on the morphology and molecular basis of segregation of Ceriops zippeliana and C . decandra ( Rhizophoraceae ) from Asia

Ceriops zippeliana, a member of the mangrove Rhizophoraceae, was first reported in 1849. It was considered to be a synonym of C. decandra, which is still widely accepted. We present morphological and molecular evidence to show that C. zippeliana is significantly distinct from C. decandra, and illustrations and an identification key to both species.

Total DNA was extracted using the CTAB method (Doyle & Doyle 1987) and the trnL intron of chloroplast DNA was amplified using universal primers (Taberlet et al.1991), following the protocols of Tsai et al. (2006).The DNA was sequenced following the method of dideoxy chain-termination using an ABI377 automated sequencer with the Ready Reaction Kit (PE Biosystems, California) of the BigDye™ Terminator Cycle Sequencing.
DNA sequence alignment was conducted using Clustal W in BioEdit (Hall 1999).Genetic relationships were determined using MEGA v2.1 (Kumar et al. 2001).A genetic distance matrix was calculated using the two-parameter model of Kimura (1980), and then used to construct the phylogenetic trees using the Neighbor-joining (NJ) method (Saitou & Nei 1987).Bootstrapping (1 000 replicates) was carried out to estimate the support for the topology (Felsenstein 1985, Hillis & Bull 1993).All characters were equally weighted.
Distribution -West coast of southern Malay Peninsula, Singapore, Bintan Island, east coast of the Malay Peninsula to the Gulf of Thailand to Vietnam, Borneo, Java, Philippines, Sulawesi, Lesser Sunda Islands, Moluccas (Ceram).
Typification -Blume (1849) did not cite any specimens for his new species C. zippeliana.Type specimens were designated by Hou (1958) as he revised the Rhizophoraceae for the Flora of Malesiana.The type specimens have four sheets collected by Zippelius from 'Nov.Guinea' (currently Moluccas and Irian Jaya) and these were deposited at L (2 sheets, one with the collection number '99/a' indicated as 'Type!', while the other one without collection number labelled as 'Type Dupl.'),K ('99/a' without label) and U ('99.a' labelled as 'Type Dupl.'), separately.However, it is apparent that two species have been included.The two type specimens at L are characterized by leaves which are elliptic-oblong in shape and fruit with a domelike calyx tube.These characters are obviously different from those of the other two type specimens at K and U, which have oval to elliptical-oval leaves and a shallow disc-like calyx tube.Blume's original descriptions of "foliis obovatis v. obovalibus" and "pedunculis brevissimis paucifloris" match the characters of the latter specimens as well as our recently collected specimens, and we therefore select the specimens at K and U as the lectotype and isolectotype of C. zippeliana.
A detailed comparison of the two species is provided in Table 2.
The phylogenetic tree is shown in Fig. 6, with bootstrap values over 50 % indicated.In this tree, accessions of C. zippeliana are separated from all those of C. decandra in a clade supported by 88 % bootstrap value.Thus, the molecular data provide further evidence that C. zippeliana is distinct from C. decandra.

dISCuSSIon
Reproductive characteristics would be useful tools to separate these two morphologically similar species of the genus Ceriops.
In the past, most taxonomists had focused on the fringe-like apex of the petals, but not on the marginal hairs of C. decan-dra petals (Griffith 1835, Arnott 1838, Hou 1958, Tomlinson 1986, Banerjee et al. 1989, Naskar & Mandal 1999).Although this character is depicted in Roxburgh's drawing (Rhizophora decandra, Roxb.Icon.Ind. 1140) at K and CAL, Arnott (1838) thought that this was probably a mistake due to "the petal appearing ciliated with scattered longish hairs or bristles round the whole margin".Based on our observations, the flower in Roxburgh's drawing is correctly depicted except that the marginal hairs are perhaps too long.By contrast, in C. zippeliana there are no visible lateral marginal hairs on the petals.Using a scanning electron microscope, only a few loosely arranged extremely short hairs (less than 50 μm) have been observed (Sheue 2003).
Except for the genus Bruguiera, all flowers of the Rhizophoraceae are enclosed in a bracteole, two flowers with their own  -----------A bracteoles are again surrounded by a bract, and a certain number of such replicates constitute a compound bifurcate cyme-like inflorescence (Sheue 2003, Naskar & Mandal 1999).
The difference in inflorescence morphology for the two compared species in this study is distinct.Ceriops zippeliana has a simple head-like structure with a single layer of (primary) bracts enclosing 3 -5 flowers, while that of C. decandra is dense, bifurcate, cyme-like, with multilayered (primary to fourth) bracts enclosing 6 -10 or more flowers.The bifurcate cymelike inflorescence of C. decandra is actually similar to that of C. tagal, but the pedicels of the former are absent and lead to a "head-like cyme" (Sheue 2003).However, due to the obscure arrangement of the small sized bracts, this difference is not easily recognized, especially for desiccated specimens.
Leaf shape and stipule length provide additional information for identification.In addition, the colleters at the adaxial base of the stipule of Ceriops could serve as a diagnostic character in the field with the help of hand lens (10×), as Sheue et al. (2003Sheue et al. ( , 2005) ) reported for the species of Kandelia and Bruguiera.But it should be mentioned that only fully expanded stipules (before dropping off the stem) can be used for comparison.
While flowering, the flowers of C. zippeliana only open slightly with the erect and reflex apex of calyx lobes pointing towards the floral axis.In contrast, the flowers of C. decandra are partially patent with oblique calyx lobes.There are marked differences in pollen grain characters, such as size and surface ornamentation.Pollen size is less in C. zippeliana (15.43 ± 1.16 μm) than in C. decandra (21.0 ± 1.49 μm), and surface ornamentation in the former is here confirmed to be the rugulate-type, while in the latter it is the scabrate-type (Das & Ghose 1990).
The ovoid fruit of Ceriops has persistent calyx tubes and lobes and their detailed and distinct surface ornamental patterns are useful for interspecific differentiation (Sheue 2003).Ceriops decandra has dome-like calyx tubes, while C. zippeliana has shallow disc-like calyx tubes.The calyx lobes of C. zippeliana are very short and erect when compared to the relative longer and ascending calyx lobes of C. decandra.It is interesting, especially for the dispersal ecology, that the hypocotyl orientation of the two species ranges from erect, to ascending or descending in both species (Sheue 2003), while those of the other species of the Rhizophoraceae are descending.Tan et al. (2005) studied the genetic structure of ten populations from the Malay Peninsula and North Australia of the so-called C. decandra using the inter-simple sequence repeat (ISSR).They concluded that the populations could be grouped into three major geographic regions, i.e., West coast of West Malaya, Southwest Malaya (including Singapore) and East Malaya, and North Australia.The populations they sampled from Southwest Malaya (including Singapore) and East Malaya are probably C. zippeliana.A similar genetic discontinuity between the Asiatic populations of Ceriops was observed in this study as well using the chloroplast trnL intron.Three stable indels and two stable transversions and the phylogenetic tree derived from these data also support the idea that C. zippeliana can be separated as a distinct species from C. decandra.We find that C. zippeliana is found throughout a large part of Malesia.Hou (1958)  Further ongoing taxonomic work pertaining to Ceriops is in progress in order to clarify the species number and the phytogeographic range of each taxon.

Fig. 2 Fig. 3
Fig. 2 Ceriops zippeliana Blume.a. Fruiting shoot; b. stipule with colleters at adaxial base; c. top view of bracteole with four colleters; d. stamens; e, f. flower lateral views, the top one showing style with perianth and stamens removed; g. inflorescence; h.sepal of adaxial side; i. petals of adaxial (left) and abaxial (right) views; j. fruit with persistent calyx tube, calyx lobes and cotyledon collar; k. tip of hypocotyl with plumule.

Fig. 4
Fig. 4 A comparison of leaves, inflorescences and fruits between Ceriops decandra (1) and C. zippeliana (2) (modified from Sheue 2003).a. Leaves, oval to obovate leaves of C. decandra; oval to elliptical-oval leaves of C. zippeliana; b. inflorescence, dense bifurcate cymelike with multilayered (primary to fourth) bracts of C. decandra; simple head-like with a single layered (primary) bract of C. zippeliana; c. fruits, calyx tube domed with longer ascending calyx lobes of C. decandra; calyx tube shallow disc-like with short and erect calyx lobes of C. zippeliana; d. hypocotyls and their orientations; hypocotyl width approximately the same tapering towards a blunt apex of C. decandra; hypocotyl width unequal thick, tappering towards an acute apex in C. zippeliana; the orientations of hypocotyl for these two taxa ranging from lifted to pendent as the diagram shows.

Fig. 5
Fig. 5The variable sites of the trnL intron.Three insertions/deletions are underlined in the site number between Ceriops decandra and C. zippeliana.

Fig. 6 Fig. 7
Fig.6The neighbour-joining tree of the seven accessions of both Ceriops decandra and C. zippeliana plus the two outgroups derived from the trnL intron sequence.Bootstrap values > 50 % are shown on each branch.
(Kitamura et al. 2004liana was not found in Sumatra and the Lesser Sunda Islands.However, we have examined specimens of C. zippeliana collected from the Lesser Sunda Islands, such as Bali and Lombok Island, and our findings are confirmed by a local mangrove handbook(Kitamura et al. 2004) which shows the photographs of C. zippeliana (labelled as C. decandra) in this region.Based on our field observations, the northern boundary of C. zippeliana may be located in Malacca, West Malaysia.According to the herbarium specimens examined, Satun, a small province in the south of Thailand that borders Malaysia, is the southern boundary of C. decandra at the West coast of the Malay Peninsula.Nevertheless, more research, especially extensive field survey, is still needed to elucidate the population boundaries of C. decandra and C. zippeliana in the Malay Peninsula.