(Wyatt et al., 1988), ferns (Werth et al., 1985a, b; Soltis et al., 1991), and many angiosperms (Soltis et al., 1992; Soltis and Soltis, 1993, 1999), and include both autopolyploids [e.g., Heuchera grossulariifolia (Wolf et al., 1989, 1990; Segraves et al., 1999) and Heuchera micrantha (Ness et al., 1989; Soltis et al., 1989)] and allopolyploids (Soltis et al., 1992; Soltis and Soltis, 1993, 1999). Recurrent formation of a polyploid species has implications for the taxonomy of polyploids, our understanding of the ease with which and rate at which polyploidization can occur, and, most relevant here, the genetic diversity of polyploid “species. ” In this section, we will address (i) the proportion of polyploid plant species that are known to have formed recurrently, (ii) the extent of recurrent formations within a species, and (iii) the genetic and evolutionary significance of these multiple origins.

Most polyploid species examined to date have shown evidence of recurrent formation (Soltis et al., 1992; Soltis and Soltis, 1993, 1999). Remarkably, these independent origins have been identified even though sampling strategies typically were not designed to investigate multiple origins but rather to test hypotheses of diploid parentage. In many cases, as few as two or three populations of a polyploid species were sampled; the genetic distinctness of these populations, coupled with additivity of diploid genotypes, strongly supported interpretations of recurrent formation. All available data suggest that nearly all polyploid species analyzed comprise multiple lineages of independent formation.

How many such lineages are present within a given polyploid species? Few studies have explicitly addressed this question. Two allotetraploid species of Tragopogon, T. mirus and T. miscellus, arose within the past century in the Palouse region of eastern Washington and adjacent Idaho from diploid progenitors that had been introduced to the region from Europe in the early 1900s (Ownbey, 1950; Fig. 1). During the past several decades, the ancestries of these two tetraploids have been investigated by using nearly every technique that has become available (Cook et al., 1998), and Ownbey's (1950) interpretations have been confirmed.

Early morphological and cytological data (Ownbey, 1950; Ownbey and McCollum, 1953, 1954) suggested multiple origins of each species, two of T. miscellus and three of T. mirus, in different locations on the Palouse. Recent isozyme and DNA analyses have supported Ownbey's (1950) original hypotheses of recurrent origin and have identified additional lineages of independent formation (Roose and Gottlieb, 1976; Soltis and Soltis, 1989b; Soltis and Soltis, 1991; Soltis et al., 1995). For example, based on the geographic distribution of isozyme multilocus genotypes, chloroplast DNA haplotypes, and rDNA markers, estimates of the number of lineages in T. mirus ranged from 4 to 9 (with an extinct population of independent origin, based on flavonoid markers; Brehm and Ownbey, 1965), and the number in T. miscellus ranged from 2 to 21 (Soltis et al.,