Genetic variation and selection for nitrogen use efficiency in maize : A synthesis

Nowadays development of varieties with a better nitrogen use efficiency (NUE, i.e. grain yield per unit of nitrogen from soil including fertilizer) is becoming a necessity both to allow the maintenance of a sufficient profit margin for the farmer and to preserve ground water from nitrate pollution. A new breeding aim could be to develop varieties adapted to low N-input. Many studies show a genetic variability for NUE at a given level of ni- trogen fertilization, with a significant genotype x nitrogen fertilization (GxN) interaction. Such an interaction is rela- tively low, as illustrated by a highly significant phenotypic or genotypic correlation between yields at low and high N-input when the yield reduction at low N-input is not greater than about 35%. However, genetic variation is ex- pressed differently at low and high N-input. Variation in NUE at high N-input is mainly related to variation in N- uptake, whereas at low N-input, both components of NUE could play a role, specifically nitrogen utilization ef- ficiency, i.e. grain yield/N-uptake. GxN interaction for grain yield appears to be due to GxN interaction for ker- nel number. To have a good NUE at low N-input, it is necessary to reduce kernel abortion just after fertilization. Genotypes with a short anthesis-silking interval and pro- lific genotypes seem to have the ability to remobilize N from the stover to the grain efficiently, particularly in the early stage of embryo development, avoiding embryo or ear abortion. Leaf area duration appears to be an impor- tant factor for post-silking N-uptake and thus grain filling and is favourable for NUE whatever the N input. Remobi- lization also plays an important role at low N-input. Re- sults from QTL detection tend to confirm that genetic variation is expressed differently at low and high N-input. Selection experiments show that to have the maximum genetic advance at low N-input, it is better to select in this N condition. Finally it appears to be quite possible to develop new varieties better adapted to low N-input than present varieties.