An overview of connectedness in animal breeding

Abstract

The connectedness was defined through a statistical approach, and under the specific animal breeding situation. A good level of connectedness, ie a balanced distribution of the records across the levels of the genetic and environmental factors, is absolutely necessary for estimating and comparing the factors (herd effect, breeding values ...) involved in a linear model of genetic evaluation. A criterion, varying continuously from 0 (complete disconnection) to 1 (perfect connection), was used for measuring the level of connectedness. The estimations provided are only useful when their accuracy is sufficient, thus only if the number of records upon which they rely is large enough. A criterion, used for measuring such accuracy, is also presented. The properties of these criterions were presented in a single theoretical design. Such an approach was also used for illustrating the decrease in the selection scheme efficiency due to a lack of  onnectedness. In the context of a selection scheme, geographical and environmental factors are known to greatly influence the degree of connectedness. A lack of connectedness is not easy to be detected by the breeder. A specific study of the connectedness and of the reliability of the data is then necessary before performing a genetic evaluation. The main situations encountered in the selection schemes (disconnectedness, insufficient or good connectedness) and their consequences on the breeders' strategy are discussed.