Relationships, origin, and diversity of Galapagos tomatoes: implications for the conservation of natural populations.

Endemic Galápagos tomatoes (Lycopersicon cheesmanii) are of great value for cultivated tomato (L. esculentum) breeding, and therefore their conservation is of significance. Although within L. cheesmanii there is heterogeneity for many traits and formal infraspecific classification is not justified, here we distinguish three forms, without taxonomic significance, of L. cheesmanii that are of interest to breeders because of their distinctive morphology and habitat preferences: L. cheesmanii 'short' (one- to two-pinnate leaves, short internodes, and coastal habitats), L. cheesmanii 'long' (one- to two-pinnate leaves, long internodes, and inland habitats), and L. cheesmanii forma minor (three- to four-pinnate leaves, short internodes, and coastal habitats). In a recent survey of tomato populations in the Galápagos Islands, we found that several populations of L. cheesmanii reported 30-50 years earlier had disappeared, mostly as a consequence of human activity. In addition, a previously unreported invasive wild red-fruited form, which we named L. esculentum 'Gal cer,' was found on the island of Santa Cruz. The total diversity (estimated with amplified fragment length polymorphisms [AFLPs]) within L. cheesmanii (H(T) = 0.051) is almost as high as that for the mainland wild species L. pimpinellifolium (H(T) = 0.072). Lycopersicon esculentum 'Gal cer,' on the other hand, has a much lower diversity (H(T) = 0.014). Comparison of AFLP fragments shared by L. esculentum 'Gal cer' with other species showed that it is closely related to weedy tomato L. esculentum var. cerasiforme and, therefore, likely of recent origin. Genetic differentiation among the three native L. cheesmanii forms is low (G(ST) = 0.235), indicating that they share a common genetic background. Nonetheless, L. cheesmanii 'short' is about twice as diverse as L. cheesmanii 'long' or L. cheesmanii f. minor. UPGMA cluster and principal components analysis distinguish four groups within Eulycopersicon: L. pimpinellifolium, cultivated L. esculentum, L. esculentum var. cerasiforme including L. esculentum 'Gal cer,' and L. cheesmanii. The geographic distance and genetic distance in the wild forms of Galápagos tomatoes were not correlated. Apart from the pressure of humans, some native L. cheesmanii populations, especially L. cheesmanii 'long,' might be displaced by invasive L. esculentum 'Gal cer' because they share a similar habitat. We did not find evidence of intercrossing of L. cheesmanii with introduced L. esculentum, but occasional hybridization that contributes to loss of genetic integrity of L. cheesmanii cannot be ruled out. Establishment of reserves of L. cheesmanii to protect this species from introduced herbivorous animals and from hybridization with L. esculentum 'Gal cer' would help to conserve L. cheesmanii. Furthermore, accessions collected by C. M. Rick and others in the 1950s-1970s and now stored in germplasm banks could be used to reinstate some extinct populations.