Variability, growth and manipulation of kernel weight within oat panicles

Two experiments were conducted between 1994 and 1997 using a milling oat, 'Drummond', and later comparison with a forage oat, 'Cashel' in 1996, at Lincoln, Canterbury, New Zealand. The relationship between kernel size, used for grading in industrial processing and weight was established. This enabled examination of how kernel position within the oat panicle related to differences in kernel weight. Autumn and spring sown crops of 'Drummond' both produced two kernel weight classes corresponding to kernels from primary and secondary florets within each spikelet throughout the panicle. Secondary kernels (21.1 mg) were roughly half the weight of the primary kernels (41.1 mg) giving a primary:secondary kernel weight ratio of 1.5. Grain filling in both 'Cashel' and 'Drummond' followed a sigmoidal growth pattern over thermal time. The lag phase was about 230 growing degree days (ºCd) and the linear phase was about 690 °Cd for both oat cultivars. Primary kernels and those from 'Cashel' were shown to be heaviest due to a faster rate of kernel filling. 'Cashel' primary kernels filled at 0.07mg/ºCd compared with 0.05mg/ºCd for 'Drummond' primary kernels and 'Cashel' secondary kernels filled at 0.04mg/°Cd compared with 0.03mg/ºCd for 'Drummond' secondary kernels. Spikelet or floret removal/emasculation at anthesis of the apical spikelet, increased final kernel weights between 2 and 20% in 'Cashel' and 'Drummond'. However, the later that spikelets were removed, the lower was the increase in final kernel weight. None of the treatments altered secondary kernel weights to a level similar to primary kernels and the primary:secondary kernel weight ratio remained constant at 1.5. The kernel weight difference between the two kernel classes was the main cause of the large kernel weight variation within oat panicles of crops. The inability to alter the primary:secondary kernel weight ratio lead to the conclusion that decreasing variability may come from the development of an oat cultivar with a single kernel per spikelet. However, to offset any yield reduction it is recommended that breeding should aim to increase the number of spikelets per panicle, decrease the lag phase of kernel growth (potentially by the introduction of a dwarfing gene) and select for maximum kernel growth rates.