3.3. Responses of cultivars to regions and fertilizer variations
A combined ANOVA carried out in two divergent regions (subhumidand semi-arid) and for four fertilizer treatments showed a significant effect for the class of cultivar _ region _ fertilizer interaction on yellow berry (YB), protein content (P) and gluten content (Gc). A comparison of means of these parameters is shown in Fig. 1. The variability of quality traits according to specific cultivars was expressed by Genotype _ Environment interaction; overall means of thousand kernel weight, gluten content and gluten index for the eight cultivars in both regions and during two cropping seasons are shown in Table 3. Significant differences amongst specific genotypes for the three traits were observed. The associations amongst quality traits, were calculated for each class of cultivars according to fertilizer treatments in both regions, and are shown in Table 4. Landraces showed a better expression of protein content than high yielding cultivars for all regionsefertilizers’ combinations. Overall means of protein content were calculated for landraces and high yielding cultivars and they were 18.32% and 15.81% respectively; these results confirmed the genotypic effect observed (Table 1) on protein content. Significant differences were observed in protein values between the semi-arid and sub-humid regions, for all fertilizersegenotypes combinations; protein content (P) (17.72% in 2007 and 18.93% in 2008) was greater in the semi-arid region than in the sub-humid region (15.47 and 16.15% in 2007 and 2008 respectively) (Table 2). Moreover, Fig. 1 shows that in the semi-arid region, both groups of genotypes expressed high protein content without fertilizers (18.80% and 15.32% for landraces and high yielding cultivars, respectively), while in the sub-humid region, similar protein values were obtained with N and NK fertilizer treatments (respectively 18.16% and 15.35% for landraces and high yielding cultivars). These results were attributed to water shortage that affected dry matter accumulation during the grain filling period. However, potential high temperature and greater disease incidence appeared to favor high grain nitrogen content and resulted in these differences in grain protein content (Debaeke et al., 1996). Protein content increased significantly with the application of Nitrogen-fertilizer in both regions and for all genotypes, this was in accordance with previous results (Lerner et al., 2006; Malik et al., 2012).