Experimental Design for Testing Inbred Lines of Swine

Abstract

In an unselected population, additive genetic differences between unrelated inbred lines are expected to average times as large as those between unrelated first generation linecrosses, 2 times those between topcrosses of unrelated inbred lines, and times those in comparisons of individual topcrosses of unrelated lines with outbred stock. As the homozygosity of the inbred lines increases, average differences between inbred lines approach a maximum which is times as large as average genetic differences between unrelated individuals in an outbred population, when all gene effects combine additively. Comparable maxima are 1 for linecrosses, for topcrosses, and for comparisons of individual topcrosses with outbred stock. Non-additive gene effects would make differences between groups larger, especially as the lines become more highly inbred. Prior selection, particularly elimination of inferior lines, would make differences between groups smaller, except in comparisons of individual topcrosses with outbred stock. Conclusions reached for swine breeding experiments concerned with weight of pigs at 180 days of age and the productivity of sows are as follows: The average numbers of animals necessary for the statistical significance of expected infra-herd and intra-season differences between two groups have the following approximate ratios to the numbers required for differences between two inbred lines: 1.5 for linecrosses; 2 to 3 for topcrosses; and about 4 for comparisons of individual topcrosses with outbred stock. For characters with as much environmental variation as weight and productivity, the homozygosity of the lines and the kind of comparison have little effect on the standard error of differences between groups and hence do not have much effect on the numbers required for the significance of a given absolute difference. When all gene effects combine additively, little further reduction in the average numbers required for the 5 or the 10 per cent level of significance is accomplished by advancing the inbreeding of the parents in unrelated lines beyond 20 to 30 per cent for comparisons between the inbred lines themselves, 30 to 40 per cent for those between linecrosses, and 40 to 50 per cent for those between topcrosses. For the 1 per cent level of significance, up to 20 per cent of additional inbreeding will accomplish appreciable further reduction in numbers required. Additional group differences due to dominance and epistasis increase in size as inbreeding progresses, so that numbers required would decline a bit more than this with further inbreeding. Inbred lines should be unrelated if maximum differences between lines are desired. Average differences between related lines are reduced to times what they would be between unrelated lines, where f is Wright's (1921) inbreeding coefficient for the lines and f0 is the average inbreeding which would result from mating animals of one line to those of another. Inter-line relationship reduces average differences between topcrosses, between linecrosses, and between individual topcrosses and outbred stock in the same proportion as those between inbred lines. About twice as many herd-season replications and considerably more work will be required to demonstrate significant differences in sow productivity than in 180-day weight, since productivity is less heritable than weight, and its observation is more expensive and is limited to less than half of the animals. In comparing groups having no parents in common, fewest litters are needed for statistical significance if no parent has more than one litter. However, the cost of sire service, the mating system, and other factors, will modify that in determining what experimental plan will provide significant results most economically.