ANALYSIS OF A LONG-TERM SELECTION EXPERIMENT WITH AN EXPONENTIAL MODEL

An exponential mode is presented for the analysis of long-term selection experiments. This model is based on the idea of ''realised heritability'' and it provides parameters which can be considered as the key parameters for the description of long-term selection experiments, such as the total response, the maximal response per generation and the half-life. In addition it provides the 'realised h2-function''. The application of this model is demonstrated, using as an example the data of a long-term selection experiment in which mice were selected for 84 generations on body weight at 60 days. The estimated selection limit for body weight at 60 days was 43.6 g and therefore 21.3 g (96 %, 7.7 phenotypic and 12.8 genetic standard deviations or 18 times the maximal response per generation) over the starting value. The half-life time was 15 to 16 generations (= 0.15 Ne). At the beginning of this experiment the realised heritability was 0.361 and decreased to 0.0004 at the end. While the genetic variance declined from 2.8 to 0.016 g2, the phenotypic variance increased from 7.7 to 39 g2, partly as a scale effect. The mean coefficient of variation for the selected trait was about 11.5 % and showed only a small increase during the experiment. The usefulness of the application of the presented model for the analysis of some types of long-term selection experiments is discussed.