Heterozygous recombinant BC6S1 individuals were allowed to self-pollinate and progeny were marker-selected to obtain homozygous BC6S2 sub-NILs. These plants underwent self-pollination to obtain ample BC6S3 seed for replicated field experiments. We evaluated 62 BC6S3 sub-NILs in the 2009 field experiments. In the 2010 field experiments, a subset of 42 of the 62 subNILs was evaluated to allow increased replication per location while reducing genetic redundancy, as explained previously by Johnson et al. . Graphical marker genotypes for the 62 selected BC6S3 sub-NILs used in field experiments for the present study are presented in Supporting Information, Table S1.The chromosome 11 BC6S3 sub-NILs and the parental NIL from which they were derived were evaluated in replicated experiments at field locations in Salinas, California and in Davis, California over 2 yr. Summer and early fall in Salinas are generally cool and humid, which is conducive to late blight disease development, whereas Davis summers are warm and dry, with no rain, as is typical of California’s Central Valley tomato production areas. Additional information about field sites has been described by Haggard et al..Sixty-five genotypes and 45 genotypes were included in the 2009 and 2010 experiments, respectively.
Experiments were arranged in randomized complete block design . For both years,vertical hydroponic garden one plot per genotype per block was included, except for controls, for which there were two plots per block. In 2009, three blocks per location were used. In 2010, use of a reduced number of 42 sub-NILs enabled replication to be increased to five blocks in locations 1 and 2 and to four blocks in locations 3 and 4. At each of the four locations, each plot consisted of five plants spaced 0.30 m apart in rows separated by 1.02 m in locations 1 and 2, and by 1.52 m in locations 3 and 4. Border rows and plots with the cultivar E6203 surrounded each experiment at each location to reduce edge effects on the experimental plots. Standard horticultural field practices for processing tomato were used at all locations. Locations 1 and 2 were sprinkler irrigated, whereas locations 3 and 4 were furrow-irrigated, as needed.All traits were evaluated on a per-plot basis, as described by Haggard et al. . Vegetative horticultural traits were evaluated in all four locations. Late blight disease was only evaluated in Salinas because, as expected, this disease did not occur in Davis due to typical warm, dry summer conditions. Reproductive traits were only evaluated at Davis due to logistics of timely sampling of ripe fruit. Vegetative horticultural traits measured were plant height and width in cm, canopy density , and plant habit . H, W, CD, and HAB were obtained at both locations at 71 and 46 days after planting in 2009 and 2010, respectively. At locations 3 and 4, these traits were evaluated at 80 DAP in 2009 and at 68 and 73 DAP in 2010. From plant height and width, two secondary traits were derived, plant size and plant shape .
The reproductive horticultural traits measured or scored were as follows. DAP to maturity was evaluated at two stages of maturity: when each plant in the plot had its first ripe fruit and when 50% of fruit in a plot were ripe . Weight of 30 ripe fruit was evaluated when 50% of fruit in a plot were ripe . Yield in kg was evaluated when 95% of the fruit in a plot were ripe. Ripe fruit were used to obtain the weight of 100 seeds , which was measured only in 2009 due to labor limitations. The ripe fruit quality traits pH and Brix were measured using a pureed sample of 10 whole fruit obtained from plots with 50% ripe fruit using a pH Testr2 and a Reichert AR200 digital refractometer , respectively. Size traits obtained on ripe fruit were perimeter , width , and height . These traits were measured on flatbed scanner images of eight longitudinally sliced fruit per plot using Tomato Analyzer software , which refers to fruit length as height and fruit longitudinal circumference as perimeter. From FH and FW, the secondary variable fruit shape was obtained. Trait names, abbreviations, and brief descriptions are provided in Table 1. On 15 September 2009, Salinas locations 1 and 2 were inoculated with a local P. infestans isolate as described in Johnson et al. . In 2010 in mid-September, a natural P. infestans infection occurred in both locations, precluding the need for inoculation. As detailed in Johnson et al. , phenotypic scoring of late blight disease symptoms was performed visually and symptom data were used to calculate area under the disease progress curve for foliar and stem disease symptom progression . Lower AUDPC values indicate less disease symptom progress and therefore are indicative of increased disease resistance.A linkage map for the chromosome 11 introgressed region was constructed using DNA marker genotype data across 21 loci for 852 BC6S1 progeny .
The map was constructed with JoinMap 3.0 using the Kosambi mapping function and a 3-LOD significance threshold. We used a comparison of marker locations between our map and to the S. lycopersicum genome sequence to estimate the physical size of the S. lycopersicum region replaced by the S. habrochaites introgression. QTL mapping for each trait was performed with the composite interval mapping module in WinQTLCartographer 2.5 using sub-NIL means obtained from ANOVA for each trait. QTL mapping was performed using CIM Model 6 and the forward and backward regression method with a walk speed of 1 cM and a window size of 2 cM. Trait-specific permuted LOD thresholds were empirically established for each trait using 1000 permutations in WinQTLCartographer. A QTL for a trait was considered significant at P # 0.05 if the peak LOD value exceeded the trait-specific permuted threshold. Multiple QTL were declared for a single trait when the LOD values between significant peaks within the introgressed region decreased below the significance LOD threshold for at least two contiguous markers. Each significant QTL was denoted by trait name, location, and year. For example, DAP1st34_2009 is a QTL detected in the analysis of DAP1st data from locations 3 and 4 in 2009. A linkage map figure showing locations of significant QTL was constructed using MapChart2.1 . QTL locations were indicated as 1-LOD bars and 2-LOD whiskers . For easy reference and purposes of discussion, QTL were assigned to QTL trait groups based on coincidence of their 1-LOD support intervals. Although a few of the QTL had 1-LOD support intervals that extended beyond the boundary of their assigned group, their peak locations supported their placement in these groups. Comparisons were made among QTL for disease resistance traits and horticultural traits for QTL coincidence by visual inspection of their chromosomal locations on the linkage map. A statistical test based on the hypergeometric probability distribution was used to calculate QTL correspondence,vertical home farming the probability of obtaining the observed number of matching QTL by chance. A QTL match was declared when the one-LOD support intervals overlapped. The number of comparison intervals was six, based on the average size of our QTL and the overall map distance of the introgression . Our QTL locations were also compared with those previously reported for both disease resistance and horticultural traits on chromosome 11 in tomato and in potato, based on common markers as well as genomic sequence data for both crop species. Sources used for QTL location comparisons included the following: tomato ; potato ; and genomic sequences . When common markers were not available, the Tomato-Expen 2000 map available on the Sol Genomics Network was used to facilitate map alignment.Truncation selection was applied sequentially for LEAF, YLD, FP, and 30Wt to identify breeding lines potentially useful for development of tomato varieties with improved resistance to P. infestans. Out of42 sub-NILs, the first round of truncation removed 9 lines with leaf resistance scores below that of E6203 in 2 years or locations. The second round removed 2 lines with YLD ,66% of E6203 in 2 years or locations, whereas the third round removed 7 lines with FP ,92% of E6203 in 2 years or locations. The final round removed 2 lines with 30Wt ,80% of E6203 in 2 years or locations. Maturity was also considered; however, two lines with significantly later maturity than E6203 in at least 1 year or location were selected due to their relatively high levels of foliar resistance to P. infestans . At the end of the process, 11 lines were chosen.There were significant differences among genotype means for all traits, except for fruit pH in 2009 . In general, subNILs with S. habrochaites introgressions at the marker loci At5g04590 and TG400 were significantly later maturing than control cultivar E6203 in at least one trait and year or location combination; however, some lines were exceptions to this trend.
Most sub-NILs with S. habrochaites introgressions at the marker loci At5g04590 and TG400 also had significantly reduced FH compared with E6203. Relative to E6203, NIL11 exhibited significantly greater foliar resistance to P. infestans , but only in 2010. NIL11 also had later maturity , reduced FH , FP, 30Wt, SW, and W, increased H and pH , taller, narrower SH, and more prostrate habit than E6203. Sub-NILs 08GH3723, 08GH3999, 08GH4018, 08GH4106, and 08GH8032 displayed significantly greater foliar resistance than E6203 in 2010; however, only 08GH8032 performed significantly better than E6203 in 2009. 08GH4228 had significantly better stem resistance than E6203, but only in one location in 2010. None of the lines with significantly greater resistance showed any significant decrease in YLD or BRIX; however, several of them had significantly delayed maturity , reduced fruit size and 30Wt, and more upright HAB.None of the QTL in Hort 11-1 was detected in all years and locations. The fruit size QTL in this group were identified at both Davis locations, but only in 2010. Similarly, the Brix QTL was mapped only in 2009. The maturity QTL in this group were only detected in a single year and location, suggesting an environmental influence on QTL expression. Similarly, none of the QTL in Hort11-2 was identified in all years and locations. The two QTL for SW were detected in both locations; however, data were only collected for this trait in 2009, so their stability over the years is unknown. Other QTL, such as those affecting Brix, pH, and FP were mapped in both locations, but only in a single year. QTL for plant size were detected over both years, but only at the two Salinas locations in 2009 and only at a single location in 2010. Their effect may have been accentuated by the narrow row spacing, relative to that at the Davis locations. QTL for SH were detected in this group with opposite effect depending on location, another manifestation of QTL · E. In 2010, the S. habrochaites allele conferred a shorter, wider phenotype at the Davis locations, but a taller, narrower phenotype at the Salinas locations. This may also be due to the differences in row spacing between the two pairs of experiments. The difference in LOD peak location and only minor overlap between these two QTL may also indicate that these are separate, linked loci of opposite effects that are most pronounced in contrasting environments. In the Hort11-3 QTL group, the DAP50, FH, and FP QTL appear to be stable, being identified in each year and location. However, the LOD peaks for the FH and FP QTL over the 2 yr are sufficiently far apart to suggest that there may be multiple linked loci contributing to the effects for each trait. QTL for H were mapped at both Davis locations in both years but only at one of the Salinas locations, and only in 2010. QTL for CD were also relatively stable, being detected in three of the four year/location combinations; however, an additional CD QTL was found at location 4 in 2009, distal to those three QTL. QTL for FW, FS, and pH were identified in both locations in 2009, but not in 2010. The HAB QTL was mapped only in the Davis locations in 2010 and was not detected in 2009.