Austrobryonia (Cucurbitaceae), a New Australian Endemic Genus, is the Closest Living Relative to the Eurasian and Mediterranean Bryonia and Ecballium

The Cucurbitaceae genus Austrobryonia, with four species endemic to Australia, is described, illustrated, and placed in a phylogenetic context based on molecular and morphological data. In the Flora of Australia (Telford 1982), all four species were provisionally included in Mukia, but not formally described. Austrobryonia argillicola, A. centralis, A. micrantha, and A. pilbarensis are adapted to arid central regions of Australia. All species are known from several (7–27) localities, and their distributional ranges are allopatric. A phylogenetic analysis of plastid and nuclear DNA sequences that includes all four species in a family-wide context revealed that Austrobryonia is the closest living relative to a Eurasian and Mediterranean clade consisting of Bryonia L. and Ecballium L. An rbcL molecular clock, calibrated with Cucurbitaceae fossils, dates this rare biogeographic disjunction to minimally 42 my ago (with an error of ca. ± 25%), while the crown group of Austrobryonia may be about 8 my old. Keywords—Australia-Eurasia disjunction, molecular clock, Mukia micrantha. The Australian continent is relatively poor in genera and species of Cucurbitaceae. The last comprehensive treatment (Telford 1982) lists 44 species, then assigned to 17 genera. Twelve genera and perhaps 30 species appear indigenous to the Australian continent (Benincasa Savi, Cucumis L., Diplocyclos (Endl.) Post & Kuntze, Luffa Miller, Momordica L., Muellerargia Cogn., Neoachmandra W. J. de Wilde & Duyfjes, Neoalsomitra Hutch., Nothoalsomitra I. Telford, Sicyos L., Trichosanthes L., and the new genus Austrobryonia described below), but only one genus was thought endemic to Australia. This is the monotypic Nothoalsomitra, a liana of SE Queensland’s rain forests, originally described in Benincaseae (Telford 1982), but placing far from them in a molecular tree of Cucurbitaceae that samples 21% of the family’s 800 species and 94% of its 130 genera, including species from all Australian clades (Kocyan et al. 2007). The most problematic group in the Flora of Australia treatment of Cucurbitaceae was the genus Mukia Arnold. As previously conceived, Mukia was thought to include nine species, five that range from China through Indo-China southeast to Java, Borneo, and the Philippines, and west through India, Pakistan, and the Yemen into subSaharan Africa, and five that were left undescribed (Telford 1982; De Wilde and Duyfjes 2006). The type species of Mukia, M. scabrella (L.) Arn. (= M. maderaspatana (L.) M. Roem.), occurs from subSaharan Africa, through tropical and subtropical Asia to Australia. The pollen of the Asian Mukia species is so different from that of the Australian M. micrantha (F. Muell.) F. Muell. that Jeffrey (1969a) suspected they might not be congeneric. He thought the affinities of M. micrantha might lie with Zehneria Endl., Kedrostis Medik., or Ibervillea Greene, but could not decide about its true taxonomic position. Earlier, Ferdinand von Mueller (1854) had first included M. micrantha in the otherwise neotropical genus Cucurbita, then in the African/ Asian Zehneria (Mueller 1858/59), and finally in Mukia (Mueller 1860/61), but without specifying the basis for these placements. Naudin (1859) redescribed the species as Cucumis muelleri, but again without providing any reasons for this generic placement. Given the extremely few collections of what appeared to be as many as five species, Telford (1982) refrained from erecting a new genus and instead provisionally discussed the new species as Mukia sp. A, Mukia sp. B, Mukia sp. C, Mukia sp. D, and Mukia sp. E (Telford 1982, pp. 182–187). Over the past quarter of a century, additional flowering and fruiting collections have become available that now permit complete descriptions, including information on phenology and natural ranges. Molecular work also clarified the affinities of Mukia, Zehneria, and the five new species from Australia. Our molecular phylogenetic analyses indicate that M. maderaspatana, M. javanica (Miq.) C. Jeffrey, Mukia sp. A, and Mukia sp. B belong in the genus Cucumis (Renner et al. 2007), while M. micrantha and Mukia sp. C, D, and E belong in a new genus Austrobryonia that forms the sister clade to a Eurasian/Mediterranean clade comprising Bryonia L. and Ecballium L. We here formally describe the new genus and species, present the first molecular phylogenies that include these new taxa, and, based on a molecular clock analysis, discuss the possible origin of this unexpected Australian/ Mediterranean/Eurasian disjunction. MATERIALS AND METHODS Morphology—Specimens were borrowed from BRI, CANB and NSW. Measurements were taken from rehydrated dried herbarium specimens. Data Generation and Retrieval—DNA extraction and sequencing followed standard procedures, using the rbcL, matK, rpl20–rps12, and trnL and trnL–F primers listed in Kocyan et al. (2007). In addition to these plastid regions, we sequenced the nuclear internal transcribed spacer region (347 nt of ITS 1, 162 nt of the 5.8S gene, and 318 nt of ITS 2) using the ITS primers of Balthazar et al. (2000). Direct PCR amplification of ITS yielded single bands and unambiguous base calls. Thirty-six sequences were newly generated for this study (GenBank accession numbers EF 487543–EF 487578; Appendix 1). Sequence Alignment and Phylogenetic Analyses—Sequences were edited with Sequencher (4.6; Gene Codes, Ann Arbor, Michigan, USA) and aligned by eye, using MacClade 4.06 (Maddison and Maddison 2003). The aligned plastid matrix comprises 4692 nucleotides after exclusion of a poly-A run in the trnL intron and a TATATA microsatellite region in the trnL–F intergenic spacer. The aligned ITS matrix comprises 829 nucleotides. Data matrices and trees have been deposited in TreeBASE (study number S1817). Equally weighted parsimony analyses for matrices of nucleotides were conducted using PAUP 4.0b10 (Swofford 2002). The search strategy involved 100 random addition replicates with TBR branch swapping, saving all optimal trees. For the parsimony analyses, gaps were treated as missing data, for maximum likelihood (ML) searches gaps were not removed. To assess node support, parsimony bootstrap analyses were performed using 1000 replicate heuristic searches, each with 10 random addition replicates and TBR branch swapping, saving all optimal trees. Systematic Botany (2008), 33(1): pp. 125–132 © Copyright 2008 by the American Society of Plant Taxonomists