Phylogenetic fragrance patterns in Nicotiana sections Alatae and Suaveolentes.

We analyzed floral volatiles from eight tobacco species (Nicotiana; Solanaceae) including newly discovered Brazilian taxa (Nicotiana mutabilis and "Rastroensis") in section Alatae. Eighty-four compounds were found, including mono- and sesquiterpenoids, nitrogenous compounds, benzenoid and aliphatic alcohols, aldehydes and esters. Floral scent from recent accessions of Nicotiana alata, Nicotiana bonariensis and Nicotiana langsdorffii differed from previously published data, suggesting intraspecific variation in scent composition at the level of biosynthetic class. Newly discovered taxa in Alatae, like their relatives, emit large amounts of 1,8-cineole and smaller amounts of monoterpenes on a nocturnal rhythm, constituting a chemical synapomorphy for this lineage. Fragrance data from three species of Nicotiana sect. Suaveolentes, the sister group of Alatae, (two Australian species: N. cavicola, N. ingulba; one African species: N. africana), were compared to previously reported data from their close relative, N. suaveolens. Like N. suaveolens, N. cavicola and N. ingulba emit fragrances dominated by benzenoids and phenylpropanoids, whereas the flowers of N. africana lacked a distinct floral scent and instead emitted only small amounts of an aliphatic methyl ester from foliage. Interestingly, this ester also is emitted from foliage of N. longiflora and N. plumbaginifolia (both in section Alatae s.l.), which share a common ancestor with N. africana. This result, combined with the synapomorphic pattern of 1,8 cineole emission in Alatae s.s., suggests that phylogenetic signal explains a major component of fragrance composition among tobacco species in sections Alatae and Suaveolentes. At the intraspecific level, interpopulational scent variation is widespread in sect. Alatae, and may reflect edaphic specialization, introgression, local pollinator shifts, genetic drift or artificial selection in cultivation. Further studies with genetically and geographically well-defined populations are needed to distinguish between these possibilities.