A Comparison of Methods for Ranking Boars from Different Central Test Stations

Abstract

Data were received from 24 test stations on 3,999 boars tested in fall 1984 and spring 1985. In an effort to increase the connectedness between stations, one reference sire was selected to produce sons (reference boars) through artificial insemination to be tested in the different stations. Fifty-two reference boars were placed across 17 of the test stations. The performance traits analyzed were average daily gain (ADG) and backfat adjusted to 105 kg (ABF). The methods used to rank boars were: 1) individual record (ADG, ABF), 2) individual record deviated from the contemporary group mean, and 3/4) individual record plus performance of relatives (including/excluding reference boars) using expected progeny differences (EPD) estimated from a reduced animal model (RAM) statistical procedure. Ranks of boars using these four methods of genetic evaluation were compared using Spearman rank correlation methodology. The ranks of the boars changed significantly as the complexity of analysis increased for both ADG and ABF. Rank correlations between individual record and contemporary group deviations for ADG and BF were .47 and .20, respectively. Rank correlations between contemporary group deviations and RAM estimates of EPD for ADG and ABF were .53 and .41, respectively. These were significantly different from 1.0. However, there was no significant difference between rankings based on EPD including vs excluding the reference boars. The importance of rank changes coupled with the increased accuracy of these more complex evaluation methods strongly suggest that best linear unbiased predictors of genetic value be utilized in comparing boars in central test stations.

Effects of errors in pedigree on three methods of estimating breeding value for litter size, backfat and average daily gain in swine

“…For data without errors, results also agree with findings of Mabry et al (1987). They compared breeding values estimated by BLUP to individual phenotype from data involving 3,999 boars in 24 central test stations.…”

Comparação de metodologias de seleção sob oscilação genética

“…Vários autores, ao compararem o BLUP e a SI, com dados simulados ou reais, observaram, nas gerações iniciais, superioridade do BLUP em relação à seleção individual, principalmente para características de baixa herdabilidade (Mabry et al, 1987;Belonsky e Kennedy, 1988; Euclydes e Guimarães, 1997;Carneiro et al, 1999 Neste estudo, a eficiência das metodologias de avaliação genética foi afetada pelo número de repetições do processo de simulação dos dados. Para as situações com apenas uma repetição do processo de simulação (Fig.…”

Comparação de métodos de seleção em populações simuladas de frangos de corte

“…Trabalhos com outras espécies também mostram a superioridade do BLUP em relação à SI. Autores como Mabry et al (1987), Belonsky & Kennedy (1988), Euclydes (1996) e Carneiro (1999) observaram superioridade do BLUP em relação à SI, principalmente nas gerações iniciais de seleção e para características de baixa herdabilidade, quando a eficiência do BLUP é comprovadamente maior (Caballero, 1994). A seleção baseada no BLUP aumenta a acurácia das avaliações genéticas, o que torna Figura 1.…”