Published online by Cambridge University Press: 27 March 2009
In five experiments with binary mixtures of a wide range of Upland cotton varieties grown at three sites in Uganda in 1969–70 and 1970–1 there was no evidence to indicate that components interacted to give better yields than would be expected from their yield in pure stand. Moreover, in district variety trials over three seasons there was no evidence that mixtures of several lines of closely related material were more stable than the individual component lines. Problems related to a breeding programme based on the development of multiline varieties are discussed. Although the variable ecological conditions in Uganda should provide an excellent opportunity of developing multilines with a high degree of population buffering, it is concluded that pure lines tested in a wide range of environments are as good as, and are less costly to produce, than mixed populations that have been specially bred and tested. Multilines with components that differ little in yielding ability or stability do, however, provide an opportunity of more rapid seed multiplication, and there is also the possibility that multiline seed issues may show progressive increases in yield with each year of growing. Present practice in Uganda of releasing as seed issues multilines with closely related components that have been adequately tested as individual lines should therefore be continued.