Genetic variation in the critical specific leaf nitrogen maximising yield among modern maize hybrids

A major aim in breeding programs is to improve nitrogen use efficiency (NUE). The critical level of specific leaf nitrogen (SLNc) maximising growth and yield is a relevant trait influencing NUE. The relationship between growth or yield and SLN is hyperbolic indicating that levels of SLN higher than the critical would tend to decrease NUE. It has been recently proposed that maize breeding may profit of this relationship by phenotyping SLN and selecting for genotypes with SLN = SLNc (SLN < SLNc would result in low yields, SLN > SLNc would bring about high yields but at the expense of low NUE); being the SLNc = 1.5 gN m−2. For this proposal to be realistic, genotypic variation in SLNc should be negligible. We aimed to determine the degree of genetic variation in SLNc in a range of modern maize hybrids. For this purpose, we carried out four field experiments with 11 contrasting maize hybrids grown under fertilized (200 kgN ha−1) or unfertilized conditions in each of the four experiments. At silking, leaf area and N concentration of leaves were determined and at maturity, yield and biomass were measured. Analysing the whole dataset together (n = 88) we found a very large range of all variables measured and calculated: yield varied between c. 6 and c. 18 Mg ha−1 and SLN from c. 0.7 to c. slightly more than 2.5 gN m−2. Within that context we found a significant bi-linear regression between yield and SLN (p < 0.001). Analyzing each hybrid individually we did find substantial variation among hybrids in SLNc, ranging between c.1.0 and c.s1.93 gN m−2. The variation was also substantial calculating SLNc with an alternative methodology and if leaf N content instead of SLN were used as independent variable. Genotypic differences in SLNc were positively related to N uptake but unrelated to other traits for which hybrids varied, including yield potential. Additionally, genotypes with low SLNc show a tendency for a better NUtE. We concluded that it does not seem appropriate to phenotype for a particular value of SLNc as any particular threshold might bring about selected lines with either a yield potential lower than possible and/or with a relatively lower than possible NUE.