Recurrent Selection for Lint Percent within a Cultivar of Cotton (Gossypium hirsutum L.)

We conducted this study to determine if a modified form of recurrent selection for lint percent within a cotton (Gossypium hirsutum L.) cultivar, ‘Deltapine 523,’ could improve lint yield. Three cycles of selection, beginning in 1965, were completed, with selection for lint percent being practiced first on a plant basis and then on a progeny-row basis. Plant populations of 430, 828, and 800 plants were used to initiate the S₁, S₂, and S₃ populations, respectively. From each population, 16 progenies were ultimately selected on the basis of high lint percent, and 8 crosses were made among them to initiate the next cycle. In each cycle, an equal number of seeds from each of the 16 selections were combined to form the bulk populations tested in 1971. Eight progenies selected in the S₃ for higher 2.5% span length were bulked and designated as the S₃L. population. In 1971, Deltapine 523 (S₀) and the four selected populations were grown in replicated experiments at four locations. Mean lint precent was 33.8, 35.4, 36.6, 38.0, and 37.1 for the S₀, S₁, S₂, S₃, and S₃L. populations, respectively. The S₂, S₃, and S₃L populations produced significantly greater lint yield than did S₀ and S₁. As lint percent was increased by selection, there were concurrent decreases in seed index and 50 and 2.5% span length. Selection for fiber length in the S₃ population restored most of the fiber length previously lost. As lint percent was increased by selection, there were simultaneous increases in lint index, seeds per boll, and Micronaire. No consistent trend or change was observed for boll size, number of bolls per plot, fiber strength, and fiber elongation. The primary cause of the increase in lint yield resulting from selection for lint percent was the association of llnt percent with increased number of seeds per boll and higher lint index. The genetic variability for lint percent within the S₁ and S₂ populations was found to be 188 and 83%, respectively, as compared to that of the S₀ population. The genetic variability for an unselected trait, boll size, was similar, with the S₁ and S₂ populations expressing 233 and 186%, respectively, as much variability as that of the S₀ population.