a final extension. The AFLP and SSR fragments were analyzed in a Li-Cor 4200 DNA Sequencer (Li-Cor, Inc., Lincoln, NE, USA).

Table 2 Allele sizes of the SSR markers that were reported to be tightly linked to QTLs for FHB resistance in Sumai 3 and Wuhan 1 for 64 wheat accessions
Full size table

Data analysis

Polymorphic DNA fragments were scored as either present (1) or absent (0) for each marker locus by using SAGA software (Li-Cor, Inc., Lincoln, NE, USA). The SAS software package was used for basic statistical analysis (SAS Institute Inc., Cary, NC, USA). Cluster analysis was performed using NTSYSpc version 2.11a (Rohlf 1998). The genetic diversity among accessions, on the basis of the AFLP and SSR data, was estimated according to Jaccard’s similarity coefficient and was calculated as \(1 - {\left[ {a / {{\left({n - d} \right)}}} \right]},\) where a is the number of bands in common between two wheat accessions, n is the number of bands in the matrix, and d is the number of bands absent in both wheat accessions. The SIMQUAL routine of NTSYSpc program was used to generate the Jaccard’s similarity coefficient matrix. The unweighted pair-group method with arithmetic mean (UPGMA) and SHAN routine of NTSYSpc program were used to construct a dendrogram. Bootstrapping (n = 500) was performed to evaluate the robustness of the branching points using Phyltools (Buntjer 2001). The neighboring and consensus modules from the Phylip program (Felsenstein 2005) were used to construct the consensus tree. Bootstrap values were percentages of number of runs showing a specific branch point in the consensus tree when the data were randomly re-sampled for 500 times. Haplotypes of the 64 wheat cultivars were determined on the basis of the allelic distribution pattern of SSR markers linked to 3BS, 5AS, 4B, and 6BS QTL in Sumai 3. The polymorphism information content (PIC) refers to the ability of a given marker to detect polymorphism within a population (Anderson et al. 1993). The PIC depends on the number of detectable alleles and their frequency. In this study, the simplified version (Lagercrantz et al. 1993) is used, which assumes that the wheat accessions are all homozygous.

$${\rm PIC}_{i} = 1 - {\sum\limits_{j = 1}^{n}} p_{ij}^{2},$$

where p ij is the frequency of the jth pattern for marker i, and n is the total number of patterns.

Results

Type II FHB resistance of Asian wheat accessions

Asian wheat accessions differed significantly in FHB severity, as reflected by their PSS (Table 1). The correlation coefficient for PSS between the two experiments was highly significant (r = 0.66, P < 0.0001). Wheat responses to FHB infection ranged from highly resistant (PSS < 10%, F60096 and Fu5114) to highly susceptible (PSS > 85%, Sanyuehuang). Approximately 67% of the Asian wheat accessions tested showed a high or moderate level of FHB resistance, with a mean PSS of less than 50% under the favorable epidemic conditions. More than half of the highly resistant accessions originated from China. Ernie from the USA also showed a high level of FHB resistance. The remaining highly resistant lines originated from Japan, including Aso Zairai II, Aso Zairai (Yuubou Kappu), Itou Komugi, and Shirasaya No.1. Taiwan Xiaomai, one of the parents of Sumai 3, had a PSS of less than 15%, which was similar to Sumai 3. Another parent of Sumai 3, Funo, showed moderate susceptibility to FHB in two experiments.

Genetic relationships among the wheat accessions

A total of 25 SSR alleles and 483 AFLP polymorphic alleles were scored. The AFLP and SSR data illustrated that the resulted groups from cluster analysis of these accessions basically matched with their geographic distribution and/or available pedigrees, with only a few exceptions (Fig. 1). In general, the cluster analysis roughly separated the 64 accessions into three major clusters, a Chinese/Japanese landrace cluster, an Avrora-related cluster, and a Funo-related cluster. Avrora- and Funo-related clusters consisted of all improved cultivars from China, with Avrora and Funo, respectively, as one parent. Most of the southwestern Japanese landraces (17 out of 19) formed a closely related subgroup within the Chinese/Japanese landrace cluster (Fig. 1). The Japanese subgroup was separated from the Chinese landraces at a similarity coefficient about 0.82, whereas the most distant Chinese landrace, Chinese Spring, was separated from the other Chinese landraces at a similarity coefficient about 0.77. It was interesting that Japanese landraces Shironankin and Shinchunaga were closer to Chinese accessions than to Japanese landraces. Overall, the genetic distance between the Chinese and the Japanese landraces was closer than between some of the Chinese landraces.

Fig. 1
figure 1

UPGMA dendrogram based on Jaccard’s similarity coefficient from AFLP and SSR marker data showing genetic relationships among the 64 wheat accessions used in this study. Bootstrap values (%) for each branch point are indicated if they are > 50%. § Accession ID

Full size image

The closest accessions in this study were Japanese accessions Nyubai, Nobeokabouzu and Nobeokabouzu Komugi, and they were separated at a similarity coefficient about 0.99. Two Chinese landraces, Huoshaobairimai and Huacandou, were also very close and were separated at a similarity coefficient about 0.96. The Avrora-related cluster and the Funo-related cluster were separated from the Chinese/Japanese landrace cluster at similarity coefficients of 0.71 and 0.73, respectively. No Japanese landraces were classified into the Avrora-related cluster or the Funo-related cluster. Accessions in the Avrora-related cluster (6 lines) and the Funo-related cluster (11 lines) were in good agreement to their pedigrees. As for FHB resistant sources, the genetic diversity within the Chinese landraces was broader than that of the Japanese landraces (Fig. 1).

Allelic variation in SSR marker loci linked to QTLs for FHB resistance

Twenty-five SSR markers linked to six putative QTLs on five chromosome arms of wheat were highly polymorphic among the wheat accessions evaluated (Table 2). The PIC values for these SSRs ranged from 0.23 (Xgwm113) to 0.92 (Xwmc612). Two (Xbarc75 and Xgwm508) to 18 (Xwmc612) alleles per SSR locus were detected across all 64 accessions (Table 2). Haplotypes based on Sumai 3/non-Sumai 3 alleles were identified—from 4 on 3BSc to 24 on 5AS (Table 2).

Twenty-one haplotypes were identified for the five SSR markers linked to the major QTL on 3BS among the 64 accessions when the marker alleles were analyzed as Sumai 3 and non-Sumai 3 alleles. Only two accessions, Ning 7840 and Taiwan Xiaomai, shared the identical haplotype with Sumai 3 (Table 2). Two Japanese accessions (Sanshukomugi and Shinchunaga) and one Chinese accession (FSW) had four of the five Sumai 3 SSR alleles; 34 accessions contained two to three Sumai 3 SSR alleles; 11 accessions carried only one Sumai 3 SSR allele; and the remaining 12 lines, including Funo (another parent of Sumai 3) and Avrora (one of Ning 7840’s parents), didn’t carry any of the five Sumai 3 SSR alleles for the major QTL on 3BS (Table 2). Among those 23 accessions with none or only one Sumai 3 allele, 7 lines expressed a similar level of FHB resistance as that of Sumai 3, and another 8 accessions demonstrated moderate FHB resistance, with less than 50% PSS under a high disease pressure. The 15 wheat lines are unlikely to carry the major QTL for FHB resistance and the resistance in those accessions may be controlled by different QTLs or alleles from that on 3BS of Sumai 3.

The QTLs on 2DL and 4B were only reported from Wuhan 1 and have not been reported in Sumai 3. Wuhan 1 was not included in this study, so it was difficult to determine the alleles of the FHB resistance related SSR markers among the accessions, but most Sumai 3 SSR alleles on the 2DL and 4B QTL regions seemed to be the same as those of susceptible controls (Tables 1, 2), indicating that these SSR are not polymorphic between Sumai 3 and FHB susceptible accessions. For SSR markers linked to QTLs on 3BS, 5AS, and 6BS, a high level of polymorphism was observed between Sumai 3 or its derivatives and susceptible controls. The overall results indicated a trend in which the more the putative Sumai 3 marker alleles for QTLs on 3BS, 5AS, and 6BS an accession had, the more likely the accession had a lower average PSS.

Discussion

Genetic relationships among Asian FHB resistant germplasm

In this study, a collection of FHB resistant germplasm was characterized on the basis of both FHB phenotypic and molecular marker data. In this collection, some are well-known FHB resistant cultivars, whereas most of them are landraces from China, Japan and Korea. Their FHB resistance has not been systematically characterized, and their genetic relationship to Sumai 3 is still unclear because their pedigrees are not available. Molecular-marker-based cluster analysis has been used as a highly reliable means for estimating the genetic relationships among cultivars, with or without known pedigrees (Barrett et al. 1998; Sun et al. 2003; Bai et al. 2003; McCartney et al. 2004). Both analysis methods were used in this study and provided consistent results that led to further understanding of genetic relationships among these Asian FHB resistant accessions. Some putative new sources of FHB resistance other than Sumai 3 were identified. This study also provided systematic pedigree information for many well-known FHB resistant wheat cultivars, and useful FHB data and molecular marker profiles for these accessions. Therefore, the information from this study will help breeders to select different sources of resistant materials for enlarging the genetic diversity of FHB resistance in their breeding programs.

Cluster analysis indicated that the genetic similarities for all pair-wise comparisons among the 64 wheat lines ranged from 0.71 to 0.99, which suggested relatively limited genetic diversity among the accessions in this collection. This result was not surprising because most of FHB resistant wheat landraces originated from the southeast China and from the Kyushu area of southwestern Japan, where FHB epidemics have been frequent and severe, and the improved cultivars were mainly related to Avrora from Russia, Mentana, and its relative Funo from Italy. The results on genetic similarities of Chinese/Japanese FHB resistance accessions from this study agreed with a previous study (Bai at al. 2003).

The genetic relationship based on cluster analysis matched well with their pedigree information and their geographic origins. The dendrogram clearly separated the Funo-related accessions and the Avrora-related accessions from the Chinese/Japanese landraces. This might be because the Russian cultivar Avrora and Italian cultivar Funo and Mentana were far from Asian landraces and were extensively used in early wheat breeding programs in China. Introduction of cultivars Funo and Avrora from Europe significantly broadened the genetic diversity of Chinese wheat. It was unexpected that the three accessions from the USA, Freedom, Ernie, and Clark, didn’t form a separate group in the dendrogram. These three cultivars were in a completely different cluster from the Chinese landraces and cultivars in a previous study when more cultivars from the USA were used (Bai et al. 2003). This could be because the small number of accessions with diverse genetic backgrounds from the USA could not provide sufficient genetic information to form their own group. Therefore, interpretation of genetic relationships between Asian landraces and the cultivars from the USA should be cautious.

The Chinese landrace Taiwan Xiaomai and the Italian cultivar Funo are two parents of Sumai 3, and Sumai 3 locates between the Funo and Taiwan Xiaomai, although it is closer to Taiwan Xiaomai in the Chinese/Japanese landrace cluster (Fig. 1). On the other hand, Sumai 3 and Avrora are the parents of Ning 7840, and Ning 7840 is closer to Avrora than to Sumai 3 (Fig. 1). The Funo cluster is further separated into three subgroups. Cultivars Nanda 2419, Emai 6, Jingzhou 1, and Wannian 2 form a subgroup that share a common Italian ancestor Mentana (Fig. 1). Yangmai 1, Yangmai 4, Fumai 3, and Yangmai 158 form another subgroup in which Funo serves as a common ancestor. Fusuihuang and Zhen 7495 form the third subgroup because Zhen 7495 was derived from a cross between Fusuihang and Youyimai, a derivative of Funo. Therefore, the results clearly confirmed the pedigree relationships within each of the three subgroups. Results for the Avrora-related cluster were similar. In this cluster, Ning 7840, Sumai 49, and Fu 5114 shared the common ancestor Avrora. But the two Chinese landraces FSW and WZHHS were close to Ning 7840, and their pedigree information was not available to verify this relationship.

The landrace accessions collected from Japan Kyushu area were stored on the Gene Bank of MAFF, Japan. They were selected as the best FHB resistant germplasm on the basis of several years of field FHB evaluations conducted in Japan and CIMMYT-Mexico. These Japanese landraces form a sub-cluster within Chinese landrace cluster, indicating the genetic bases of FHB resistant landraces from southwestern Japan is narrower than that of the Chinese landraces. It is interesting that the Japanese landraces Shinchunaga and Shironankin are closer to Sumai 3 and Caizihuang, respectively, than to the other Japanese landraces (Fig. 1). Shironankin might directly originate from China because “Shironankin” means ‘White Nanjing’ in Japanese and Nanjing is a Chinese city where many FHB resistant Ning lines were developed. The same might also be true for Shinchunaga, which has been a major source of FHB resistance widely used in Japanese breeding programs for decades (Ban 2000). Shinchunaga had similar banding patterns to those in Sumai 3 at most SSR marker loci in the three QTL regions of Sumai 3 (Table 2). The results suggested that some of FHB resistant QTLs in Japanese germplasm might also originate from Chinese landraces.

Another Japanese landrace Nobeokabouzu Komugi was reported to have the best resistance in Japanese germplasm (Ban 2000). In our study, two other accessions, Nobeokabouzu and Nyubai, are very close to Nobeokabouzu Komugi, with 99% identity according to AFLP marker data and 100% identity according to the 25 SSR marker alleles scored in this study. In our collection, Nyubai and Nobeokabouzu Komugi were originally from the Gene Bank of Japan, and Nobeokabouzu was obtained from China. Therefore, it is possible that Nobeokabouzu Komugi and Nobeokabouzu are the same landrace with different identification because ‘Komugi’ means ‘wheat’ in Japanese and it can be omitted from its name. Nyubai may also be the same landrace as Nobeokabouzu. Therefore, any one of them should be able to represent the same accession in breeding programs.

Origin of QTLs for FHB resistance from Sumai 3

The major FHB resistance in Sumai 3 was once assumed to be from Funo, or from transgressive segregation of resistance genes from both parents (Liu and Wang 1990). More recent studies suggested that Taiwan Xiaomai might be the donor of 3BS major QTL from Sumai 3 by comparing haplotypes of 3BS markers from Sumai 3 and Funo (Bai et al. 2003; Liu and Anderson 2003). But, marker data for other QTLs and phenotypic data from Taiwan Xiaomai were not available in those studies. The FHB and SSR marker data from this study provided more solid evidence to support the recent assumption. Taiwan Xiaomai demonstrated the same high level of FHB resistance as Sumai 3, whereas Funo, the other parent of Sumai 3, was moderate susceptible to FHB. The five SSR marker alleles closely linked to the 3BS FHB resistant QTL in Taiwan Xiaomai were exactly the same as those of Sumai 3, whereas those from Funo were completely different. In addition, Taiwan Xiaomai had most of Sumai 3 SSR alleles at 5AS and 6BS loci, whereas Funo has only a few Sumai 3 SSR alleles in the two QTL regions. Therefore, Taiwan Xiaomai was also likely the donor of the QTLs on 5AS and 6BS in Sumai 3. This study reveals the origin of 5AS and 6BS QTLs for FHB resistance in Sumai 3.

Haplotype pattern and FHB resistance

The haplotype of SSR markers that flank QTL can help to predict whether an accession carries known or different QTL. The results from this study show a trend in which the more the putative Sumai 3 marker alleles an accession has for QTLs on 3BS, 5AS and 6BS, the more likely the accession has lower average PSS, suggesting that QTLs on 3BS, 5AS and 6BS are important in most of the resistant accessions. For example, the accessions with less than six Sumai 3 SSR alleles for the three QTLs had an average PSS of 43%, whereas those with more than 10 Sumai 3 alleles had an average PSS of 17%. This result also suggested that marker-assisted selection based on one marker per QTL might not be sufficient, and two or more flanking markers per QTL could provide better selection progress.

On the other hand, haplotype information could only roughly predict if an accession had one or more of the putative QTL(s) and predict its FHB-resistance performance in general. For instance, Fu 5114 was a descendant of Sumai 3 and carried three Sumai 3 SSR alleles at 3BS, three Sumai 3 SSR alleles at 5AS and one Sumai 3 SSR allele at 6BS. The haplotyping data suggested that Fu 5114 might inherit both the 3BS and the 5AS QTL (Table 2). Therefore, its good FHB resistance was consistent with the anticipation based on its haplotype information. But several factors may affect the accuracy of QTL predictions, including the genetic relationship between a target line and the line with known QTLs, the phenotypic effect of the target QTLs, non-precise locations of QTLs, and the genetic distance between the QTL and the markers used for the prediction. If an accession is not genetically related to the QTL donor, haplotype information may not provide a valid prediction for the presence of target QTL. If the markers are far from the target QTL, the prediction may also not be accurate. For example, the Chinese landrace Sanyuehuang carries two Sumai 3 3BS alleles, including Xgwm533, the most closely linked SSR maker for 3BS QTL in Sumai 3, and four Sumai 3 5AS alleles, but it showed high susceptibility to FHB. In contrast, Huoshaomai, which carries one 3BS marker allele (Xgwm389) of Sumai 3, and four 5AS marker alleles of Sumai 3, showed high FHB resistance. Although both had similar haplotypes, their reactions to FHB were completely different. These two landraces might not relate to Taiwan Xiaomai or Sumai 3 and the high susceptibility of Sanyuehuang is possibly due to epistasis or has several other QTLs that increase susceptibility. Excellent FHB resistance from Huoshaomai might be contributed by other QTLs than those on 3BS and 6BS or by different alleles of the QTLs on 3BS and 6BS.

Potential new QTL for FHB resistance

Haplotyping wheat accessions with SSR markers flanking FHB resistant QTLs may also provide useful information for predicting novel QTLs by comparison of haplotypes of target accessions with known cultivars. The underlying assumption is that if a wheat line has the same allelic pattern for marker loci flanking the QTL as that in the known resistant line, the two lines most likely have the same QTL (Bai et al. 2003; Sun et al. 2003; McCartney et al. 2004); on the other hand, if a wheat line has a different allelic pattern from that in the known resistant line, the two lines most likely have different alleles of the QTL.

In this study, 21 wheat lines expressed a similar level of FHB resistance as that of Sumai 3. Among them, Ning 7840, Fu 5114, and Sumai 49 were descendents of Sumai 3, and haplotyping data also supported a conclusion that they carry one to three of these QTLs on 3BS, 5AS, and 6BS. The other 18 accessions did not have clear genetic relationships with Sumai 3 or Taiwan Xiaomai. Among them, 12 accessions including F60096, Wangshuibai, Asozairai II, Huoshaobairimai, Shirasaya No 1, and Huangfangzhu had two to three Sumai 3 marker alleles at 3BS, one to four Sumai 3 marker alleles at 5AS, and one to six Sumai 3 marker alleles at 6BS. The haplotyping data indicated that the 12 accessions might carry at least one of the three Sumai 3 FHB resistant QTL, and the QTLs from the 12 lines were possibly either the same as, or allelic to, QTLs on 3BS, 5AS, or 6BS of Sumai 3. The remaining six accessions, Fumai 3, Yangmai 1, Haiyanzhong, Huoshaomai, Huangcandou, and Ernie, carried none or only one Sumai 3 SSR allele at 3BS and 6BS, and no more than four Sumai 3 SSR alleles at 5AS; therefore, these lines might not have the 3BS and 6BS QTL, and might have the 5AS QTL in a few accessions. Novel FHB resistant QTL might contribute to their high level of FHB resistance because the 5AS QTL only had a minor effect on Type II resistance (Somers et al. 2003). These accessions can be used to enlarge the wheat FHB resistance gene pool and enhance genetic diversity by incorporating different types of resistance in FHB resistant cultivars.