Reduction of the Hawaiian genus Platydesma into Melicope section Pelea (Rutaceae) and notes on the monophyly of the section

Abstract Platydesma, an endemic genus to the Hawaiian Islands containing four species, has long been considered of obscure origin. Recent molecular phylogenetic studies have unequivocally placed Platydesma within the widespread genus Melicope as sister to the rest of the Hawaiian species of Melicope. This makes submerging Platydesma into Melicope necessary. We make the necessary new combinations: Melicope cornuta (Hillebr.) Appelhans, K.R. Wood & W.L. Wagner, M. cornuta var. decurrens (B.C.Stone) Appelhans, K.R. Wood & W.L. Wagner, M. remyi (Sherff) Appelhans, K.R. Wood & W.L. Wagner, and M. rostrata (Hillebr.) Appelhans, K.R. Wood & W.L. Wagner. An additional species that has been recognized within Platydesma should now be recognized under its original name M. spathulata A. Gray. All Hawaiian species belong to Melicope section Pelea. Our molecular phylogenetic studies also showed that in addition to merging Platydesma into section Pelea, five species described from New Caledonia need to be excluded from the section in order to achieve monophyly of section Pelea.


Introduction and discussion
The genus Melicope J.R. Forst. & G. Forst. is the largest genus within Rutaceae, with approximately 230 species ranging throughout the Malagasy and Indo-Himalayan regions, Southeast Asia, Australasia, and across the Pacific Islands (Hartley 2001). One of the centers of diversity is the Hawaiian Islands, where Melicope (including Platydesma) is the fourth largest radiation with 54 species after the Hawaiian Lobeliads (Campanulaceae), Cyrtandra J.R. Forst. & G. Forst. (Gesneriaceae), and mints (Lamiaceae) (Hartley and Stone 1989, Wagner et al. 1990, Hartley 2001, Wood et al. 2017. It furthermore represents the largest radiation of woody plants on the Hawaiian Islands (Wagner et al. 1990). Hawaiian Melicope are an example of an adaptive radiation, as the genus has undergone spectacular morphological and ecological diversification (Stone 1966, Carlquist 1974. Hawaiian Melicope taxa, and another rutaceous genus Platydesma H. Mann have been widely decimated throughout the Hawaiian Islands due to habitat alteration and introduced organisms; many extant species of Melicope and all taxa within Platydesma except for P. spathulata  (Wagner et al. 1999, Wood 2011, 2014, Wood et al. 2016. The immediate relationships of the Hawaiian endemic genus Platydesma (Rutaceae) have puzzled taxonomists due to the divergent floral morphology and hermaphroditic breeding system of Platydesma (Stone 1962a, Wagner et al. 1990, Funk and Wagner 1995. Engler (1931) placed Platydesma between the North American (Mexico and southwestern USA) genus Choisya Kunth and the New Caledonian genus Dutaillyea Baill., while Stone (1962a) hypothesized that the genus was derived from the Australian, New Caledonian and New Guinean genus Medicosma Hook. f. Even though Dutaillyea proved to be part of Melicope, and Medicosma is the sister genus of Melicope, neither of these taxa are immediate relatives of Platydesma and the Hawaiian Melicope species (Appelhans et al. 2014a). While most Melicope species have a dimorphic breeding system (Sakai et al. 1995, Hartley 2001, Platydesma is a peculiar taxon for its monadelphous stamens and bisexual flowers (Hillebrand 1888, Kubitzki et al. 2011). In addition, Platydesma has unique chemical characteristics; the leaves, bark and wood emit a semeniferous odor due to the unique alkaloid platydesmine (Werny and Scheuer 1963).
Recent molecular phylogenetic studies provided unequivocal evidence for the placement of Platydesma within Melicope, as sister to the Hawaiian taxa in Melicope section Pelea (Harbaugh et al. 2009, Appelhans et al. 2014a, 2014bFig. 1). These data are consistent with Carlquist's (1974) hypothesis of a Hawaiian origin of Platydesma, as well as with Asa Gray's classification of Platydesma as Melicope taxa (Stone 1962b). This represents an example of divergent evolution in an insular setting. Despite the dif- ferences in morphology and breeding system, however, the seedling stages of Melicope and Platydesma are homologous (Stone 1962b).
The Hawaiian Islands have the highest rate of dioecy in the world, evolving in situ in at least 12 lineages, possibly as a selective force to avoid inbreeding depression, affect resource allocation, and sexual selection (Sakai et al. 1995). Sakai et al. (1995) hypothesized that Hawaiian Melicope and Platydesma both arose from separate colonizations; the Melicope colonist was dimorphic while the ancestor of Platydesma was monomorphic.
The results of molecular phylogenetic analyses (Harbaugh et al. 2009, Appelhans et al. 2014a demonstrate that Hawaiian Melicope and Platydesma most likely arose from a single ancestor. The ancestor of Hawaiian Melicope + Platydesma was likely dioecious, because the closest relatives of the Hawaiian species (=the remainder of Melicope section Pelea excluding species from New Caledonia) are almost exclusively dioecious (Hartley 2001). Only the widespread M. triphylla (Lam.) Merr. and the New Guinean endemic M. conjugata T.G. Hartley, which are normally dioecious, are in rare cases monoclinous (M. triphylla, M. conjugata) or andromonoecious (M. triphylla). Therefore, Platydesma represents a rare reversal from dioecy to synoecy as Carlquist (1974) had hypothesized. Despite the differences in flower, seed and fruit characteristics (Melicope has dehiscent fruits while Platydesma has indehiscent or tardily dehiscent fruits), Sakai et al. (1995) hypothesized that both genera are insect pollinated, and have undergone long distance dispersal through bird ingestion. Therefore, the mechanisms that may have lead to this reversal in breeding system are unknown, but with discovery of large quantities of nectar produced in the flowers it is likely the breeding system change is part of the shift to bird pollination.
The copious nectar production and the stamens connate into a cup-like structure that holds accumulating nectar in Platydesma flowers suggest adaptations to bird-pollination. A similar case of adaptation to bird-pollination can be found in Hawaiian Schiedea/Alsinidendron (Caryophyllaceae) (Weller et al. 1998, Golonka et al. 2005, Wagner et al. 2005. Like in Melicope/Platydesma, adaptation to different pollination vectors has resulted in differences in flower morphology between Schiedea and Alsinidendron that are virtually identical to those in Melicope/Platydesma, which led to the separation of the group into two genera. To preserve the monophyly of Melicope section Pelea, Platydesma must be merged with Melicope and the New Caledonian species of section Pelea have to be excluded (Fig. 1). Hawaiian Melicope have been subdivided into the four groups Apocarpa, Cubicarpa, Megacarpa and Pelea, which were regarded as sections within the genus Pelea A. Gray (Stone 1969, Wagner et al. 1990). Now that Pelea occupies the rank of a section within Melicope, these four groups perhaps should be regarded as subsections. Due to the significant morphological differences between Hawaiian Melicope and Platydesma (Wagner et al. 1990), a new subsection would need to be created to accommodate Platydesma. However, since our molecular phylogenetic studies (Appelhans et al. 2014b) showed low resolution concerning some of the Hawaiian groups, it is premature to establish a new subsectional classification. Instead, we will await the results of our recently initiated Next-Generation Sequencing project focused on Hawaiian Melicope. Hartley (2001) placed the five New Caledonian species of Melicope in section Pelea. He mentions that their "immediate and broader relationships within sect. Pelea are not clear" (Hartley 2001;p. 139) and that they "are probably relicts" (Hartley 2001;p. 31).
Hartley listed some characters that might connect the New Caledonian and the Hawaiian species (e.g., persistent petals in three New Caledonian and two Hawaiian species, infertile antepetalous stamens in three New Caledonian species and rare occurrences in Hawaiian species). He also mentioned characters of the New Caledonian species that do not occur in other species of the section or the genus, among which are: abruptly acuminate flower buds, and persistent and accrescent sepals and/or petals (Hartley 2001;p. 140). Molecular phylogenetic studies (Appelhans et al. 2014a,b; Fig. 1) have shown that the New Caledonian species of section Pelea are not directly related to the Hawaiian species and the remainder of section Pelea, but that they belong to a clade of taxa from New Caledonia, Australia, New Zealand and the South Pacific, including Melicope sections Melicope p.p. and Vitiflorae (F. Muell.) T.G. Hartley p.p., as well as the genera Comptonella Baker f., Dutaillyea, Picrella Baill. and Sarcomelicope Engl. (Appelhans et al. 2014b). In line with the results of the phylogenetic reconstruction and the unusual characters mentioned by Hartley (2001), we propose to exclude the New Caledonian species from Melicope section Pelea.

Taxonomy
With four species of Platydesma included and five New Caledonian species excluded, Melicope section Pelea consists of 86 currently recognized species (Hartley 2001, Wood et al. 2017. The distribution of the section ranges from Borneo, the Philippines, Taiwan and the Ryukyu Islands to the Hawaiian and Marquesas Islands. In the Pacific, the section occurs on the Ryukyu Islands, Pohnpei (Caroline Islands), Bismarck Archipelago, Solomon Islands, Wallis and Futuna, Tonga, Samoa, Niue, Hawaiian Islands, and Marquesas Islands (Hartley 2001 ; Fig. 2). In the following synopsis sheet numbers, when available are cited for holotype specimens, and barcode numbers are provided for specimens in brackets [].
We propose the following taxonomic changes for Platydesma:   (when someone other than himself made the collection). In this case he cited three localities (Halemano, Wailupe, and Pauoa) indicating he collected or saw material from each one of them. As such these collections must be considered syntypes. Stone (1962a) indicated the K sheet and the GH sheet cited below as holotype and isotype. They represent inadvertent selection of a lectotype and isolectotype. A number of additional syntypes from all three cited localities are known but can with certainty not be considered isolecotypes because the two specimens selected by Stone  Note. These two sheets are the only type material of this collection located. They were labelled as co-type, which was used at the time Rock published for the equivalent of isotype. Rock stated in his book that most of the specimens mentioned in the text were in the "college of Hawaii Herbarium", which are now incorporated into BISH. He specifically indicated the type of this new variety to be there, so the presumed holotype should be at BISH, but could not be located. Note. At the time Hillebrand described this species there was a single specimen known, the type. When Rock brought fragments from B back to BISH the label information was not well transcribed, but all have printed labels indicating they were from the B collection. There were no other collections in the B herbarium so despite the lack of information this must be a fragment of the holotype.
Type. Kaua'i: V. Knudsen 68 (holotype: B-destroyed, fragment BISH-581794 [BISH1016395, specimen!]). 343. 1854) and also used part of the description (fruit). Hillebrand cites among the three collections the type of Pelea auriculifolia A. Gray as well as a specimen of Platydesma (Hillebrand s.n.), and the third one can't be located (Lydgate s.n.). So his cited collections are a mixture as is the description. This Hillebrand name must be taken as a new combination. The critical facts are: Hillebrand cited the basionym, used the same epithet, and did not exclude the holotype of the basionym. It does not matter that Hillebrand cited a later publication of the cited basionym. Since this is a pre-1953 publication, it is not mandatory to cite the original publication of the basionym.