Inoculation methods under greenhouse conditions for evaluating soybean resistance to sudden death syndrome

The objectives of this work were to evaluate two greenhouse screening methods for sudden death syndrome (SDS) and to determine which one is best correlated with fi eld resistance of soybean genotypes. The evaluations were done with three sets of genotypes that were classifi ed as partially resistant, intermediate, and susceptible to SDS based on previous fi eld evaluations. These three sets were independently evaluated for greenhouse SDS reactions using cone and tray inoculation methods. Plants were infected using grains of white sorghum [Sorghum bicolor (L.) Moench] infested with Fusarium solani f. sp. glycines. Foliar symptom severity was rated 21 days after emergence. The cone and fi eld SDS ratings were signifi cantly correlated and ranged from 0.69 for set 1 to 0.51 for set 3. Correlations of SDS ratings of genotypes between fi eld and greenhouse tray ratings were signifi cant for set 1 and not signifi cant for set 2. The cone method showed the highest correlation with fi eld results and is recommended to screen soybean genotypes for SDS resistance.


Introduction
Sudden death syndrome (SDS) is a soybean disease caused by the soilborne fungus Fusarium solani f. sp.glycines (FSG) (syn.Fusarium virguliforme Akoi, O'Donnell, Homma and Lattanzi) (Roy, 1997).Severe losses due to SDS have been reported in soybeans grown in the USA (Hartman et al., 1995), Argentina, and Brazil (Rupe & Hartman, 1999).The fungus infects plants through the roots, and severely infected plants exhibit blackened and rotted taproots with few lateral roots (Stephens et al., 1993).Symptoms also include interveinal chlorosis and necrosis of leaves, vascular discoloration of the lower part of the stem, premature defoliation and pod abortion (Rupe, 1989).Recently, Aoki et al. (2005) reported that sudden death syndrome of soybean in North and South Americas is caused by phylogenetically and morphologically distinct species.In North America, SDS is caused by Fusarium virguliforme sp.nov., formally known as F. solani Pesq. agropec. bras., Brasília, v.43, n.11, p.1475-1482, nov. 2008 f. sp.glycines, and in South America, SDS is caused by four species of Fusarium: Fusarium brasiliense sp.nov., F. cuneirostrum sp.nov., F. tucumaniae, and F. virguliforme.
Screening for SDS reactions of soybean genotypes has been done under fi eld conditions (Hartwig et al., 1996;Schmidt et al., 1999;Farias Neto et al., 2006, 2007), and under greenhouse conditions (Hartman et al., 1997;Mueller, 2001;Njiti et al., 2001).Selection for SDS resistance in the fi eld is diffi cult, because the disease occurrence is unpredictable due to the sensitivity of symptom development to environmental factors (Njiti et al., 1996).In addition, the evaluation of a great number of lines in the fi eld is time-consuming and expensive.
Researchers have evaluated SDS resistance under greenhouse conditions by inoculating plants with FSG infested oat (Stephens et al., 1993) or sorghum seeds (Hartman et al., 1997;Mueller, 2001), cornmeal (Njiti et al., 2001) or toothpicks (Klingelfuss et al., 2002).Stephens et al. (1993) evaluated the SDS reactions of 12 soybean cultivars, infected with FSG infested oat seeds, in pots, in a greenhouse and in the fi eld, under natural and under inoculation conditions.The correlation between fi eld and greenhouse SDS ratings of cultivars ranged from 0.60 to 0.90.Njiti et al. (2001) tested soybean genotypes for SDS reactions in a greenhouse, using different rates of inoculum mixed to the growth medium.They found that an inoculum rate of 4,000 FSG spores cm -3 of plant growth medium resulted in the best correlation with the fi eld results.Hashmi et al. (2005) compared the SDS fi eld reactions of soybean genotypes with greenhouse reactions.The genotypes were infected in a greenhouse by layering inoculum in trays, mixing inoculum to the soil, in trays, and layering inoculum in tubes that were kept under precise soil temperature control with a water bath.They obtained the greatest correlation between fi eld and greenhouse results with the water bath system.
Although successes in predicting SDS fi eld reactions with greenhouse methods has been reported; there is a need to evaluate other greenhouse methods that could predict fi eld reactions more effi ciently.Inoculation process, generally, is time-consuming and diffi cult to be implemented if many genotypes are being evaluated.
The objective of this study was to compare two greenhouse inoculation methods to determine if either can effi ciently predict fi eld SDS reactions of soybean genotypes.

Materials and Methods
Three sets of genotypes were used in this study.Set 1 included 30 recombinant inbred lines (RILs), selected by Njiti et al. (2001) from a population of 100 lines derived from a cross of the SDS susceptible cultivar Essex with the SDS partially resistant cultivar Forrest (Hartwig & Epps, 1977).The RILs were selected based on previous evaluations of foliar symptoms in fi ve fi eld environments naturally infested with FSG as reported by Njiti et al. (2001) and Hnetkovsky et al. (1996).In that study, the fi eld plots were rated for disease incidence (DI) and disease severity (DS).Disease index (DX) values were calculated from these two scores according to Njiti et al. (1998).Disease incidence was taken as an estimate of the plants percentage in each plot with foliar symptoms.Foliar DS was recorded as: 1, 0-10% chlorosis or 1-5% necrosis; 2, 10-20% chlorosis, or for 6-10% necrosis; 3, 20-40% chlorosis, or for 10-20% necrosis; 4, 60% chlorosis, or for 20-40% necrosis; 5, rates greater than 60% chlorosis or greater than 40% necrosis; 6, defoliation 33%; 7, defoliation up to 66%; 8, defoliation greater than 66%; 9, premature death of the plant (Njiti et al., 2001).A disease index (DX; 0-100) was calculated as (DIxDS)/9.The ten most resistant lines from the population were placed in a partially resistant (PR) class, eight of these lines were signifi cantly more resistant than Forrest; the intermediate (IN) class was composed of the ten lines with resistance ratings closest to the population mean; and susceptible (S) class included the ten least resistant lines, all of them signifi cantly more susceptible than Essex (Njiti et al., 2001).
Set 2 included 24 soybean cultivars and lines with characterized SDS fi eld resistance.These genotypes were previously evaluated for resistance in at least three fi eld environments, and were rated for DI as described previously.Because not all cultivars and lines were evaluated in the same fi eld tests, due to maturity differences among lines, the DI scores were adjusted relatively to resistant and susceptible controls.The relative DI scores were employed to place the lines into partially resistant, intermediate and susceptible classes.Pesq. agropec. bras., Brasília, v.43, n.11, p.1475-1482, nov. 2008 Set 3 included 30 RILs selected from a set of 92, which were derived from the cross of the SDS partially resistant cultivar Ina (Nickell et al., 1999) with the SDS susceptible experimental line LN91-1695.The RILs were evaluated for SDS symptoms in a fi eld in Urbana, IL, during 2003and 2004(Farias Neto et al., 2006).
The FSG-1 isolate (Hartman et al., 1995), originated from Monticello, IL, was used to produce all inoculum used in this study, according to Huang & Hartman (1996), with modifi cations described by Farias Neto et al. (2006).Since 2000, the isolate was inoculated onto soybean and reisolated annually.The isolate was grown on white sorghum seeds, which were soaked in water overnight and autoclaved twice in 1-L fl asks.Each fl ask with 300 g of sorghum seeds was infested with 4 mm diameter plugs of fungal mycelium and, then, incubated for two weeks.The colony forming units (CFU) of the infested sorghum inoculum was determined, as previously described on hairy roots (Li et al., 2008), with modifi cations described by Farias Neto et al. (2006).Briefl y, 1 g of sorghum inoculum was soaked in a 250-mL Erlenmeyer fl ask containing 100 mL of sterile distilled water.The fl asks were shaken at 150 rpm, on an orbital shaker, for 30 min and, then, serially diluted 10 fold with sterile distilled water, for two times, resulting in a 100 fold dilution.From each dilution, 100 µL of inoculum was spread on an agar plate (100x15 mm) containing FSG semiselective medium (Huang & Hartman, 1996).Six plates were used for each inoculum dilution.The plates were incubated at room temperature (25±2ºC) for 10 days.Colonies of FSG were identifi ed as described previously (Li et al., 2000).The number of colonies on each plate was counted and used to calculate the number of colony-forming units per gram of sorghum.The experiment was run twice.The infested sorghum used as inoculum in the greenhouse experiments averaged 2.4x10 5 CFU g -1 .
In the cone method, all three sets of soybean genotypes were evaluated for SDS reactions in SC-10 type cones, containing a layer of FSG inoculum, as described by Farias Neto et al. (2007).The cones were fi lled with 100 mL of steam-treated soil mix (2:1 sand:soil) topped with 5 mL (3 g) of FSG infested white sorghum seeds.Twenty mL of soil mix were added to cover the infested sorghum seeds, and three soybean seeds were added to each cone, which were covered with another 20 mL of soil mix.After emergence, seedlings were thinned, and one seedling was left per cone.The soil was maintained near to water-holding capacity by fl ooding the cones twice daily.
Each cone was an experimental unit, and was arranged in a randomized complete blocks design.The sets were evaluated in separate experiments: sets 1 and 3 were each tested in two experiments.Set 2 was tested in a single fi ve-replication experiment.
For the tray method, the SDS reactions of plants from sets 1 and 2 were tested in 37x52 cm galvanized trays, according to Hartman et al. (1997), with modifications.The trays were filled with a steamtreated soil mix (2:1 sand:soil) to a depth of 4 cm.A template was used to make 7 furrows -36 cm long, 2 cm deep, and 7 cm apart -, and 10 mL of infested sorghum seed was evenly distributed in each furrow.Soil mix was added to cover the infested seeds to a depth of 2 cm.The template was reapplied to make a 2-cm deep furrow directly over the inoculum.Three 12-cm long experimental units were placed in each furrow resulting in each tray holding 21 experimental units, each one sown with five soybean seeds covered with soil mix to a depth of 2 cm.
The soil was maintained near to water-holding capacity by fl ooding the trays twice daily.The experimental units were arranged in a randomized complete blocks design, with four replicates for set 1 and three replicates for set 2. The plants were rated for SDS symptoms 21 days after germination, using the GDS scale previously described with score based on the mean of the plants per experimental unit.
For both cone and tray greenhouse experiments, noninfected controls were included.The greenhouse experiments were conducted in Urbana, IL, during the winter of 2002/2003, with a 12-hour photoperiod and air temperatures at 25±2ºC.Pesq. agropec. bras., Brasília, v.43, n.11, p.1475-1482, nov. 2008 The analysis of variance was computed for the greenhouse data using PROC MIXED (SAS Institute, 2000).All factors were considered fi xed except for blocks.Means were separated using least signifi cance differences (LSD) at 5% probability.Normality and homogeneity of data variances were verifi ed.Preplanned contrasts were computed between the treatments.The CORR PROCEDURE of SAS was used to calculate Pearson correlations between fi eld DX and DI, and greenhouse disease severity (GDS) data and between rankings of genotypes in the fi eld and greenhouse.

Results and Discussion
Typical SDS foliar symptoms for both cone and tray methods were observed on plants of all three sets grown in the greenhouse in soil with inoculation of infested sorghum.The noninoculated control plots showed no SDS foliar symptoms.
For the cone method, analysis of variance for GDS scores across experiments showed that there were signifi cant differences among the resistance classes defi ned based on fi eld results.Signifi cant differences were detected among genotypes within classes for set 2, but not for sets 1 and 3. Nonsignifi cant differences between experiments were detected for sets 1 and 3, each one grown in two separate experiments.
Contrasts between score means of the resistance classes (Table 1), defi ned by fi eld results, showed that the cone method was able to signifi cantly separate the means of PR and S classes and of PR and IN classes, for the three sets (Tables 1 and 2).However, for sets 2 and 3, the IN class did not differ signifi cantly from the S class.Set 2 was composed by cultivars and lines from different maturity groups, which were not evaluated in the same fi eld experiments; this may have introduced inconsistencies in the fi eld classifi cations of these genotypes.Besides, genotypes in this set are also from different backgrounds and may respond differently to inoculations under greenhouse conditions, compared to fi eld reactions to the disease.These factors could have made separation of IN and S classes diffi cult for this set.
Correlations between fi eld DX or DI scores and greenhouse cone GDS scores were signifi cant for the three sets (Table 3).The greatest correlation between Table 2. Estimates from contrasts between greenhouse disease severity score ratings obtained using the cone and tray inoculation methods for of soybean genotypes placed into different Fusarium solani f. sp.glycines resistance classes.
(1) Genotypes in each set were rated into resistance classes as partially resistant (PR), intermediate (IN) and susceptible (S), based on eld ratings.
Table 1.Sudden death syndrome (SDS) disease severity ratings of genotype resistance classes, grown in cones and trays infested with Fusarium solani f. sp.glycines, in greenhouse.
Table 3. Correlation coeffi cients between greenhouse disease severity (GDS) ratings and ranking, for cone and tray methods, and fi eld genotypic rankings (Njiti et al., 2001) and means of disease index (DX) and disease incidence (DI) ratings and rankings.
Resistance class (1)  Greenhouse disease severity Pesq.agropec.bras., Brasília, v.43, n.11, p.1475Brasília, v.43, n.11, p. -1482Brasília, v.43, n.11, p. , nov. 2008 greenhouse and fi eld scores was observed for set 1, while the lowest ones was observed for set 2. The correlation for set 3 genotypes was 0.68, when only the selected 30 lines were used in the analysis, and dropped to 0.51 when all 94 lines in the population were included.This lower correlation was expected, because the 30 selected lines were weighted towards those having the greatest and the least resistance.The rank correlation values were greater than the correlations of the scores.The rank correlations between the fi eld and GDS scores were 0.74 for set 1, 0.61 for set 2 and 0.65 for set 3. A rank correlation of 0.50 was detected for set 3, when all the 94 lines from the cross Ina by LN91-1695 were included in the analysis.
The good association between fi eld and greenhouse cone scores is illustrated by a comparison of scores for genotypes in set 1 (Table 4), set 2 (Table 5) and set 3 (Table 6).For example, of the 11 genotypes placed in the partially resistant class in set 2 based on fi eld DI, eight were rated as partially resistant based on the cone ratings.Of the seven genotypes placed in the susceptible class based on fi eld DI, fi ve were rated as susceptible, based on the cone ratings (Table 5).
For the tray method, analysis of variance of GDS scores revealed signifi cant differences an disease severity among fi eld based resistance classes for set 1, and nonsignifi cant effects for set 2. This method signifi cantly separated PR class from S class and IN from PR classes, for set 1 (Table 1).For set 2, the DS averages of the three classes were similar (Table 2), and there were no signifi cant differences among these classes.No signifi cant differences among genotypes Table 4. Means and rankings of soybean lines for fi eld sudden death syndrome (SDS) disease index (DX) and greenhouse disease severity (GDS) for cone and tray inoculation methods for set 1 genotypes. (1) From Njiti et al.. ( 2201). (2)Greenhouse disease severity ratings ranging from 1 (no symptoms) to 6 (severe symptom); adapted from Hartman et al. (1997).

Pesq
When cone and tray GDS scores were compared, the correlation between these methods was signifi cant for set 1 (0.48) but not for set 2. These correlations between greenhouse methods were lower than the ones between fi eld DI and GDS scores, for either method.
The greenhouse inoculation methods, especially the cone method, were successful in predicting fi eld SDS ratings.Because of the diffi culty in achieving consistent SDS symptoms in the fi eld, these greenhouse methods could prove useful for evaluating the level of SDS resistance of soybean genotypes.For example, breeding populations could be fi rst screened for resistance in the greenhouse, followed by the verifi cation of resistance ratings of selected lines in the fi eld.
The correlation between GDS scores obtained using the tray and cone methods was lower than the correlation of the GDS scores for either method with fi eld scores.It is surprising that GDS scores from the two greenhouse methods were not more highly correlated since for both methods, the same inoculum, greenhouse, planting depth, and watering regime were used.
Table 5. Means, ranks, and disease classes for sudden death syndrome (SDS) fi eld disease incidence (DI) and greenhouse disease severity (GDS) for cone and tray inoculations of set 2 soybean genotypes.
(1) Disease incidence percentage related to susceptible check, according to previous study. (2)Greenhouse disease severity ratings ranging from 1 (no symptoms) to 6 (severe symptom); adapted from Hartman et al. (1997). (3)Based on previous study of eld ratings. (4)Least signi cant difference, at 5% probability.Table 6.Means and rankings of set 3 soybean genotypes for fi eld sudden death syndrome (SDS) disease index (DX) and greenhouse disease severity (GDS) for the cone inoculation method.

Genotype
Field A greater correlation was observed between GDS scores and fi eld scores for the cones method than for the trays method.A reason for the lower correlation for the tray method may be that roots too frequently escaped infection on this method compared to the cone method.This is because a continuous layer of infested grain was placed in the cones, whereas for the tray method, infested grain was only placed under the seed.This allowed the roots to potentially grow around the inoculum in the tray method, thus escaping the disease.
Some lines partially resistant in the fi eld were susceptible with both greenhouse methods, showing that the inoculum overcame the resistance of these lines, as observed by Njiti et al. (2001).In addition, genotypes in set 2 were from different backgrounds and may respond differently to inoculations under greenhouse conditions compared to fi eld reactions to the disease.
The cone method requires more resources than the tray one for conducting resistance evaluations.This is because each cone has to be prepared separately and contains only one plant.In contrast, 21 experimental units, planted with fi ve seeds each, were grown in each tray.Because of the fewer plants in each experimental unit with the cone method, we used more replications with this method than for the tray one.
The associations obtained between the greenhouse cone method and fi eld scores were not as great as observed by Njiti et al. (2001) or Hashmi et al. (2005).Set 1 lines and fi eld data used in the present work were the same used by Njiti et al. (2001).They evaluated these lines in a greenhouse test in which plants were grown in pots with inoculum mixed into the growth medium with low (3.3 x 10 3 spores cm -3 of grow medium), moderate (5 x 10 3 ), and high (10 4 ) inoculum levels.The moderate inoculum level resulted in the greatest correlation with fi eld results and the R value from the regression was 0.60 which is greater than the R 2 from the cone test, which was 0.48.Hashmi et al. (2005) achieved a correlation of 0.81 between fi eld and greenhouse inoculations using the same set 2 genotypes and fi eld data used in this work.In cone method used in this work and in the method used by Hashmi et al. (2005), plants were grown in tubes and inoculated with a layer of inoculum that the roots needed to grow through.The main difference between these two methods is that in Hashmi et al. (2005) greater soil temperature control were obtained by the use of a water bath system whereas our soil temperature was regulated only by the air temperature in the greenhouse.This greater soil temperature regulation may be a major factor leading to the high correlation between fi eld and greenhouse results observed by Hashmi et al. (2005) Further research is needed to investigate the role of soil temperature on SDS development in the greenhouse.
The cone method used in this study has a number of advantages compared to other methods.Although the correlations with fi eld results were not as great compared to the methods described by Nijti et al. (2001) or Hashmi et al. (2005), the cone method is less complicated to set up than these other methods and does not require a water bath system.This makes it a good choice when researchers need a relatively simple system to rate the SDS resistance levels of genotypes in genetic mapping studies, breeding programs and cultivar testing.

Conclusions
1.The cone method showed the highest correlations with fi eld results and can be used to screen soybean genotypes for sudden death syndrome resistance.
2. The tray method is a good option for screening soybean germplasm for sudden death syndrome resistance, when many genotypes need to be tested.