Resistance to Multiple Soil-Borne Pathogens of the Pacific Northwest, USA Is Colocated in a Wheat Recombinant Inbred Line Population

Soil-borne pathogens of the Pacific Northwest decrease yields in both spring and winter wheat. Pathogens of economic importance include Fusarium culmorum, Pratylenchus neglectus, P. thornei, and Rhizoctonia solani AG8. Few options are available to growers to manage these pathogens and reduce yield loss, therefore the focus for breeding programs is on developing resistant wheat cultivars. A recombinant inbred line population, LouAu (MP-7, NSL 511036), was developed to identify quantitative trait loci (QTL) associated with resistance to P. neglectus and P. thornei. This same population was later suspected to be resistant to F. culmorum and R. solani AG8. This study confirms partial resistance to F. culmorum and R. solani AG8 is present in this population. Six major and 16 speculative QTL were identified across seven measured traits. Four of the six major QTL were found within the same genomic region of the 5A wheat chromosome suggesting shared gene(s) contribute to the resistance. These QTL will be useful in breeding programs looking to incorporate resistance to soil-borne pathogens in wheat cultivars.

Rhizoctonia root rot and bare patch disease are common in notill farming systems Paulitz et al. 2002), a practice that has increased steadily in the PNW, ranging from 11 to 70% of planted acreage across PNW counties (Young 2009). Fungal infection causes rotting of the seminal and crown root tissue, resulting in brown lesions, spear tipping of roots, and stunted seedlings (Paulitz et al. 2002;Weller et al. 1986). In the field, R. solani AG8 causes Rhizoctonia bare patch disease, characterized by large circular dead patches of wheat. Yield losses by Rhizoctonia have been reported between 10 and 30%, and are highly correlated with disease severity Mahoney et al. 2016;Paulitz et al. 2002).
Cultural methods to reduce soil-borne diseases include tillage, reducing nitrogen fertilizer, or rotation to less profitable crops, all of which are used, but not favored, in dryland wheat farming systems. Fusarium crown rot damage can be reduced by delayed planting and management of nitrogen fertilizer applications (Paulitz et al. 2002). Rhizoctonia root rot can be reduced through tillage, and managing weeds and volunteer wheat well before planting (Gill et al. 2002;Paulitz et al. 2002;Smiley et al. 1992). RLNs have been more difficult to manage, they have an extensive host range, reducing the effectiveness of crop rotations, and can be found as low as 90 cm in soil profiles, essentially evading tillage disturbance Smiley and Machado 2009). Therefore, the focus for management of these diseases is on developing resistant and/or tolerant wheat cultivars.
The search for resistance to these soil-borne pathogens has been challenging. Phenotyping is time-consuming and difficult, and subject to spatial and environmental variation in the field (Collard et al. 2005;Schmidt et al. 2005;Schneebeli et al. 2016;Sharma et al. 2011). Often, the identified resistant or partial-resistant germplasm have been wild relatives, landraces, and synthetic hexaploids that also possess nondesirable traits (Li et al. 2010;Mahoney et al. 2016; Thompson et al. 1999Thompson et al. , 2016. To improve breeding efficiency, it is desirable to define the genetic architecture (number and location of loci) for resistance to these pathogens. The LouAu recombinant inbred line (RIL) population was developed by Thompson et al. (2015). The resistant parent IWA8608077 (PI621458), an Iranian landrace, was previously identified as resistant to both Pratylenchus species by Sheedy et al. (2007) and A. L. . The LouAu population was phenotyped for resistance to P. neglectus and P. thornei, but the whole population has not yet been phenotyped for resistance to other soil-borne pathogens.
The objectives of this study were to (i) phenotype the LouAu RIL population for response to F. culmorum and R. solani AG8; and (ii) identify quantitative trait loci (QTL) associated with reduced infection to F. culmorum, both Pratylenchus species, and R. solani AG8.

Plant material and genotyping
The LouAu RIL population (MP-7, NSL 511036) was derived by crossing the PNW cultivar "Louise" (PI634865) with the Iranian landrace IWA8608077 (PI621458). Louise is a soft white spring wheat cultivar, with good milling and baking characteristics (Kidwell et al. 2006), and high-temperature adult plant resistance to stripe rust (Carter et al. 2009). IWA8608077 is a late-maturing hexaploid, and a member of an 11,000 accession collection that has been phenotyped for resistance or tolerance to many biotic stresses (Thompson et al. 2016). The development, genotyping, and linkage map of the LouAu RIL population was previously described by Thompson et al. (2015).
Briefly, the population was genotyped at F 5 :F 6 with simple sequence repeat markers, and also with SNP markers using the Illumina wheat 9K SNP chip (Cavanagh et al. 2013). The population was also genotyped for known major developmental genes. Linkage groups were generated using JoinMap software (Kyazma B.V., The Netherlands). Linkage groups were assigned to genome-specific chromosomes by comparison to the wheat 9K SNP consensus map (Cavanagh et al. 2013). The linkage map contains 2038 markers mapped to 26 linkage groups, representing all 21 of the T. aestivum chromosomes. The total map distance is 2426.8 cM, with an average intermarker spacing at 1.33 cM.
Greenhouse plant disease phenotyping The population was phenotyped for response to multiple soil-borne pathogens at generations F 5 -F 8 as described below. Screening for response to soil-borne pathogens can be time-consuming and variable (Schneebeli et al. 2016;Sharma et al. 2011;Smiley and Patterson 1996); therefore, the phenotyping assays included several susceptible and resistant checks (Table 1).
Fusarium culmorum: Inoculum was prepared following the procedure of Smiley et al. (2005c) with twice-autoclaved millet seed. Five PNW native isolates of F. culmorum (F70110023, F70110097, F70110022, F70110019, and F70110126) collected from a survey in 2011 (T.C. Paulitz, unpublished data) were maintained on either 0.5-strength potato dextrose agar (Sigma-Aldrich) or water agar (Sigma-Aldrich). Each isolate was individually seeded onto the autoclaved millet in 4 l Erlenmeyer flasks and kept at 22-24°in the dark for 3-4 wk. After 1 wk, flasks were shaken every week to ensure even colonization of the millet seed. Colonized millet seed was broken up to remove clumps, air dried, and stored at room temperature (22-24°) in paper bags. Isolates were then blended in equal proportions before application in the greenhouse assays.
The LouAu RILs, parent lines, and checks (Table 1) were evaluated in two experiments with five replicates each in a randomized block design. All plants were grown in D40H Deepots (Stuewe and Sons, Oregon) with Sunshine Mix#4 potting soil (Sungro, Canada), and Osmocote 14-14-14 slow release fertilizer (Scotts Co. LLC) as described by Poole et al. (2012) for growth room assays. The experiments were conducted in the greenhouses at the Washington State University Plant Growth Facility (WSU-PGF) in Pullman, WA. Temperatures were maintained at 22-24°with 12-hr day length supplemental lighting. Two-wk-old seedlings were inoculated with $1 g of the blended colonized millet at the base of the seedling, then covered with 1 g of potting soil. Plants were evaluated for F. culmorum symptoms 12 wk after infection. F. culmorum infection was determined by visually scoring the crown and internodes for browning using a 0-10 rating system, where 0 = no disease, and 10 = dead plant with severe symptoms, as described by Poole et al. (2012). The rating for these symptoms will be referred to as Fc-Score for the remainder of the paper.
Pratylenchus species: Inoculum was prepared following a modified monoxenic carrot culture protocol described by Castillo et al. (1995). Prepared carrot disks were inoculated with a single nematode, and kept in a dark growth cabinet (Percival Scientific) at 22°for at least 2 months prior to extraction in water. Nematodes were collected from a 2010 survey of dryland wheat fields in eastern WA (Kandel et al. 2013), and maintained in water until carrot disk inoculation. Pratylenchus inoculum for each species was quantified using a nematode counting slide (Chalex Corporation) with 1.5 ml aliquots, and species-specific primers developed by Yan et al. (2008) were used to identify P. neglectus or P. thornei.
The LouAu RILs, parent lines, and checks (Table 1) were evaluated in separate experiments for each Pratylenchus species with two replicates each in randomized block designs. All plants were grown in D40H Deepots with 150 g of 3:1 pasteurized Palouse silt loam soil to sand and Osmocote 14-14-14 slow release fertilizer (Scotts Co. LLC) as described by Thompson et al. (2016). All nematode experiments were conducted in a Conviron GR48 (Controlled Environments LTD) chamber at the WSU-PGF. Temperatures were maintained at 24°with 14-hr day lengths. Two-wk-old seedlings were inoculated with $100 nematodes (P. neglectus or P. thornei) in water at the base of the seedling, then covered with $1 g of the 3:1 soil mixture. Plants were evaluated for Pratylenchus damage 12 wk after inoculation. Pratylenchus damage was determined by visually scoring root browning on a rating scale of 1-5, where 1 = little to no root browning, and 5 = complete root browning (Thompson et al. 2016). The rating for these symptoms will be referred to as Pn-Score or Pt-Score for P. neglectus or P. thornei, respectively for the remainder of the paper. After damage was assessed, soil and root samples were collected in plastic bags, chilled to 4°overnight and sent to Western Laboratories Inc. (Parma, ID; http://www.westernlaboratories. com) for nematode extraction using a modified Oosterbrink elutriator (Smiley et al. 2011), followed by enumeration with a Chalex counting slide. The protocol followed by Western Laboratories concentrates the final extracted volume of nematodes to $10 ml of water, and a single 1 ml aliquot is taken for nematode enumeration. The total count for this assay will be referred to as Pn-Count or Pt-Count for P. neglectus or P. thornei, respectively for the remainder of the paper.
Rhizoctonia solani AG8: Inoculum was prepared following the procedure by Paulitz and Schroeder (2005) with autoclaved oat seed. A R. solani AG8 isolate (isolate C1, Weller et al. 1986) was grown on water agar, and plugs were used to seed the autoclaved oats in Erlenmeyer flasks. Flasks were kept in the dark at 22-24°for 3-4 wk, then dried for 2 d in a laminar flow hood. Colonized seeds were ground with a coffee grinder, and passed through a 1-mm and 250-mm sieve. Ground oat particles were collected between the two sieve sizes.
Dilution plating was used to quantify inoculum using Rhizoctonia selective media (Paulitz and Schroeder 2005).
The LouAu RILs and parent lines were evaluated in three experiments, with three replicates each, in a completely randomized design. All experiments were conducted in a Conviron GR48 (Controlled Environments LTD) at WSU-PGF at 15-18°for 14 d, with a 12-hr day length. To synchronize germination, seeds of each RIL and parent line were cold stratified for 24 hr at 4°on premoistened paper, then germinated in the dark for 24 hr at 22-24°. All germinated seeds were transplanted to D27L Deepots (Stuewe and Sons, Oregon) in each of two treatments; with either $60 g pasteurized Ritzville silt loam soil inoculated with colonized ground oats at 200 propagules per gram of soil, or $60 g pasteurized Ritzville silt loam soil with sterile ground oats. Seeds in each treatment were covered with noninoculated soil. Seedlings were evaluated for Rhizoctonia symptoms 2 wk after transplanting. Previous experiments had indicated that seedling stunting by Rhizoctonia infection is a good indicator of resistance or tolerance at the seedling stage (Mahoney et al. 2016). Seedlings were measured from the crown to the top of the first and tallest leaf lengths. Because leaf length varied according to genotype, leaf length was measured as a percent reduction from the control (noninoculated) plant, by comparing the first and tallest leaf of each RIL. The leaf length reduction formula was y = (a 2 b)/a · 100, where y = percent leaf length reduction; a = the leaf length of the noninoculated plant; and b = the leaf length of the corresponding inoculated plant. The calculated percent reduction will be referred to as Fl-Stunt, or Tl-Stunt, for first leaf and tallest leaf reductions, respectively, for the remainder of the paper.

Statistical analysis
The SAS software MIXED procedure (SAS Institute Inc 2012) was used to fit a linear model to each trait for outlier removal and least squared means (lsmeans) estimates; the models for each trait are below. The data for each assessed trait was first examined for significant outliers that could influence an estimated mean. Outliers were determined by n setting an upper and lower limit for the studentized deleted residuals calculated from the population size with a criterion of a = 0.05 (Kutner et al. 2004). After outliers were removed, the lsmeans were estimated across experiments for each RIL for each trait assessed.
Fusarium culmorum: The model for analysis of the F. culmorum infection symptoms (Fc-Score) was: where Y ijkl is the Fc-Score of an individual Deepot; entry i is the fixed effect of the ith entry (RIL or check under evaluation); exp j is the fixed effect of the jth experiment; rep(exp) jk is the random effect of the kth replication nested within the ith experiment; tray(rep Ã exp) jkl is the random effect of the lth tray (vessel in which Deepots were held) nested within the kth replication within the jth experiment; (entry · exp) ij is the fixed effect of the of the interaction between the ith entry and the jth experiment; and e ijkl is the error term.
Pratylenchus species: The statistical model for analysis of the nematode root damage (Pn-Score or Pt-Score) was: where Y ijkl is the Pn-Score or Pt-Score of individual Deepot; entry i is the fixed effect of the ith entry; trt j is the fixed effect of the jth Pratylenchus treatment (either P. neglectus or P. thornei); rep(trt) jk is the random effect of the kth replication nested within the jth treatment; tray(rep Ã trt) jkl is the random effect of the lth tray nested within the kth replication within the jth treatment; and e ijkl is the error term. The statistical model for analysis of the total nematode counts (Pn-Count or Pt-Count) was: where ƒ(y) ijkl is the log transformation of Pn-Count or Pt-Count of an individual Deepot; entry i is the fixed effect of the ith entry; trt j is the fixed effect of the jth Pratylenchus treatment (either P. neglectus or P. thornei); rep(trt) jk is the random effect of the kth replication nested within the jth treatment; tray(rep Ã trt) jkl is the random effect of the lth tray nested within the kth replication within the jth treatment; and e ijkl is the error term. The log transformation was used because the Pratylenchus count data were highly right skewed so a natural logarithmic transformation ln x + 1 (x = count) was applied before data were analyzed.

Rhizoctonia solani AG8:
The model for analysis of the R. solani infection symptoms (Fl-Stunt or Tl-Stunt) was: where Y ij is the calculated Fl-Stunt or Tl-Stunt of an individual Deepot; entry i is the fixed effect of the ith entry; exp j is the fixed effect of the jth experiment; and e ij is the error term. Due to the way in which Rhizoctonia symptoms were calculated (noninoculated minus inoculated measurements), the tray and replication effects became collapsed within the experiment and could not be separated for statistical analysis. The collapse of terms also limited the available degrees of freedom, and so an interaction effect between entry and experiment could not be determined.
To determine if the assessed traits were related, correlation coefficients were determined using the lsmeans from the analyses described above. Pearson's r and calculated P-value were calculated using the SAS software CORR procedure. Correlations were considered significant at probability P # 0.05. The heritability [h 2 = Var(G)/Var(P)] for each trait on a family-mean basis was estimated for each trait using the LouAu RILs only; the parent lines and checks were removed from the data sets. The statistical models for each trait were the same as described above except that each term in the model was considered random. The asycov option in the SAS software MIXED procedure (Self and Liang 1987) was used to calculate heritability with the SAS code provided by Holland et al. (2003).
The lsmeans for each trait were used to identify QTL with inclusive composite interval mapping (ICIM) (Li et al. 2007) for biparental populations with the ICIM-ADD function in QTL IciMapping v4.1.0 software (http://www.isbreeding.net). Seven markers were dropped from the linkage maps due to colocalization. Mapping parameters were set to stepwise regression with a walk interval of 1 cM. Significance thresholds were calculated by permutation (1000 permutations at P , 0.05) for major QTL. To detect speculative QTL the significance threshold was changed to manual input setting with LOD score to 1.5. Where QTL were detected, the percent variation explained by the QTL and additive effect were also calculated.

Data availability
Lsmeans from the linear models for all soil-borne pathogen traits used in QTL analysis are in Supplemental Material, File S1. Linkage map information for the LouAu population is available in File S2. Remaining genotypic data for the 150 RILs are available in File S3.

Greenhouse plant disease phenotyping
The root disease symptoms caused by F. culmorum were moderate, and there was not a significant difference in response between the two parents. Nevertheless, there was a significantly different distribution of response to F. culmorum among the RILs ( Figure 1A). The P. neglectus counts (Pn-Count) in both experiments were lower than P. thornei (Pt-Count) when averaged over the RIL population (Table 2). Both Pn-Count and Pt-Count showed a normal distribution after total count data were log transformed in the RIL population (Figure 1, B and D). The root damage due to nematode infection (Pn-Score and Pt-Score) were lower in IWA8608077 than Louise, as expected, and showed a similar range in the RIL population for both Pratylenchus species (Table 2). The Pn-Score distribution followed a normal curve, while the Pt-Score is skewed right toward increased root damage (Figure 1, C and E). The difference in Pt-Count and Pt-Score distributions indicate P. thornei caused more root damage than P. neglectus. Similar to the Pratylenchus traits, the Rhizoctonia symptoms were lower in the IWA8608077 parent than the Louise parent ( Table 2). The distribution for first leaf stunting (Fl-Stunt) was skewed left toward decreased leaf reduction ( Figure 1F). The range for tallest leaf stunting (Tl-Stunt) was slightly bigger than Fl-Stunt (Table 2), and showed a normal distribution in the RIL population ( Figure 1G).
Of the possible correlations between all seven traits, only five were found to be significant at a # 0.05. The Fc-Score trait was significantly related to three of the seven traits measured, although correlation values were low (,30%; Table 3). Estimated heritability for the traits ranged from 11.4 to 85.5% (Table 2). The estimated Fc-Score heritability is somewhat low compared to other Fusarium greenhouse assays reported by Poole et al. (2012), but is not out of range for soil-borne pathogen traits. The Pn-Count and Pt-Count heritability estimates (Table 2) fall within those previously reported by Thompson et al. (2012) and Sheedy et al. (2012) using total count data. The Pn-Score and Pt-Score estimates were higher than those for the count data (Table 2) indicating this is a more consistent phenotype to measure plant-nematode interactions. The estimated heritability for R. solani AG8 was lower than previously reported by Mahoney et al. (2016), which were calculated using shoot length reduction.

QTL analysis
The calculated significance threshold by permutation (1000 permutations at P , 0.05) was 3.1842. A total of six QTL with LOD . 3.0 were  found across the seven traits, and 16 QTL with LOD , 3.0 (Table  4 and File S4). These 22 QTL were found on 11 of the 21 wheat chromosomes, with LOD scores ranging from 1.51 to 9.40. The Pn-Count trait had the greatest number of QTL identified (7) over five chromosomes, and had the QTL with the highest LOD score.
Only speculative QTL (LOD , 3.0) were found for traits Pt-Count and Fl-Stunt. The percent variation explained by the identified QTL ranged from 2.5 to 17.1%. The QTL marker spacing was ,10.0 cM for 17 of the 22 identified QTL indicating the marker coverage, and linkage maps were sufficient to detect important resistance loci. The 5A chromosome was the location of six QTL associated with decreased soil-borne pathogen symptoms. The 5A QTL identified for Pn-Score (QRlnn.wsu-5A.2) and Tl-Stunt (QRs.wsu-5A.2) share the same cM position, and map within 4 cM of the Fc-Score QTL (QFs.wsu-5A.2) ( Table 4). These QTL map within 33 cM of the centromere, closer than the Pt-Score 5A QTL (QRlnt.wsu-5A.2). The allele (A) with reduced disease symptoms came from the IWA8608077 parent for all four of these QTL (Table 4). These same traits also possessed significant correlations with each other (Table 3) indicating shared resistance gene(s) coming from the IWA8608077 parent. Sequence for the significant SNP markers can be found in File S5.
Three QTL were identified on chromosome 2A that were within 42 cM marker spacing to each other (QRlnn.wsu-2Aa, QRlnn.wsu-2Ab, and QRs.wsu-2A) (Table 4 and File S4). The Pn-Count and Fl-Stunt traits were not significantly correlated (Table 3), but this 2A genome region could contain gene(s) associated with resistance to all three pathogens. Two of these three QTL showed the allele (B) with reduced disease symptoms was derived from the Louise parent. Louise was also the source of the allele associated with reduced disease symptoms for 12 of the 16 speculative QTL identified (File S4) suggesting transgressive segregation in this population.

DISCUSSION
The LouAu RIL population was initially developed to identify QTL associated with resistance to P. neglectus and P. thornei. Differences in root architecture traits between the IWA8608077 and Louise parents indicated this population may also be useful as sources of resistance or partial-resistance to other soil-borne pathogens. Particularly interesting was the significant increase of root lignin content in the IWA8608077 parent compared to the Louise parent (Thompson et al. 2015), and the distribution to the outermost root cells (A.L. Thompson, unpublished data). Increased lignification has been associated with decreased appressorium penetration by fungal species in grasses, including wheat (Vance et al. 1980), and failure to penetrate wheat leaves by the foliar fungal pathogen Botrytis cinerea (Ride and Pearce 1979). Since IWA8608077 was found to be resistant to the Pratylenchus species, F. culmorum, and R. solani AG8, it is possible the increased root lignin is contributing to the resistance, and would be an excellent breeding target for multi-pathogen resistance introgression.
Heritability estimates for three of the seven traits evaluated were below 50%, which is common in soil-borne diseases. These traits are difficult to phenotype, which increases environmental variation, limiting the ability to accurately estimate genetic effects. The low heritability calculated for F. culmorum in this study could be because of low phenotypic variation, or the scoring method used to quantify disease symptoms. The 0-10 scale used to rate F. culmorum symptoms was developed for F. pseudograminearum (Poole et al. 2012), yet F. culmorum shows more root browning and fewer stem symptoms than F. pseudograminearum (Smiley et al. 2005c). The low heritability estimates for the R. solani AG8 traits Fl-Stunt and Tl-Stunt is n most likely due to the calculated percent reduction used to measure stunting for these two traits. The random experimental variation was confounded with the experiment model term, limiting the ability to quantify error, and therefore estimate the genetic effects. Chromosome 5A had four major QTL associated with reduced disease incidence to multiple different soil-borne pathogens in this study. Resistance QTL have not previously been identified on chromosome 5A for F. culmorum, R. solani or either Pratylenchus species. A QTL for resistance to P. neglectus was identified on barley chromosome 5H, which shares synteny with wheat chromosomes group 5 (Sharma et al. 2011). The 5A chromosome region that contributed resistance to F. culmorum and R. solani, also decreased damage from P. neglectus, and probably accounted for the correlation between these three traits in the RILs. It is therefore possible that all three traits are controlled by the same gene, or closely linked genes, which may confer a nonspecific defense to soil-borne pathogens.
For many of the speculative QTL identified in this study, the allele with reduced disease symptoms was derived from Louise, which was considered to be the susceptible parent. Transgressive segregation has been shown to occur in wide crosses, and can be attributed to the complementary action of genes from the two parents, or the unmasking of recessive genes in the wild parent (deVicente and Tanksley 1993). For this study, the genes within the speculative QTL regions could be acting complementary to resistance genes from the IWA8608077 parent but further study is needed. Increased root lignin content is suspected to be involved as a potential defense mechanism contributing to IWA8608077 resistance to soil-borne pathogens. The lignin biosynthesis pathway is highly conserved throughout the plant kingdom as it is a major contributor to plant cell walls (Cosgrove 2001). It would be of great interest to determine if the QTL found in this study contain genes associated with the lignin biosynthesis pathway, and if alleles associated with resistance increase lignin content.
The incorporation of landrace material into breeding programs for genetic and trait diversity has become of great interest, as concerns of narrow genetic pools in wheat breeding programs increase (Able et al. 2006;Hoisington et al. 1999;Reif et al. 2005;Warburton et al. 2006). Landraces have been shown to be sources for resistance and tolerance to complex traits (Ehdaie et al. 1988;Jafari-Shaberstari et al. 1995;Waines and Ehdaie 2007), and so are candidates for mapping population development. Difficulties that can be associated with the use of landraces in breeding programs emphasize the importance of identifying cost effective tools, like marker assisted selection, for resistance introgression and to reduce linkage drag. In this study, useful markers were identified for multiple soil-borne pathogen traits from a single landrace parent.

ACKNOWLEDGMENTS
We thank members of the Garland-Campbell group, Tricia DeMacon, and Jocelyn Bowser for help with trial harvest; Adrienne Burke for help with growth room maintenance and nematode assays; members of the Smiley research team, Jennifer Gourlie and Guiping Yan for providing the P. neglectus nematodes and guidance; and Yvonne Manning-Thompson for providing F. culmorum cultures. We thank the funding agencies for this project, the Washington Grains Commission project no. 3949, United States Department of Agriculture-Agricultural Research Service (in house) projects: 5348-21220-003-00D and 5348-22000-013-00, and the Washington State University O. A. Vogel grant 8334.