Sequence variation at cpDNA regions of watermelon and related wild species: implications for the evolution of Citrullus haplotypes

Sequencing analysis of one coding and four noncoding cpDNA regions was conducted to infer biogeographic and evolutionary relationships in the genus Citrullus. Eighteen taxa from diverse geographical areas were included. A low number of parsimony informative characters (1.1%) was observed at the ×4 kb section of cpDNA. Variability within Citrullus was detected primarily at noncoding regions of high A + T content. Substitution rates varied from 0–0.48% for ndhF with A + T content of 68.4% to 0.39– 1.69% for the intergenic region of atpA with A + T content of 82.8%, mainly resulting in indels and transversions. Indels at several regions acted as valuable parsimony informative markers. Citrullus lanatus var. lanatus, the cultivated watermelon, and C. ecirrhosus and C. rehmii from Namibia, lacked molecular variability. The genus Citrullus is supported monophyletically and shows two main clades, one of which contains C. colocynthis. In the other clade, C. rehmii is sister to a clade containing C. ecirrhosus and C. lanatus. Two clades were recovered within C. lanatus, consisting of domesticated watermelon and wild citron, var. citroides. Five haplotypes within C. colocynthis were used to deduce colonization routes of the species. Biogeographic patterns point to separate colonization events into Africa and the Far East.