DNA sequencing was performed using ABI BigDye Terminator according to the manufacturer’s protocol, except that 5-μl reactions were performed with 0.25 μl of BigDye on an ABI 3730xl DNA sequencing machine with 50 cm arrays. DNA sequences were analyzed with CodonCode Aligner v. 2.0.6 . We detected three types of polymorphism in our sequences – single feature polymorphism , insertions and deletions and variable number tandem repeats . Most of the SFPs that had been detected using the Affymetrix GeneChip were due to a single nucleotide polymorphism , but in five cases due to a single base indel. Since Haploview cannot handle missing values, missing bases were substituted prior to data analysis with an appropriate single nucleotide. Because all single-base indels could be tagged with SNPs from the same marker locus , we use the term SNP throughout the text. Both indels and VNTRs were excluded from data analysis, unless otherwise noted in the text.One hundred and ninety two RILs derived from a cross between an F1 of cv. Valmaine × cv. Salinas 88 and cv. Salinas were genotyped with EST-derived markers. Selection of markers for this first round of genotyping was based on the molecular linkage map developed from an interspecific cross between L. sativa cv. Salinas and Lactuca serriola accession UC96US23. Twenty markers were selected to evenly cover linkage group 2 in intervals of approximately 10 to 20 cM. After preliminary mapping of the resistance gene, the region containing Tvr1 was saturated with markers originating from a microarray-based study also carried out on the Salinas × UC96US23 population.

Marker polymorphism was tested with HRM analysis,25 liter pot plastic unless the difference between segregating alleles could be visually observed using gel electrophoresis. If polymorphism could not be observed with HRM analysis, PCR products from the two parental genotypes were sequenced and new primers were designed for HRM. Statistical analysis of the linkage between molecular markers and die back resistance was performed by MapManager QTX software. Dieback resistance for each RIL was considered as a bi-allelic qualitative trait and used for linkage analysis.Association mapping was performed on a set of 68 accessions from seven horticultural types of lettuce . In the first step, markers closely linked to the Tvr1 gene were amplified from each accession and sequenced. In the second step, the sequenced amplicons were analyzed for polymorphism with the CodonCode software and inputted into Haploview v. 4.2. Intra-locus SNPs were tagged in Haploview with the Tagger function at r2 = 1. Untagged SNPs from all markers and a representative SNP for each tag were then entered into TASSEL v. 2.0.1. TASSEL was subsequently used to test for association between individual SNPs and resistance to die back while accounting for the population structure. Both p-values for each SNP and percent of phenotypic variation explained by the model were calculated with TASSEL after 100,000 permutations. Prior to association analysis, the population structure in the set of 68 accessions was assessed with thirty EST-SSR markers distributed throughout the genome using the computer program STRUCTURE 2.2. Ten runs of STRUCTURE were done by setting the number of populations from 1 to 15. For each run, the number of iterations and burn-in period iterations were both set to 200,000. The ad hoc statistic was used to estimate the number of sub-populations. The optimum number of sub-populations was subsequently used to calculate the fraction of each individual’s genome that originates from each of the five sub-populations.

The qk values obtained from STRUCTURE were used as covariates in the statistical model given by TASSEL.The level of genetic variation at the nucleotide level was estimated as nucleotide polymorphism and nucleotide diversity . To test the neutrality of mutations, we employed Tajima’s D test, which is based on differences between π and θ. Analyses of genetic variation and estimates of haplotype diversity were carried out using DnaSP v. 5.00.04 software. Linkage disequilibrium between pairs of SNP loci in the genome was calculated with Haploview and the values were pooled over the entire data set. Decay of LD with distance was estimated using a logarithmic trend line that was fitted to the data. Distances between markers were calculated from their respective positions on the consensus molecular linkage map. The consensus map was created with JoinMap v. 2.0 from the Salinas × UC96US23 map and the × Salinas map . SNPs with frequency < 5% were excluded from the analysis.The resistance-SNP association observed in the set of 68 accessions was detected through sequencing of PCRamplicons from individual accessions. In order to accelerate and simplify the test of association, we developed a primer pair that allowed detection of polymorphism in the marker Cntg10192 through high-resolution melting analysis. These primers amplify a 185 bp product that contains all three SNPs detected in the marker Cntg10192 at the positions 54, 72, and 100. The first two SNPs match perfectly with the resistance allele, while the third SNP explains 40.9% of the trait variation. As with the first two SNPs, the third SNP has a C ⇔ T substitution. All susceptible genotypes carry the T allele, while resistant genotypes have either the T or C alleles at the third SNP. It appears that the T allele in the resistant material is associated with the resistance present in cv. Salinas and most of the other iceberg cultivars, whereas the C allele is associated with the resistance present in the three lines that originate from the romaine-like primitive accession PI491224.

Marker Cntg10192, therefore, not only allows for the detection of alleles associated with die back resistance, but also separates alleles of different origins. To further investigate polymorphism in this genomic region we sequenced two accessions from L. serriola, a wild species closely related to cultivated lettuce. One of the accessions is resistant to the disease, while the other one is susceptible.The resistant accession has a haplotype similar to cv. Salinas but instead of the T allele at position 54, it carries the C allele. The three SNPs at the marker Cntg10192 can thus distinguish four different haplotypes; three resistant and one associated with susceptibility . Haplotype R1 has the T-T-T allele combination at positions 54, 72, and 100. Haplotype R2 carries the T-T-C combination, while haplotype R3 carries the C-T-T alleles. Disease susceptibility was always associated with the S1 haplotype that carries the C-C-T combination. All four haplotypes can easily be separated through high-resolution melting analysis .Nucleotide polymorphism was observed in all nine markers that were sequenced from the region flanking the Tvr1gene. The rate of nucleotide substitutions in a set of 68 accessions translates into ~1 SNP per 149 bp between pairs of randomly selected sequences. This SNP frequency was somewhat lower when only coding regions were considered . These values are well within the range observed for other plant species. For example,25 litre plant pot the average SNP frequency is 60 bp in aspen, 87 bp in potato, 104 bp in maize, 130 bp in sugar beet, 232 bp in rice, 435 bp in sorghum, 585 bp in tomato, and 1030 bp in soybean. Both nucleotide polymorphism and nucleotide diversity of lettuce are similar to that observed in maize , potato , and sugar beet , but larger than in tomato , and soybean. If results from the analyzed region correspond to those for the whole genome, sequence variation in lettuce is relatively high for a selfing species. It was previously observed that selfing species have generally lower levels of sequence variation than outcrossing species because of smaller effective population sizes. Although polymorphism in lettuce appears to be considerably larger than in selfing soybean and tomato, it is similar to that observed in rice, which is also a self-pollinating species. The ratio of nucleotide diversity in coding and non-coding sequences was not analyzed in detail, since data from only four markers are available. However, the ratio haplotype frequencies, and calculated using the across these markers appears to be smaller than in DNAsp software.

This measure of diversity is analogous to Arabidopsis , soybean the heterozygosity at a single locus, and is at its maximum , maize , and potato . when haplotypes observed in the sample occur at equal This difference is likely due to a higher level of functional frequencies. Diversity based on haplotypes ranged constraint on the perigenic sequence of lettuce. Meas- from 0.593 in QGG19E03 to 0.809 in marker Cntg11275, ures of haplotype diversity were based on estimated with an average value of 0.732 ± 0.024. These values are higher than in rice , soybean , and human . It is possible that the high level of diversity is related to the way that selection of the 68 accessions was performed. We included die back resistant and susceptible material from all predominant horticultural types, thereby selecting haplotypes at similar frequencies. It would be interesting to observe how haplotype diversity changes in different genomic regions and/or for a different set of accessions. To test the neutrality of mutations, Tajima’s D was calculated for all surveyed markers. The average Dwas larger than in soybean , potato , and sorghum . A positive D value indicates a deficit of low-frequency alleles relative to what is expected. Since large D values can be caused by a population subdivision, it is possible that the presence of sub-populations in the analyzed set of lettuce accessions affects both haplotype diversity and the D values. When neutrality of mutations was tested in individual markers, three markers closely linked to the Tvr1 gene had Tajima’s D values significantly higher than expected . Again, the population structure or selection at the Tvr1 locus or the marker itself could have caused departures from neutrality.The decay of LD for the Trv1 region was relatively slow when measured both within individual markers and between markers flanking Tvr1. Estimated values of r2 were ~0.322 at 900 bp, and ~0.124 at 8 cM. A fitted logarithmic curve shows that the r2 value of 0.2 is reached somewhere between 0.5 cM to 1 cM. LD of SNP markers observed in some other selfing species was similar; LD in Arabidopsis was 250 kb or 1 cM and in soybean was ~50 kb. Intra-locus LD decayed very little in tomato, with the log trend showing r2 > 0.6 at 900 bp. However, it is problematic to compare decay of LD across species due to the large variability in LD quantification. LD depends on a combination of many factors, such as the origin of the population, selected set of accessions, analyzed genomic region, molecular marker system, and presence of unidentified sub-populations. Hyten compared four different soybean populations for levels of LD decline. While in the domesticated Asian G. max population LD did not decline along the 500 kb sequenced region, the wild Glycine soja population had a large LD decline within the LD block size averaging 12 kb. Comparable observations were not only made in the selfing Arabidopsis, but also in the out crossing maize and aspen. Our results show a large difference between estimates of LD when analyses were performed across all horticultural types or within each individual type. While the estimate of r2 at a distance of 8 cM was 0.124 for the whole set, it was 0.247 for romaine type and 0.345 for crisp head lettuce. Because only a relatively small part of the genome was analyzed in the present work, it is not possible to calculate LD at distances over 8 cM. However, the trend for the logarithmic curve suggests that LD could reach more than 15 cM in romaine and probably more than 25 cM in crisp head types before declining to the value of r2 < 0.2. When only iceberg types were included in the analysis, LD was still at its maximum at a distance of 8 cM . Although these observations come from a limited number of individuals, they are supported by the fact that the modern iceberg-type lettuce has an extremely limited genetic diversity that is frequently associated with extensive LD.We identified the genomic region carrying resistance against die back and nine markers closely linked with the Tvr1 gene through linkage analysis. We subsequently used this information to test the linked markers for association with the disease resistance on a set of 68 diverse accessions.