Effectiveness of Index Selection Relative to Selection on Yield Alone at Several Levels of Soil Phosphorus 1

Expected genetic advances for oat grain yields were calculated at four levels of soil phosphorus using selection indices composed of various combinations of the following attributes: plant height, plant weight, panicle number, average spikelet number per panicle, weight per 100 seeds, and yield itself. The genetic materials used were 24 oat cultivars grown as subplots in a splitplot design in which the phosphorus levels were whole. plot treatments. The phosphorus was applied at rates of 0, 22.5, 45, and 90 kg/ha of actual P, and the subplots were hills spaced 30.5 cm within and between rows. Genetic and phenotypic variances and covariances used to calculate the selection indices were obtained from variance and covariance component estimates derived from randomized‐block analyses performed on each phosphorus level. Among the indices calculated, only the index composed of panicle number plus yield had any appreciable advantage compared with selection on yield alone. The relative value of the index diminished at higher phosphorus levels because, as phosphorus was increased, heritability of the index increased less rapidly than heritability of yield itself. The results suggested that selection for yield would be more efficient at high levels of plant productivity, but that panicle number might be an effective aid to yield selection at low productivity levels.