*3.1. Variation for Biomass Production and Root Morphologies in Seedlings Cultivated In Vitro*

Seedlings grew on vertical dishes filled with agar medium, as illustrated in Figure 1a. The germination characteristics of seedlings grown with 0.01 mM (N−) or 10 mM (N+) nitrate are presented in Figure 2. In this case, 12 selected cultivars were ranked from the poorest to the greatest total root length at N−: Cardiff (CAR), SY Carlo (CLO), Exocet (EXO), NK Aviator (AVI), DK Exquisite (EXQ), Recordie (REC), Limone (LIM), SY Saveo (SAV), Hertz (HER), Troy (TRO), Battz (BAT) and ES Jason (JAS). On average for the diversity panel cultivated in vitro, the root biomass (R, +28%) and the root-to-shoot biomass ratio (R:S, +39%,) were superior, while the shoot biomass (S, −18%) and the total biomass (R + S, −8%) were inferior at N− compared to N+. The length of primary root (LPR, +18%), the lengths of primary root zone 2 and zone 4 (LZ2 and LZ4, +24%), the number of lateral roots (NLR, +38%), the sum of lateral root lengths (∑LLR, +63%), the total root length (TLR, +41%), the density of lateral roots in zone 1 (DLR-Z1, +22%) and the specific root length (SRL, +15%) increased, while the length of primary root zone 3 (LZ3, −351%) and the density of lateral

roots in zone 2 (DLR-Z2, −8%) decreased at N− compared to N+ treatment. The cultivars showed considerable variation for biomass production and root morphology in response to N supply (Figure 3a,b). Variations for NLR and ∑LLR were generally the largest ones, and they were amplified at N+ compared to N−. For instance, the percentage differences for ∑LLR between extreme cultivars were more than ten-fold (CAR vs. JAS at N− and CLO vs. LIM at N+). The responsiveness of one cultivar to N depletion (i.e., increase/decrease of one trait value in response to N− compared to N+) was evaluated (Figure 3c). A large variation in lateral root phenotypic plasticity was observed, with cultivars being poorly (CAR showing ∑LLR increase less than 10%) or greatly (BAT showing ∑LLR increase more than 500%) responsive to N−. The genotype (cultivar) effect, the environment (N treatment) effect and the interaction (cultivar × N) were significant (*p* < 0.05) for all traits, except the (cultivar × N) for S biomass (ns) (Table S2). The in vitro observations highlight an important variability among cultivars in terms of root morphological traits, and this leaves space for genetic selection targets in breeding programs.

A principal component analysis was used to compress and classify the biomass and root morphology data (Figure 4a,b). The two first principal components (PCs) explained together 71% of the total phenotypic variation. The PC1 (53.1%) was influenced the most by root morphological traits, while PC2 (17.8%) by the shoot biomass production (Figure 4a). The distribution of the cultivars across PC1 and PC2 permitted to clearly distinguish between the two N treatments (Figure 4b). Cultivars with low PC1 scores exhibited long primary and lateral roots, while those with elevated PC2 scores produced important aerial biomass.

**Figure 2.** Root morphologies of oilseed rape cultivars grown in vitro with two divergent nitrate supplies. Pictures of representative root organs of 12 contrasting oilseed rape accessions grown with 0.01 mM (upper row, N−) or 10 mM (lower row, N+) nitrate supplies. Accessions are ranked from the left to the right by increasing total root length (TLR) measured at N−. The TRL values were ranging between 12.1 ± 0.9 cm (CAR) and 41.5 ± 2.9 cm (JAS). Cultivar full names are listed in Table S1. Scale bar: 2 cm.

Spearman correlation coefficients were calculated between the traits measured in vitro (Figure S1). Some correlations were found between derivative traits and their components (Table 1) but also between unrelated traits. The strongest and most significant correlation was found between LPR and ∑LLR (*-*<sup>2</sup> = 0.57; *<sup>p</sup>* < 0.001) at N−. Furthermore, S biomass showed weak but significant positive correlations with NLR and ∑LLR (0.15 < *-*<sup>2</sup> < 0.26, *p* < 0.001) during both N treatments.

**Table 1.** Abbreviations and definitions of the measured traits.


**Figure 3.** Relative variation of phenotypic traits measured in 55 oilseed rape cultivars cultivated in vitro (**a**–**c**) and in 12 cultivars in hydroponics (**d**–**e**). The spider plots (**a**,**b**,**d**,**e**) show the percentage variation of a given trait for every cultivar, normalized by the mean value of the panel, measured during N+ (**a**,**d**) or N− (**b**,**e**) conditions. Zero percent (blue circle) indicates no difference compared to the mean value of the panel in one condition. The spider plots (**c**,**f**) show the percentage variation of a given trait for every cultivar grown during N−, normalized by the value observed during N+ conditions. This defines the responsiveness of one trait to nitrate depletion. Zero percent (blue circle) indicates no difference compared to elevated nitrate conditions. n = 15 plants per genotype and per N condition cultivated in vitro, and 5 plants in hydroponics. Traits are defined in Table 1.

**Figure 4.** Principal component analysis of 14 phenotypic traits measured in 55 oilseed rape cultivars cultivated in vitro (**a**,**b**) and of 13 phenotypic traits in 12 cultivars in hydroponics (**c**,**d**). (**a**,**c**) Principal component (PC) biplot showing the compositions of the first two PCs, with cumulative variance; (**b**,**d**) Representation of the cultivars. Symbols indicate position of the cultivars as determined by their trait values in the two first PCs. Black symbols refer to low nitrate (N−) and yellow symbols to high nitrate (N+) conditions. Traits are defined in Table 1 and cultivars listed in Table S1.
