GENETIC ARCHITECTURE FOR THE ADAPTIVE ORIGIN OF ANNUAL WILD RICE,ORYZA NIVARA

Abstract

The wild progenitors of cultivated rice, Oryza nivara and Oryza rufipogon, provide an experimental system for characterizing the genetic basis of adaptation. The evolution of annual O. nivara from a perennial ancestor resembling its sister species, O. rufipogon, was associated with an ecological shift from persistently wet to seasonally dry habitats. Here we report a quantitative trait locus (QTL) analysis of phenotypic differentiation in life history, mating system, and flowering time between O. nivara and O. rufipogon. The exponential distribution of effect sizes of QTL fits the prediction of a recently proposed population genetic model of adaptation. More than 80% of QTL alleles of O. nivara acted in the same direction of phenotypic evolution, suggesting that they were fixed under directional selection. The loss of photoperiod sensitivity, which might be essential to the survival of the ancestral populations of O. nivara in the new environment, was controlled by QTL of relatively large effect. Mating system evolution from cross- to self-fertilization through the modification of panicle and floral morphology was controlled by QTL of small-to-moderate effect. The lack of segregation of the recessive annual habit in the F2 mapping populations suggested that the evolution of annual from perennial life form had a complex genetic basis. The study captured the genetic architecture for the adaptive origin of O. nivara and provides a foundation for rigorous experimental tests of population genetic theories of adaptation. Darwin (1859) considered adaptation by natural selection a pri- mary driving force for the origin of new forms and species. The following century witnessed noteworthy theoretical advances in elucidating the genetic basis of adaptation. The earliest theory traces back to Fisher's (1930) geometric model predicting that favorable mutations fixed during adaptation should have small phenotypic effect, because mutations of large effect were more likely to generate negative, pleiotropic effects. Kimura (1983) later suggested that mutations of small effect could suffer from stochastic loss and concluded that mutations of intermediate sizes should be most frequently fixed. More recently, Orr (1998a) pro-