Consistency over time of spatial patterns of fibre diameter and staple length variation over sheep fleeces

Abstract

Macro-patterns of wool variation over sheep fleeces are markedly diverse among individuals. Assessing the consistency over time (i.e. repeatability) of those patterns would be relevant for sampling, individual selection, and raw wool classing, but there are no canonical procedures available to estimate the repeatability of 2D traits. We devised an approach to evaluate consistency over time as a proxy for repeatability in a 2D domain and applied it to a dataset of fibre diameter (FD) and staple length (SL) measurements. Data were collected over a regular grid of 128 sampling points from 10 Corriedale ewes sampled before shearing in two consecutive years. We scaled down the dimensionality of the data set from 128D to 2D, projected the new synthetic data (n=20, one per animal per year) on principal coordinate axes, extracted all relevant Euclidean distances between pairs of data points, and applied an ad hoc coefficient (Csp) designed to assess consistency of spatial patterns over time in a scale from zero (random dispersion independent of individual animal origin) to one (perfect overlapping of data points from the same animals in different years). Point estimates of Csp (±bootstrap-estimated SE) were 0.78±0.06 and 0.64±0.08 for FD and SL, respectively. Estimated 95% confidence intervals excluded zero for both traits, FD: [0.60, 0.84]; SL: [0.43, 0.75]. Contributions of individual animals to overall Csp coefficients were independent between traits (r=−0.11, P=0.76). Main conclusions were that considering the spatial variability of wool traits over fleeces may help avoiding biases of phenotyping, estimation of genetic and phenotypic parameters, and selection decisions, that spatial patterns of variation of FD and SL over sheep fleeces are similarly repeatable across years, and that FD and SL pattern expressions may be independently regulated.