Xu S, Martin T G, Muir W M
Department of Genetics, North Carolina State University, Raleigh 27695, USA.
J Anim Sci. 1995 Mar;73(3):699-710. doi: 10.2527/1995.733699x.
Methodology for selection index updating was developed to allow multistage selection. The program determines truncation points for each stage of selection that will maximize either profit or the ratio of aggregate economic gain to cost (R = delta H/C). Either maximum profit or R may be attained by reducing the cost of performance testing in later stages of a multistage program. In order to eliminate the need for multiple integration and assure convergence, a piecewise algorithm was developed. Examples of beef bull selection compared single-stage selection at 1 yr of age, two-stage selection at birth and 1 yr, two-stage selection at 205 d and 1 yr, and three-stage selection at birth, 205 d, and 1 yr. Selection based on three traits (birth weight, gain birth to 205 d, and gain 205 to 365 d) was compared with selection based on four traits (the above three plus ultrasound fat depth) and selection based on five traits (the above four plus feed:gain ratio). Five scenarios were used that allowed variation in proportion of candidates selected for breeding, number of progeny per selected bull, and proportion of profit returned to the nucleus herd. General conclusions based on the examples were 1) multistage selection reduced aggregate economic gain relative to that attained by single-stage selection, 2) inclusion of feed conversion in the index of traits resulted in reduced profit and aggregate economic gain, 3) measurement of feed conversion could be justified when selected bulls produced a large number of progeny, and 4) three-trait selection produced greater profit in all five scenarios than did four- or five-trait selection. Use of the selection updating program described here provides a new source of information that can be used in developing economically sound performance testing and selection programs.
开发了选择指数更新方法以实现多阶段选择。该程序确定每个选择阶段的截断点，以最大化利润或总经济收益与成本的比率（R = ΔH/C）。通过降低多阶段程序后期性能测试的成本，可以实现最大利润或R值。为了消除多次积分的需要并确保收敛，开发了一种分段算法。肉牛选择的示例比较了1岁时的单阶段选择、出生时和1岁时的两阶段选择、205日龄和1岁时的两阶段选择以及出生时、205日龄和1岁时的三阶段选择。比较了基于三个性状（出生体重、出生至205日龄的生长速度和205至365日龄的生长速度）的选择、基于四个性状（上述三个性状加上超声脂肪深度）的选择以及基于五个性状（上述四个性状加上饲料转化率）的选择。使用了五种情景，这些情景允许在选择用于繁殖的候选者比例、每头选定公牛的后代数量以及返还给核心牛群的利润比例方面有所变化。基于这些示例得出的一般结论是：1）与单阶段选择相比，多阶段选择降低了总经济收益；2）在性状指数中纳入饲料转化率会导致利润和总经济收益降低；3）当选定的公牛产生大量后代时，饲料转化率的测量是合理的；4）在所有五种情景中，三性状选择产生的利润都高于四性状或五性状选择。使用此处描述的选择更新程序提供了一种新的信息来源，可用于制定经济合理的性能测试和选择程序。
