AcmenA , AcmenospermA , cleistocAlyx , piliocAlyx And WAterhouseA FormAlly trAnsFerred to syzygium ( myrtAceAe )

Based on morphological and molecular data, we have concluded that Acmena, Acmenosperma, Cleistocalyx, Piliocalyx and Waterhousea should be combined with Syzygium. The five genera are formally transferred to Syzygium and new names provided for those of their accepted representatives for which names are as yet not available in Syzygium, with the exception of the new Caledonian members of Piliocalyx as these currently are being revised.


inTrOdUCTiOn
The myrtle flora of the Old World rain forests is dominated by Syzygium Gaertn.and its generic allies.Current estimates are that these genera, Acmena dC., Acmenosperma Kausel, Cleistocalyx Blume, Piliocalyx Brongn.& Gris, Syzygium Gaertn.and Water housea B. Hyland, together contain in the order of 1200 species. in habit, they range from canopy-emergent trees to canopy trees, understorey trees, treelets and shrubs, and occur from littoral communities at sea level through swamp forests, lowland and montane forests to subalpine shrubberies.They have high ecosystem significance as their (often massed) nectariferous flowers and (usually) fleshy fruit are food sources for a wide range of animals, from small insects through to cassowaries and primates.Several species, e.g. S. aqueum (Burm.f.) Alston, S. cumini (L.) Skeels and S. malaccense (L.) Merr.& L.M. Perry are widely cultivated, especially in the Malesian region, for their edible fruit, and one species, S. aromaticum (L.) Merr.& L.M. Perry, is extensively cultivated for its flower buds that are dried and used as the spice, clove. in some areas, especially New Guinea, species of the group can form a significant proportion of the logs extracted during forestry operations although the timber is not highly valued for cabinet or veneer work and is utilised for general purpose construction and as a filler in laminated boards.
The basis for floristic work on the Syzygium generic group within the Indo-Pacific region during the past 50 years has been the taxonomic studies by E.d.Merrill and/or L.M. Perry (references given in Schmid, 1972a, andin Craven, 2001).The generic concepts of Merrill & Perry were adopted by the majority of taxonomists with the notable exceptions of Henderson (1949), Kochummen (1978) and some others who continued to treat all species within Eugenia L. as that genus had been circumscribed by Bentham & Hooker (1865).Merrill & Perry's concepts largely are congruent with those of niedenzu (1893) and diels (1922) in that the Old World species of the Syzy gium alliance generically were separate from the Eugenia generic alliance.Schmid's research into generic delimitation of the Eugenia-Syzygium complex indicated that, while Eugenia and the Syzygium group were well distinct, there was little support for maintenance of the several generic segregates from Syzygium (Schmid, 1972a, b).The diagnostic features of Acmena, Acmenosperma, Cleistocalyx, Piliocalyx, Syzygium and Waterhousea are given in tabular form in Craven (2001: Table 1).
The species of the various syzygioid genera, especially within a local region, are generally clear and easy to distinguish.It is higher in the classification that problems arise as attempts are made to inter-relate the species groups and/or to classify them on the basis of features such as presence/absence of a calyptra, stamen number, anther sac orientation, and the presence of intercotyledonary tissue in the seed.Systematists have increasingly been taking advantage of the opportunities provided by analysis of molecular sequence data to gain fresh perspectives on the phylogenetic relationships between the groups of species of their particular interest and the evolution and distribution of morphological traits.Phylogenies inferred from molecular sequence data have now been obtained for comprehensive samplings of the Syzygium group of genera (Harrington & Gadek, 2004, ITS and ETS regions of nuclear ribosomal DNA; Biffin et al., in press a, in press b, matK, ndhF and rpl16 regions of chloroplast dnA).These provide support for the view that the conventionally accepted genera of the Syzygium group are not robust and that homoplasy exists in conventional characters. in particular, the occurrence of calyptra, the sole defining character for Cleistocalyx, is widely distributed across the Syzygium group (fig.1).That Cleistocalyx is not a monophyletic group is supported also by numerical and phylogenetic analysis of morphological data.Parnell (1999) analysed morphological data obtained from Thai species of Acmena, Cleistocalyx and Syzygium.Even with such a geographically restricted sampling of species, Cleistocalyx was not demonstrated by phylogenetic analysis to be monophyletic and in the numerical analysis its species did not form a cohesive grouping separate from the Syzygium species.
The distribution of other characters used in the definition of genera similarly are not reflective of monophyly in higher level species groupings (Biffin et al., in press b).developmental studies indicate that the intercotyledonary tissues, that are such a distinctive feature in the seeds of Acmena, Piliocalyx and Waterhousea, may be homologous with tissues that surround the cotyledons in many Syzygium species and the situation with respect to the condition in Acmenosperma is also being investigated (Biffin, unpublished data).The distribution of intercotyledonary tissues is indicated in fig. 1.
Our conclusions are that Acmena, Acmenosperma, Piliocalyx and Waterhousea are better included in Syzygium than being maintained as distinct genera.Even if Acmena and Piliocalyx were to be merged, the resultant grouping is still nested within Syzy gium.That monophyletic groupings equating to the generic concepts of Piliocalyx and Waterhousea occur at a low topological level in a cladogram is not in itself an argument for the continued taxonomic recognition of these groups at the rank of genus.There are other equally strongly supported monophyletic groupings that similarly could be argued as warranting generic recognition, e. .The morphological markers available to characterise these monophyletic groups are few in number and may not hold up across the whole complex, in which case unambiguous taxonomic circumscription would not be possible.Should classification proceed in this direction, the end result would be the establishment of a very large number of finely delimited genera and a residue of species that would be difficult to fit into any of the microgenera.Two botanical dicta are worth mentioning in this context: 1) genera are evolutionary units and consequently exhibit variation, i.e. the concept of patio ludens applies; and 2) the ability to distinguish associations of species does not lead to an automatic conclusion that they should be given taxonomic recognition.valid names already exist in Syzygium for several species of the genera concerned, i.e. in some species of Acmena, Acmenosperma, Cleistocalyx and Waterhousea, and names are provided below for most of those species not yet transferred.The exceptions are the new Caledonian species of Piliocalyx.J.W. dawson, Wellington, presently is revising the new Caledonian representatives of Piliocalyx for 'flore de la nouvelle-Calédonie' (dawson, pers. comm.) and it would be premature to make new combinations before the species level taxonomy is settled.Syzygium, even without the addition of the relatively small number of species involved in the present recircumscription, is extremely speciose and devising an infrageneric classification will be a priority once current studies into embryology and fruit development are concluded.
Unless indicated by '!', the specimens cited below have not been seen.note -A leaf rubbing made by Merrill of Meyer 2598, presumably from the holotype, is at A (Emily Wood, pers.comm.; xerox seen).The rubbing is accompanied by a leaf of Cuming 1275 which, along with Cuming 1325 and Cuming 1710, represents the same species (fide Merrill, in herb.sched.at A).An isotype of Meyer 2598 is not at K and a search needs to be made in other herbaria in the effort to locate duplicate material.failing the existence of such material, it would be possible to lectotypify the name Eugenia arcuatinervia Merr.with the leaf rubbing at A (on the basis that this constitutes part of the original materials used by the author), and designate another gathering as the iconotype.for the latter purpose, Cuming 1275, 1325 and 1710 may be worthy of consideration given that Merrill identified them as representing the species in question.