Cooperative Second-generation Breeding in the Us Pacific Northwest

The question has arisen whether to use “backwards” selections (i.e. tested parents), “forwards” selections (in this case, mainly untested progeny from openpollinated tests), or a combination of parents and progeny, both in seed orchards and in breeding. Those are the three main options currently available for cooperative breeding programs and seed orchards in the US Pacific Northwest. Which option will give the most gain, and which is the most reliable? The questions are important, since several 1.5 generation orchards are being established and we are in a period of intense activity for second-generation breeding and testing. They have been explored in some publications (e.g. Burdon and Kumar in press, Hodge 1985 and 1997, Hodge and White 1993, Ruotslainen and Lindgren 1998) ; however, a study focusing on the situation in the US PNW is also worthwhile. We therefore tried to shed some light on this issue for the benefit of NWTIC members. When using parents, we have a good idea of what to expect because we have already tested their progeny. We also have good precision when selecting parents since across-site family mean heritabilities (hf) tend to be between 0.6 and 0.8 for many cooperative first-generation test series. The expected gain from selecting families is a direct function of family mean heritability (expected genetic value = hf x family mean). Assuming a familymean heritability of 0.75, the correlation between the family mean and its genetic value is r = ÷ hf = hf = ÷0.75 = 0.87. If we deploy seed by family (i.e., seed orchard parent) and want to match families and sites, then it is always good to be sure we have the appropriate family. We can also select progeny for seed orchards. In theory we increase gain every generation of breeding so the next generation of selections should be better than the last. A concern with forward Executive Summary