Allelic-sensitivity vs gene-regulatory model
Sixty-three plasticity QTLs were identified among which 22 (35%) were also identified when using the genotypic means; and 41 (65%) were specific to plasticity. Via et al. (1995) proposed two genetic models – the allelic-sensitivity and gene-regulatory models – among the mechanisms involved in the genetic control of phenotypic plasticity. These two models are distinguishable through QTL analysis (Ungerer et al., 2003) with the expectation that allelic-sensitivity model will lead to co-localization of genotypic means and plasticity QTLs, while a distinct location of QTLs affecting mean and plasticity will likely correspond to the gene-regulatory model (Kusmec et al., 2017). Regarding our results, both models are suspected to regulate tomato plasticity, even though the gene-regulatory model is predominant with 65% of the plasticity QTLs that did not co-localize with genotypic means QTLs for the same trait. In maize, using a larger number of environments and traits, Kusmec et al. (2017) found similar results and even a higher rate of distinct locations of plasticity and mean QTLs. Studying plasticity as a trait per se is therefore of a major interest since breeding in both direction (considering the mean phenotype and its plasticity) is achievable. Through transcriptomic analyses, Albert et al. (2018) observed that genotype x water deficit interaction was mostly associated to trans- acting genes which could be assimilated to the gene-regulatory model in agreement with our results.
Although the distinct location of QTLs detected on plasticity and genotypic means could be confidently assigned to the action of genes in interaction, their co-localization is not necessarily a case of allelic-sensitivity regulation, especially if the QTL is in a large region. Indeed, the allelic-sensitivity model assumes that a constitutive gene is directly sensitive to the environment regulating its expression across different environmental conditions, inducing hence phenotypic plasticity. This is a very strong hypothesis regarding the QTLs since the overlapping region between genotypic means and plasticity could carry different causal variants in strong linkage disequilibrium affecting either mean phenotype or plasticity. Thus, co-locating mean and plasticity QTLs should be not automatically imputed to the allelic-sensitivity model. We found a total of 22 constitutive QTLs between genotypic means and plasticity for all 10 measured traits (Supplemental Table 4). Considering the estimated QTL effects, the variation patterns of the eight parental allelic classes were compared between mean and PP QTL of the same trait. Only ten QTL showed consistent allelic effects (Spearman correlation significant at 0.05 threshold level) strengthening the hypothesis of the allelic-sensitivity model for these QTLs (Figure 6). Further studies should help to elucidate and validate the candidate plasticity genes and to clarify their functional mechanism.