Natural hybridisation occurs commonly in Rosaceae. It has often been suggested, based on for example morphological or cytogenetic data, that hybridisation, sometimes in combination with polyploidy, has led to the formation of new species in Rosaceae. Reticulate historical patterns would complicate the interpretation of phylogenetic trees, and lead to contradicting results based different types of data, for example morphology compared with molecules or nuclear data with organellar data. Here, we focus on examples from Rosoideae where molecular data indicate that hybridisation may have influenced phylogenetic estimates. Some species in our analyses of Geum and Potentilla appear in different well supported groups in trees based on nrDNA and cpDNA, respectively. This discordance may indicate that these species are of hybrid origin and have inherited cpDNA from the ovulate parent while the nrDNA of the pollen parent has become fixed. Other processes, for instance introgression or chloroplast capture, are additional possible interpretations. Also, phylogenetic analysis of sequences from paralogous copies of the low copy starch synthase gene indicate that a group of species, which are hexaploid or above, in for example Geum and Erythrocoma, have originated through allopolyploidy. Paralogues from these species appear in different well supported groups in the tree. An hypothesis of reticulate organismal relationships, involving two consecutive allopolyploidisation events, which gave rise to the majority of the species in Colurieae, is proposed based on the starch synthase phylogeny.

Key words: Colurieae, Geum, hybridisation, Rosaceae, waxy