EFFECT OF THE PURPOSIVE CHOICE OF FAMILIES ON THE ESTIMATES OF THE JUVENILE MATURE CORRELATION DERIVED FROM RETROSPECTIVE TESTS

Retrospective tests have been widely used during the last decade to ad dress early selection of forest trees. The main goal of retrospective testing is to derive estimates of juvenile-mature correlations for a g roup of families that have already been tested in the field at the mat ure stage. The families were sometimes purposefully chosen to represen t contrasted growth abilities. The aim of the present study is to show that the purposive selection of the families substantially influences the value of the estimate of the juvenile-mature correlation. We assu me that the regression between the mature and juvenile stages is linea r with constant residual variance. Under this assumption, a formula is derived that permits the correlation estimated through retrospective tests with purposive selection of the individuals to be related to the correlation in the whole population. An example with forest trees is given, and the implications of the results for future research are dis cussed. It is concluded that correlations estimated from retrospective tests whose families are purposefully chosen should be used with caut ion or that their computation should simply be avoided. Alternative ap proaches are suggested.