Determination of characters for yield selection in spring wheat breeding programs
McVetty, Peter B. E.
Three crosses of spring wheat (Triticum aestivum) currently included in the University of Manitoba wheat breeding program were used to evaluate the use of physiological and/or morphological parameters alone or in combination on F2 plants or F3 families as selection criteria to identify high yielding F4 bulks. The F2 generation was handled as spaced plants grown in the field in two types of environments, (a) a stress-free environment and (b) a normal environment. Approximately 200 plants of each cross were measured for physiological, morphological, yield component and phenological traits as well as being visually rated for yield potential by 5 selectors. a reandom 50-seed sample from each of these plants was sent to Mexico in the winter of 1976-77 to generate F4 bulks. Groups of 49 for stress-free and 81 for normal environment grown F2's were returned from Mexico as F4 bulks to be yield - tested in a six - replicate partially balanced lattice design. The same 49 and 81 groups as tested in F4 were grown simultaneously from F3 remnant seed in the summer of 1977. The results indicated that the most common single F2 parameter which identified high yield potential in F2 was visual yield selection, however the selection intensity had to be low in order to retain a majority of the high yielding lines. A multiple regression analysis approach on a cross - specific basis was found to be better than a single F2 parameter approach because a majority of the high yielding lines could be retained with a much higher selection intensity. The results were simmilar for the F3 generation. A combined cross analysis using multiple regression indicated that productivity and peduncle length in F2 were the two most important common parameters. A prediction of high yielding lines based on these two parameters retained in excess of one - third of the high yielding lines with high selection intensity. However, no acceptable common parameters were found in F3 and it was concluded that replicated F3 yield tests would be more suitable in such cases. Harvest index evaluation in a productivity and height framework retained a majority of the high yielding lines while allowing a selection intensity of approximately 15%. This technique involved the use of an optimum harvest index in productivity and height extremes of statistically characterized populations. It is concluded that the harvest index - productivity - height approach to selection will enhance the effectiveness of plant breeders in their search for high yield in spring wheat.