Populations Sonora x CBdeM , Sonora x Foisy , and CBdeM x Foisy have 146, 141, and 128 lines respectively. The DH lines were phenotyped using a modified cigar roll method of Zhu et al. . The system is similar to the Cyg germination growth pouches and the gel based system of Bengough et al. . It consists of two plexi-glass plates 20 cm x 30 cm fitted with spacers, germination paper, racks holding the plates upright and tubs used to hold water. One hundred seeds were counted and weighed to estimate average seed weight. Seeds of similar size and weight for each genotype were imbibed in water for 24 hr prior to planting. Germination paper wetted with deionized water was placed on one of the two plates, and two seeds of the same genotype were placed with embryos down 5 cm below the top edge of the paper and 8 cm apart. This set up was covered by a second sheet of wet germination paper and a second sheet of plexiglass. The entire set-up was clipped together and placed upright into tubs of water about 8cm deep; this water level was maintained constant throughout the experiment. Seedlings were grown for 7 days at room temperature without supplemental lighting in a head house at the University of California, Riverside between February 2014 to May 2014 and November 2014 to February 2015. Experiments were setup in a randomized complete block design with four replications where replications were treated as blocks. Each replication had two plants from every genotype. After 7 days plates were removed from the tubs, disassembled, 4x8ft rolling benches and seminal roots were imaged using a hand held digital scanner set to 300 DPI.
To acquire images, the top sheet of germination paper was carefully removed and the top plexi-glass plate was laid over the seedlings so the scanner could be passed over from above. Seminal root angles were measured using the angle tool in ImageJ . For each plant, the angle between the first pair of seminal roots was measured at approximately 3cm below the embryo of the seed , as shown in Figure 1. The analysis of variance for seminal root angle and number was based on mean values of the experimental units. Broad sense heritability was calculated on a mean basis across four replications. Genotype means were used to calculate Pearson‟s correlation coefficients for seminal root number, seed weight, and seminal root angle. Genomic regions associated with traits of interest were detected by the software package IciMapping using linkage maps for these populations as described previously and the mean value of 8 seedlings of each genotype from four replicates. The composite interval mapping method with a step of 1cM was used and the threshold for QTL detection was determined using 1000 permutations where α = 0.05. Markers in the linkage maps were renamed using the index number provided by Wang et al. preceded by the chromosome designation. QTL consistent between years within populations and/or consistent between populations were considered as verified QTL and named according to McIntosh Catalogue of gene Symbols for Wheat . Following the format of previous publications an uppercase “Q” in the name signifies strong verification of the QTL and lowercase “q” was used to name QTL that were consistent but warrant further investigation. In most cases associations of root system characteristics with specific genome region varied between populations and within populations from one year to the next. Over two years of the experiment and with all three populations taken together, a total of 31 genomic regions showed statistically significant associations with the seminal root angle and number . Seminal root angle was associated with 12 chromosome regions in the SC population, located on chromosomes 2D, 3B, 4A, 5A, 6A, 6B, 6D, and 7B. In the SF population, five regions on 2D, 5B, 6B, and 7B were identified and another five regions were identified in the CF population, on chromosomes 5B, 6A, and 7A.
The chromosome region with the single largest effect for the seminal root angle was located on chromosome 2D in the SC population. Its estimated effect was equivalent to 7.33° of the total root angle, and it was responsible for 21.42% of the population variation. The region with the lowest, but statistically significant effect for root angle was identified in the CF population, accounting for an estimated 2.90° of the root angle and explaining 9.40% of the variation observed in this population. For the seminal root number, nine genomic regions were identified in the three populations. Of these, four were identified in the SC population, on chromosomes 4A, 5B and 7A. The SF population had only one region, on chromosome 4B. The remaining three regions were identified in the CF population on chromosomes 1B, 6B, and 7D. The region with the largest effect was on chromosome 4A in the SC population, with an estimated effect of -0.25 roots per seedling explaining 17.32% of the total variation. The region with the lowest but statistically significant effect was identified on chromosome 6B in CF, with an estimated effect of 0.15 roots per seedling, explaining 8.41% of the population‟s variation. For the purpose of this study, only those genome regions that showed consistent associations with specific traits within a given population over both years were considered as verified QTLs . In the SC population three such regions were identified, located on chromosomes 2D, 6A, and 7B . The region on chromosome 2D was 4.17 cM region with a peak at 113 cM between markers 2Dx_79444 and 2Dx_77420 in 2014. It accounted for 25.99% of the phenotypic variation seen in the population that year. In 2015, the region was located between markers 2Dx_32130 and 2Dx_79444 covering a 0.67 cM with a peak at 112 cM. That year it explained 21.42% of the phenotypic variation seen. The allele for wider seminal root angle was contributed by Sonora. The second QTL was located on chromosome 6A. In 2014 it was between markers 6A_72189 and 6A_55084 covering a 4.90 cM region with a peak at 151 cM. It explained 7.04% of the phenotypic variation that year. In 2015, this QTL formed a peak at 155 cM between markers 6A_55084 and 6A_21174, it coved 1.35 cM and explained 7.21% of the variation for the trait. The allele for wider seminal root angle was contributed by Sonora.Each of the three tested population showed large phenotypic variation for both seminal root traits measured in this study.
The largest range in seminal root angle was between Sonora and CBdeM with average seminal root angles of 108.73° and 63.31° respectively . The least difference, but still statistically significant,flood and drain table was between CBdeM and Foisy which have more similar seminal root angles of 63.31° and76.95° respectively. The distribution patterns among progenies imply considerable trait complexity.All three parents typically had five seminal roots with few variations between replication giving averages of 4.36, 4.38, and 4.49 seminal roots for SC, SF, and CF respectively. The occurrence of less than five seminal roots is likely explained by environmental interaction and associations with seed weight. Since all parents typically develop five seminal roots it is not surprising that the three populations have similar means and ranges. As will be discussed later, the lack of consistent QTLs for seminal root number may suggest that this trait is heavily influenced by the environment and seed weight. However, one consistent QTL was identified which also suggests that there is a genetic component as well. Additionally, heritability values were relatively high for both traits in all populations but it does not seem to promise any ease of selection for breeding efforts. As will be discussed it certainly doesn’t hint at simplicity for the genetics of these traits.The 90K SNP array was used on eight mapping populations of doubled haploids to order SNPs along individual chromosomes and 44,345 of those were mapped to one or more of 46,977 loci . Due to differences in polymorphism among different sets of parents, only a fraction of all mapped markers can be expected to be useful in any given pairwise combination. Moreover, as distribution of crossover can vary substantially between different pairs of parents the actual genetic map position of any given marker may also differ . To facilitate utilization of the maps generated using the 90K SNP chip, Wang et al. 2014 created a consensus SNP map of wheat, based on the tested eight populations. In essence, this map provides average marker positions for all polymorphic markers of their study and may be used to coordinate maps generated for different populations. As it was explained in an earlier chapter, total lengths of maps for each of the three populations here varied but more importantly, at times very few common markers were present in specific chromosome regions. For verified QTLs, that is for consistent associations between specific DNA markers and genome regions consistently showing up in replications, the consensus map was used to allocate those to specific regions and used DNA sequence data of the closest associated marker to blast against the wheat sequence survey on the URGI database and determine its actual location. In this fashion, relative locations of QTLs identified in this study can be compared to all previous results and can be verified in the future. This approach makes it possible to use even those DNA markers that were not polymorphic between two parents of a given population increasing the resolution of a mapping exercise. This study identified 31 genomic regions associated with seminal root angle and seminal root number in three populations. Most of these regions were unique to specific populations and varied from year to year. This implies that these traits are far from simple, as proposed by Oyanagi and do not appear to be controlled by single loci. It must be pointed out that compared to other studies on seminal root traits, the results presented here appear to be better supported by experimental data. Using a single population of 103 doubled haploids Hamada et al. were unable to identify a QTL for seminal root angle; two QTLs for deep root ratio appeared on chromosomes 1B and 5D. Another QTL, for seminal root, was found on chromosome 5A. None of the regions consistently identified in this study appear to be located on chromosomes of Hamada et al . In another study, Christopher et al. identified 12 QTLs for seminal root angle and number in a single mapping population of bread wheat consisting of 184 individuals. The QTLs for seminal root angle were located on chromosomes 2A, 3D, 5D, 6A, and 6B; those for seminal root number on chromosomes 1B, 3A, 3B, 4A, and 6A. While some chromosomes are the same as those identified in this study, none are on chromosomes verified as valid QTL in this study: 2DS, 6AL and 7BS for seminal root angle and 4BL for seminal root number. In another study Liu et al. again identified a total of 12 QTLs for seminal root angle and number. Seven of those, for seminal root angle were on chromosomes 1A, 2B, 3A, 3B, and 7D and five for seminal root number on 2B, 3B, 3D, 5A, and 7A. Again, there are some genome regions in common with this study but none appear to be are similar to our verified QTL. Most studies employ a single mapping population. Beavis demonstrated that in populations numbering 100 progeny, the QTL effects were greatly overestimated, in populations with 500 progeny the QTL effects were slightly overestimated while populations with 1000 individuals produced estimates close to the actual magnitude of QTL effects. That study highlighted the necessity for larger populations and the need for verification of QTL across populations. Beavis did not address the issue of mapping in parallel populations sharing common parents. To the best of our knowledge only a couple of studies made use of two or more populations in studying root system traits: Zhang et al. used three related recombinant inbred line populations with a single common parent and Kabir et al. used two unrelated populations. Zhang et al. identified QTLs for seminal root number on chromosomes 1D, 2A, 2B, 2D, 3A, 3B, 4A, 4D, 5A, 5D, 6A, 6B, and 7B.