Effect of Artificial Selection on Grain Dormancy in Eleusine (Gramineae)

Abstract

The effect of artificial selection on grain dormancy was studied in six populations belonging to three species of Eleusine. Selection was effective in all pop- ulations but to different degrees. Genes for non-dormancy approached fixation at the fourth generation in one population of E. indica. Threshed grains exhibited higher degrees of germination in all selection cycles when compared with non- threshed grains. This resulted from the presence of inhibitors in the lemma and palea. Populations of E. coracana subsp. africana underwent an occasional drop in germination. This was attributed to a concomitant selection of ill-adapted genes or to fluctuation in allele frequency. It was concluded that man probably eliminated dormancy in finger-millet under domestication within a relatively short time period, particularly in cultivars used for malting. In response to selection by man, cereals and other seed crops have undergone well-defined evolutionary changes during their domestica- tion. In general, these changes were toward larger inflorescences, larger seeds, higher seed number, more determinate growth, free threshing, loss of seed dormancy, and others (Harlan et al. 1973; de Wet 1975). These evolutionary trends have not yet been well-documented experi- mentally. In this paper the efficiency of artificial selection on seed dor- mancy was examined in order to assess the dimensions of this aspect of cereal domestication. MATERIALS AND METHODS Selection against dormancy was applied to three populations of Eleu- sine indica (L.) Gaertn., two of E. coracana subsp. africana (Kennedy- O'Byrne) Hilu & de Wet, and one of E. tristachya (Lam.) Lam. These species are annual wild relatives of the African cereal Eleusine coracana subsp. coracana (L.) Gaertn. (finger millet). They were used in this study because 1) Eleusine species have a short life cycle that is insensitive to day length and a high seed set per individual plant due to extensive tillering, 2) populations of the wild progenitor of finger millet, E. coracana subsp. africana, are extant, and 3) the crop is used for both brewing and bread- making and requires immediate seed germination after harvesting to produce malt. These taxa are predominantly self-pollinated with 1-3% outcrossing. Plant materials were obtained from wild stands (table 1) by Professors J. R. Harlan and J. M. J. de Wet of the University of Illinois. Two 500-grain samples were taken from a mixed collection of each pop- ulation. The size of this initial sample was determined by low frequency of non-dormant seeds in natural populations, the average seed set per individual plant, and our intention to establish population densities com-