Requirements for advanced generation breeding populations in forest trees are considered in terms of initial selection of parent genotypes, mating designs, and the nature of progeny plantings.
Large numbers of parent genotypes, 200 or more, are considered desirable, to minimise inbreeding, and to avoid loss of uncommon genes which might eventually prove valuable.
Mating designs should combine immediate efficiency of genetic gain with maintenance of effective population size. Several mating designs, commonly used to test parent genotypes or to estimate genetic parameters, are unsuitable for producing advanced generation breeding populations.
The most promising designs appear to be single-pair matings, and some modifications of the polycross. These are considered in detail, but further study is needed of the effects of non-additive genetic variance on their expected efficiencies. Final choice of mating design, however, may depend on availability of certain information from existing progeny trials, and on the possible need to fulfil other objectives, such as testing parental genotypes.
Progeny plantings could, in some circumstances, be designed entirely for efficient evaluation of individuals in relation to their family means. Clonal replication of seedlings should be explored as a procedure in selecting for traits with "all-or-none" expression.