A new Australian species of Luffa (Cucurbitaceae) and typification of two Australian Cucumis names, all based on specimens collected by Ferdinand Mueller in 1856

Abstract As a result of his botanical explorations in northern Australia, Ferdinand von Mueller named several Cucurbitaceae that molecular data now show to be distinct, requiring their resurrection from unjustified synonymy. We here describe and illustrate Luffa saccata F. Muell. ex I.Telford, validating a manuscript name listed under Luffa graveolens Roxb. since 1859, and we lectotypify Cucumis picrocarpus F. Muell. and Cucumis jucundus F. Muell. The lectotype of the name Cucumis jucundus, a synonym of Cucumis melo, is mounted on the same sheet as the lectotype of Cucumis picrocarpus, which is the sister species of the cultivated Cucumis melo as shown in a recent publication.


Introduction
Ferdinand von Mueller (1825Mueller ( -1896 was the botanist on the North Australian Exploring Expedition that in 1855 and 1856 explored North Australia under the command of A. C. Gregory (Gregory 1858; maps of the expedition are available at http://nla.gov. au/nla.map-rm2807). In mid-September 1855, the expedition's two ships reached the mouth of the Victoria River, and the explorers then spent eight months exploring the surrounding country. They started their return journey on 21 June 1856. Mueller is not known to have visited the Northern Territory again later (see Orchard 1999, and literature cited there), and although many of his specimens are undated they can be associated with confidence with the North Australian Exploring Expedition on the basis of the locality data.
Many new taxa were collected on that expedition, including two new species of melon described as Cucumis jucundus F. Muell. and C. picrocarpus F. Muell. (Mueller 1859). Both names were subsumed into C. trigonus Roxb. by Bentham (1866), who however noted that they might be forms of C. melo L. In 1993 (pp. 104 and 114), Kirkbride included both as synonyms under C. melo without assigning them to a definite infraspecific taxon. Some of the specimens in CANB and MEL thus referred to C. melo had earlier (in 1986) been annotated by Charles Jeffrey as C. melo subsp. nov., and Kirkbride (1993) also commented on the Australian material's polymorphism in the degree of leaf dissection and the indumentum of the hypanthium of female flowers.
A phylogenetic reconstruction of Cucumis that includes over 100 accessions from Asia and Australia now indicates that some of the Australian material previously referred to C. melo constitutes distinct species (Sebastian et al. 2010). The molecular data show that Australia harbors seven native species of Cucumis, five of them new to science and described elsewhere (Telford et al. 2011). Examination of Mueller's collections and protologues indicates that Mueller's name C. picrocarpus applies to the Australian sister species of the worldwide crop C. melo (and its wild progenitor forms native in India), while Mueller's C. jucundus is a synonym of C. melo. The possible importance to plant breeders of this Australian sister to C. melo makes it expedient to designate the types of Mueller's Australian Cucumis names, a task carried out here.
Ongoing molecular-phylogenetic work on Luffa vindicates another of Mueller's suspected new species, this one never formally described by him. Like the two melon species, he discovered it in the Victoria River region, and there are at least two specimens labeled by Mueller as 'Luffa saccata.' Mueller's manuscript name was listed as a synonym under L. graveolens Roxb. by Naudin (1859), a famed Cucurbitaceae specialist, with the result that Mueller's name went unnoticed for the next 150 years. Luffa graveolens occurs in India, Nepal, Bangladesh, and Burma, and is morphologically distinct from the Australian species. We here validate the name L. saccata, describe the morphological differences between L. graveolens and L. saccata, and provide illustrations. . A second sheet at GH from "Depot Creek" and one at MEL000592947 from "Victoria River, Depot Creek" are possible further isolectotypes. Kirkbride (1993) designated a sheet in the Kew Herbarium (now K000634446; our Fig. 1) as 'neotype' of Cucumis jucundus and another one (now K000634445) as 'neoisotype' [sic]. In fact, the material on K000634446 represents two species, C. jucundus and C. picrocarpus, both discovered by Mueller during the North Australian Exploring Expedition and described by him (Mueller 1859) on the basis of his own collections. Kirkbride's 'neotype' thus comprises original material for both C. jucundus and C. picrocarpus. Under the Vienna Code (McNeill et al. 2006: Art. 9.8), the term neotype is, in such cases, correctable to lectotype. However, the sheet referred to by Kirkbride is not a specimen as defined in the Code (Art. 8.2), because it does not consist of 'a single species'. A "second stage lectotypification" as provided for in Art. 9.12 is therefore necessary, to ensure that the name Cucumis jucundus F. Muell. remains attached to those plant parts on K000634446 that "correspond most nearly with the original description or diagnosis." At least five other sheets annotated by Mueller as Cucumis jucundus or Cucurbita jucunda and coming from different collecting localities (without collecting dates) are kept at MEL, the herbarium of Mueller's home institution, and the protologue statement about the geographic range of C. jucundus is accordingly broad: 'In Arnhem's Land and on the Gulf of Carpentaria, particularly on the banks of rivers, also in eastern tropical Australia, and in Central Australia observed with certainty as far south as Cooper's River' (Mueller 1859: 45). The sheet MEL000592946, with male flowers, was collected at 'Victoria River' and is undoubtedly a duplicate of our lectotype, same as one of the two specimens kept at GH. A second sheet at GH from "Depot Creek" and one at MEL from "Victoria River, Depot Creek" are possible further isolectotypes.

Description of
Representative specimens examined. AUSTRALIA. Western Australia: Fitzroy River floodplain, river road from Minnie River bridge to Udialla homestead, 27 Apr.  Habitat. Luffa saccata grows in riverine or littoral habitats on sand or clay, sometimes on rocky ridges of limestone or sandstone to 300 m of altitude. Associated species recorded include Eucalyptus camaldulensis, Melaleuca leucadendra and Barringtonia acutangula in gallery forest or woodland, and Eucalyptus miniata, Adansonia gregorii, Brachychiton spp. and Triodia spp. on ridges and littoral Cenchrus grassland.
Phenology. Flowers and fruits March to October. Conservation Status. The species is widespread and common, and we therefore do not consider it at risk. Conserved in Mitchell River and Bungle Bungle Ntional Parks in Western Australia and Gregory National Park in the Northern Territory.
Etymology. From Latin saccatus, bag-like, obviously in reference to the fruit (Figs 2, 3). Notes. The MEL holotype has two labels in Mueller's handwriting, one with 'Luffa saccata Baines Creek, May 1856', the other with 'L. graveolens, Tributaries of the Victoria River, N.W. Australia, May 1856,' the latter obviously attached after communication with, or reading of, Naudin (1859). It is surprising that Naudin failed to accept Mueller's Australian Luffa as a good species, since C.B. Clarke , who knew the Indian cucurbits well, made a note on one of the three Kew specimen, saying 'not near [Luffa] graveolens which has the males [male flowers] on very short subfasciculate pedicels.' This is indeed one of the differences between the Indian and the Australian species, the latter having the male flowers mostly in elongate racemes. Detailed measurements of living Indian L. graveolens plants, black and white photos, and observations on their chromosome numbers are contained in Roy (1969, 1971).
No material of Luffa graveolens from India is held in the following major herbaria: CGE, E, GH, L, MO, NY, US. This lack of material in western herbaria probably contributed to the Indian and Australian species having been confused for so long. The confusion also affected a recent treatment of Cucurbitaceae in the Flora Malesiana series (De Wilde and Duyfjes 2010), which states that Luffa aegyptiaca forma sylvestris (Miq.) W.J.de Wilde & Duyfjes is common in Australia (and elsewhere) and comprises "all wild-growing and naturalized small-fruited feral forms" of Luffa. Several Australian specimens of Luffa saccata, such as Sands 4499 (Fig. 3), thus are annotated as Luffa aegyptiaca forma sylvestris. The Luffa specialist C.B. Heiser, on the other hand, cultivated both Australian species, L. aegyptiaca and L. saccata (under the name L. graveolens), and distinguished them without hesitation (Heiser and Schilling 1988).