The efficiency of seedling selection by visual preference in a potato breeding programme.

TLDR: It was concluded that selection of seedlings was not very efficient and this was not, in the main, a result of a high error variance in visual assessment but rather due to poor association between the performance of Seedlings and first clonal year plants.

Abstract: The efficiency of visual selection in a glasshouse of 1600 seedlings compared with visual selection in the field of the same clones is examined. Also the amount of agreement in assessment between four different potato breeders screening the same clones is investigated. Seedlings were grown from true seed in four-inch pots in a glasshouse and all the clones which produced tubers were grown in the field the following year. Clones which produced more than one tuber from a seedling were grown at two locations in the first clonal year. One of the sites used in the first clonal year is normally used for potato yield trials (i.e. a ware site) and the other is normally used for the production of healthy seed tubers (i.e. a seed site). In all three environments, the tubers produced from each plant were assessed by four breeders independently on a 1–9 scale of increasing attractiveness. From the data it was found that the repeatability of assessment between the glasshouse and the first clonal year was low in that the correlation between the average preference score of the four breeders in the glasshouse and each of the two first clonal year sites accounted for only 8·24 and 6·84% of the total variation. Many clones which had low scores in the glasshouse were subsequently given high scores in the first clonal year. It was therefore concluded that selection of seedlings was not very efficient. Although, in general, the weight of the tuber that was planted greatly influenced whether a clone was selected in the first clonal year, a large number of clones which produced only small tubers in the glasshouse were subsequently selected in the first clonal year. Within each environment the four breeders were either all selecting, or all rejecting, a much higher proportion of clones than would be expected if selection had been made completely at random. The breeders were in most agreement when assessing clones at the ‘ware’ site, and in most disagreement when assessing them grown in the glasshouse from true seed. Therefore the poor efficiency of selection of seedlings grown in the glasshouse was not, in the main, a result of a high error variance in visual assessment but rather due to poor association between the performance of seedlings and first clonal year plants.