Plant breeding progress and genetic diversity from de novo variation and elevated epistasis

Breeding programs in major crops normally restrict the use of parents to those improved for a variety of traits. Gain from utilizing these good x good crosses appears to be high, and improvements are sufficient to encourage continued breeding within narrow gene pools even though each cycle is expected to lead to reduced genetic variability. These finely tuned programs have gradually limited the amount of new diversity introduced into the breeding gene pool. This breeding strategy has led to a genetic gap where there is a large difference in the favorable gene frequency between the improved and unimproved lines and to a narrowing of genetic diversity within elite gene pools. At the same time, evidence has accumulated in plant breeding programs and long-term selection experiments in several organisms that the genome is more plastic and amenable to selection than previously assumed. In the barley (Hordeum vulgare L.) case study reported here, incremental genetic gains were made for several traits in what appears, based on pedigree analysis, to be a narrow gene pool. Given this situation, we call for an examination of the generally held belief that the variation on which selection is based in elite gene pools is provided almost exclusively from the original parents. Classical and molecular genetic analyses have shown that many mechanisms exist to generate variation de novo, such as gene amplification and transposable elements. Accordingly, we put forward the hypothesis that newly generated variation makes an important contribution. We also hypothesize that gene interaction, epistasis, is more important than commonly viewed and that it arises from de novo generated diversity as well as the original diversity.