Quantitative Disease Resistance Loci towards Phytophthora sojae and Three Species of Pythium in Six Soybean Nested Association Mapping Populations

Comparison of quantitative disease resistance loci (QDRL) towards the diverse array of soilborne pathogens that affect soybean [Glycine max (L.) Merr.] is key to the incorporation of resistance in breeding programs. The water molds Phytophthora sojae (Kauffman & Gerdmann), Pythium irregulare (Buisman), Pythium ultimum var. ultimum (Trow), and Pythium ultimum var. sporangiiferum (Drechsler) contribute to soybean yield losses annually. Six Soybean Nested Association Mapping (SoyNAM) populations were evaluated for resistance to one or more of these pathogens. Four were screened with a tray test to measure lesion length after inoculation with Ph. sojae; cup assays were used to screen three, three, and two populations for resistance towards Py. irregulare, Py. ultimum var. ultimum, and Py. ultimum var. sporangiiferum, respectively. There were two to eight major or minor QDRL identified within each SoyNAM population towards one or more of these water molds for a total of 33 QDRL. The SoyNAM populations evaluated for resistance to two or more water molds had different QDRL towards each pathogen, indicating that within a source of resistance, mechanisms are potentially specific to the pathogen. Only 3 of the 33 QDRL were associated with resistance to more than one pathogen. There was a major QDRL on chromosome 3 associated with resistance to Py. ultimum var. ultimum and Py. ultimum var. sporangiiferum, and QDRL on chromosomes 13 and 17 shared a flanking marker for both Py. irregulare and Py. ultimum var. ultimum. The SoyNAM population can serve as a diverse resource to map QDRL and compare mechanisms across pathogens and isolates. K. Scott, D. Veney, and A.E. Dorrance, Dep. of Plant Pathology, The Ohio State Univ., Ohio Agricultural Research and Development Center, 1680 Madison Ave., Wooster, OH 44691; C. Balk, Dep. of Plant Pathology, The Ohio State Univ., Ohio Agricultural Research and Development Center, 1680 Madison Ave., Wooster, OH 44691, current address, The Davey Tree Expert Company, 1500 N Mantura St., Kent, OH 44240; L.K. McHale, Dep. of Horticulture and Crop Science, The Ohio State Univ., 2021 Coffey Rd., Columbus, OH 43210; L.K. McHale and A.E. Dorrance, Center for Applied Plant Sciences and Center for Soybean Research, The Ohio State Univ., Columbus, OH 43210. Received 20 Sept. 2018. Accepted 28 Nov. 2018. *Corresponding author (dorrance.1@osu.edu). Assigned to Associate Editor Candice Hirsch. Abbreviations: BLUP, best linear unbiased predictor; LOD, logarithm of the odds; NAM, nested association mapping; QDRL, quantitative disease resistance locus/loci; QTL, quantitative trait locus/loci; RCBD, randomized complete block design; RIBD, randomized incomplete block design; RIL, recombinant inbred line; SoyNAM, Soybean Nested Association Mapping. Published in Crop Sci. 59:605–623 (2019). doi: 10.2135/cropsci2018.09.0573 © Crop Science Society of America | 5585 Guilford Rd., Madison, WI 53711 USA This is an open access article distributed under the CC BY license (https:// creativecommons.org/licenses/by/4.0/). Published February 7, 2019

Phytophthora sojae populations have continued to adapt to the Rps genes that have been deployed in cultivars across the north-central region (Nelson et al., 2008;Dorrance et al., 2016), and consequently there is now a greater focus on using partial resistance, which is conferred by QDRL (Dorrance et al., 2003(Dorrance et al., , 2009;;Wang et al., 2010Wang et al., , 2012aWang et al., , 2012b;;Stasko et al., 2016).In addition to the consistent presence of P. sojae in production fields in Ohio (Dorrance et al., 2003;2016), surveys also identified >35 different species of Pythium associated with seed and seedling rot of soybean (Dorrance et al., 2004;Broders et al., 2007Broders et al., , 2009;;Ellis et al., 2013;Dorrance Laboratory, unpublished data, 2015-2018).Pythium irregulare Buisman, Pythium ultimum var.ultimum Trow, and Pythium ultimum var.sporangiiferum Drechsler are among the most commonly found and most aggressive towards soybean (Broders et al., 2007(Broders et al., , 2009;;Balk, 2014;Eyre, 2016).Additionally, all three of these Pythium species were found on soybean throughout the north-central region (Jiang et al., 2012;Radmer et al., 2017;Rojas et al., 2017a), as well as Arkansas (Kirkpatrick et al., 2006a).The environmental conditions and mode of initial infection are very similar for all of these water molds, although P. sojae is a well-known hemibiotroph and species of Pythium are necrotrophs (Lévesque et al., 2010;Schroeder et al., 2013).Thus, the key questions for breeders are: can all of these different species of soilborne pathogens be managed with resistance, and will any of the QDRL be effective against multiple pathogen species to aid in cultivar development?
The Soybean Nested Association Mapping (SoyNAM) Project developed and made available to the soybean research community a large nested association mapping (NAM) population composed of 40 RIL populations derived from crosses between one common, high-yielding hub parent (IA3023) and exotic germplasm, plant introductions, and high-yielding breeding lines with the goal of identifying QTL associated with yield and other desirable traits (Grant et al., 2009;Diers, 2014;Xavier et al., 2017aXavier et al., , 2017bXavier et al., , 2018;;Diers et al., 2018;www.soybase.org/SoyNAM).The parents and the 40 related F 5 RIL populations were previously genotyped with the SoySNP6K Illumina Infinium BeadChip genotyping array and the linkage maps developed for each NAM RIL population (Song et al., 2017), which are publicly available at SoyBase (http://soybase.org,accessed April 2017).
Previous studies identified pathogen resistance in the donor parent lines and hub parent IA3023 used to develop these NAM populations; 14 of the 40 donor parents were resistant to P. sojae race 1 (vir 7), indicating the presence of Rps genes, seven had high levels of partial resistance to P. sojae, and 15, 12, and 10 parents had moderate levels of resistance to Py. irregulare, Py. ultimum var. ultimum, and Py. ultimum var. sporangiiferum, respectively (Wickramasinghe et al., 2012;Balk, 2014).Thus, the objective of this research was to identify and compare QDRL for resistance to P. sojae, Py. irregulare, Py. ultimum var.ultimum, and Py.ultimum var.sporangiiferum in six of the individual NAM RIL populations where there was resistance to more than one of these pathogens.Since resistance to each pathogen was evaluated in only one or four of the RIL populations within the full SoyNAM population, each of these was treated as an individual RIL population.

Plant Materials
From among the 40 available SoyNAM RIL populations, we selected six for which the parents differed significantly for a resistance response to more than one of the oomycete pathogens targeted in this study.Seed for the 140 RILs for each from one P. sojae isolate was placed over the wound.The trays were placed in a 26.5-L bucket and moved to a growth chamber at 25°C with 60% relative humidity and a 14-/10-h light/dark cycle.For each plant, the lesion lengths were measured from the inoculation site to the edge of the lesion margin 7 d after inoculation.There was a total of two replicate trays over time for each RIL (20 plants from each RIL total).Each of the buckets contained 24 trays, and within each replication, there were three sets of check lines and the parents of the RIL population.The check lines were cultivars with different levels of resistance to P. sojae and included moderately resistant cultivars Conrad (Fehr et al., 1989) and Resnik (McBlain et al., 1990), moderately susceptible cultivars Sloan (Bahrenfus and Fehr, 1980) and Williams (Bernard and Lindahl, 1972), and highly susceptible genotype OX20-8 (Mideros et al., 2007).
The data for the mean lesion length of the 10 seedlings from one tray for each RIL in each replicate was used in the analysis.The best linear unbiased predictor (BLUP) of each RIL was calculated with PROC MIXED in SAS 9.1 (SAS Institute, 2003) (Stroup, 1989).The model was where m is the overall mean, R i is the effect of the ith replication, B(R) ij is the effect of the jth bucket in the ith replication, C k is the effect of the kth class of entry (checks: Conrad, OX20, Resnik, Sloan, Williams, parents, and RILs), G(C) kl is the effect of lth genotype within class for recombinant inbred lines only (genotypic variance, s 2 G ), and e ijkl is the experimental error (s 2 ).Class of entry was treated as a fixed effect, and all other terms were considered random effects.The model of this analysis allowed checks and parents to be considered fixed effects and the RILs considered as random effects.
Resistance to three species of Pythium was evaluated in separate experiments (Table 1).Three NAM RIL populations were evaluated for resistance to Py. irregulare isolate Br 2-3-5 and two towards Py.ultimum var.sporangiiferum isolate Will 1-6-7.Three populations were evaluated for resistance to Py. ultimum var.ultimum with one or two isolates; the RILs from NAM HS6-3976 and S06-13640 were evaluated with Py. ultimum var.ultimum isolate Miami 1-3-7 in the first experiment, then with Py. ultimum var.ultimum isolate N201.2(2) in the second, whereas population 4J105-3-4 was evaluated only with isolate Miami 1-3-7.Each of these isolates was used in preliminary studies to evaluate the parents for resistance.Each of these populations was provided by Brian Diers (University of Illinois) and used for seed increase.For each of the six families, 100 seeds from each RIL was planted in separate plots for seed increase at the Ohio Agricultural Research and Development Center, Wooster, OH, on 25 to 26 May 2016.Each RIL was bulk hand-harvested and individually threshed to obtain fresh seed to be used in greenhouse phenotyping experiments in the following year.

Phenotypic Data
The six donor parents were previously identified as having resistance based on greenhouse and laboratory experiments to one or more of the following pathogens: P. sojae, Py. irregulare, Py. ultimum var.ultimum, and Py.ultimum var.sporangiiferum (Wickramasinghe et al., 2012;Balk, 2014).Four NAM populations were evaluated for quantitative resistance to P. sojae with a tray test.Two of the four donor parents have Rps genes, which would mask the quantitative resistance, Rps1k and Rps3a in HS6-3976 and Rps1a in LG00-3372 (Wickramasinghe et al., 2012).Thus, two isolates of P. sojae, with complex pathotypes, collected in 2010 in Ohio, were selected to be used in the tray assay.The NAM populations HS6-3976 and LD02-9050 were inoculated with P. sojae isolate Win371 (vir 1a,1b,1k,2,3a,7), and LG05-4832 and LG00-3372 were inoculated with Day739 (vir 1a,1b,1c,1k,3a,3c,8).The tray test was performed as previously described (Tucker et al., 2010;Wang et al., 2010Wang et al., , 2012a;;Stasko et al., 2016).Briefly, 10 7-d-old seedlings from each RIL were placed on a cotton wicking pad on a plastic tray, and a 10-mm scratch was made on the tap root with a scalpel 20 mm below the crown.An agar-mycelial slurry made from a 7-d old culture population was screened separately using a greenhouse cup assay that was previously described (Ellis et al., 2013;Stasko et al., 2016).Briefly, the isolates were grown on potato carrot agar at 20°C for 3 d.Eight 10-mm agar plugs were placed into a sterilized Spawn bag (Myco Supply) containing 950 mL play sand (Quikrete), 50 mL corn (Zea mays L.) meal (Quaker Oats Company), and 250 mL deionized water.These bags were closed with an electrical-impulse sealer (Harbor Freight Tools) and placed in a 20°C incubator for 10 d while shaking every other day to ensure even growth of the pathogen.The sandcornmeal mixture was then mixed with fine vermiculite in a 1:4 ratio and placed into 600-mL polystyrene cups with drainage holes.The cups were watered three times over 24 h with deionized water to ensure a suitable environment for the pathogen.Eight healthy seeds of each RIL were placed on the inoculum, and the cups were arranged in a randomized complete block design (RCBD) with three replications per experiment.To verify that the experiment was working as intended and the pathogen being screened was having an impact on the root morphology, at least one noninoculated cup per RIL-pathogen combination was included in the screening of each population.The data from the noninoculated control cups were not used in the root phenotype calculations and instead acted as a check to verify the quality of the seed and success of each experiment.The cups were watered twice per day to maintain a conducive environment for root disease.There was a minimum of two experiments for each population and pathogen combination.
Experiments to evaluate resistance towards Py.ultimum var.ultimum and Py.irregulare were done in a greenhouse at temperatures ranging between 18 and 23°C, whereas those for Py.ultimum var.sporangiiferum were performed in a growth chamber set at 20°C and 60% relative humidity with a 16-/8-h light/dark cycle.Within each replication in each experiment, the following cultivars with differing levels of resistance to the three Pythium spp.were included as check lines: Clermont (Ohio Agricultural Research and Development Center, The Ohio State University), Dennison (St. Martin et al., 2008), Kottman (St. Martin et al., 2001), Lorain (Ohio Agricultural Research and Development Center, The Ohio State University), Sloan (Bahrenfus and Fehr, 1980), and Williams 82 (Bernard and Cremeens, 1988).The cultivar Clermont was susceptible to Py. ultimum var.ultimum, whereas Dennison, Kottman, and Sloan were all moderately susceptible in previous studies (Balk, 2014).Clermont was moderately resistant to Py. ultimum var.sporangiiferum, Dennison was susceptible, and Kottman and Sloan were moderately susceptible.Lorain was previously found to be moderately resistant to Py. irregulare, whereas Dennison, Kottman, and Sloan were all moderately susceptible.The experiment was considered successful and the data were used for analysis only if the check cultivars developed the anticipated levels of disease towards the pathogen.
For each cup assay experiment, plants were grown for 14 d, then the plants were gently removed from the cups and the roots were washed under running tap water.Data for disease severity for root rot, percentage seed germination, total plant weight, and fresh root weight were collected.The disease severity was based on a root rot score from 1 to 5, where 1 = healthy roots with no infection, 2 = lesions covering 1 to 25% of roots, 3 = 26 to 75% of roots with lesions, 4 = 76 to 100% of roots with lesions, and 5 = total colonization of the seed with no germination.All roots from one cup were dried in an oven for 7 d and then weighed to obtain the total dry root weight.Adjusted root weight was calculated based on the fresh root weight of all plants in a single cup divided by the final number of plants in that cup.
The data for the percentage germination, root rot score, average root weight, and total dry root weight from the inoculated cups were analyzed for each population separately.The BLUP (Stroup, 1989) for the response of each RIL population to each pathogen was separately calculated by using the PROC MIXED in SAS 9.1 (SAS Institute, 2003).Due to overall goal of the project, space, and time, data from the noninoculuated cups were not used, other than to verify the quality of the individual RIL in the present study.The parents in each of these populations were adapted and high yielding, so we would not expect many differences.Additionally, in an earlier study with a similar cup assay where lines differed in morphology, genetic signals for root and shoot weight were dissimilar under inoculated and noninoculated conditions (Schneider et al., 2016).
The cup assays for Py.ultimum var.ultimum and Py.irregulare were set up in a RCBD.Three replicate cups for each RIL were included in a single experiment; two experiments were performed for each NAM population ´ pathogen species combination for a total of six replicate cups for each RIL.The model for the RCBD experiments was where m is the overall mean, E i is the effect of the ith experiment, R(E) ij is the effect of the jth replication in the ith experiment, C k is the effect of the kth class of entry (checks: Clermont, Dennison, Kottman, Lorain, Sloan, Williams 82, parents, and RILs), G(C) kl is the effect of lth genotype within class (genotypic variance, s 2 G ), and e ijkl is the experimental error (s 2 ).Class of entry was treated as a fixed effect, and all other terms were considered random effects.The model of this analysis allowed checks and parents to be considered fixed effects and the RILs considered as random effects.
The cup assays for Py.ultimum var.sporangiiferum were arranged in a randomized incomplete block design (RIBD).Due to limited growth chamber space, two NAM populations were evaluated simultaneously with a set of lines evaluated at one time including each RIL for one NAM population, plus one third of the RILs from another NAM population in RIBD.This design was repeated over time in the same growth chamber until population HS6-3976 had a total of four replicates for each RIL and population LG05-4832 had a total of five replicates for each RIL.The model for the RIBD experiments was where m is the overall mean, R i is the effect of the ith growth chamber experiment, R(E) ij is the effect of the jth growth chamber experiment in the ith replicate, C k is the effect of the kth class of entry (checks: Clermont, Dennison, Kottman, Lorain, Sloan, Williams 82, parents, and RILs), G(C) kl is the effect of lth genotype within class (genotypic variance, s 2 G ), and e ijkl is the experimental error (s 2 ).Class of entry was assumed to be a fixed effect, and all other terms were considered random effects.The model of this analysis allowed checks and parents to be considered fixed effects and the RILs considered as random effects.

Correlations among Phenotypic Traits
Within each pathogen, Pearson correlation coefficients were calculated to compare the relationship of the BLUP values among percentage germination, root rot score, adjusted root weight, and molds in a tray test towards P. sojae, and cup assays for Py.irregulare, Py. ultimum var.ultimum, and Py.ultimum var.sporangiiferum (Tables 2-5).Overall, the check lines in each experiment responded as expected (Fig. 1-3).
All four NAM populations evaluated for resistance responded quantitatively to P. sojae, but only three populations had significant differences among the RILs for lesion length 7 d after inoculation (Fig. 1, Table 2).The heritability for lesion length was low for the LD02-9050 and LG00-3372 populations and moderate for the HS6-3976 and LG05-4832 populations (Table 2).The BLUP values calculated from the mean lesion length data were inverted for presentation of data; in this study, a lower BLUP value corresponds to a higher level of resistance (Fig. 1).
After inoculation with Py. irregulare isolate Br2-3-5, the RILs in the three NAM RIL populations were all highly significantly different (P < 0.001) for percentage germination, root rot score, adjusted root weight, and mean dry root weight (Fig. 1 [2A-2D and 3A-3D] and 2 [4A-4D], Table 3).However, there were a large number of missing data points for root rot score in HS6-3976 and dry root weight in the LD02-9050 RIL population; these data were not used in the analysis but were included in Fig. 1.The heritability was moderate (0.3703-0.7808) across all of the measured traits in a cup assay (Table 3).
The checks and RILs (genotype) of two NAM populations were inoculated with Py. ultimum var.ultimum isolate Miami 1-3-7 in one experiment followed by isolate N201.2.2 in a second experiment.Based on the combined data from the two experiments, the root rot score (P = 0.2831), adjusted root weight (P = 0.8914), or total dry root weight (P = 0.1390) for the HS6-3976 population's parents and checks the genotype ´ experiment interaction was not significant, although the interaction for percentage germination was significant (P = 0.0387).For the S06-13640 population's parents and checks, findings were similar with no significant genotype ´ experiment interaction for percentage germination (P = 0.2507), root rot score (P = 0.6566), adjusted root weight (P = 0.1396), and total dry root weight (P = 0.1396).Thus, the data for both experiments was combined for further analysis within each population.
dry root weight across all of the RIL populations evaluated, as well as individually.The p values were adjusted using the Bonferroni multiple test correction for each Pythium species analysis.

Quantitative Disease Resistance Locus Identification
The published linkage maps previously generated for each of the six NAM populations (Song et al., 2017) were used even though there were missing RILs in each population.The program MapQTL 5 (van Ooijen, 2004) was used for QDRL identification for each phenotypic trait for each pathogen, separately.Each marker had a logarithm of the odds (LOD) score that estimated the likelihood of that marker contributing to the observed phenotypic variation.Interval mapping was used to identify preliminary regions of interest, followed by the multiple-QTL method and automatic cofactor selection with the walking speed set to 1 cM (van Ooijen, 2004).The LOD thresholds for chromosome-and genome-wide analysis were calculated by doing a permutation test with 1000 permutations at a = 0.05.The genome-wide LOD thresholds were used as a basis for identifying significant QDRL; if a marker's LOD score surpassed the genome-wide LOD threshold, is was considered a significant QDRL.The LOD thresholds calculated at the individual chromosome level were lower than the genome-wide LOD threshold and are a useful way of identifying suggestive QTL or comparing QTL between experiments (van Ooijen, 1999).In this study, QDRL that did not surpass the genome-wide LOD threshold but did surpass the calculated chromosome-wide threshold were considered "suggestive QDRL" and are useful for comparative purposes only.
The QDRL interval size was determined by the markers flanking the peak with a single LOD confidence interval.The QDRL were considered co-localized if there was overlap in the physical position of markers flanking the QDRL or the same flanking marker was identified in the same population or in comparison across populations.The percentage of phenotypic variation contributed by a QDRL was calculated; if this value was 15.0% or above, the QDRL in question was considered a "major QDRL," and if it was below, then it was considered a "minor QDRL," similar to St. Clair (2010).

Phenotypic Data
These six NAM populations all responded with a wide range of resistant to highly susceptible phenotypic responses after inoculation to one or more of the water  number that emerged from the eight seeds that were planted in each cup.Root rot score was evaluated on a the basis of 1 to 5, where 1 was no evidence of infection and 5 was no emergence.Adjusted root weight was the total fresh weight (g) of the plants 14 d after planting divided by the final number of plants.Mean dry root weight was the total weight (mg) remaining after all roots in a single cup were dried in an oven for 7 d.‡ nd, no data.number that emerged from the eight seeds that were planted in each cup.Root rot score was evaluated on a basis of 1 to 5, where 1 was no evidence of infection and 5 was no emergence.Adjusted root weight was the total fresh weight (g) of the plants 14 d after planting divided by the final number of plants.Mean dry root weight was the total weight (mg) remaining after all roots in a single cup were dried in an oven for 7 d.number that emerged from the eight seeds that were planted in each cup.Root rot score was evaluated on a basis of 1 to 5, where 1 was no evidence of infection and 5 was no emergence.Adjusted root weight was the total fresh weight (g) of the plants 14 d after planting divided by the final number of plants.Mean dry root weight was the total weight (mg) remaining after all roots in a single cup were dried in an oven for 7 d.The RILs in the HS6-3976 and S06-13640 NAM populations were highly significantly different (P < 0.0002) after inoculation with two isolates of Py. ultimum var.ultimum for percentage germination, root rot score, adjusted root weight, and mean dry root weight (Table 4).For the NAM population 4J105-3-4, the root rot score (P = 0.0024) and mean dry root weight (P = 0.0034) were significantly different among RILs, and percentage germination and adjusted root weight were both highly significantly different among RILs (P < 0.0001).Heritability was also low for the mean dry root weight in the 4J105-3-4 population, whereas the remaining traits for all three populations were moderate (0.3286-0.6540,Table 4).
Two RIL populations, HS6-3976 and LG05-4832, had similar overall means for percentage germination, root rot score, and adjusted root weight, but the differences among the RILs in each population were not significant after inoculation with Py. ultimum var.sporangiiferum isolate Will 1-6-7 (Table 5).The RILs in population HS6-3976 were not significantly different for the root rot score nor mean dry root weight (Table 5), and the overall heritability for all traits was low.In the population LG05-4832, adjusted root weight and dry root weight were significant (P < 0.02), whereas the remaining traits, percentage germination and root rot score, were highly significant (P < 0.005) (Table 5).Heritability for the traits measured in this population was moderately low (0.3115-0.5770,Table 5).

Comparison of Phenotypic Measures for Species of Pythium
Several different measures were evaluated for resistance to three different species of Pythium including percentage germination, root rot score, adjusted fresh root weight, and dry root weight.Overall, the correlations for the comparisons were highly significant, although low to moderate among measures for percentage germination, root rot score, and adjusted root weight (Table 6).As expected, comparisons of adjusted root weight to dry root weight were high (>0.73).Interestingly, the Pearson correlation coefficients for comparisons between percentage germination and dry root weight ranged from 0.69 to 0.81 for the three isolates, which was higher than the comparison of percentage germination with adjusted root weight of 0.26 to 0.59.

Quantitative Disease Resistance Loci
There were four QDRL identified in two populations which were inoculated with P. sojae isolate Win371 (Table 7, Fig. 4).Two major QDRL (>15.0%phenotypic  # GW LOD, genome-wide threshold for LOD. variation) in the LD02-9050 population were identified on chromosome 6 with the resistant allele from the susceptible parent IA3023 and on chromosome 18 with the resistant allele from LD02-9050 (Table 7).Both a major and minor QDRL were identified on chromosome 13 in the HS6-3976 population (Table 7).The major QDRL on chromosome 13 contributed >42% of the phenotypic variance.There were no major or minor QDRL identified in the two populations that were screened with P. sojae isolate Day739 even though there were significant differences among the RILs in the LG05-4832 population (P = 0.0008, Table 2).However, two suggestive QDRL for P. sojae were identified in this population for 11 total suggestive QDRL across all four populations that were significant at the chromosome-wide LOD threshold.Ten of the suggestive QDRL were contributed by the resistant parent (Supplemental Table S1).Interestingly, the major, minor, and suggestive QDRL for P. sojae isolate Win371 all mapped to different loci between the HS6-3976 and LD02-9050 populations; this indicates potentially different mechanisms for quantitative disease resistance in these populations.There were two or more QDRL identified to each of the phenotypic measurements in the two populations screened for resistance to Py. irregulare isolate Br2-3-5 in a cup assay.The resistance allele for five of the seven minor QDRL (Table 8, Fig. 4) mapped in the LG00-3372 population were contributed by the more resistant parent LG00-3372.In addition, the loci were different for adjusted root weight, percentage germination, and dry root weight except for one QDRL on chromosome 10 for both adjusted root weight and dry root weight (Table 8).There was one major QDRL on chromosome 4 towards Py.irregulare in the LD02-9050 population, which contributed 20.0% of the phenotypic variance for percentage germination from the susceptible parent IA3023 (Table 8).Additionally, four of the eight minor QDRL were also contributed by IA3023, as were six of the 12 suggestive QDRL (Table 8, Supplemental Table S1).There were no dry root data for this population, but there were three minor QDRL mapped for each of the remaining phenotypic traits.In this population, the major QDRL for percentage germination overlaps the minor QDRL detected for root rot score on chromosome 4 and may account for the higher heritability of this trait in this population (Table 3).Seven suggestive QDRL were identified on chromosomes 5, 6, 10, 13, and 19 in the HS6-3976 population (Supplemental Table S1), in which several were detected for more than one phenotypic trait.Additionally, four of the seven suggestive QDRL were contributed by the susceptible parent IA3023.
Five major QDRL were identified in the 4J105-3-4 and S06-3640 populations for resistance towards Py.ultimum var.ultimum in a cup assay (Table 9, Fig. 4).Interestingly, four of the major QDRL were all contributed by IA3023 for percentage germination and root rot score.All of the major and minor QDRL were contributed by IA3023, whereas the majority of the suggestive QDRL were contributed by the more resistant parent 4J105-3-4 (Supplemental Table S1).There are indications that this resistance towards Py.ultimum var.ultimum is complex, as there were two suggestive QDRL with the resistance alleles contributed by 4J105-3-4 (Supplemental Table S1) mapped between the major and minor QDRL with Table 8.Comparison of quantitative disease resistance loci (QDRL) to Py. irregulare isolate Br2-3-5 significant at the genomewide threshold mapped in the Soybean Nested Association Mapping (SoyNAM) recombinant inbred line (RIL) derived from the cross of IA3023 ´ LG00-3372 and IA3023 ´ LD02-9050.Traits shown are total dry root weight (DRW), adjusted root weight (ARW), and percentage germination (%G).(van Ooijen, 2004).¶ LOD, logarithm of odds.
resistance alleles contributed by IA3023 on chromosome 13 (Table 9).Due to the challenges posed in performing QTL mapping using data from a small population, these QDRL that mapped nearby each other are difficult to discern.In the S06-13640 population, the major QDRL with the resistant allele contributed by parent S06-13640 mapped to chromosome 7 using root rot score data, whereas the two QDRL identified on chromosomes 5 and 17 had the resistance allele contributed by IA3023.One of the two minor QDRL in the HS6-3976 population was mapped to the same locus for adjusted root weight and dry root weight (Table 9).The second minor QDRL on chromosome 17 had resistance contributed by HS6-3976, but four of the 13 total suggestive QDRL were contributed by IA3023 (Supplemental Table S1).
The susceptible parent, IA3023, contributed the most resistance alleles towards resistance to Py. ultimum var.sporangiiferum with major and minor QDRL on chromosome 3 in the LG05-4832 population and two minor QDRL on chromosomes 5 and 11 in the HS6-3976 population (Fig. 4).The major QDRL for percentage germination mapped to the same markers as the minor QDRL on chromosome 3 for root rot score in the LG05-4832 population.There was only one minor QDRL with the resistance allele contributed from HS6-3976 for root rot score (Table 10).None of the loci for resistance to Py. ultimum var.sporangiiferum were similar between these two populations, despite most of the resistance alleles coming from the susceptible parent (Table 10, Supplemental Table S1).

Key Quantitative Disease Resistance Loci
Overall, 24 of the 33 major and minor QDRL identified in these six NAM RIL populations (Tables 7-10, Fig. 4) contributed resistance to only one of these four pathogens, and these loci were uniquely mapped on 13 of the 20 chromosomes in soybean.Surprisingly, there were only a few QDRL identified that were associated with resistance towards more than one pathogen or from more than one population to the same pathogen, thus confirming their role in resistance.The QDRL with the resistance alleles contributed by IA3023, one on chromosome 13 (Gm13: 22901190-25230180 and Gm13: 25230180-26955004) and the second on chromosome 17 (Gm17: 6517544-7100289 and Gm17: 4949843-6517544), were associated with resistance to Py. irregulare and Py.ultimum var.ultimum in four of the six populations (Tables 8 and 9).This indicates that these two regions are particularly important for disease resistance and demonstrates that IA3023, even as the susceptible parent, can contribute to resistance.
There were also two QDRL identified to different pathogens that mapped to the same locus and were contributed by the same parent.The first is the major QDRL contributed by IA3023 for percentage germination mapped to chromosome 3 (Gm3: 510431-853885): in the 4J105-3-4 population to Py. ultimum var.ultimum and (Gm3: 425209-587640) the LG05-4832 population to Py. ultimum var.sporangiiferum.The second QDRL is the region on chromosome 17 where a suggestive locus for P. sojae and minor locus for Py.ultimum var.ultimum from parent HS6-3976 overlaps a suggestive locus to Py. ultimum var.ultimum from parent 4J105-3-4.This locus may be an example of resistance that could be considered broad-spectrum resistance.The contribution to resistance was 3 to 12% of the phenotypic variance for the region on chromosome 17.
There were also instances where resistance loci were mapped to the same region for different pathogens but Table 9.Comparison of major and minor quantitative disease resistance loci (QDRL) for Py.ultimum var.ultimum which were significant at the genome-wide threshold (GW LOD) mapped in the nested association mapping (NAM) F 5 recombinant inbred line (RIL) population derived from the crosses of HS6-3976 ´ IA3023, IA3023 ´ 4J105-3-4, and IA3023 ´ S06-13640.Traits shown are total dry root weight (DRW), percentage germination (%G), and root rot score (RRS).originated from different parents.A QDRL for resistance to P. sojae with resistance from parent HS6-3976 mapped to chromosome 13 (Gm13: 40233656-42919730), which overlaps the QDRL for resistance to Py. ultimum var.ultimum from parent IA3023 (Gm13: 40935278-41953362).Although these could be controlled by one or different mechanisms, this overlap indicates that combining these two loci may be challenging for plant breeders.Resistance to Py. irregulare mapped to the same region on chromosome 16 as a minor QDRL from LG-00372 and a suggestive QDRL from LD02-9050 (Supplemental Table S1).Most of the instances of colocation of a QDRL contributed very low levels to the phenotypic variance and were primarily suggestive QDRL.For example, adjusted root weight as a measure of resistance towards Py.irregulare mapped to the same location on chromosome 20 from parents LD02-9050 and LG00-3372.

DISCUSSION
Multiple disease resistance is heritable resistance conferred to two or more pathogens, from one or more loci in which the loci may or may not be distinct to each pathogen (Wiesner-Hanks and Nelson, 2016).Co-localization of QDRL and similar mechanisms for multiple disease resistance would be useful in developing cultivars with broad-spectrum resistance, especially if these were incorporated into the germplasm base.This study mapped resistance to four different soilborne water molds that have hemibiotrophic and necrotrophic lifestyles and are very commonly found in the north-central region.Prior resistance screenings of the SoyNAM parents with P. sojae, Py. irregulare, Py. ultimum var.ultimum, and Py.ultimum var.sporangiiferum suggested that quantitative resistance to these pathogens could be identified in select RIL populations of the SoyNAM population (Wickramasinghe et al., 2012;Balk, 2014).In this study, to maximize the chance of resistance segregating in a population, we only evaluated a SoyNAM RIL population if the two parents were significantly different from each other in their resistance to that pathogen in those earlier studies.
The resistance identified in each of these populations to each of the pathogens was quantitative-most contributed 10% or less to the phenotypic variance.This was expected, as most quantitative resistance to plant pathogens is based on several loci, each contributing a small proportion to the total resistance level (St. Clair, 2010;Niks et al., 2015).This is a critical finding, as breeding for resistance to traits controlled by minor QDRL is different from those of major gene or few major QDRL (Bernardo, 2008;St. Clair, 2010) Additionally, there were major QDRL that contributed >15% of the phenotypic variation towards each of the four pathogens that were evaluated in this study.The major QDRL for P. sojae identified on chromosome 13 with the resistance allele from HS6-3976 was identified previously from Conrad (Wang et al., 2010), PI 408105A (Nguyen et al., 2012), and in a genome-wide association study for quantitative resistance to P. sojae (Schneider et al., 2016).The percentage variance explained was 20.1 for the PI 408105A mapping population in 2010 (Nguyen et al., 2012) and was 42.2 in this study, indicating that this could be a major contributor to partial resistance.This region on the soybean genome is also near a QDRL towards Sclerotinia sclerotiorum (Guo et al., 2008).A similar trend also occurs for the region on chromosome 18 with the resistance contributed by LD02-9050, which overlaps previous reports for QDRL towards P. sojae with resistance contributed by V71-370 (Tucker et al., 2010), PI 427106, and PI 427105B (Lee et al., 2014).A cluster of resistance gene analogs (McHale et al., 2012) were also reported in this genomic region on chromosome 18, and these could be providing the specificity to P. sojae in this region, as none of the QDRL for Pythium have been identified in this region to date.
The QDRL identified in this study towards Py.irregulare are new compared with those reported from two previous studies (Ellis et al., 2013;Stasko et al., 2016).
However, there is one potential overlap with chromosome 11 recently reported by Lin et al. (2018) from Michigan breeding line E09088.In this study and the previous reports, the identified QDRL were minor, with the exception of QDRL on chromosome 13 (BARC-900926-00961), which explained 17.7% of the phenotypic variance in a cross with Dennison (Ellis et al., 2013).Interestingly, this maps below the resistance reported in this study on chromosome 13 towards Py.irregulare with resistance from IA3023 in two populations, towards Py.ultimum var.ultimum with resistance from 4J105-3-4, and towards P. sojae with resistance from HS6-3976.
Resistance to Py. ultimum var.sporangiiferum has not been mapped in soybean previously, and only recently, minor QDRL were identified on chromosomes 6 and 8 towards Py.ultimum var.ultimum (Klepadlo et al., 2018).Thus, the QDRL identified on chromosomes 1, 2, 3, 5, 7, 13, and 17 towards Py.ultimum var.ultimum and on chromosomes 3, 5, 11, and 17 towards Py.ultimum var.sporangiiferum are novel.Resistance in the region on chromosome 17 was contributed by HS6-3976 for root rot score to both pathogens but is separated by ?2.6 MB.Resistance towards Sclerotinia sclerotiorum, another necrotroph, also mapped to this region (Arahana et al., 2001).Resistance to Py. ultimum var.ultimum was contributed primarily (five of six QDRL) by the more susceptible parent, IA3023, and to a lesser extent towards Py.ultimum var.sporangiiferum and Py.irregulare.Resistance alleles contributed from the susceptible parent have been reported previously (Vales et al., 2005;Ulloa et al., 2010;Stasko et al., 2016).Hnetkovsky et al. (1996) identified a QDRL towards Fusarium solani with the resistance allele being contributed from the susceptible soybean parent, as well as F. graminearum and Py.irregulare (Stasko et al., 2016).It is generally recognized that biological complexity of quantitative resistance has made it difficult to clone genes and characterize mechanisms involved (St. Clair, 2010;Niks et al., 2015;Corwin and Kliebenstein, 2017).Pathways or networks may be more important and may account for the ability of susceptible genotypes to contribute to the overall resistance (Kliebenstein, 2009;Corwin et al., 2016).
In populations where there were not significant differences among the RILs or significant LOD values, population size may have been a factor.Recombinant inbred line populations are considered a valuable and powerful means of identifying QTL of many different phenotypic traits.However, the power of RIL populations to identify QTL with precision decreases with the number of lines included in the population (Ferreira et al., 2006;St. Clair, 2010).Due to heavy losses in the field, populations, 4J105-3-4, LD02-9050, and S06-13640 from the full NAM population each had <100 RILs, whereas the linkage maps were developed from 140 RILs.The QTL analysis may have been more precise or able to identify QDRL with significantly higher LOD values if the full set of 140 RILs had been phenotyped for each population.The large number of suggestive QDRL identified in this study may also be a reflection of the small population sizes used in this experiment.If these experiments were repeated with the entire 140 RILs per population, some of the previously identified suggestive QDRL may surpass the genome-wide LOD threshold.
Combining populations and revaluating with a joint linkage analysis approach has proven useful when parents are closely related and the QTL exist in several of the populations (Buckler et al., 2009).However, others (Li et al., 2011;Chandler et al., 2013;Yang et al., 2013;Lee et al., 2014) have generally identified the same QTL in mapping resistance in a single population vs. joint linkage mapping approach when there were 2 to 10 populations in a NAM design.Additionally, both in maize (Chandler et al., 2013) and soybean (Lee et al., 2014), joint linkage QTL analysis failed to detect QTL when the parentage among the populations was diverse, such that one QTL would be found in one population and not in another.Based on these previous studies and since most of the QTL in this study mapped to such different loci, we did not attempt the joint linkage mapping approach.However, given that numerous resistance alleles were identified from the susceptible IA3023 hub parent, it would be reasonable for future research to assay SoyNAM RIL populations in which the parents did not differ significantly for resistance and carry out a joint linkage analysis on a larger number of nested RIL populations.
Our goal was to identify multiple disease resistance in these genotypes similar to that reported in common bean genotypes resistant to Py. aphanidermatum, as well as Py.splendens (Binagwa et al., 2016).In two earlier studies, where the parents of the NAM were screened for resistance, there was limited overlap in resistance to the eight different species of Pythium (Balk, 2014;Lerch, 2017).Only one parent, HS6-3976, was resistant to three of the four pathogens.Although there were multiple populations that had QDRL mapped to more than one pathogen, there was little overlap in the QDRL.This was surprising, as these four pathogens have very similar processes of colonization and infection in the necrotrophic phase.Stasko et al. (2016) compared regions of the genome that were associated with resistance to Py. irregulare with those that were associated with resistance for necrotrophic pathogen F. graminearum and the hemibiotroph P. sojae in a Conrad ´ Sloan RIL population.There were different QDRL for each, indicating that the mechanism for resistance to Py. irregulare is most likely different than the mechanisms for resistance towards F. graminearum and P. sojae in this population.
The differences among the QDRL for each pathogen may be a consequence of how the phenotypic traits to indicate resistance were measured.For example, the root lesion size used to evaluate resistance to P. sojae may not translate completely to other root characteristics, like the root weight and root rot score used for the three species of Pythium.Mapping for specific components of quantitative or partial resistance can measure unique aspects of where resistance may be expressed the strongest.This was recently reported in another cross where resistance in the seed rot phase was different from the root rot phase in soybean to Py. lutarium, Py. oopapillum, Py. sylvaticum, and Py.torulosum (Lerch, 2017).This is well known for sudden death syndrome of soybean caused by Fusarium virguliforme (Chang et al., 2018), where QDRL to the foliar phase are distinct from those for the root rot phase.Similar findings were also reported in maize for three necrotrophic foliar pathogens (Belcher et al., 2012).Some of the QDRL for P. sojae were different when lesion size was mapped compared with those for root rot score, especially those that were contributed by the susceptible parent (Wang et al., 2012a).This was attributed, in part, to the different components of partial resistance to P. sojae being measured with the different assays (Wang et al., 2012a).This may also explain why the Pearson correlation coefficients for most of the measured parameters were moderate, as different components of the resistance response may be acting in different ways.The components of resistance are, for the most part, unknown for the Pythium spp.used in this experiment and should be a focus of research in the future, as this may guide future evaluations of resistance.
A similar pattern of quantitative resistance controlled by different loci to each pathogen isolate is emerging in studies that have examined QDRL to different isolates of the same pathogen (González et al., 2012;Corwin et al., 2016;Debieu et al., 2016), including P. sojae (Stasko et al., 2016) and Sclerotinia sclerotiorum (Wilbur et al., 2017) in soybean.Parlevliet and Zadocks (1977) were the first to propose that individual host genes interact with virulence genes of the pathogen whether the resistance is inherited as single-gene-mediated resistance or quantitatively.As mapping of resistance towards multiple pathogens in the same populations continues, this pattern is beginning to emerge.
Multiple disease resistance where a single locus from a single source confers resistance to one or more diseases would be a great asset to soybean breeders.However, this study indicates that these soilborne pathogens, which have very similar pathogenic styles and sites of infection, have different QDRL and potentially mechanisms of resistance.The parents of the SoyNAM population were not originally selected for resistance to biotic stress (Diers et al., 2018) but will serve as a rich resource as these studies continue to target and dissect QDRL for resistance to soilborne pathogens.

Table 1 .
Summary of the six Soybean Nested Association Mapping (SoyNAM) recombinant inbred line (RIL) populations developed with crosses with IA3023 evaluated for resistance in a tray test towards isolates of Phytophthora sojae and cup assays towards isolates of Pythium irregulare, Py. ultimum var.ultimum, and Py.ultimum var.sporangiiferum.All isolates were collected in Ohio as part of surveys or from samples that were submitted to the laboratory and are part of the soybean pathology isolate collection.‡ Two isolates of Py. ultimum var.ultimum were used, one experiment with isolate Miami 1-3-7 and the second experiment with isolate N201.2.2.§ Indicates that the donor parent was not significantly different from IA3023 in preliminary tests, and the population was not evaluated for resistance to one or more of the pathogens.

Table 2 .
Summary of four Soybean Nested Association Mapping (SoyNAM) F 5 recombinant inbred line (RIL) populations evaluated for lesion length and heritability after inoculation with Phytophthora sojae † in a tray test.

Table 3 .
Summary of three Soybean Nested Association Mapping (SoyNAM) F 5 recombinant inbred line (RIL) populations evaluated for resistance towards Pythium irregulare isolate Br2-3-5 in a cup assay and significance and heritability of each of the measured traits for percentage germination (%G), root rot score (RRS), adjusted root weight (ARW), and dry root weight (DRW).

Table 4 .
Summary of three Soybean Nested Association Mapping (SoyNAM) F 5 recombinant inbred line (RIL) populations evaluated for resistance towards Pythium ultimum var.ultimum isolates Miami 1-3-7 and N201.2.2 in a cup assay and significance and heritability of each of the measured traits for percentage germination (%G), root rot score (RRS), adjusted root weight (ARW), and dry root weight (DRW).

Table 5 .
Summary of four Soybean Nested Association Mapping (SoyNAM) F 5 recombinant inbred line (RIL) populations evaluated for resistance towards Pythium ultimum var.sporangiiferum isolate Will 1.6.7 in a cup assay and significance and heritability of each of the measured traits for percentage germination (%G), root rot score (RRS), adjusted root weight (ARW), and dry root weight (DRW).
† Percentage germination was calculated from the total

Table 10 .
Comparison of major and minor quantitative disease resistance loci (QDRL) to Py. ultimum var.sporangiiferum significant at the genome-wide threshold mapped in the nested association mapping (NAM) recombinant inbred line (RIL) population generated from the cross of IA3023 ´ LG05-4832.Traits shown are total dry root weight (DR), adjusted root weight (AR), percentage germination (%G), and root rot score (RRS).