Relationships between large-spike phenotype, grain number, and yield potential in spring wheat
Our objective was to investigate the physiological basis of grain number per square meter (GN) and yield in two CIMMYT spring wheat (Triticum aestivum L.) lines of large-spike phenotype (LSP), LSP1 and LSP2, and one check cultivar, Bacanora, when grown as single plants in the growth room and at normal sowing densities in high radiation, irrigated field conditions. In the growth room, rachis length, spikelets per spike, and grains per spike were increased by 14 to 39%, 12 to 31% and 8 to 19%, respectively, compared to Bacanora. Increased spikelet number was associated with a longer thermal duration for spikelet primordia production. In the field, averaged over three seasons, 2003–2004 to 2005–2006, spikelets per spike and grains per spike were increased by 4 to 6% and 4 to 5%, respectively, in LSP lines compared to Bacanora (P < 0.05). Overall GN (–23%) and yield (–8%) were decreased in LSP2 compared to Bacanora associated with a lower spike number per square meter (–26%) (P < 0.05). The GN (–9%) and yield (+2%) of LSP1 overall were not different to Bacanora, although LSP1 yielded more in one season, 2004–2005 (+9%) (P < 0.05). LSP1 produced grains about 10% heavier and LSP2, about 20% heavier, than Bacanora, in both growth-room and field experiments. LSP1 showed a positive departure from the overall negative relationship between grains per gram of spike dry matter (at anthesis) and grain weight among the genotypes. LSP1 may represent a source of novel spike morphology for use in breeding programs aimed at boosting grain size at high GN to enhance yield potential.