Correlations within and between characters between and within generations as a measure for the early generation selection in potato breeding

Summary The main objective was to determine the reliability of selection in early generations of a potato breeding programme. To that end data were recorded in the seedling generation, an early and a late harvested experiment of the first clonal generation and the late harvested second clonal generation on up to 30 genotypes of 55 populations from crosses of three Solanum tuberosum ssp. tuberosum cultivars and seven selected S. tuberosum ssp. andigena clones, adapted to the northern temperate zone. Phenotypic correlations were computed per population in almost all possible combinations within and between characters between and within generations. The mean across 55 comparable correlation coefficients was considered to be the best estimate of the relationship concerned.In all four experiments data were collected on number of tubers, tuber yield and mean tuber weight; in the three experiments of the two clonal generations in addition on date of emergence, number of stems, plant height, haulm type, and general impression of the underground plant parts and in the two late harvested clonal generations on date of maturity and under water weight.The results confirmed that general impression which is largely decisive for the selection in early generations, is a little reliable measure. However, from the magnitude of the mean correlations between experiments within characters it must be concluded that selection for most of these separate characters is hardly more reliable. An exception was date of maturity for which a mean correlation was determined of r=0.78 between the two clonal generations. The second strongest mean correlations, perhaps surprisingly, were for tuber yield, viz. r=0.52 between the seedling generation and the late harvested first as well as the second clonal generation and r=0.63 between the late harvested clonal generations. From the correlations between characters it appeared that for a high tuber yield, besides its components, in particular an early emergence, a large plant height, an attractive haulm type and a late maturity are of importance.The main conclusion is that, as long as not enough seed tubers are available for the layout of statistically sound experiments, a negative instead of a positive selection should be applied to avoid the rejection of most of the valuable genotypes in an early stage. On the basis of the results directives for the application of such a negative selection are given.