In 2009 and 2010, four maize diversity lines (CML103, Mo18W, NC350, and Tx303) were evaluated at Champaign, IL for their responses to fertilizer N. These evaluations were made at the test-cross level at five levels of N availability (0 – 225 kg N ha^-1). The objectives of this study were to (i) document genotypic differences in grain yield at low and high N, and (ii) to measure differences in yield components, biomass accumulation, and N uptake and utilization.

The average check plot yields (grain yield at 0 kg N ha^-1) were 1.8 and 3.3 Mg ha^-1, while the average maximum responses to N fertilizer were 7.3 and 3.3 Mg ha^-1, in 2009 and 2010, respectively. Despite the clear effect of year on average grain yields, we found that genetic differences in check plot yield ranked consistently in both years, while there were no clear trends for fertilizer N response. Averaged across both years, the high check plot yield genotypes (CML103 and NC350) had a mean check plot yield of 3.6 Mg ha^-1, while the low check plot yield genotypes (Mo18W and Tx303) had a mean check plot yield of only 1.6 Mg ha^-1. Improved grain yield at low N by CML103 and NC350 did not appear to be caused by better N uptake or biomass accumulation at flowering. On average, the two pairs of genotypes had similar N uptake (mean = 0.62 g N plant^-1) and biomass accumulation (mean = 78 g plant^-1) at flowering. This result implies that the high check plot yield genotypes utilized N more efficiently (genetic N utilization, GU) to establish sink capacity (i.e., kernel number) at flowering. Accordingly, the high check plot yield genotypes had an average GU that was 35% greater than that of the low check plot yield genotypes.

We theorize that differences in GU are reflective of differences in earshoot N metabolism at flowering. Unpollinated earshoots were sampled in 2009 to address this hypothesis.  Global metabolite profiling of 267 compounds in unpollinated earshoots from the four genotypes revealed that amino acid biosynthesis in early ear development is highly associated with end yield.  In particular, glutamine is the most highly correlative metabolite of the 267 with yield, with a correlation coefficient of 0.89 across N treatments.  The possibility of using glutamine as a predictive biomarker for yield and breeding is being explored.

Collectively, these results indicate that grain yield at low N (check plot yield) is a trait that is under strong genetic control. As such, evaluation of diverse germplasm under managed N environments using appropriate phenotyping methods (e.g., measurement of grain yield along with key metabolites) is a promising strategy for developing genotypes that are productive under reduced N inputs.