Genetic Mapping of Loci underlying Field Resistance to Soybean Sudden Death Syndrome (SDS) N. Hnetkovsky, S. J. C. Chang, T. W. Doubler, P. T. Gibson, and D. A. Lightfoot* ABSTRACT Sudden death syndrome (SDS) is a fungal disease of soybean [Glycine max (L.) Merr.], caused by Fusarium solani (Mart.) Sacc. f. sp. phaseoli (Burk.) Snyd. & Hans., type A (FSA), that reduces crop yields in the USA. Quantitative partial resistance to SDS does exist; therefore, one method of controlling the disease is to select cultivars with genetic resistance. The objective of this study was to use molecular markers to identify and locate alleles of chromosomal segments associated with field resistance to SDS in adapted soybean genotypes. Seventy polymorphic DNA markers were compared with SDS response among 100 Fs:9 recombinant inbred lines derived from a cross between SDSresistant 'Forrest' and SDS-susceptible 'Essex'. SDS disease incidence (D1), disease severity (DS), and yield were determined in replicated, FSA-infested test sites during 4 yr encompassing five locations with recombinant inbred lines from the F5:7 to Fs-.n- Two separate chromosomal segments identified by two RAPD markers, OOOSjso and OCOltfo, were found to be associated with mean SDS response across five locations as well as within each of the five locations. These two quantitative trait loci (QTL) jointly accounted for 34% of total phenotypic variability in mean D1. OCOlaso was significantly associated with mean DS and yield and was putatively assigned to linkage group N. The beneficial allele was derived from the resistant parent Forrest. OO0525o was not significantly associated with mean DS or yield and was putatively assigned to linkage group C. The beneficial allele was derived from the susceptible parent Essex. Molecular markers can be used to define alleles of chromosomal segments conferring resistance to SDS in several environments and may allow efficient selection of resistant genotypes with good yield potential for FSA-infested fields.
S
OYBEAN sudden death syndrome is causing increasingly severe losses to soybean production (Mulrooney et al., 1988; Gibson et al., 1993; Yang and Rizvi, 1994). The cause of SDS is Fusarium solani (Mart.)
S.J.C. Chang, T.W. Doubler, P.T. Oibson, and D.A. Lightfoot, Dept, of Plant and Soil Science, Molecular Science Program, Southern Illinois Univ., Carbondale, IL 62091; and N. Hnetkovsky, Olney Central College, Olney IL 62450. Received March 19, 1995. Corresponding author (
[email protected]). Published in Crop Sci. 36:393-400 (1996).
Sacc. f. sp. phaseoli, type A (Rupe, 1989; Roy, 1989, O'Donnell and Gray, 1995). To date, host plant resistance has been the most effective control measure for SDS (Gibson et al., 1994). However, SDS resistance is difficult to determine since resistance is incomplete (partial), polygenie in nature, and can be environmentally sensitive. Classical genetic inheritance studies in the field estimated that SDS resistance (expressed as disease incidence or disease severity) is conditioned by a minimum of five loci in each of two progeny populations of adapted soybean (Matthews et al., 1991; Njiti et al., 1996). In a third population, a genetic study using a greenhouse assay indicated resistance to leaf symptom development, expressed as disease severity, may be conditioned by a single gene, rfs (Stephens et al., 1993). However, in the field rfs confers only partial SDS resistance (Gibson et al., 1994). Allelic relationships among resistance loci in the three separate progeny populations are unknown. All soybean genotypes tested show some SDS disease symptoms in infested fields during conditions conducive to disease expression (Gibson et al., 1994). Therefore, resistance to SDS in soybean germplasm is unlikely to show the classical race-cultivar-specific resistance manifest in the hypersensitive response. Rather, useful field resistance appears to be partial and polygenic (Matthews et al., 1991; Gibson et al., 1994; Njiti et al., 1996). Breeding for SDS resistance is complicated by the absence of a satisfactory quantitative greenhouse assay predictive of field performance, thereby necessitating extensive field selections with many lines. Factors such as rainfall, soil fertility, maturity date, date of planting, plant genotype, and cyst nematode (Heterodera glycines Ichinohe) infestation all significantly alter the degree of disease progression in the field (Rupe et al., 1991; Rupe Abbreviations: cM, centimorgan; D1, disease incidence; DS, disease severity; FSA, Fusarium solani f. sp. phaseoli, type A; kbp, kilobase pairs; QTL, quantitative trait loci; RFLP, restriction fragment length polymorphism; RAPD, random amplified polymorphic DNA; RIL, recombinant inbred line.
394
CROPSCIENCE, VOL. 36, MARCH-APRIL 1996
et al., 1994; Gibson et al., 1994). Therefore, reliable field evaluations require multiple replications in many infested sites with accurate disease progression data collected at 3- to 4-d intervals (Gibson et al., 1994; Njiti et al., 1996). Consequently, conventional breeding for SDSresistance is time consuming, difficult, and expensive. By means of DNAmarkers, molecular biotechnology makes it possible to identify chromosomalsegments containing agriculturally importantgenes as quantitative trait loci (QTL). DNAmarkers can expedite the identification of SDS-resistant soybean lines based on genotype instead ofphenotype, which should reduce the need for replicated field tests. Potentially, lines containing several resistance genes or combinations of SDS-resistance loci with different modesof action can be selected with molecular markers. In studying disease resistance in plants, the use of DNAmarkers has focused on genes involved in the race-cultivar specific hypersensitive response (complete resistance). Theseinclude single gene resistance to fungal pathogens in tomato (Lycopersicum lycopersicon L.) (Sarfati et al., 1989); lettuce (Lactuca sativa L.) (Michelmore et al., 1991); rice (Oryza sativa L.) (Yu al., 1991); and soybean (Diers et al., 1992). With DNAmarkers, cyst nematode resistance has been shownto be a race-cultivar specific partial resistance mediated by the combinedactivity of two to three loci in potato (Solanumspegazzinii L.) (Kreike et al., 1994) and soybean (Concibido et al., 1994; Webbet al., 1995). However, in some cases, resistance may be considered a specialized exampleof race-cultivar specific complete resistance mediated by a hypersensitive response (Endo et al., 1964; Webbet al., 1995). DNAmarkers have not been widely used in studying partial disease resistances which are based on interactions between several loci in plants. Those reported include 12
-- ( Essex = 58.9
10 ( Meanof Lines= 48.5 ) ( Forrest = 16.5
partial resistance to fungal pathogens in corn (Zea mays L.) (Bubeck et al., 1993), mungbean (Vigna radiata L.) (Young et al., 1994), rice (Wanget al., 1994), and barley (Hordeumvulgarae L.) (Backes et al., 1995), and partial resistance to bacterial pathogens in bean (Phaseolus vulgaris L.) (Nodari et al., 1993) and tomato (Danesh et al., 1994; Sandbrink et al., 1995). However, breeding for partial resistance to fungal pathogens is relatively commonbecause of the stable nature of the disease resistance (Eddington et al., 1994; Gingera et al., 1994) and because genes conferring partial disease resistance can afford cross protection to several fungal pathogens (Zhanget al., 1994). Cross protection is also manifest in systemic acquired resistance (Ward et al., 1991) and is mediated in part by genes conferring partial resistance to fungal pathogens (Alexander et al., 1993). Selection for partial resistance can be preferable to single gene resistance since partial resistance may be stable to genetically variable field populations of pathogens (Eddington et al., 1994), including F. solani f. sp. phaseoli, type A. DNAmarker studies indicate that somepartial disease resistance loci are environmentally insensitive in the field and maybe race insensitive (Bubeck et al. 1992; Wanget al., 1994). This paper demonstrates the utility of DNAgenetic markers for identifying chromosomalsegments contributing partial resistance to SDSin the field. The implications for the use of DNAmarkers for improved selection of SDSresistance in soybean are considered. MATERIALS
AND METHODS
Plant Material The Essex × Forrest (E × F) F5 derived population of soybean recombinantinbred lines (RILs) wasconstructedby cross-
ing Essex (Smith and Camper,1973) and Forrest (Hartwig and Epps, 1973). Essex is SDSsusceptible, while Forrest is SDSresistant (Fig. 1; Gibsonet al., 1994). About4500 plants wereinbred to the F5 generationby a single poddescent method(Brim, 1966). In 1988, a random bulk of seed was planted to obtain 500 F5 plants of which 150 were randomly selected, only intentionally excluding a few agronomically undesirable extremes. In 1989, 100 Fs-derivedlines of modal maturity(mid-maturitygroupV) with sufficient seeds for field testing wereretained. All inbreedingoccurredin fields with minimal incidence of soybean cyst nematode(SCN)and history of SDS. SDS Disease Scoring
5 15 25 35 45 55 65 75 85 95 10 20 30 40 50 60 70 80 90 100 Disease Incidence Fig. l. Frequencydistribution of the meanscores across five field locations for SDSdisease incidence among100 recombinantinbred lines froma cross betweencultivars Essex and Forrest. The mean DI score for each parent in the same environments is shown. The least significant difference between~parentand line meanswas16.5 (P< 0.05).
Eight field experimentswere conductedover a 4-yr period encompassing a total of eight locations in southernIllinois, of these useful data could be derived from five locations. The other three were excluded because of poor stand and absent or minimal SDSleaf symptomsresulting from drought. The five were Villa Ridge1990 (vg0), Cora 1991 (C91), Pulaski 1991 (Pgl), Cora 1992 (C92), and Ridgway1993 (R93) and C92were separate fields, 2 kmapart). All fields were selected basedon a history of visually uniformSDSinfestation and managedas described by Gibsonet al. (1994). The 100 E×FF5 derived lines were scored for DI, DS, and yield as described (Matthewset a1.,1991; Gibsonet al., 1994). Two-row plots in a partially balanced simple 11 × 11 lattice design(two replications, eight duplicationsper parent
HNETKOVSKY ET AL.: LOCI UNDERLYING FIELD RESISTANCE TO SOYBEAN SDS and five checks) were used (Gomezand Gomez,1984). Disease was rated weekly and the last score before and the first score after R6 (full pod) were used to standardize DI and DSto the R6 stage. DI was defined as the percentage of plants in the plot with visible leaf symptoms. DS was rated as the degree of leaf damageon diseased plants and was scored on a scale of 1 to 9 (1 = 0-10%/1-5%, 2 = 10-20%/6-10%, 3 20-40%/10-20%, 4 = 40-60%/20--40%, 5 = >60%/40% chlorosis/necrosis whereas 6 = 1-33%, 7 = 34-66%, 8 = 66-100% premature defoliation and 9 = premature death) (Gibson et al., 1994). Yield was determined at harvest from the two-row plots trimmed to 3 m (4.3 m planted). There was no intra-plot bordering since the progeny population displayed uniform growth habit and maturity dates. Lattice adjusted means were used whenever the lattice analysis was more efficient than the randomized complete block. To detect transgressive segregants, an LSD(P = 0.05) for across environment meansof individual lines vs. parents was calculated using the genotype × environment interaction as the error variance.
DNA Clones Bacterial strains containing cloned soybean PstI genomic DNA inserts were obtained from Dr. R. Shoemaker, USDAARS, Ames, IA (Shoemaker and Olson, 1993). Plant DNAExtraction, Restriction Digestion, Electrophoresis, and Southern Transfers Leaf and seed material were obtained from Dr. Paul Gibson and Dr. Michael Schmidt, Southern Illinois University at Carbondale. Leaf material for DNAextraction was collected from 20 F5:9 field grownplants per genotype in July 1992 (Agronomy Research Center, Southern Illinois University at Carbondale) and frozen (-70°C). DNAextraction from leaf tissue was according to Dellaporta et al. (1983) with modification for legumes(Guo et al., 1991). Aliquots of soybean DNAwere digested with one of five restriction enzymes: TaqI, DraI, EcoRI, EcoRV, or HindIII based on informative probe enzyme combinations. DNAfrag-~ ments were resolved by gel electrophoresis on a 10 g L agarose gel, then transferred and fixed to Hybond-N+ (Amersham Corporation, Arlington Heights, IL). The informative probe-enzyme combinations were obtained from Dr. P. Keim (N. Arizona University, Flagstaff, AZ), Dr. R. Shoemaker (USDA-ARS, Ames, IA), Dr. D. Webb (Pioneer Hybrid, Johnston, IA), or were generated in this study. In total, 243 RFLPprobes were tested for the ability to detect polymorphism between Essex and Forrest. Polymorphic RFLPloci were referred to according to the naming convention suggested by Michelmore et al. (1991). DNA Hybridizations Plasmid DNAwas digested with PstI and inserts were purified by gel elcctrophoresis and elution for use in radiolabeling reactions as described by Guoet al. (1991). Briefly, 50 to 100 ng of probe DNAwas labeled with 32P-dCTP by random hexamer primed synthesis. Hybridization was according to Guoet al. (1991) except for the following modifications: filters were hybridized at 65°C in the rotator incubator and polyethylene glycol 8000 at 5 % (w/v) was used in hybridization media in place of 5 % (w/v) dextran sulfate 500 000.
RAPDProtocol-Polymerase
395 Chain Reaction
The amplification reactions were performed after Williams et al. (1990) with 180 separate primers from kits A, B, D, E, F, G, H, and O from Operon Technologies, Inc. (Alameda, CA). The total volume of 25 ~tL contained 100 I~M dATP, dCTP, dGTP, and dTTP, 0.2 ~Mof a 10-base-pair primer, 10 to 40 ng genomic DNA,and either 1.0 unit Taq polymerase with 1.5 mMMgC12(Promega Corporation, Madison, WI; Pcrkin-Elmer Corporation, Norwalk, CT) or 2.5 units of the Stoffel fragment of Taq polymerase with 3 mM MgC12.DNAwas amplified in a Savant TL49 thermal cycler (Farmingdale, NY) programmed for 45 cycles of 1 min 94°C, 1 min at 36°C, and 2 min at 72°C. Amplification -~ products were analyzed by electrophoresis on a 14 g L agarose gel stained with ethidium bromide. Primers which generated distinct polymorphic bands between Essex and Forrest across a range of soybean DNAconcentrations (10-40 ng) were identified and used to amplify DNAsamples from the F5:9 lines to generate segregation data for mappinganalysis. RAPDmarkers amplified with Stoffel fragment reported here were OO05450and OO05250.All others were amplified by Taq polymerase. RAPDmarkers associated with SDS resistance were amplified independently on three or more separate occasions to assure reproducibility. Polymorphic RAPDloci were referred to according to the naming convention suggested by Michelmoreet al. (1991). Mapping Quantitative
Resistance
Loci
To discover genomicregions associated with SDSresistance the RILs were classified as Essex type or Forrest type (heterozygotes were excluded) for each marker and compared with SDS-disease response scores by analysis of variance (ANOVA). One-way ANOVAwas performed with SAS (SAS Institute Inc., Cary, NC). The probability of association of each marker with each trait was determined and a significant association was declared if P < 0.005 (unless noted otherwise in the text) to minimize the detection of false associations (Lander and Botstein, 1989). Mapmaker-EXP 3.0 (Landers al., 1987) was used to calculate map distances (cM) between linked markerswith the RIL (ri-self) genetic model. To identify intervals containing QTLgoverning SDSresponse, the marker map and disease data were simultaneously analyzed with Mapmaker/QTL1.1 (Paterson et al., 1988) with the F2-backcross model for trait segregation (Webbet al., 1995; M.J. Daly, Whitehead Inst., Cambridge, MA,1995, personal communica: tion). Putative QTLwere inferred when LOD(log,0 of odds ratio) scores exceeded 2.0 at somepoint in each interval. RESULTS
AND DISCUSSION
Polymorphism
and Linkage
A total of 243 selected DNAclones and 180 random RAPD primers were surveyed for polymorphism between Essex and Forrest. Only 58 RFLP probes and 32 RAPDprimers gave clear polymorphism. Since Essex and Forrest are more closely related than the parents of most mapping populations as judged by their co-variance coefficient (0.21) (Allen and Bhardwaj, 1987) and polymorphism index (0.29) (Keim et al., 1992; Skorupska et al., 1994), the low RFLP frequency (0.24) RAPDfrequency (0.18) was expected. Twenty of the 58 RFLP markers were selected as likely to represent separate loci (>10 cM apart) and likely to identify the
396
CROPSCIENCE, VOL. 36, MARCH-APRIL 1996
loci mappedin G. max × G. soja (Shoemaker and Olsen, 1993). These 20 RFLP markers were scored in the segregating population and and identified 22 loci. The 32 RAPDprimers were scored and identified 48 loci. In total, 70 loci were mapped, 57 mapped to eight coherent linkage groups encompassing 1065 cM. There were 13 unlinked markers which together with the linkage groups would allow detection of associations with SDS resistance QTLover about 1551 cM with mean intervals of’l 8.7 cM. This compareswith a recombination distance for the soybean genome of nearly 3000 cM within 25 linkage groups (Shoemaker and Olson, 1993). The F5:9 lines were heterogeneous and this was detected with RFLPmarkers. Heterogeneity is a product of the heterozygosity present in each F5 plant used for line derivation. Theoretically, the progeny line mean heterogeneity would be 6.25% of the genome. Mean heterogeneity was calculated from data for all co-dominant RFLP probes within all the RILs and was 8.04% which was not significantly different from the 6.25 %expected (P 0.05).
Analysis of AgronomicTraits The population meansfor trait data based on the combined progeny line means and the progeny line means in each of the five environments in which SDSresistance was measuredare shownin Table 1. The infection levels as reflected by environment mean disease incidence were high (49-64%) in all environments except C91 (16%). However,the individual line data in C91 were consistent with environments in which SDS leaf symptoms were more severe (data not shown). For disease severity, environment means ranged from 1.1 to 1.7 and were broadly consistent with the DI environment means. Mean yield in FSA-infested environments ranged from 2410 to 2810 kg ha-1 except in C91 (4190 kg ha-~). The higher yield in C91was partly a reflection of the inherent yield potential of this particular field and partly of the association of SDS leaf symptoms with reduced yield (Gibson et al., 1994). Frequency Distribution
of SDS Disease Incidence
MeanDI showed an approximately normal (P = 0.13), continuous distribution, although there is evidence for a peak below 20%DI and a significant kurtosis (-0.65) reflecting a flattened distribution (Fig. 1). Of 36 lines with a mean DI higher than Essex (Fig. 1) there were Table 1. Meansof disease incidence (D1), disease severity (DS), yield and days to R6 of 100 Fs derived lines from the cross Essex by Forrest tested in five locations. -t) Env.’~ DI (%) DS(1-9) Yield (kg ha Days to R6 V90 C91 Pgl C92 R93 Mean
53 (33)~ 16 (19) 57 (29) 64 (25) 52 (30) 49 (23)
1.7 (0.5) 1.1 (0.1) 1.4 (0.4) 1.5 (0.5) 1.6 (0.4) 1.5 (0.3)
2440 (310) 4190 (430) 2590 (330) 2420 (410) 2410 (360) 2810 (290)
104.4(1.4) 115.8(2.2) 105.5(1.5) 110.1(1.6) 105.5(1.9) 108.1(1.4)
V90, Villa Ridge 1990; C91, Cora 1991; P91, Pulaski 1991; C92, Cora 1992; R93, Ridgway1993. Standarddeviation amongprogenyline meansin each location andover locations is shownin parentheses.
16 statistically significant susceptible transgressive segregants (P < 0.05). Ten lines had a mean DI lower than Forrest, but none had a significantly lower DI than Forrest. However,eight of these lines were significantly transgressive when the LSDwas determined from the genotype × environment of the subset of the 10 lines numerically superior to Forrest and excluding data from Forrest. The smaller LSD0.05(4.8) involved in this comparison reflects the consistently low DI scores of these lines in all environments. The eight transgressive lines may be useful starting points for developing soybean cultivars with stronger SDSresistance. DNAMarkers Associated
with Disease
Incidence
To detect loci conditioning partial resistance to SDS we tested associations betweenF5:9 genotypic classes for each DNAmarker and the corresponding mean SDS DI across five locations. A significant association was declared if P < 0.005 and if that marker was linked to a second marker that was also associated with the trait with P < 0.01. This minimized the detection of false associations (Landers and Botstein, 1989) and enabled QTLanalysis. Twochromosomal regions had significant effects on DI (Table 2). The position of SDS-resistance loci within each region relative to the DNAmarkers was determined by QTLanalysis (Paterson et al., 1988) with linked DNAmarkers (Fig. 2). The first locus, identified RAPDmarker OO05250, accounted for 26% of total variation, with P = 0.001 (Table 2). This RAPDmarker mapped within 6 cM of RFLP marker K455D-1 (DraI bands, Essex = 5.5 kbp and Forrest = 3.0 kbp). The polymorphic bands were generated with the same restriction enzyme and were of similar size as K455D-1that mappedto the soybean linkage group C, locus 1 (Shoemaker and Olson, 1993). Therefore, the QTLwas tentatively assigned to linkage group C of the soybean RFLP map (Shoemaker and Olson, 1993). The K455D-1 marker was also strongly associated with DI (R2 = 0.22, P = 0.001) (Table 3). The chromosomal region was flanked by distal and proximal RAPDmarkers, with neither significantly associated with SDS response, probably because 000545o and OF05490 mapped 20 and 24 cM from 0005250 and consequently were not close to the QTLconditioning SDSresponse (Fig. 2). The second locus, identified by RAPDmarker OC01650,accounted for 30%of the total variation in DI, with P = 0.001 (Table 2). The RAPDmarker was Table 2. RAPDmarkers showing significant associations mean SDS disease incidence across five locations.
with
QTLMean DI % for RILs percentage with alleles from
Linkage DNA Marker group’t R2 P > F LODe;variation§ OO05280 OC016s0
C 0.26 N 0.30
0.001 0.001
2.78 1.80
18.1 15.6
Essex
Forrest
27.3 5:7.0 61.0 5:6.1 82.3 5:2.4 38.5 + 7.5
from Soybase, based on Shoemakerand Olson (1993). LODis the loglo of the oddsratio that supportsevidencefor the presence of the QTLat the locus from Mapmaker/QTL 1.1. the percent variation associated with the interval fromMapmaker/QTL 1.1.
397
HNETKOVSKY ET AL.: LOCI UNDERLYING FIELD RESISTANCETO SOYBEANSDS
SDS
QTL N
C END
END
SDS
K455D-1
2.45
0005
2.62
t SDS QTL-C
2.78
1.98
SDS
0004
0.48
OF04
1.33
OC01
2,05 1.80
A71T-1
0.48
--
I
SDS QTL-N
~10cM 1 Fig 2. Locations of RFLP markers, RAPDmarkers and two QTL conditioning SDS response. The QTLwere putatively assigned to linkage group C and linkage group N on the soybean genetic mapby anchored RFLP(Shoemaker and Olson, 1993). ENDindicates the likely position of the telomere on that linkage group, the disjunct bar indicates the rest of the linkage group. Marker namesare given on the left and marker-QTLLODscores are given on the right of each linkage group. LODscores at markers were from single-locus analyses of additive gene effects using MAPMAKER/QTL 1.1. Genetic distances (cM) were from the recombinant inbred line function of MAPMAKER/EXP 3.0. A cM distance scale is shown below group C. The estimated position of the QTLis shown based on 2 cM interval mapping using MAPMAKER/ QTL1.1. Boxes indicate the region with 2-LOD(100 fold) likelihood intervals.
mapped within 14 cM of the RFLP marker A071T-1 (TaqI bands, Essex = null, Forrest = 3.0 kbp). The Forrest band was polymorphic with the same restriction enzymeand was of similar size as the G. max allele of A071-1 that mapped to the soybean linkage group N, locus 1 (Shoemaker and Olson, 1993). Therefore, the QTLwas tentatively assigned to linkage group N of the soybean RFLP map (Shoemaker and Olson, 1993). The RFLPmarker was not significantly associated with DI. OC01650 was linked to two additional RAPDmarkers, Table 3. Markers associated each of five locations.
DNA Marker OC016s0
K455D-1
00052~o
with DI in single environments at MeanDI %for lines with allele from
Env/f
2R
P> F
Essex
Forrest
vg0 C91 P91 C92 R93 V90 C91 Pgl C92 R93 V90 C91 P91 C92 R93
0.25 0.43 0.23 0.27 0.20 0.15 0.23 0.12 0.08 0.22 0.22 0.14 0.25 0.24 0.29
0.003 0.0001 0.004 0.002 0.008 0.007 0.0006 0.02 0.05 0.0007 0.003 0.02 0.001 0.002 0.0005
90.0 + 4.0 56.0 + 7.0 88.3 + 4.6 92.9 + 2.7 84.0 _+ 4.1 42.9 _+ 6.7 8.3 + 2.7 49.1 + 5.5 60.5 + 4.9 40.8 + 6.4 27.8 _+ 9.5 9.4 _+ 4.8 33.1 + 7.2 39.5 + 8.0 26.4 _+ 8.8
39.6 _+ 7.9 14.0 + 4.1 45.0 + 7.2 51.0 _+ 6.3 43.0 + 7.4 67.9 + 5.9 27.7 + 4.5 68.5 + 1.2 73.4 _+ 4.2 70.3 +_ S.0 67.1 _+ 7.9 29.1 _ 5.7 69.5 5:7.1 72.1 5:5.7 67.1 + 6.4
V90, Villa Ridge 1990; C91, Cora 1991; P91, Pulaski 1991; C92, Cora 1992; R93, Ridgway1993.
OO04~o75and OF04~60o. Of these only OF04~60o was associated with DI (R2 = 0.19, P = 0.009). OF04~60o mapped within 14 cM of OC01650. A071T-1 was 20 cM and OO041075 28 cM from OC01650and consequently were not close to the QTLconditioning SDSresponse (Fig. 2). The MAPMAKER/QTL data suggest that together the two chromosomal regions may explain 33.6% of the total variation in DI. The heritability of SDSDI in Essex × Forrest has been estimated at 89% on a line mean basis so that additional chromosomal regions affecting SDSresistance remain to be discovered in this progeny population. An analysis with 70 molecular markers which encompassjust 1551 cMis unlikely to detect all SDS-resistance genes segregating in the progeny population. In addition, one hundredplant lines is a relatively small sample to detect all genes controlling SDSdisease response. Onlythose partial resistance loci with relatively large effects would be expected to be detected (Landers and Botstein, 1989). The alleles at the two mapped QTLwith beneficial effects on DI were derived from Essex at one locus (linkage group C) and Forrest at the secondlocus (linkage group N). Therefore, resistance in Forrest and other resistant varieties is apparently incomplete and can be improved, whereas, some SDS-sensitive varieties, such as Essex, will be useful sources of SDS-resistance alleles given marker assisted QTLdiscovery. Favorable alleles for partial resistance to fungal pathogensderived from the susceptible parent are commonlydetected with molecular
398
CROPSCIENCE, VOL. 36, MARCH-APRIL 1996
markers (Bubeck et al., 1992; Wanget al. 1994; Young et al., 1994). The extensive transgressive segregation observed here may result from recombination between SDS-resistance alleles from both Forrest and Essex in the progeny population. Associations
between Locations
FSAinfestation in the five locations was consistent and moderate to severe except in C91 where infestation was moderate (Table 1). The locus on linkage group identified by OC01650, was associated with DI (P 0.005) in four of the five locations tested (Table 3). the fifth location, R93, OC01650was weakly associated with DI (P = 0.008). The locus on linkage group identified by OO05250, was associated with DI (P 0.005) in four of the five locations tested (Table 3). the fifth location, C91 where DI was lowest, 0005250 was weakly associated with DI (P = 0.02) whereas the linked marker K455D-1was strongly associated with DI in both C91 and R93 (P < 0.0007) but was weakly associated with DI in the other three environments (P 0.005). With both OO05250and K455D-1, the presence of the QTLcould be detected in all locations tested. The stability of resistance conferred by the two chromosomalregions was clear. No evidence for consistent location-specific resistance was apparent in this genetic material, so no evidence for either pathologically distinct strains of F. solani in the locations tested or race-specific cultivar resistance genes were apparent from this study. In addition the Essex x Forrest progeny behave as expected in SDS-infected fields in Arkansas (C. Sneller, University of Arkansas, Fayetteville, AR, 1994, personal communication). Therefore, these markers will be useful for selecting stable SDS-resistant genotypes throughout the FSA-infested soybean production area. The environmental stability of both of these loci contrasts with many loci conferring partial resistance to fungal pathogens of corn and rice (Bubeck et al., 1992; Wanget al., 1994). DNA Markers Associated
with Disease
Severity
As judged by the ANOVA,the chromosomal region identified by marker OC01650was associated with DS (R2 = 0.28, P = 0.001). In contrast the chromosomal region identified by 0005250did not qualify as significantly associated with DS (R2 = 0.18, P = 0.007). Neither chromosomalregion was significantly associated (LOD < 2.0) with mean DS by Mapmaker/QTL. HowTable 4. A Markerassociated with yield in SDS infested
DNA Marker OCOl~so
fields.
Meanyield (kg ha-1) for lines with alleles from Env.~"
2R
P >F
V90 C91 P91 C92 R93 Mean
0.31 0.05 0.10 0.34 0.31 0.27
0.0008 0.2 0.07 0.0004 0.0007 0.002
Essex 2547 + 4891 + 2943 + 2291 + 2399 + 3016 +
128 222 168 114 141 155
Forrest 3050 _+ 60 5248 _+ 128 3286 + 87 3211 +_ 114 3144 + 94 3588 + 101
V90, Villa Ridge 1990; C91, Cora 1991; P91, Pulaski 1991; C92, Cora 1992; R93, Ridgway1993.
ever, both QTLconditioning SDSDI may also condition DSsince the greater significance of the associations with DI may reflect the higher heritability and lower environmental interaction of DI compared to DS (Matthews et al., 1991; Gibson et al., 1994; Njiti et al., 1996). In ANOVA, the stability of marker associations between locations with DSwas similar to those with DI, therefore DI was a useful indicator of effects of these two loci on DS. None of the 70 markers tested identified loci strongly associated with DSthat were not associated with DI. Therefore, DI was a useful indicator of the effects of all loci tested on DS. Interactions between Maturity Date and SDS Response SDSleaf symptomsoften do not develop until several days after flowering and in general late maturing lines tend to be more resistant than early lines (Gibson et al., 1994). Essex and Forrest are very similar in maturity date and the progeny population also shows a restricted variability in maturity date (Table 1). There is about 10d variation in date of the R6 stage (R6 date), the critical point in SDSrating. Weinvestigated the location of QTLconditioning the mean R6 date averaged across five locations. Two unlinked markers, OF20~o25and OC01570,were strongly associated (P < 0.002) with mean R6 date, each accounting for 27%of the variability. However, as measured by Mapmaker/QTL,neither chromosomalregion described above contained a significant QTLfor mean R6 date in FSA-infested fields (LOD< 2). However, OF20~025and OC01570were both significantly associated with the R6 date (P < 0.005) in three of the five locations (C91, P91, and C92, data not shown). Neither was associated with SDSresponse nor linked to markers encompassing QTLconditioning SDS response. Therefore QTLconditioning the R6 maturity date and SDS DI did not correspond, which contrasts with the correspondence between QTLfor partial resistance to gray leaf spot (Cercospora zea-maydis Tehon & E.Y. Daniels) and maturity date in corn (Bubecket al., 1992). DNA Markers Associated
with Yield
In ANOVA,the chromosomal region identified by OC01650on linkage group N was significantly associated with mean yield in FSA-infested test sites, accounting for 27%of variation (P = 0.002, Table 4). The association was also significant in three of five locations (P 0.0008, Table 4). Therefore this chromosomal region may condition both SDS DI and yield, as a result of variation in DI, in FSA-infested fields. The chromosomal region identified by 0005250 on linkage group C was not significantly associated with mean yield (P = 0.12) or yield in single locations (P > 0.01). However, measured by Mapmaker/QTL,neither chromosomal region described above contained a significant QTLfor mean yield in FSA-infested fields (LOD< 2). Significant but weakcorrelation coefficients (r < 0.6) between yield and DI or DShave been reported (Gibson et al., 1994). Therefore, some loci which reduce the development of leaf SDS symptoms (DI and DS) may
HNETKOVSKY ET AL.: LOCI UNDERLYING FIELD RESISTANCE TO SOYBEAN SDS
not substantially improve yield potential. Such loci might reduce the symptoms of infection (leaf chlorosis) without alleviating the causes of yield reduction such as root rot, xylem plugging, and flower and pod abortion. The effects of individual partial resistance loci on yield in multiple environments have not previously been reported. We have observed that one locus associated with SDS resistance confers significantly improved yield potential, while a second does not. That differences are observed between the two loci reported, emphasizes the power of the molecular marker technique in dissecting the components of poly genie disease resistance. Implications for Genetic Dissection and Marker Assisted Breeding The markers K455D-1, OO0525o, OC0165o, and OF04i6oo can be used for marker assisted selection for partial SDS resistance. This will accelerate the breeding of SDS resistant soybean lines and may reduce the need for field tests of SDS response, which are slow and expensive. These loci are currently being used to determine allelic relationships among different sources of resistance, to assist SDS resistance gene introgression into susceptible germplasm, and to enable resistance gene pyramiding. However, since the currently identified favorable alleles are derived from Essex and Forrest their unequivocal identification and co-selection will be possible only in restricted instances, such as when crossing with Essex, Forrest, or one of their progeny lines. These loci may have several alleles, some of which may confer greater or lesser SDS resistance than the alleles carried by Essex and Forrest. Therefore, defining allelic associations with resistance at these loci in a wider selection of soybean germplasm is important. In the future, through genetic dissection, the effects of single resistance loci will be determined. F5:n lines that were derived from ¥5 plants heterozygous in the region of an SDS resistance QTL are heterogeneous for the region, while individual plants within a line are largely homozygous within the region by the FS : U. Lines heterogeneous for each resistance locus have been identified and sub-line populations extracted from forty individual plants within a line. With these sub-line populations the genetic and physiological effects of the segregation of an individual resistance locus will be isolated, the region saturated with markers and the QTL localized further. These studies will provide a basis for map based isolation of a gene conferring partial resistance to SDS. ACKNOWLEDGMENTS We thank Dr. M.E. Schmidt for developing the Exp progeny population. We thank Dr. R. Shoemaker for the RFLP probes and some parent RFLP polymorphism data. We thank Dr. P. Keim for much parent RFLP polymorphism data. We thank Dr. D. Webb for some novel parent RFLP data and verification of the RFLP polymorphism data. We thank all the workers on the SDS field team at Southern Illinois University at Carbondale from 1987 to present. Thanks to Dr. V. Njiti and Dr. O. Myers, Jr. for critical reading of the manuscript. This work was supported in part by grants from the Illinois
399
Soybean Program Operating Board Nos. 92-18-125-3, 93-19132-3, 94-20-143-3.
400
CROP SCIENCE, VOL. 36, MARCH-APRIL 1996
