4.3. Phenotypic Traits

Four sets of phenotypic traits were used in the present QTL analysis. One set of observed traits (e.g., measured in the field or the lab), and three sets of derivative traits: traits that were adjusted for phenology, or drought plasticity traits I and II (with and without adjustment for phenology). The observed set included 17 traits of which 13 were previously measured in the population under WL and WW conditions [10,43]: grain yield (GY); thousand kernel weight (TKW); kernel number per spike (KNSP); harvest index (HI); spike dry matter (SpDM); total dry matter (TotDM); carbon isotope ratio (δ¹³C); osmotic potential (OP); chlorophyll content (Chl); flag leaf rolling (LR); culm length (CL); days from planting to heading (DP-H); days from heading to maturity (DH-M). The four additional traits that were not analyzed earlier included: (i) vegetative dry matter (VegDM), comprised of stems and leaves, weighed after drying at 80 °C for 48 h; (ii) spike length (SpL) (cm) measured from the base of the spike to the start of awns at maturity stage; (iii and iv) flag leaf length (FLL) (cm) and flag leaf width (FLW) (mm), of the longest and widest parts of the flag leaf, respectively. Three representative plants were measured in each plot for each trait.

The three derivative sets of traits that were used for QTL mapping were obtained by calculating the deviations (residuals) from the regression line (Figure 4).

Graphical representation of the regression approach for the calculation of derivative traits. (A) Linear regression was calculated between the means of corresponding observed traits and DH to get predicted values of the trait (V_DH). Obtained residuals between observed and predicted values of traits were used as “adjusted phenology traits” for each environment separately. (B) Linear regression was calculated between means of the observed trait values in the WW and WL conditions (V_WL) and linear regression between means of the corresponding observed trait values in the WL treatment and trait values in the WW conditions and DP-H valuesin the dry treatment (V_WL^DH). Obtained residuals were used as “drought plasticity traits I” and “drought plasticity traits II”, respectively.

The first derivative set defined here as ‘adjusted phenology traits’ was obtained, for each environment separately, by calculating the residuals of the linear regression between the means of the corresponding observed trait values and DP-H values (Figure 1A) to exclude the effect of differences in flowering phenology on these traits (prefix ’df‘ was added to the observed trait name):

where V is a value of the observed trait, DH is DP-H value, V_DH is a predicted value of the trait based on linear regression, and Ε_DH is a residual from the regression line.

The second derivative set defined here as ‘drought plasticity traits I’ (Figure 1B) was obtained by calculating the residuals of the linear regression between means of the observed trait values in the WW and WL conditions (prefix ’d‘ was added to observed trait name), to get a deviation between trait value in WL stress and WW condition, adjusted for the differences in trait values in the population under normal conditions:

where V_WW is a value of the observed trait under WW conditions, V_WL is a value of the observed trait under WL conditions, prV_WL is a predicted value of the trait based on the linear regression, and E_WL is a residual from the regression line.

The third derivative set defined here as ‘drought plasticity traits II’ was obtained by calculating the residuals of the linear regression between means of the corresponding observed trait values in the WL treatment and trait values in the WW conditions and DP-H values in the WL (prefix ‘ddf’ was added to the observed trait name), to exclude the effect of drought escape mechanisms in ‘drought plasticity traits I’ by taking into account the effect of heading time:

where V_WW is a value of the observed trait under WW conditions, V_WL is a value of the observed trait under WL conditions, DH_WL is a value of DP-H under WL conditions, V_WL^DH is a predicted value of the trait based on the linear regression, and E_WL^DH is a residual from the regression line.