3.4 | Relationship between multilocus genotype and traits related to quantitative pathogenicity
Analysis of multilocus genotypes was conducted on the 63 isolates individually sequenced across the 28 regions in which haplotypes could be defined. No significant multilocus disequilibrium was detected in the two populations considering the two haplotypes previously defined across the 28 regions (0.008 > \(\overline{r}\)D> 0.009), the 12 regions only associated with DLA-R (0 > \(\overline{r}\)D > 0.007), or the eight regions only with DLA-S (0 >\(\overline{r}\)D > 0.009). Furthermore, based on the same haplotypes, no significant pairwise linkage disequilibrium (LD) was detected between the 28 regions in the two populations (CU1S2 and CU1R2). However, the number of advantageous haplotypes in the resistant cultivar accumulated per isolate ranged from 2 to 10 (out of 12 possible advantageous haplotypes) and the correlation of this number with the level of aggressiveness in the resistant cultivar (measured with DLA-R) was positive (Spearman’s correlation coefficient ρ=0.39, p=2.21e-16). On the other hand, the number of advantageous haplotypes in the susceptible cultivar accumulated per isolate ranged from 0 to 7 (out of 8 possible advantageous haplotypes) and the correlation of this number with the level of aggressiveness in the susceptible cultivar (measured with DLA-S) was also positive (Spearman’s correlation coefficient ρ=0.38 and p=4.093 e-9). These results suggested that the most aggressive isolates in both cultivars tend to accumulate more advantageous haplotypes in the corresponding cultivars across the candidate regions but in different combinations.
The frequencies of haplotype 1 or haplotype 2 defined from individual sequencing in the 17 genomic regions associated with DLA-R and/or DLA-S (Table 3) were estimated in the 14 study pools with the software harp (71, Table S4). This program uses an expectation-maximization (EM) algorithm to infer the maximum-likelihood estimated frequencies of a known haplotype in a pool of individuals. To validate this approach, we first compared the frequencies of the two haplotypes between estimates resulting from individual and pool sequencing in samples from location 1 for which both kinds of data were available (Chi2 test). Haplotype frequencies differed significantly in pool and individual sequencing in 4/17 regions (S1R3-Cu, S2R3-Cu, S4R2-Cu, S4R4-Cu). Visual examination of the alignments from the sequencing of individuals revealed a large number of recombinant haplotypes, missing data, and/or point mutations in these regions that could have biased our estimation of haplotype frequencies in pools. These four regions were consequently not used for further analysis. A significant difference in haplotype frequencies (Fisher’s exact test) was detected between population pairs in some of the 13 remaining regions and in some of the locations, indicating high heterogeneity between populations but a tendency was nevertheless observed (Table 4). For the seven regions correlated with DLA-R, haplotype 1, which was advantageous in the resistant cultivars, was always significantly more frequent in the populations of this cultivar sampled, with one exception (region S8R1-Cu). In the four regions correlated with DLA-S, the advantageous haplotype 2 in the susceptible cultivar was always more frequently found in the populations of this cultivar sampled. In the S2R1-Cu region correlated with both DLA-R and DLA-S, haplotype 1 was more aggressive considering both traits but was found to be more frequent only in the population of the resistant variety sampled in location 3 in 2013. Fot the region S9R3-Cu haplotypes 1 and 2 were considered to be advantageous in the resistant and susceptible cultivar, respectively. In this region in location 1 in 2011 and 2013, haplotypes 1 and 2 were significantly more frequent in populations of resistant cultivars and susceptible cultivars, respectively, suggesting diversifying selection. Overall, our results concerning multilocus haplotypes suggested a low degree of convergence between the different locations, and the regions involved in the two traits are not necessarily the same from one location to another. However, some populations accumulated several haplotypes across regions that are advantageous in their cultivar of origin.