Abstract

The gourd family, Cucurbitaceae, is among the economically most important families of plants, with many crop species that form the basis of multi-million dollar industries. Knowledge of these species’ geographic origin and their closest wild relatives is fundamental to breeding efforts, genetic improvement, and conservation. Surprisingly, these aspects have been unknown or misunderstood for many widely cultivated species, even though plant material that could have been used for broad phylogenetic studies has long been available in herbaria. For the thesis presented here, I focused on the phylogenetic relationships within two clades of Cucurbitaceae that comprise cultivated species: the genus Cucumis, to which cucumber (Cucumis sativus) and melon (Cucumis melo) belong, and the New World Sicyoeae, which contain vegetable pear or chayote (Sicyos (Sechium) edulis) and tacaco (Frantzia tacaco), locally important in Mexico and Costa Rica, and the former also cultivated worldwide. I used a combination of DNA sequence data from up to 175-year old herbarium specimens and molecular phylogenetic methods as well as traditional morphological and ecological data from my own fieldwork in Asia and Australia to infer the phylogenetic relationships among these clades. I also discovered and described several new species, and reconstructed plausible scenarios for the two clades’ geographical unfolding over time. Until recently, only two species of Cucumis, namely cucumber and its closest relative C. hystrix, were thought to be of Asian origin, and melon was thought to have originated in Africa, from where 30 species were known. Using DNA sequences from plastid and nuclear markers for some 100 Cucumis accessions from Africa, Australia, and Asia, I have shown that cucumber and melon both are of Asian (probably Indian) origin and form a clade with 23 previously overlooked species-level relatives in Asia, Australia, and around the Indian Ocean, at least nine of them new to science and some described as part of this thesis. Fieldwork I carried out in Thailand and Australia contributed new knowledge about the life forms and habitats of some of these species and resulted in fertile material essential for the descriptions. My study furthermore revealed that the sister species of melon is the re-discovered C. picrocarpus from Australia. Future breeding efforts and investigations of wild species related to melon and cucumber should therefore concentrate on Asia and Australia, instead of Africa. In my second study group, the Sicyoeae, my aim was to test long-problematic generic boundaries and to reconstruct the history of the tribe’s name-giving genus, Sicyos, which has an exceptional geographical distribution. Using a densely sampled molecular phylogeny that included type species of 23 currently or formerly accepted genera of Sicyoeae, I showed that morphology-based concepts did not result in monophyletic genera, and that species from numerous smaller genera, including chayote, need to be part of Sicyos if monophyly is to be established. Sicyos, in its new circumscription, has a center of distribution in the Neotropics, where c. 50 species occur, but long-distance dispersal has resulted in the group’s presence on Hawaii (where it radiated into 14 species), at least two arrivals on the Galápagos archipelago (but no radiations), and one arrival in Australia and New Zealand, now with three species, two of them new to science. Using molecular clock models, I dated these four trans-Pacific dispersal events, all from the American mainland, to the last 4.5 to 1 million years. The mode of dispersal may have been adherence of the small, spiny fruits to birds, which would fit with the documented occurrence of Sicyos plants near seabird nesting colonies. The rapid diversification on Hawaii may have followed the loss of the fruit spines in the ancestor of the 14 Hawaiian species, leading to lower dispersal ability and faster allopatric speciation in the diverse habitats of the archipelago.