Selection of meat sheep on CT measurements could increase economic response by up to 50% per annum, compared to selection on ultrasonic measurements. However, it would be impractical to obtain CT measurements on the number of sheep that currently are ultrasound scanned. It is likely that much of the benefit of CT measurement of all animals could be obtained at a fraction of the cost, by the use of two-stage selection - with most animals scanned ultrasonically 'in the field', but only the best of these then getting CT measurements.
The overall aim of this project is to design optimum two-stage selection programmes for meat breeds of sheep, incorporating CT measurements. To do so we need (i) information on which CT measurement protocols and sites give the best prediction of traits of interest in meat breeds (e.g. total carcass lean and fat weights, muscle distribution), and (ii) estimates of the heritabilities of these CT measurements, and their phenotypic and genetic correlations with ultrasonic measurements in the field. Area (i) above was addressed in a detailed study of a small number of animals from the three major meat breeds in the UK - the Suffolk, Texel and Charollais. Area (ii) involved more limited CT scanning, for parameter estimation, of 3000 lambs from breeders' flocks.
The SAC-BioSS CT Unit was established in April 1997. Work immediately began on developing methods for transfer of data from the scanner system to a PC where scan images could be readily archived and manipulated. In tandem with this, existing software, was adapted for use with the image file format produced by the model of scanner at the unit. This software was made available through the generous offer of assistance by Dr Neville Jopson of AgResearch, Invermay in New Zealand. Subsequently, work undertaken by BioSS staff developed semi-automated procedures to emulate the "manual" delineation of carcass tissue boundaries required by the early generation software. These new algorithms have been incorporated into a suite of tools for image file manipulation and analysis named STAR (Sheep Tomograph Analysis Routines). Advantages of the new software include speed of use, improved electronic data capture, one unified user interface and enhanced functionality of tools. Furthermore, the new software works with the scan image data file rather than an image "bitmap" so information on bone density is not lost.
Efficient scanning protocols were developed from study of seven anatomically positioned "reference" scans. A calibration trial found that only three of these were required to achieve highly accurate predictions of tissue weight. Multiple regression equations were derived so that carcass tissue weights could be predicted using data from these three scan positions with very high accuracy (up to 99%).
Almost three thousand animals, in three terminal sire breeds, from commercial breeding flocks were CT scanned to allow estimation of the parameters (heritabilities and correlations) required to develop new generation selection indices. Changes in the values for parameters used previously were such that, generally, the differences between breeds were smaller than for those previously used.
Incorporation of CT data into a 2-stage selection programme was examined from a theoretical viewpoint and this work developed to determine practical approaches suitable for UK sheep breeding populations. The recommendations are that 15% of ram lambs go forward for CT scanning on the basis of superior ultrasound scan performance, and that intense selection pressure be applied to maximise benefits from the two-stage selection procedure. Gains can be in the order of £1.2M over 20 years, or greater if schemes using CT increase their share of the ram sate market.