Genetic variability in tomato response to environmental variation
Genotype x environment interaction is a long-standing challenge for breeders and the predicted climate change has encouraged plant geneticists to devote more attention into understanding its genetic basis. Tomato is a widely cultivated crop adapted to a variety of environmental conditions (Rothan et al. 2019). However, important incidences of abiotic stress in the final productivity, fruit quality and reproductive performance have been noticed (Albert et al. 2016; Estañ et al. 2009; Mitchell et al. 1991; Xu et al. 2017). We quantified the level of GxE and the subjacent phenotypic plasticity in a multi-environment and multi-stress trial – involving induced water-deficit, salinity and heat stresses – in a highly recombinant tomato population. An important genetic variability was observed for the plasticity traits related to yield, fruit quality, plant growth and phenology (Supplemental Figure 6). This highlights the interest of the MAGIC population as a valuable resource for tomato breeding in dynamic changing environments. Tomato wild species have been also characterized as an important reservoir for abiotic stress tolerance genes (Foolad, 2007). However, their effective use in breeding programs could be difficult due to undesirable linkage drag, notably for fruit quality. Unlikely, the MAGIC population characterized here is an intra-specific population with high diversity regarding fruit quality components, which provides a great advantage as a breeding resource compared to wild populations.
Several statistical models are available to explore, describe and predict GxE in plants (Yan et al., 2007; Malosetti et al., 2013). Factorial regression model is among the most attractive as it allows to describe the observed GxE regarding relevant environmental information. We used the factorial regression model with different environmental covariates that are readily accessible from year to year, which allowed us to predict a variable proportion of the observed GxE (Supplemental Figure 4). Besides, each MAGIC line was characterized for its sensitivity to the growing climatic conditions opening avenues to effectively select the most interesting genotypes for further evaluation in breeding programs targeting stressful environments.
Interestingly we found significant correlation between the genotypic sensitivities to the different environmental covariates and slopes from the Finlay-Wilkinson regression model (Supplemental Figure 10). This emphasizes the adequacy of the selected environmental covariates to explain differences observed in the average performance of the genotypes across environments. Conversely, slope and VAR showed less significant correlations, although they were both correlated to mean phenotypes in the same direction – except for SSC (Figure 2). This may be induced by distinct genetic regulation of these two plasticity parameters which reflect different types of agronomic stability (Lin et al. 1986). Indeed, we identified 7 and 14 plasticity QTLs that were specific to VAR and slope, respectively (Supplemental Table 4). The correlation pattern of the different plasticity parameters evokes a complex regulation of plasticity which besides is seemingly trait specific.
Significant correlation at phenotypic level might result from the action of pleiotropic genes. The Figure 2 displays the correlations between genotypic means and plasticity which were significant for almost every trait at variable degree. These correlations were reflected at the genetic level by 22 QTLs overlapping between genotypic mean and plasticity parameters, representing about 21% of all identified QTLs. However, a high proportion of the QTLs were specific either to genotypic means or plasticity parameters (Supplemental Figure 11), hence suggesting the action of both common and distinct genetic loci in the control of mean phenotype and plasticity variation in tomato.