Diallel Analysis of Fusarium Head Blight Resistance in Soft Red Winter Wheat M. D. Hall and D. A. Van Sanford* ABSTRACT Fusarium graminearum Schwabe, the causative agent of Fusarium head blight (FHB), is an economically important pathogen of wheat (Triticum aestivum L.). Breeding FHB resistant wheat requires knowledge of the underlying genetic control of FHB resistance. While soft red winter (SRW) wheat breeders have begun to characterize germplasm for resistance, little information is available on the value of cultivars as parents in breeding for FHB resistance. Two diallel series of crosses, each involving nine SRW parents, were evaluated in both the greenhouse and field near Lexington, KY. The progeny and parents from the diallels were point inoculated with F. graminearum in both environments. FHB severity, deoxynivalenol (DON) level, and kernel damage were recorded. FHB severity ranged from 4.9 to 84.3% in the greenhouse and from 16.4 to 86.7% in the field. Significant general and specific combining ability effects were observed in both diallels in both environments. The nature of the resistance varied with diallel and with screening environment. On the basis of ratios of variance components, resistance in diallel 1 depended primarily on additive genetic effects, while in diallel 2 in the field dominance effects prevailed. Correlations between the greenhouse and field screening environments were low (0.36 in diallel 1 and 0.14 in diallel 2). Genotype environment interactions were highly significant and due almost exclusively to rank change interactions. These results do not support indirect selection in a greenhouse environment for resistance expressed in the field. F usarium head blight, caused by Fusarium spp. is a devastating disease of wheat and barley (Hordeum vulgare L.) throughout the world. The damage attributed to FHB has been well documented periodically throughout the past 110 yr (Arthur, 1891; Windels, 2000). In addition to reducing grain yield and test weight, the fungus produces mycotoxins. The most prevalent of these, deoxynivalenol (DON), has been linked to livestock feed refusal (Meronuck and Xie, 2000) and depression of the immune system, nausea, and vomiting in humans (Peraica et al., 1999). Concerns over food safety have led the FDA to impose a strict 1 g g 1 (1 ppm) standard for finished wheat food products (www.usda.gov/gipsa; verified 2 April 2003). In 1997, after several years of severe FHB epidemics in the northern Great Plains and in the eastern wheat region, the U.S. Wheat and Barley Scab Initiative was conceived and funded (http://www.scabusa.org; verified 2 April 2003). One of the key avenues of research within the Initiative is the development of FHB-resistant wheat and barley cultivars. Significant genetic variation for FHB resistance has been documented in wheat (Buerst- Dep. of Agronomy, Univ. of Kentucky, Lexington, KY 40546. The investigation reported in this paper (03-06-013) is in connection with a project of the Kentucky Agric. Exp. Stn. and is published with the approval of the director. Received 1 Sept. 2002. *Corresponding author (firstname.lastname@example.org). Published in Crop Sci. 43:1663–1670 (2003). Crop Science Society of America 677 S. Segoe Rd., Madison, WI 53711 USA mayr et al., 1996, Bai et al., 2001, Mesterhazy et al., 1999). Resistance to FHB is considered to be a quantitative trait and is therefore likely governed by several genes (Buerstmayr et al., 1999). A number of studies have indicated that resistance is controlled by a few major genes and numerous genes with minor effects (Bai et al., 1999, Waldron et al., 1999, Snijders, 1990). Bai et al. (2000) suggested that additive genetic effects are prevalent in controlling resistance to FHB, although some dominant and epistatic effects were found. Depending on the genotypes used, many researchers have reported that resistance to FHB is controlled by one to three genes (Bai et al., 2000), one to six genes (Snijders, 1990), or two genes (Van Ginkle et al., 1996). Screening for FHB resistance is performed in many different ways and by many national and international breeding programs. With up to seven different types of resistance identified (Mesterhazy, 1995, Mesterhazy, personal communication, 2001), researchers have developed numerous methods to study each different type of resistance. Research is most often focused on Type I resistance (resistance to initial infection) and Type II resistance (resistance to spread). Both greenhouse and field screenings are utilized to select genotypes with potential resistance. Type II resistance is routinely screened for in greenhouse experiments utilizing the point inoculation technique with a macroconidial spore suspension and humidity chambers. Field techniques usually include some type of irrigation and the mode of inoculation can differ widely from program to program. Common methods of field inoculation include colonized grain spawn made from either corn or wheat kernels, or macroconidial sprays that are used to assess other potential types of resistance along with Types I and II. Few studies in the literature have reported both greenhouse and field screenings of the same genotypic material and even fewer use the same inoculation technique in both environments. This is the first study to report a diallel analysis in both the greenhouse and field. In spring wheats, several FHB resistant cultivars have been developed through the use of the highly resistant Chinese Spring cultivar Sumai 3 or derivatives from it. ‘McVey’ (Busch et al., 2001) and ‘Alsen’ (www.ag.ndsu. nodak.edu/alsen.htm; verified 2 April 2003) are two such spring wheat cultivars. The resistance from Sumai 3 has also been incorporated into some winter wheats including Pioneer Brand ‘25R18’ (B. Laskar, personal communication, 2001). The narrow use of resistance sources may become problematic in the future assuming that the pathogen contains genetic diversity within its natural population. The other problem associated with use of Chinese Spring wheats as resistance sources is Abbreviations: FHB, Fusarium head blight; DON, deoxynivalenol; SRW, soft red winter; GCA, general combining ability; SCA, specific combining ability. 1663 1664 CROP SCIENCE, VOL. 43, SEPTEMBER–OCTOBER 2003 that it is difficult to recover adapted SRW types rapidly. Thus, among SRW wheat breeders there is great interest in identifying FHB resistance in the adapted SRW germplasm. To this end, two uniform FHB nurseries have been established (http://www.scabusa.org). However, there is a scarcity of information on the potential of SRW wheats as parents in an FHB resistance breeding program. The objectives for this study were to: (i) identify parents which produced low FHB severity and/or low DON progeny in two diallel sets of crosses among adapted SRW wheats and (ii) compare field and greenhouse response for FHB severity and DON concentration in the parents and their diallel progeny. MATERIALS AND METHODS F1 Crossing Cycle Cultivars or breeding lines well adapted to Kentucky with acceptable agronomic characteristics and end use qualities were used to construct two nine-parent diallels. Parents were chosen on the basis of their range in severity of FHB infection (diallel 1) or for their range in DON concentration (diallel 2) when grown in an inoculated FHB nursery. Each of the diallels contained parents that ranged from susceptible (‘2555’, 53.4% severity) to resistant (‘25R18’, 5.2% severity). Field Screening F1 progeny and parents were planted in five-seed hill plots on 15 Oct. 2000 at the Spindletop Research Farm near Lexington, KY. Reciprocal crosses were combined to ensure that enough seed was available to represent all F1 crosses. Plots were hand-planted in a randomized complete block design with three replications. Row spacing between the hill plots was 61 cm. Unfortunately, because of low F1 seed numbers, two parents and their corresponding progeny were eliminated from diallel 1, and one parent and its corresponding progeny were eliminated from diallel 2. An overhead mist irrigation system on an automatic self timer was installed to provide adequate moisture and humidity to create an FHB epidemic. Between 0600 and 0800 h the irrigation ran for 5 min in 15-min intervals. The irrigation schedule also included a 10-min misting every 20 min between 0800 and 1000 h. The field inoculation procedure was modified from Gilchrist et al. (1996). Ten spikes per plot were injected with a macroconidial spore suspension at anthesis as described earlier. Approximately 30 injections (3 replicates 10 spikes/ replicate) were made for all parents and F1s in each diallel. Injected spikes were covered with glassine bags at anthesis to guard against natural infection from wind-blown ascospores emanating from the nearby FHB screening nursery. The bags were removed to record spikelet severity at 21 d post injection and a new glassine bag was fastened over the spike where it remained until harvest. At harvest maturity individual injected spikes and auxiliary spikes within the hill plots that were not injected were harvested. The auxiliary spikes were exposed to the wind-blown ascospores and developed FHB symptoms. Control spikes were not injected but were covered with the glassine bags to investigate the effectiveness of the glassine bags at excluding wind-blown ascospores. Macroconidial Spore Suspension Twelve isolates of F. graminearum were obtained from scabby wheat seed by surface sterilization and plating onto acidified potato dextrose agar. Ten of these isolates were obtained from different geographical regions of Kentucky; one isolate was obtained from Indiana and one from Virginia. Macroconidial suspensions were prepared by placing two mycelial plugs from a culture of F. graminearum in 100 mL of carboxymethylcellulose (CMC) liquid media. Flasks were placed on a shaker (115 rpm) for 2 wk at 24 C. Spore suspensions were prepared by filtering the cultures through a 3.0-mm Millipore (Bedford, MA) filter. Macroconidia were resuspended in sterile water and streaked onto mung bean agar plates. The plates were incubated for 7 d, and then washed with sterile water. The washed suspension from each of the 12 isolates was then combined and calibrated to 600 000 spores/ mL with the aid of a hemocytometer. Deoxynivalenol Analysis After harvest, injected spikes from each hill plot were hand threshed in bulk. The auxiliary spikes within each hill plot were also harvested and threshed with a stationary thresher with the fan set at the lowest setting to minimize loss of severely shriveled “tombstone” kernels. All harvested samples were hand cleaned. DON analysis was completed only on diallel 2. A 5 g sample of grain from each F1 and parent was analyzed for DON with the EZ-Quant Vomitoxin Test Kit (Diagnostix Company, Mississauga, ON). Each sample was ground in a coffee grinder for 15 s. The coffee grinder was vacuumed between samples to prevent cross-contamination. Twenty-five milliliter of distilled water was added to each ground sample and the remainder of the test was completed following the test kit protocols. Two aliquots from each DON extraction were pulled to provide replication when analyzing the injected spikes that were threshed in bulk. Two field replications were sampled for the DON analysis on the auxiliary spikes. Greenhouse Screening The F1 progeny, reciprocals, and their parents were vernalized beginning on 17 Aug. 2000 and then transplanted into the greenhouse on 12 Oct. 2000. Ten seeds of each F1 cross and parent were planted in a completely random design. Spikes were inoculated by injecting a macroconidial spore suspension. As each wheat spike reached anthesis, a central floret was marked using a permanent marker. This marked floret was then injected by dispensing 3 L of the spore suspension from a digital microliter pipette. Approximately 10 inoculations (1 spike per plant) were made for all parents, F1 progeny, and reciprocals. After plants had been injected, they were moved into a mist humidity chamber for three consecutive nights. Plants were removed from the chamber on the fourth day and returned to the greenhouse bench where they were scored for FHB spikelet severity 21 d post injection. FHB spikelet severity was calculated as the number of FHB infected spikelets over the total number of spikelets. Seed Quality Evaluation Harvested grain from the auxiliary spikes was visually inspected and evaluated for percentage of Fusarium damaged kernels (FDK) or tombstones. Visual estimates of FDK are well correlated (r 0.92) with actual counts of FDK (DillMacky et al., 2001). Statistical Analysis The data from the individual spikes in both the greenhouse and field studies were averaged to give genotype means. These HALL & VAN SANFORD: DIALLEL ANALYSIS OF FUSARIUM HEAD BLIGHT RESISTANCE 1665 Table 1. Mean values of Fusarium head blight (FHB) severity from diallel 1, a diallel based on SRW wheat parental differences in FHB severity, as measured in the greenhouse (GH) and field (F), Lexington, KY, 2000 to 2001. Parent Parent Environment Freedom Patton Ernie 2555 Clark Patterson Foster CK 9663 25R18 Percent infection Freedom Patton Ernie 2555 Clark Patterson Foster CK 9663 25R18 GH F GH F GH F GH F GH F GH F GH F GH F GH F 7.8 † 11.6 7.5 9.9 20.1 8.4 5.7 11.2 6.7 11.3 † 10.4 34.0 7.1 8.9 6.8 7.3 8.7 13.4 5.5 7.0 † 11.9 17.9 11.8 27.9 25.2 37.6 14.5 11.2 25.6 6.3 13.7 † 10.6 46.8 25.6 31.8 53.4 53.5 25.3 9.2 14.5 24.9 9.7 5.7 † 24.4 42.3 58.4 24.8 46.3 44.4 76.9 43.0 41.1 37.2 29.4 8.7 † † 11.4 36.8 35.4 35.3 36.6 47.9 26.8 60.5 12.4 60.4 11.8 14.0 9.9 8.4 † 5.7 16.4 11.4 37.3 13.1 37.8 21.7 28.5 23.2 46.6 27.6 33.1 18.7 5.0 21.3 † 11.6 29.1 17.4 35.5 34.5 60.6 51.8 43.6 21.9 54.0 17.5 41.9 43.6 57.5 10.3 † † 13.9 † 6.4 † 4.9 † 11.7 † 11.9 † 5.6 † 6.3 † 5.2 † † Not tested because of insufficient F1 seed. genotype means were used in all diallel analyses. Analysis of variance was performed on all variables of interest by means of a completely random design in the greenhouse experiment and a randomized complete block design in the field experiment. Correlations of interest were estimated by SAS procedure CORR (SAS Institute, 1990). General (GCA) and specific combining abilities (SCA) were calculated for traits of interest using methods as described in Griffing (1956). In the greenhouse experiment in which reciprocals were planted, Griffing’s Method 1 was used. In the field experiment Griffing’s Method 4 was used. The parents used in the diallel crosses were regarded as a fixed set of parents with respect to the trait for which they were selected. For example, the parents in diallel 2 were specifically chosen based on their range in DON without any regard to severity of infection. Thus when analyzing diallel 2 for DON, we treated the parents as a fixed set. When analyzing severity, however, we treated the parents as a random set. As a rough estimate of the relative importance of additive effects, we used the variance component ratio 2 2/(2 2 2 2). According to g g s Baker (1978), this ratio can be used with either a fixed or random set of parents. effects for FHB severity were significant at P 0.05 (Table 3). GCA effects indicate that the cultivars Freedom, Patton, and 25R18 have good general combining ability for Type II resistance to FHB (Table 2). In fact, Freedom, Patton, and 25R18 produced superior hybrids that were more resistant than the parents themselves. Furthermore, the most resistant hybrids were progeny of one of these three cultivars. The cultivars Clark, CK 9663, and 2555 produced inferior hybrids as noted by their positive GCA effects. One observation of interest is that the cultivar Ernie did not impart any resistance to its progeny despite reports of its tolerance to FHB (McKendry, 1995). One hybrid (Ernie Clark) produced a significant ( 1 SE) positive SCA effect (data not shown). In this specific hybrid, a mean severity of 58.4% was much higher than the severity in the rest of Ernie’s progeny (Table 1). In two cases, the susceptible check Clark combined well with Freedom and 25R18 to produce hybrids with low mean severities (5.7 and 4.9%, respectively; Table 1). Table 2. Mean values and general combining ability (GCA) effects of Fusarium head blight (FHB) severity in nine soft red winter wheat parents from diallel 1, a diallel based on SRW wheat parental differences in Fusarium head blight (FHB) severity, as measured in the greenhouse and field, Lexington, KY, 2000 to 2001. Greenhouse Parent Freedom Ernie Patton Foster Clark Patterson CK 9663 2555 25R18 Overall Mean SE Severity 7.8 11.8 10.4 27.6 76.9 12.4 43.6 53.4 5.2 18.0 1.9 GCA Effect % 7.8 0.7 7.1 2.5 15.9 0.2 5.4 5.6 9.9 0 3.9 † 27.9 34.0 33.1 43.0 60.4 57.5 53.5 † 38.4 4.2 † 10.3 8.9 5.1 1.9 9.4 6.1 7.0 † 0 7.9 Severity Field GCA Effect RESULTS AND DISCUSSION Diallel 1 (Severity Diallel) Greenhouse Experiment Parents in this diallel ranged from the most susceptible ‘Clark’ (76.9%) to the most resistant 25R18 (5.2%) (Table 1). Overall mean severity in the greenhouse experiment was 18.0% for diallel 1 (Table 2). In this set of crosses, two data points were missing (‘Freedom’ ‘Patterson’ and Freedom 25R18). For these two data points, the cross means were simply repeated and substituted for the missing reciprocal means and then two degrees of freedom were dropped from the error term. Highly significant differences (P 0.01) existed among the genotypes for FHB severity (Table 3). Both GCA and SCA effects were also highly significant (P 0.01). The GCA mean square was nearly 10 times the magnitude of the SCA mean square. Reciprocal † Not tested because of insufficient F1 seed. 1666 CROP SCIENCE, VOL. 43, SEPTEMBER–OCTOBER 2003 Table 3. Mean squares and their significance in two diallels screened for Fusarium head blight resistance and deoxynivalenol levels in the greenhouse and field, Lexington, KY, 2000 to 2001. Greenhouse Severity Diallel 1 Rep Genotypes GCA Effects SCA Effects Reciprocal Effects Error 1 744.5** 10 396.2** 1 061.9** 504.4* 320.1 Diallel 2 1 828.21** 7 179.3** 2 190.52** 276.76 418.3 Diallel 1 4 847.1** 3 901.9** 8 877.1** 1 769.6** 372.6 Field Severity Diallel 2 4 544.5** 3 464.2** 4 055.5** 3 257.25** 354.1 DON Levels Diallel 2 3.5 11.0** 19.6** 8.0* 3.9 * Mean squares significant at the 0.05 probability level. ** Mean squares significant at the 0.01 probability level. In each case, the SCA effect was significant (data not shown). These hybrids were as resistant as the most resistant parent 25R18, which had a mean severity of 5.2%. The Freedom Clark cross demonstrates that FHB resistance in winter wheat can be attained without the use of parents such as 25R18 which carries resistance derived from Chinese sources. When crossed with Freedom, Clark may have some merit as an acceptable parent in breeding for FHB resistance. This is similar to the situation observed in the cross Sumai 3 by ‘Stoa’ in which the susceptible parent Stoa actually contributed some resistance alleles (Waldron et al., 1999). The resistant Chinese line Sumai 3 was itself the result of crossing two moderately susceptible cultivars (Bai and Shaner, 1994). The difficulty for the breeder is that characterization of potential parents cannot be made without some analysis based on crosses (like a diallel) or on molecular marker data as was the case in the work of Waldron et al. (1999). Reciprocal effects were significant in the overall analysis, albeit of a much smaller magnitude than GCA effects (Table 3). Some maternal effects for FHB resistance in winter wheat may exist, yet progress in breeding for FHB resistance should primarily focus on the choice of the two parents and not on the choice of the maternal parent. Field Experiment The mean FHB severity in the field was 38.4%, much higher than the greenhouse severity mean of 18.0% (Table 2). In general, the field environment provided much greater disease pressure than the greenhouse envi- ronment. White mycelium developed on the injected spikes in the field that were covered with glassine bags. FHB-infected wheat spikes within the same field under the same irrigation schedule but not covered with a glassine bag did not produce this mycelial growth, nor was such growth noted in the greenhouse experiments. Control spikes that were not injected but were covered with glassine bags also did not produce any mycelial growth or symptoms. The ANOVA revealed highly significant differences (P 0.01) among replications and genotypes. Both GCA and SCA effects were also highly significant (Table 3). A comparison of the greenhouse and field mean square errors (320.1 vs. 372.6) indicates that the two environments had similar error variances. On the basis of the GCA effects estimated in the field environment, Patton and Ernie were the best parents to use to reduce FHB severity (Table 2). Patterson was the only parent with a significant positive GCA effect. The SCA effects for FHB severity in the field were calculated, but none exceeded the standard error (data not shown). Diallel 2 (DON Diallel) Greenhouse Experiment The overall mean severity for diallel 2 (18.5%; Table 4) was very similar to the mean severity for diallel 1 in the greenhouse (18.0%; Table 2). Mean severity values for parents and progeny in this diallel are given in Table 5. Highly significant variation among genotypes was ob- Table 4. Mean values and general combining ability (GCA) effects of Fusarium head blight (FHB) severity and deoxynivalenol (DON) levels in nine soft red winter wheat parents from diallel 2, a diallel based on SRW wheat parental differences in DON levels, as measured in the greenhouse and field, Lexington, KY, 2000 to 2001. Greenhouse Parent 25R26 CK 9663 CK 9474 Roane KY89C-804-14-2 KY86C-127-3 Freedom Kaskaskia Patton Overall Mean SE Severity 45.2 69.5 20.4 7.2 84.3 33.5 7.4 19.9 22.5 18.5 2.3 GCA effect % 2.2 9.7 2.9 6.2 7.7 1.4 8.8 6.8 4.3 0 4.6 66.2 71.6 73.8 56.8 78.1 81.7 54.6 † 75.6 67.7 3.5 5.9 5.9 2.9 8.2 0.6 0.4 3.2 † 1.1 0 7.2 5.8 9.5 7.3 6.0 11.7 11.0 2.8 † 6.5 7.8 0.4 Severity GCA effect Field DON mg kg GCA effect 1 2.2 2.2 0.3 0.4 0.5 0.9 0.7 † 0.4 0 0.8 † Not tested because of insufficient F1 seed. HALL & VAN SANFORD: DIALLEL ANALYSIS OF FUSARIUM HEAD BLIGHT RESISTANCE 1667 Table 5. Mean values of Fusarium head blight (FHB) severity from diallel 2, a diallel based on SRW wheat parental differences in deoxynivalenol (DON) content, as measured in the greenhouse (GH) and field (F), Lexington, KY, 2000 to 2001. Parent Parent Environment 25R26 CK 9663 CK 9474 Roane KY89C-804-14-2 KY86C-127-3 Freedom Patton Kaskaskia Percent infection 25R26 CK 9663 CK 9474 Roane KY89C-804-14-2 KY86C-127-3 Freedom Patton Kaskaskia GH F GH F GH F GH F GH F GH F GH F GH F GH F 45.2 66.2 † † 24.6 18.9 18.0 9.7 7.9 16.1 17.8 66.2 69.5 71.6 28.2 49.4 31.7 23.7 17.4 19.3 10.7 15.7 56.6 16.6 71.2 20.4 73.8 29.8 18.6 15.7 6.2 8.0 5.2 20.9 46.1 58.0 70.3 22.6 54.4 7.2 56.8 42.8 25.8 5.3 18.7 22.1 5.4 66.1 28.9 78.8 6.8 86.7 32.2 66.9 84.3 78.1 18.5 19.5 28.8 5.0 9.9 75.9 21.0 69.9 29.5 66.5 40.2 50.3 25.4 47.1 33.5 81.7 11.9 16.5 13.1 5.3 54.2 9.6 81.4 14.2 81.6 19.9 66.3 17.6 67.6 12.9 77.2 7.4 54.6 5.5 4.2 10.5 68.8 26.4 67.4 9.5 69.9 12.8 65.6 14.5 59.4 14.1 84.7 5.8 60.2 22.5 75.6 13.3 5.9 † 9.7 † 14.4 † † † † † † † 12.6 † 16.3 † 19.9 † † Not tested because of insufficient F1 seed. served for FHB severity (Table 3). General combining ability effects and SCA effects were also significant (P 0.01). Reciprocal effects were not significant, in contrast to what was found in diallel 1. The magnitude of the GCA mean square to the SCA mean square (approximately a 8:1 ratio) indicates that additive effects were prevalent. Although this set of diallel parents may be regarded as a random set of lines with regard to severity, GCA, SCA, and reciprocal effects were calculated (Table 4). Freedom was identified as an excellent parent to use in reducing FHB severity. ‘Kaskaskia’ also had a significantly negative GCA effect. Surprisingly, ‘Roane’ had a significantly positive GCA effect. Roane is considered to be tolerant to FHB (Griffey et al., 2001) and could even be noted as resistant based on the mean FHB severity of only 7.2% observed in the greenhouse. However, Roane’s progeny ranged from 12.8 to 49.4% in the greenhouse (Table 5). Only two SCA effects exceeded the estimate of the standard error. The cross CK 9663 Roane had a significantly positive SCA effect. The cross KY86C-804-142 Kaskaskia had a significantly negative SCA effect (data not shown). Field Experiment—Severity Mean severity of parents and progeny are presented in Table 5. The overall mean severity in the field environment was much higher than the greenhouse (67.7 vs. 18.5%, respectively; Table 4). There were significant differences among genotypes for FHB severity. GCA effects and SCA effects were highly significant (P 0.01, Table 3). The GCA mean square and SCA mean square were of the same order of magnitude. This indicates that along with additive effects, dominance effects may also control some of the variation expressed in the field. Roane’s GCA effect for severity in the field was sig- nificantly negative while in the greenhouse screening the GCA effect was significantly positive (Table 4). One would expect a negative GCA effect for Roane based on its reputed tolerance to FHB (Griffey et al., 2001). This drastic switch from a positive GCA in the greenhouse screening to a negative GCA in the field screening is an unusual but not unique finding. Several other researchers have reported inconsistent results with Roane as screened in the Uniform Southern Fusarium Head Blight Nursery (J. Chen, personal communication). In this study, a low FHB severity was observed on Roane in the greenhouse, yet Roane’s progeny were not as resistant (in fact they were more susceptible than Roane). In the field, however, Roane imparted good Type II resistance to its progeny (Table 5). The reverse was true for Freedom. In the greenhouse (as in Diallel 1) Freedom showed good GCA for reduced severity of infection. In the field, however, Freedom’s positive GCA effect indicates that it was not an effective parent for reducing severity in its progeny. These two cultivars and their performance as parents illustrate clearly the difficulties in trying to resolve greenhouse and field data with one another. Furthermore, they demonstrate that no single parent or combination of parents will ensure success when breeding for FHB resistance in SRW wheat. Field Experiment—DON There were no significant differences among the genotypes for DON levels as sampled from the injected spikes (data not shown). The bagged heads were considered to be under extremely high disease pressure which made it impossible to perceive any differences in DON levels. Therefore, the auxiliary heads that were not covered with glassine bags or injected were tested for DON levels. The mean DON concentrations for each parent are presented in Table 6. Freedom resulted in the lowest DON concentration with a reported mean value of 1668 CROP SCIENCE, VOL. 43, SEPTEMBER–OCTOBER 2003 Table 6. Mean values of deoxynivalenol (DON) concentration in a field – grown wheat diallel, Lexington, KY, 2000 to 2001. Parents were chosen on the basis of differences in DON concentration (Diallel 2). Parent Parent 25R26 CK 9663 CK 9474 Roane KY89C-804-14-2 KY86C-127-3 Freedom Patton 25R26 CK 9663 5.8 CK 9474 4.9 9.5 Roane 5.7 10.9 7.3 KY89C-804-14-2 mg kg 1 KY86C-127-3 6.2 10.6 10.0 9.4 11.7 11.0 Freedom 7.9 8.6 8.1 8.3 7.4 3.7 2.8 Patton 7.5 12.0 6.1 9.7 6.4 5.6 5.4 6.5 3.0 8.9 7.2 6.0 2.8 g g 1 (2.8 ppm). Although this is the lowest value in this study, this and all other mean DON concentrations reported still exceed the USDA 1 g g 1 (1 ppm) standard for finished wheat products and most exceed the USDA standard for feed (http://usda.gov/gipsa; verified 8/19/02). Significant variation among the genotypes was detected. GCA effects were highly significant (P 0.01) and SCA effects were significant (P 0.05, Table 3). Based on GCA effects (Table 4), 25R26 and the breeding line KY86C-127-3 were identified as superior parents for reducing DON levels. DON levels less than 5 g g 1 (5 ppm) were observed on two hybrids produced from 25R26 and one hybrid from KY86C-127-3 (Table 6). Significant negative SCA effects for DON levels were observed on four hybrids: 25R26 CK 9663, 25R26 Roane, KY86C-127-3 Freedom, and Freedom Patton. The variance component ratio 2 2/(2 2 2 2) disg g s cussed by Baker (1978) offers some insight into the relative importance of additive and dominance genetic variance for a given set of parents or a defined population. A value of 1 indicates that all genetic variance is additive. What is interesting in this study is the difference in the magnitude of this ratio that was observed between diallels 1 and 2. In diallel 1 the ratio was 0.72 in the greenhouse, and 0.67 in the field, indicating that additive effects predominated in the expression of resistance in both environments. In diallel 2, in which the parents differed for DON concentration (but were random with respect to severity), the ratios were considerably lower: 0.08 in the field and 0.36 in the greenhouse. The variance component ratio was 0.48 for DON concentration in diallel 2. 0.36 (P 0.05). These low correlations do not support the selection of FHB resistant lines on the basis of DON data alone, nor do they agree with the literature. Bai et al. (2001) reported a higher correlation coefficient of 0.65 (P 0.01) between proportion of infected spikelets and DON in a greenhouse screening where only injected spikes were analyzed for DON. The correlation between FHB severity and DON levels found in this study could be lower than previously reported estimates on the basis of the fact that the parents studied here were chosen on the basis of range in DON levels and not FHB severity. The high mean severities in the field environment were therefore not too surprising. It was surprising to observe the relatively low DON levels in this material. It is important to note that these low DON levels were observed only when the auxiliary heads were tested. The injected heads were under so much disease pressure that it was impossible to perceive any differences in DON levels (data not shown). Severity data was not taken on the auxiliary heads; thus, we were only able to estimate the correlation between severity (as observed on the injected heads) and DON levels (as observed on the auxiliary heads). Progress toward breeding low DON producing lines may be possible, but this effort should not be restricted to selection based solely on low severity. For example, the cross 25R26 CK 9663 between two susceptible parents had a mean severity of 66.2% in the field, yet it was one of three genotypes to produce less than 5 g g 1 (5 ppm) DON in diallel 2. There was essentially no correlation between FDK and DON (r 0.02, NS). This low correlation may be an artifact of the relatively narrow distribution in FDK data. Greenhouse versus Field Screening Disease intensity was highest in the field. The overall mean FHB severity for the field experiment (53.0%) was almost three times the greenhouse mean severity of 18.3%. The microenvironment provided by the glassine bags in the field was obviously more favorable for pathogen growth and infection. The higher disease severity and greater consistency observed in the field made it a better screening environment than the greenhouse. One difference between the two environments was the number of spikes per genotype that were screened: 30 in the field vs. 10 in the greenhouse. To evaluate reciprocal Correlations between FHB Severity, DON Level, and FDK Most correlations were not significant (data not shown). In diallel 1, the correlation between field severity and FDK was significant (P 0.01) but only moderate (r 0.48). Bai et al. (2001) reported a similar correlation coefficient of 0.54 (P 0.01) between severity and FDK in a field screening environment. In diallel 2, the correlation coefficient between DON level and field FHB severity was r 0.39 (P 0.05) while the correlation between greenhouse severity and DON was r HALL & VAN SANFORD: DIALLEL ANALYSIS OF FUSARIUM HEAD BLIGHT RESISTANCE 1669 crosses in the greenhouse, however, we were limited to 10 observations per genotype. Correlation coefficients between the two environments for severity data were low at r 0.36 in diallel 1 and r 0.14 in diallel 2. To further explore the relationship between greenhouse and field environments a combined analysis of variance was completed on each diallel. The combined analysis revealed highly significant genotype by environment interactions within each diallel. To evaluate the nature of the genotype by environment interactions, variance components were estimated by equating mean squares to their expectations under a random effects model. Knowing these variance components the genetic correlation coefficient can be estimated using the formula 2 2 [( g1 (1 rg) g1 g2 (Robertson, 1960). ge g2) /2] The resulting genetic correlation coefficients were low in diallel 1 (rg 0.38) and negative in diallel 2 (rg 0.02). These low genetic correlation coefficients argue against the greenhouse as an indirection selection environment for developing resistance expressed in the field. Genotype environment interactions are comprised of two factors, interactions due to differences in scale and interactions due to changes in genotype rank. In this study, genotype rank change interactions outweighed the interactions due to differences in scale 100 fold. The performance of the progeny as tested in the greenhouse was of no value when predicting field performance. The two environments yielded totally different results. Several investigators have commented on the relationship between greenhouse and field screening data. These comparisons, however, have typically involved two disparate types of data: spikes in greenhouse-grown wheat are injected with conidia and are compared with spikes in field grown wheat that has been infected by ascospores and conidia from a grain spawn inoculum source. Bai et al. (2001) for example, evaluated 33 genotypes in both field and greenhouse in such a study. They reported a moderate correlation (r 0.55, P 0.01) in percent FHB severity between the greenhouse and field environments but other traits such as seed weight, seed grade, and DON levels were not correlated between the two environments. The present study is the only one (with which we are familiar) in which greenhouse and field screening methods were essentially identical. The significance of this fact is that it provides a more accurate estimate of genotype environment interaction than studies in which the screening method differs according to environment. Scab researchers must determine the association between the single spikelet injection method and the damage likely to be incurred in a farmer’s field under a natural epidemic. To answer this question in the present study, we looked at the correlation between spikelet severity in injected spikes and FDK in a sample of grain from auxiliary, noninjected spikes in the same hill plot. The strength of this design is that it allowed us to assess the value of the injection method in the field within the same experimental unit, thus eliminating noise caused by plot-to-plot environmental variation. Our assump- tion was that the noninjected spikes were subject to infection from ascospores and conidia emanating from the scabby corn used as the inoculum source in the adjacent nursery. This correlation coefficient ranged from 0.16 (NS) in diallel 2 to 0.48 (P 0.01) in diallel 1. An area of concern that was highlighted by our study was the lack of agreement between greenhouse and field results. While the greenhouse offers certain conveniences and provides an off-season screening environment, the critical evaluation must be done in the field. There is little support in the data that we report for selection of low DON and low FDK lines on the basis of greenhouse screening. What is less clear is the value of the injected spike method versus the grain spawn method in assessing resistance under field conditions. Additional studies are underway to answer that question. ACKNOWLEDGMENTS The authors thank Brenda Kennedy and Dennis TeKrony for their many suggestions and insights, and Carl Griffey for his helpful review of the manuscript. REFERENCES Arthur, J.C. 1891. Wheat scab. Indiana Agric. Exp. Stn. Bull. 36: 129–132. Bai, G., and G. Shaner. 1994. Scab of wheat: Prospects for control. Plant Dis. 78:760–766. Bai. G., F. L. Kolb, G. Shaner, and L. L. Domier. 1999. Amplified fragment length polymorphism markers linked to a major quantitative trait locus controlling scab resistance in wheat. Phytopathology 89:343–348. Bai, G., G. Shaner, and H. Ohm. 2000. Inheritance of resistance to Fusarium graminearum in wheat. Theor. Appl. Genet. 100:1–8. Bai, G., R. Plattner, A. Desjardins, and F. Kolb. 2001. Resistance to Fusarium head blight and deoxynivalenol accumulation in wheat. Plant Breed. 120:1–6. Baker, R.J. 1978. Issues in Diallel Analysis. Crop Sci. 18:533–536. Buerstmayr, H., M. Lemmens, S. Berlakovich, and P. Ruckenbauer. 1999. Combining ability of resistance to head blight caused by Fusarium culmorum in the F1 of a seven parent diallel of winter wheat. Euphytica 110:199–206. Buerstmayr, H., M. Lemmens, H. Grausgruber, and P. Ruckenbauer. 1996. Scab resistance of international wheat germplasm. Cereal Res. Commun. 24:195–202. Busch, R., D. McVey, G. Linkert, J. Anderson, J. Wiersma, R. DillMacky, and G. Hareland. 2001. Registration of ‘McVey’ wheat. Crop Sci. 41:926–927. Dill-Macky, R., R.W. Stack, and J.V. Wiersma. 2001. Comparison of two methods for estimating scabby kernels in Fusarium-infected spring wheat. 2001 National Fusarium Head Blight Forum. Erlanger, KY. 8–10 December 2001. Michigan State University. Gilchrist, L., S. Rajaram, A. Mujeeb-Kazi, M. van Ginkel, H. Vivar, and W. Pfeiffer. 1996. Fusarium Scab Screening Program at CIMMYT. p. 7–12. In H.J. Dubin et al. (ed.) Fusarium Head Scab: Global Status and Future Prospects. Proc. of CIMMYT Workshop on Fusarium Head Scab, El Batan, Mexico. 13–17 Oct. 1996. CIMMYT, Mexico. Griffey, C., T. Starling, A. Price, W. Sisson, M. Das, T. Pridgen, M. Vaughn, W. Rohrer, and D. Brann. 2001. Registration of ‘Roane’ wheat. Crop Sci. 41:1359. Griffing, B. 1956. Concept of general and specific combining ability in relation to diallel crossing systems. Aust. J. Biol. Sci. 9:463–493. McKendry. A. L., J. E. Berg, D. N. Tague, and K. D. Kephart. 1995. Registration of ‘Ernie’ soft red winter wheat. Crop Sci. 35:1513. Meronuck, R., and W. Xie. 2000. Mycotoxins in feed. Feedstuffs 72: 95–100. 1670 CROP SCIENCE, VOL. 43, SEPTEMBER–OCTOBER 2003 Mesterhazy, A. 1995. Types and components of resistance to Fusarium head blight of wheat. Plant Breed. 114:377–386. Mesterhazy, A., T. Bartok, C.G. Mirocha, and R. Komoroczy. 1999. Nature of wheat resistance to Fusarium head blight and the role of deoxynivalenol for breeding. Plant Breed. 118:97–110. Peraica, M., B. Radic, A. Lucic, and M. Pavlovic. 1999. Toxic effects of mycotoxins in humans. Bull. World Health Organ. 77:754–766. Robertson, A. 1960. Experimental designs on the measurement of heritabilities and genetic correlations, p. 101–106. In O. Kempthorne (ed.) Biometrical genetics. Pergammon Press, New York. SAS Institute. 1990. The SAS system for Macintosh. Release 6.12. SAS Inst., Cary, NC. Snijders, C.H.A. 1990. The inheritance of resistance to head blight caused by Fusarium culumorum in winter wheat. Euphytica 50: 11–18. Van Ginkel, M., W. Van Der Schaar, Y. Zhuping, and S. Rajaram. 1996. Inheritance of resistance to scab in two wheat cultivars from Brazil and China. Plant Dis. 80:863–867. Waldron, B.L., B. Moreno-Sevilla, J.A. Anderson, R.W. Stack, and R.C. Frohberg. 1999. RFLP mapping of QTL for Fusarium head blight resistance in wheat. Crop Sci. 39:805–811. Windels, C.E. 2000. Economic and social impacts of Fusarium head blight: Changing farms and rural communities in northern Great Plains. Phytopathology 90:17–21.