3.8 Fitness costs and gains associated with VfmB alleles
We used the same eight isolates carrying either VfmBPro(IPO2222, MIE35, AM3a and 3337) or VfmBSer (Ds0432.1, RNS10-27-2A, Sp1a and M21a) to constitute VfmBPro and VfmBSer experimental populations for comparing their fitness in plant assays.
In soft-rot assays (Figure 8a-b ), each of the two experimental populations and their mix was inoculated on 10 tubers and a symptom class was assigned to each tuber. Two inoculation loads were used (107 CFU or 105 CFU per tuber). This assay was performed in triplicate (3 x 10 tubers per treatment). Kruskal-Wallis tests revealed differences between the three treatments, i.e., inoculation by D. solani VfmBPro, D. solani VfmBSer and their mixture, and the differences were significant at both high (k=17.6; DF=2; p=1.5x10-4) and low (k=7.2; DF=2; p=0.03) pathogen loads. Whatever the bacterial load, pairwise comparisons (Post-hoc Tukey tests) showed that symptom severity was different between D. solani VfmBPro and D. solaniVfmBSer (at a high load: F=4.6; p=9.5 x 10-6; at a low load: F=0.8; p= 0.01). The D. solani VfmBSer population was more aggressive than theD. solani VfmBPro population (DSI mean value ± SE of 92 ± 2 for VfmBSer vs 73 ± 3 for VfmBPro at a high load and 45 ± 2 for VfmBSer vs 26 ± 9 for VfmBPro at a low load; Figure 8a-b) . VfmBSer D. solanistrains thus more efficiently damaged tubers than VfmBPro D. solani strains.
When the VfmBSer and VfmBProD. solani experimental populations were co-infected in tubers, shotgun sequencing of the pathogens recovered from symptoms allowed to calculate CI values (Figure 8a-b ). With a load of 107 CFU per tuber, the CI values (8 values; median = 1.3) were not different from one (Kruskal Wallis test; k=0.8; DF=1; p= 0.4). With a load of 105 CFU per tuber, the CI values (3 values; median = 2) differed from one (Kruskal Wallis test; k=4.4; DF=1 ; p= 0.04). In tubers, the vfmB allele thus appeared neutral or to confer a moderate advantage as compared to the VfmBPro allele.
In blackleg symptom assays (Figure 8c ), the D. solaniVfmBPro and D. solani VfmBSerexperimental populations and their mix were inoculated on non-wounded plants and the number of symptomatic and asymptomatic plants were counted. The assay was performed in triplicate: two assays with 25 plants and one assay with 15 plants per treatment. Kruskal-Wallis tests revealed no difference between the three treatments, i.e., the D. solani VfmBPro strain mixture, the D. solaniVfmBSer strain mixture and their mix (k=0.29; DF=2; p=0.86). Shotgun sequencing of bacteria recovered from symptoms allowed calculation of CI values in 16 co-infected plants (Figure 8c ). These CI values were different from one (Kruskal-Wallis test; k=6.6; DF=1; p= 9 10-3) with a median CI of 14, showing a fitness advantage of the VfmBPro allele in stem symptoms.
Altogether, the plant assays indicated that the VfmBSerallele was associated to a higher aggressiveness in the rotted tubers compared to the VfmBPro allele. This suggested that the VfmBSer allele could more efficiently initiate maceration of tubers in the field and storage conditions. In contrast, the VfmBSer allele seemed to confer a lower competitive fitness than the VfmBPro allele when D. solaniproliferates in rooted tubers and, particularly, in stem symptoms. Such opposite gains across different conditions could contribute to the maintenance of balanced frequencies of the two alleles in D. solani populations.