The plants for our experiments were artificially inoculated with race 1 or 2 isolates of F. oxysporum f. sp. fragariae using previously described protocols . The AMP132 isolate originated in California, whereas the MAFF727510 isolate originated in Japan . To produce spores, the pathogen was grown on potato dextrose agar or Kerr’s broth under continuous fuorescent lighting at room temperature, as previously described . Crude suspensions were passed through two layers of sterilized cheesecloth to remove hyphae. Spore densities were estimated using a haemocytometer and diluted with either sterile DI water or 0.1% water agar to a final density of 5 × 106 spores/ml. Seedling and bare-root plants were inoculated by submerging their root systems up to the crown in the spore suspension for 7–8 min prior to planting. The individuals in these studies were visually phenotyped for resistance to Fusarium wilt over multiple post-inoculation time points using an ordinal disease rating scale from 1 to 5 . For our field studies, individual plants were phenotyped once per week for four to eight consecutive weeks beginning in early June. Symptoms were observed on plants 26- to 36-weeks post-inoculation. For greenhouse and growth chamber studies, entries were phenotyped weekly for 6 to 12 weeks post-inoculation. For field, greenhouse,square black flower bucket and growth chamber experiments, the onset and progression of disease symptoms among resistant and susceptible checks were used as guides for initiating and terminating phenotyping.Our race 1 resistance screening experiments were conducted over a three year period at the UC Davis Plant Pathology Farm. The plants for these experiments were artiificially inoculated with the AMP132 isolate of the pathogen. Strawberries had not been previously grown in the fields selected for our studies.

The fields were tilled and disked prior to fumigation and were broadcast fumigated in October of each year with a 60:40 mixture of chloropicrin:1,3-dichloropropene at 560.4 kg/ha. The entire field was sealed with an impermeable plastic film for one-week post-fumigation before shaping 15.3 cm tall × 76.2 cm center-to-center raised beds. Sub-surface irrigation drip tape was installed longitudinally along the beds followed by black plastic mulch with a single row of planting holes spaced 30.5 cm apart. Artifcially inoculated plants were transplanted in mid-November both years. The fields were fertilized with approximately 198 kg/ha of nitrogen over the growing season and irrigated as needed to prevent water stress. For the 2016–17 field experiment, 344 germplasm accessions were screened for resistance to AMP132 and were part of a study that included 565 germplasm accessions developed at UC Davis, which is hereafter identified as the ‘California’ population. The resistance phenotypes for the latter were previously reported by Pincot et al. . Collectively, 981 germplasm accessions were screened in the 2016-17 field study. These were arranged in a square lattice experiment design with four single-plant replicates per entry . The experiment design and randomizations of entries within incomplete blocks were generated with the R package agricolae . For the 2017-18 and 2018- 19 field experiments, 144 ‘host diferential panel’ individuals were screened for resistance to AMP132 . These individuals were arranged in a 12 × 12 square lattice experiment design with four single-plant replicates per entry as described above. Guardian, Wiltguard, and Earliglow S1 and 61S016P006 S2 populations were screened for resistance to AMP132 in the 2016-17 field study. Ninety-nine Guardian S1 and 98 Wiltguard S1 individuals were phenotyped and genotyped and 85 Earliglow S1 and 77 61S016P006 S2 individuals were phenotyped. Nine-month-old S1 or S2 plants started as seedlings and asexually multiplied bare-root plants of the parents were artificially inoculated with AMP132, transplanted to the field in March 2018, and visually phenotyped weekly for six to 11 weeks post-inoculation.

The 12C089P002 × PI602575 , PI552277 × 12C089P002 , and PI612569 × 12C089P002 full-sib families, 17C327P010 S1 family, and parents of these families were screened for resistance to AMP132 in greenhouse experiments at UC Davis. Two- to four-month-old seedlings of the progeny and bare-root plants of the parents were artificially inoculated with AMP132 and planted in February 2019 , June 2019 , or November 2019 into 10.2 × 10.2 × 15.2 cm plastic pots filled with 3 parts coir : 1 part perlite and phenotyped weekly for six to 12 weeks post-inoculation. Four uninoculated and four inoculated single-plant replicates of the parents were arranged in completely randomized experiment designs. The plants were irrigated with a dilute nutrient solution as needed to maintain adequate soil moisture. The 12C089P002 × PI602575 and PI552277 × 12C089P002 populations were genotyped with a 50K Axiom SNP array .We screened a host diferential panel for resistance to the MAFF727510 isolate of Fof race 2 in a growth chamber at the UC Davis Controlled Environment Facility in 2018-19. Two single-plant replicates/individual were arranged in a randomized complete block experiment design. The entire experiment was repeated twice, resulting in four clonal replications/individual. The bare-root plants for these experiments were produced in high-elevation nurseries, preserved in cold storage, artificially inoculated with the MAFF727510 isolate, transplanted into 10.2 × 10.2 × 15.2 cm plastic pots filled with a 4 parts sphagnum peat moss : 1 part perlite , and phenotyped weekly for six to 12 weeks post-inoculation. The plants were grown under a 12-hour photoperiod with a 20 °C night temperature and 28 °C day temperature and irrigated with a dilute nutrient solution as needed to maintain adequate soil moisture. Because these experiments utilized a non-California isolate of the pathogen, the experiments were quarantined and conducted in compliance with federally-mandated bio-safety regulations .

DNA was isolated from newly emerged leaves harvested from field grown plants using a previously described protocol . Leaf samples were placed into 1.5 ml tubes or coin envelopes and freeze-dried in a Benchtop Pro . Approximately 0.2 g of dried leaf tissue/sample was placed into wells of 2.0 ml 96-well deep-well plates. Tissue samples were ground using stainless steel beads in a Mini 1600 . Genomic DNA was extracted from powdered leaf samples using the E-Z 96®Plant DNA Kit according to the manufacturer’s instructions. To enhance the DNA quality and yield and reduce polysaccharide carry-through, the protocol was modifed by adding Proteinase K to the lysis bufer to a final concentration of 0.2 mg/ml and extending lysis incubation to 45 min at 65 °C. Once the lysate separated from the cellular debris, RNA was removed by adding RNase A. The mixture was incubated at room temperature for 5 min before a final spin down. To ensure high DNA yields, the sample was incubated at 65 °C for 5 min following the addition of elution bufer. DNA quantification was performed using Quantifor dye on a Synergy HTX . The individuals phenotyped in these studies were genotyped with either 50K or 850K Axiom® SNP arrays . SNP markers on the 50K Axiom array are a subset of those on the 850K Axiom array. The probe DNA sequences for SNP markers on both arrays were previously physically anchored to the ‘Camarosa’ and ‘Royal Royce’ reference genomes . The ‘Camarosa’ genome assembly has been deposited in the Genome Database for the Rosaceae and Phytozome . The ‘Royal Royce’ genome assembly has been deposited in the Genome Database for the Rosaceae and Phytozome . The assemblies for each ‘Royal Royce’ haplotype have been deposited in a Dryad repository . The physical addresses for the SNP markers are provided in our online resources . We utilized both reference genomes as needed to cross-check and compare statistical findings and search genome annotations. The results presented in this paper utilized the haplotype-resolved ‘Royal Royce’ reference genome FaRR1 unless otherwise noted. SNP genotypes were called using the Affymetrix Axiom Suite . Samples with call rates exceeding 89-93% were included in genetic analyses. The haplotypes for 71 50K Axiom array-genotyped SNPs within a 1.60 Mb haploblock on chromosome 2B were imputed and phased using BEAGLE software version 5.3 for 651 individuals phenotyped for resistanceto Fusarium wilt race 1. The individuals were classified as resistant or susceptible ,square black flower bucket wholesale where ̄y is the estimated marginal mean for resistance score calculated from replicates. We ran BEAGLE with 10 burnin iterations for imputation and 25 subsequent iterations for phasing on sliding windows of 5 Mb with a window overlap of 2 Mb.The R package lme4 was used for linear mixed model analyses of the germplasm screening experiments . LMMs for square lattice experiment designs were analyzed with entries as fixed effects and incomplete blocks, complete blocks, years, entries × years, and residuals as random effects . LMMs for randomized complete block experiment designs were analyzed in parallel to estimate the relative efficiency of the square lattice to the randomized complete block experiment designs .

We did not observe an increase in efficiency by using incomplete blocks; hence, the statistics reported throughout this paper were estimated using LMMs for randomized complete block experiment designs . Estimated marginal means for entries were estimated using the R package emmeans . Genome-wide assocation study analyses were carried out to search for the segregation of loci affecting resistance Fusarium wilt races 1 and 2 among individuals genotyped with either the 50K or 850K Axiom SNP array . GWAS analyses were applied to estimated marginal means for resistance phenotypes using physical positions of SNP markers in the ‘Camarosa’ and ‘Royal Royce’ reference genomes . SNP marker genotypes were coded 1 for AA homozygotes, 0 for heterozygotes, and -1 for aa homozygotes, where A and a are the two SNP alleles. GWAS analyses were performed using the GWAS function in the R package rrBLUP. The genomic relationship matrix was estimated from SNP marker genotypes for each population using the rrBLUP A.mat function . The genetic structure of the GWAS population was investigated using hierarchical clustering and principal components analysis of the GRM as described by Crossa et al. . To correct for population structure and genetic relatedness, a Q + K linear mixed model was used where Q is the population stratification structure matrix and K is the GRM . The first three principal components from eigenvalue decomposition of the GRM were incorporated into the Q + K model. Bonferroni-corrected significance thresholds were calculated for testing the hypothesis of the presence or absence of a significant effect. GWAS was repeated in the California population by ftting a SNP marker in LD with FW1 as a fixed effect using the rrBLUP::GWAS function .SNP markers with ≤ 5% missing data, high quality codominant genotypic clusters, progeny genotypes concordant with parent gentoypes, and non-distorted segregation ratios were utilized for genetic and quantitative trait locus mapping analyses. The linkage phases of the SNP markers were not known a priori. The arbitrarily coded SNPs in the original data were in mixed coupling and repulsion linkage phases. The linkage phases of the SNP markers were ascertained using pair-wise recombination frequency estimates, and recoded so that the 100% of the SNP markers were in coupling linkage phase. This was only necessary in the S1 populations. The recoded SNP markers were genetically mapped in S1 populations using phase-known F2 mapping functions. SNP markers were genetically mapped in full-sib populations using phase-known back cross mapping functions from the subset of SNPs that were heterozygous in the resistant parent and homozygous in the susceptible parent. Genetic maps were constructed using the R packages onemap and BatchMap and custom PERL scripts for binning co-segregating SNP markers, calculating pairwise recombination frequencies, and grouping markers using LOD threshold of 10 and maximum recombination frequency threshold of 0.05. The custom PERL scripts are available in the Dryad repository for this paper . Linkage groups were aligned and assigned to chromosomes using inter-group linkage disequilibrium statistics and percent-identity against the reference genome . Marker orders and genetic distances were estimated in parallel using the RECORD algorithm in Batchmap with a 25-marker window, window overlap of 15 markers, and ripple window of six markers . For smaller linkage groups, the window size was reduced incrementally by five to ensure at least two overlapping windows. We used the checkAlleles, calc.errorlod, and top.errorlod functions of the R package qtl and custom R scripts to identify and eliminate spurious SNP markers and successively reconstruct linkage groups as described by Phansak et al. . Genetic distances were estimated from recombination frequencies using the Kosambi mapping function .