Relationships among clean wool weight and its components. I. Changes in clean wool weight related to changes in the components

Abstract

This paper is the first of a series presenting results from selection experiments with medium-wool Merino sheep, initiated by the Commonwealth Scientific and Industrial Research Organization (C.S.I.R.O.) in 1947. Clean wool weight (W) is discussed in relation to its five components: smooth body surface area (S), degree of skin wrinkling (R), fibre number per unit skin area (X), fibre cross-sectional area (A), and staple length (L). Statistical difficulties are encountered in relating a whole to one of its parts. Various methods have been used in the past, a common one being the gross correlation between clean wool weight and a component. It is shown that, given the variances and covariances of the components, and assuming that all relationships are mainly linear, this gross correlation can be predicted. The influence on the gross correlation of the relationships among the components is discussed, and it is shown that N and L will give the highest correlations with W for a range of values of phenotypic variances and covariances of the components which have been reported in the literature. Some selected combinations of extreme values, which might be encountered in specific strains, will increase the importance of other components. A study of the phenotypic and genetic variances and covariances of the components, and of the factors influencing them, is therefore likely to be more rewarding than a study of the gross correlations between clean wool weight and a component. One difficulty with the gross correlation technique is that if the variances of some components were more inflated by sampling errors than others, the correlation coefficients would be biased. A simple method, the "percentage deviation technique", is presented, which analyses means, and so avoids this possible bias. It has the added advantage of apportioning variation in clean. wool weight among the five components. Use of this technique on data from the C.S.I.R.O. selection experiments shows that in this flock N and L are more closely associated with W than are the other components. In each case the association is positive and accounts, on the average, for 0.3-0.4 of the phenotypic variation in W. There is little difference between the two components in their phenotypic association with W, but there is an indication that N may have a higher genetic association. Selection for W has increased N and L, as well as W itself, but has not changed body size.