Centaurea corymbosa, a cliff-dwelling species tottering on the brink of extinction: a demographic and genetic study.
In order to detect the evolutionary potential of two endangered species, Brassica insularis (Brassicaceae) and Centaurea cor-ymbosa (Asteraceae), within and among-population genetic variation for both quantitative traits and allozymic markers was examined. Four populations of each species were studied, representing a large proportion of extant populations. High values of ST (0.213 and 0.364 for B. insularis and C. corymbosa respectively) suggested that low amounts of gene ¯ow occur among the study populations. In each species, the genetic distance based on allozymes (estimated by the ratio (ST /1À ST)) was positively correlated with the geographical distance, indicating isolation by distance. In contrast to previous studies in either outcrossing or sel®ng plant species, and especially for B. insularis, population dierentiation for quantitative traits (Q ST) was generally found lower than differentiation for allozymes (ST), suggesting that the populations studied were experiencing similar selective forces acting upon the quantitative traits measured. Such forces would be strong enough to counteract local genetic drift. Interestingly, for both species Q ST 's were statistically independent of geographical distance, in contrast to the marginally signi®cant positive isolation by distance shown by ST. Altogether, these results suggest that ST 's might not always be used as conservative estimates of Q ST 's, and might instead overestimate the evolutionary potential of endangered species. This would be especially expected in narrow-endemic species, whose ecological niche is often so restricted that indeed homogeneous selective forces are likely to occur, whereas small population sizes and restricted dispersal are likely to produce strong dierentiation for neutral variation. In fact, knowledge of both neutral and quantitative diversity patterns allows identi®cation of those traits undergoing natural selection, and could be useful in designing reinforcement or reintroduction programs. However, this approach might have limitations too, in the presence of outbreeding depression due to locally coevolved gene complexes.