“…A high level of correlation was observed for durum wheat for both the two sites: in particular six couples of elements (Cu/Zn, Mg/Zn with significance at p < 0.001 and Cu/Mg, Cu/Mn, Zn/Mn, Cr/Fe with significance at p < 0.01) for site A and seven couples of elements (K/Cu, K/Mg, K/Mg, Cu/Mg, Cu/Mn, Cr/Fe with significance at p < 0.001 and Ca/Cd with significance at p < 0.01) for site B (Tables 3 and 5), while wild wheat showed a low number of correlations in both the two sites (Cr/Cu, Fe/Mn with significance at p < 0.001 and Cr/Fe, Cr/Mg, Cd/Mn and Ca/Mg with significance at p < 0.01 for site A; Cr/Fe with significance at p < 0.001 and Ca/Cd, K/Zn, K/Cr, K/Fe, Cr/Fe with significance at p < 0.01 for site B, Tables 4 and 6). It should also be noted that in some cases, more significant or positive correlation values for other elements (i.e., Mn/Mg) were expected, although comparable results to those that were obtained in this work were also reported in literature [28,37,47,48]. Multivariate statistical analysis (PCA and OPLS-DA) was used to deeply investigate the variation in macronutrients (Ca, Mg, K), micronutrients (Zn, Fe, Cu, Mn) and toxic trace elements (Cd, Pb) for the whole dataset of 25 wheat accessions and cultivars (three replicates for each cultivar studied) that is representative of durum and wild species in the two cultivation sites.…”