Our analyses show that false positive and ‘of-target’ GWAS signals that arise because of erroneous physical addresses can typically be identified and rectified by fitting DNA markers associated with causal loci as fixed effects, which is effectively equivalent to fitting a multilocus genetic model in a QTL mapping or candidate gene analysis study using mixed linear models . A preponderance of the erroneous physical addresses are typically going to be found on homoeologous chromosomes, as was the case in our study, and hence could be misconstrued as signals associated with the effects of homoeologous loci.The sheer numerical abundance of sources of resistance to Fusarium wilt race 1 in strawberry did not shed light on the diversity of R-genes that they might carry, if any,or genetic mechanisms underlying resistance . Was resistance to race 1 conferred by dominant R-genes? How many unique Fusarium wilt R-genes exist in wild and domesticated populations of strawberry? To explore these questions, we developed and undertook genetic analyses of S1 populations developed by self-pollinating highly resistant F. × ananassa heirloom cultivars and individuals and full-sib populations developed by crossing highly resistant ecotypes of F. chiloensis subsp. chiloensis , F. virginiana subsp. virginiana , and F. virginiana subsp. grayana with a highly susceptible F. × ananassa individual .
The phenotypes of of spring in each of the segregating populations spanned the entire range from highly resistant to highly susceptible with bimodal distributions . When individuals within each population were classified as resistant or susceptible using 2.0 as the cut of on the disease symptom rating scale,plastic planters the observed phenotypic ratios perfectly ft the expected phenotypic ratios for the segregation of dominant resistance genes . The Guardian and Wiltguard S1 and 12C089P002 × PI602575 and PI552277 × 12C089P002 full-sib populations each appeared to segregate for a single dominant resistance gene, whereas the Earliglow and 17C327P010 S1 populations appeared to segregate for two dominant genes with duplicate epistasis, where a single dominant allele at either locus was sufficient to confer resistance . The statistical inferences were not affected by shifting the cut of downward to 1.5 or upward to 2.5; hence, we concluded that dominant R-genes segregated in these populations .Genome-wide searches for SNPs associated with loci segregating for resistance to Fusarium wilt uncovered a single tightly linked cluster of statistically significant SNP markers in each population . The SNPs most strongly associated with differences in resistance phenotypes were tightly linked to partially to completely dominant R-genes that segregated in these populations and mapped to three non-homoeologous chromosomes . The putative R-genes are hereafter designated FW2 , FW3 , FW4 , and FW5 . FW2 and FW5 genetically mapped proximal to FW1 on chromosome 2B . The effects and positions of the these loci were identified by GWAS or genetic mapping alone .
The SNP markers most strongly associated with each of these loci were different and spanned a 1.6 Mb haploblock . To explore the structure of this haploblock in greater depth, we imputed and phased the haplotypes for 71 50K Axiom array-genotyped SNPs among 653 individuals in the California population . These included the parents and more distant relatives and common ancestors of the S1 and full-sib progeny that were both genotyped and phenotyped . Although FW1, FW2, and FW5 could be alleles, haplotypes for 71 SNPs within the haploblock predicted to harbor these loci were insufficient to rule out paralogs, and confidence intervals for the estimated positions of these loci spanned the gene-rich haploblock . The SNP haplotypes for Fronteras and Portola were identical except for three consecutive SNPs in a short haploblock slightly downstream of the location predicted to harbor FW1 . The haplotype associated with the dominant FW1 allele for that haploblock was ascertained from the genotypes of Fronteras and Portola . The haplotypes observed for the other resistant parents difered from each other and Fronteras and Portola; hence, from SNP haplotypes and approximate physical positions, we could not unequivocally show that the putative R-genes associated with these phenotypically mapped loci were allelic. Finally, the KASP assays we developed for informative SNPs associated with these loci were not cross predictive . The paralog hypothesis seems plausible for the underlying R-genes; however, our data were in sufficient to rule out the single locus, multiple allele hypothesis. Although additional studies are needed to resolve this question, the classes of R-genes hypothesized to underlie these loci are commonly found in tandemly duplicated clusters in plants .
The genotypic means, effects, and PVE estimates for SNP markers tightly linked with FW2 and FW5 were nearly identical to estimates for SNP markers associated with FW1 in the Fronteras and Portola S1 populations . FW2 was nearly completely dominant . The additive and dominance effects of the FW2 locus were 1.5- to 1.9- fold greater than those reported for the FW1 locus, partly because unfavorable allele homozygotes were more strongly susceptible in the Guardian S1 population than in the Fronteras and Portola S1 populations. The estimated marginal means for favorable allele homozygotes ranged from 1.07 to 1.25 in the three populations . The EMM for resistant homozygotes was ̄yAA = 1.1, whereas the EMM for susceptible homozygotes was ̄yaa = 4.5. We could not estimate the degree of dominance for FW5 because the AA homozygote was not observed in the fullsib population; however, the EMM for the heterozygote was 1.06, which implies that the FW5 allele might be completely dominant. Although the statistical evidence for the segregation of a single dominant R-gene on chromosome 1A was strong in the Wiltguard S1 population, the effects of SNPs associated with FW3 were weaker than those associated with FW1, FW2, and FW5 on chromosome 2B . The most significant FW3-associated SNP was AX-123363542 , which only explained 23% of the phenotypic variation for resistance to race 1. Despite this, the EMM for FW3 homozygotes was only slightly greater than the EMMs for FW1 and FW2 homozygotes . Although the PVE estimate was greater for FW4 than FW3 , the EMMs for SNP marker heterozygotes were virtually identical: 1.76 for FW3 and 1.69 for FW4; hence, FW4 appears to be as strong as FW3 .With the genomic locations of FW1, FW2, and FW5 narrowed to a short haploblock on chromosome 2B , we searched annotations in the ‘Royal Royce’ reference genome to identify genes encoding proteins known to play an important role in race-specific disease resistance via pathogen recognition and activation of defense responses, e.g., pathogen-associated molecular pattern -triggered immunity or effector triggered immunity . Eight of 1,208 annotated genes found in the 0.0-5.0 Mb haploblock on chromosome 2B encode proteins with known R-gene domains and functions . These included one coiled-coil domain NLR encoding gene and two tightly linked Tollinterleukin 1 receptor domain type NLR encoding genes . Hence, the most promising candidate genes for FW1 encode NLR proteins. The approximate 95% Bayes confdence interval for the genomic location of FW4 on chromosome 6B was fairly wide and consequently harbored 197 annotated genes in the ‘Royal Royce’ reference genome . Nine of these 197 annotated genes are predicted to encode R-proteins that mediate gene-for-gene resistance in plants . These included multiple NBS-LRR R-proteins . Finally,plastic nursery plant pot the approximate 95% Bayes confdence interval for the genomic location of FW3 on chromosome 1A was slightly wider than that observed for the other mapped loci because the effect of the locus was weaker. There were 535 annotated genes within that interval, of which seven were predicted to encode NLR or other R-proteins . This was the locus with the weakest support for the segregation of a race-specific R-gene; however, as noted earlier, homozygous resistant of spring in the Wiltguard S1 population were highly resistant . Hence, even if FW3 does not encode a race-specific R-protein, this locus merits further study, in part because the favorable allele can be deployed and pyramided to increase the durability of resistance to Fusarium wilt.To accelerate the introduction and selection of Fusarium wilt resistance genes in breeding programs, we developed a collection of high-throughput Kompetitive Allele Specifc PCR markers for SNPs in linkage disequilibrium with FW1–FW5 . Collectively, 25 KASP markers were designed for the five loci using PolyOligo 1.0 .
The genotypic clusters for 17 of these were codominant , co-segregated with the predicted resistance loci, and were robust and reliable when tested on diverse germplasm accessions . For each target locus, at least one KASP-SNP marker had a prediction accuracy in the 98-100% range when tested in the original populations where they were discovered . To further gauge their accuracy when applied in diverse germplasm, they were genotyped on 78 California and 66 non-California individuals, mostly cultivars . Because the causal genes and mutations underlying FW1–FW5 are not known, the SNPs we targeted are highly population-specific . They are strongly predictive when applied in populations where specific genes are known to be segregating and moderately predictive when assayed among random samples of individuals because of recombination between the SNP markers and unknown causal mutations.The deployment of Fusarium wilt resistant cultivars has become critical in California since the early 2000s when outbreaks of the disease were first reported . This disease has rapidly spread and become one of the most common biotic causes of plant death and yield losses in California, the source of 88-91% of the strawberries produced in the US . The scope of the problem was initially unclear, as were the solutions, because the resistance phenotypes of commercially important cultivars, genetic mechanisms underlying resistance, and distribution and race structure of the pathogen were either unknown or uncertain when the disease unexpectedly surfaced in California . A breeding solution instantly emerged with the discovery of FW1 , and was further strengthened with the discovery of additional homologous and non-homoeologous resistance genes in the present study . Genetic and physical mapping of these race-specific R-genes has enabled the rapid development and deployment of Fusarium wilt resistant cultivars through marker-assisted selection. The transfer of R-genes from race 1 resistant donors to susceptible recipients via MAS has been rapid because the resistant alleles are dominant, found in both heirloom and modern cultivars, and identifiable without phenotyping using SNP markers tightly linked to the causal loci . Once FW1 was discovered, we knew that we had a robust solution to the race 1 resistance problem; however, we had virtually no knowledge of the diversity of Fusarium wilt R-genes in populations of the wild octoploid progenitors and heirloom cultivars of cultivated strawberry that might be needed to cope with pathogen race evolution . We did not purposefully set out to identify redundant R-genes but rather to scour global diversity for ancestrally diverse R-genes, both to facilitate R-gene pyramiding and inform future searches for sources of resistance to as yet unknown races of the pathogen, in addition to assessing the frequency, diversity, and distribution of R-genes in the wild and domesticated reservoirs of genetic diversity . Our results paint a promising picture for the identification of genes for resistance to race 2 and other as yet unknown races of the pathogen. As our phenotypic screening studies showed, the frequency of resistance to race 2 was comparable to that observed for race 1 . Similar to our findings for race 1, the sources we identified for resistance to race 2 were symptomless, which suggests that gene-for-gene resistance might underlie their phenotypes. The genetic basis of resistance to race 2 and other races of the pathogen, however, has not yet been elucidated. There is empirical evidence that resistance to Australian isolates of the pathogen might be quantitative . Henry et al. showed that the non-chlorotic symptom syndrome caused by Australian Fof isolates differs from the chlorotic symptom syndrome caused by California and Japanese Fof isolates . Hence, the genetic basis of resistance to the wilt- and yellowsfragariae diseases could be markedly different. We identified several strong sources of resistance to Australian and other non-California isolates of the pathogen that should accelerate the discovery of novel race-specific R-genes, elucidation of genetic mechanisms, and development of resistant cultivars . Growing resistant cultivars is indisputably a highly effective and cost-free method for preventing losses to Fusarium wilt race 1 in strawberry . We estimate that approximately two-thirds of the cultivars grown in California since the earliest outbreaks in 2005 were highly susceptible, whereas the other one-third were highly resistant .