Fifty-four Lethrinus olivaceus individuals were sampled in 2006–2011 from ten localities in the Indo-West Pacific (Fig. 2) as part as an ongoing program on the multiple-species phylogeography of the Lethrinidae. Identification to species was made on the basis of the distinctive morphological and meristic features given by K.E.C. and G.R. Allen [2], which include the ratio of head length to body depth and the count of scale rows between the lateral line and the basis of the first dorsal fin. Voucher specimens (Fig. 1, Appendix 1) included one individual from Sorong (West Papua) shipped frozen to LIPI-Oseanografi in Ancol (Indonesia), one specimen from Moruroa (French Polynesia) shipped frozen to IRD UR 227 in Nouméa (New Caledonia) and subsequently fixed in formalin and deposited at Museum national d’histoire naturelle, Paris (France) under collection no. MNHN 2009-0016, and the skulls and jawbones of 36 individuals from the Maldives, Bali, West Papua, New Caledonia, and Moruroa, now preserved at IRD-UR 227 in Montpellier (France). Two L. microdon individuals were also sampled from the Baa atoll (Maldives) and the Kedonganan fish market (Bali), respectively.

The samples analyzed for morphometric variation consisted of heads or carcasses of L. olivaceus purchased from the Male (Maldives) (N = 1), Kedonganan (N = 4), Sorong (N = 1), and Nouméa (N = 22) fish markets in 2007–2008. To prepare the neurocrania and the jawbones, the specimens were boiled, and the boiled skin and flesh were then separated from the bone using forks, fingers, twizzles, toothpicks, toothbrushes and a wash bottle. The specimens sampled in the Moruroa lagoon, June 2006 (N = 8) were shipped frozen to Noumea and the heads were spared for similar processing. Neurocrania of individuals from Bali ranged in condylo-basal length from 81.2 mm to 111.9 mm; those from New Caledonia, from 80.9 mm to 102.2 mm; and those from Moruroa, from 86.6 mm to 126.3 mm. The neurocranium of the single specimen examined from the Maldives was 130.8 mm long; that of the single specimen examined from West Papua was 129.8 mm long. Nine morphometric variables (Fig. 3A) were measured on the neurocranium and jawbones. To normalize the variables and standardize them for size, prior to analysis, each was log-transformed and divided by the log-transformed condylo-basal length [11,12]. We used principal component analysis (PCA) [13] to determine whether longface emperor populations were craniometrically distinct. Principal component analysis was run using the ADE-4 software [14].

The choice of the neurocranium (instead of, e.g., the whole fish) to study geographic variation in morphology in L. olivaceus was dictated by site accessibility and size of specimens. In some of the places where they were sampled (Maldives and New Caledonia), emperors were filleted and the carcasses were sold separately, hampering the provision of entire specimens. Specimens of this species are usually of such large size that preserving them whole and shipping them poses a challenge when operating from remote field sites. Neurocrania, however, have the advantage that once cleaned and dried, they do not necessitate specific preservation conditions nor much storing space.

Fin clips of a number of samples (N = 28; Appendix 1) were preserved dry or in 95% ethanol and an aliquot of these was shipped to the Marine Phylogenomics Laboratory, Taipei, for DNA extraction and sequencing analysis. DNA was extracted from subsamples of the fin clips using the Qiagen DNAeasy extraction kit (Qiagen, Valencia CA, USA) according to the manufacturer's instructions. The concentration of extracted DNA was measured by spectrophotometry using an Eppendorf spectrophotometer (Eppendorf, Hamburg, Germany). Conditions for amplification by polymerase chain reaction (PCR) were as follows: 0.5 U GoTaq^® Flexi DNA polymerase (Promega, Madison, U.S.A.), 1× reaction buffer, 2 mM MgCl₂, 200 μM each dNTP, 0.2 μM each primer, and 20–50 ng genomic DNA in a 25-μl final reaction volume. Primers MT14673F (5′-TAACCAGGACTAATGGCTTGAAA-3′) and Let15856R (5′-GAGCTACTAGTGCATNGTCAKCC-3′) designed by W.-J.C. (this study) were used for amplifying the entire cytochrome b gene. Thermocycling conditions for PCR were: initial denaturing step at 95 °C for 4 min followed by 35 cycles of 95 °C (for 40 s), annealing at 55 °C (for 40 s), and 72 °C (for 1.5 min), and a final extension step of 72 °C (for 7 min) before thawing at 4 °C. The PCR product was purified using AMPure magnetic beads (Agencourt Bioscience Corp., Beverly, MA, USA) and resuspended in 30 μL of sterile water. Nucleotide sequences of the PCR-amplified DNA fragments were then determined using the sequencing facilities of the National Taiwan University and Genomics Inc. (Taipei, Taiwan) using ABI 3730xl analyzers (Applied Biosystems, Foster City CA, USA). Nucleotide sequences were edited and managed using the CodonCode Aligner v. 3.0.7 (CodonCode Corporation, Centerville MA) and Se-Al v. 2.0a11 [15] together with sequences AF381265 (L. microdon), AF381252 (L. olivaceus), AF381253 (L. reticulatus), and AF24075 and AF381263 (L. rubioperculatus) retrieved from GenBank (http://www.ncbi.nlm.nih.gov). The final matrix of aligned sequences used for phylogenetic analysis, representing a total of 33 haplotypes, was 1107 base pairs in length. L. microdon, L. reticulatus and L. rubrioperculatus were selected as outgroups because of their close evolutionary affinity to L. olivaceus [16]. Phylogenetic analyses were based on a partitioned maximum likelihood (ML) method as implemented in RAxML 7.0.4 [17]. A mixed model analysis was used that allows the independent estimation of individual models of nucleotide substitution for each partition (by codon position). A GTR+G model for each partition was used and an Optimal ML tree search was made with 100 separate runs using the default algorithm of the program from a random starting tree for each run. The final tree was selected among suboptimal trees in each run by comparing likelihood scores under the GTR+G model. To evaluate the robustness of the internal branches of the ML tree, 1,000 bootstrap replications were run for the data set under the GTR+G model. Ancillary descriptive statistics were computed using the Free Statistic Software (Calculator) software [18] and the Mega5 software [19].

Values for cranial morphometric variables in 36 L. olivaceus specimens are presented in Appendix 2. Principal component (PC) 1 (not shown) from PCA on cranial morphometric variables was correlated with condylo-basal length (Spearman's rank correlation [18,20]: ρ = 0.838; P < 0.001). This suggests a strong and expected effect of growth allometry. The pattern defined by PC 2 and PC 3 (Fig. 3B) segregated the individuals by geographic origin. All individuals from New Caledonia and Moruroa (Pacific Ocean) were grouped into a single cluster, albeit with some statistical differences between the two populations. Individuals sampled in Bali were clustered separately (Fig. 3B). The single specimen from West Papua and the single specimen from the Maldives were positioned close to the Bali cluster along PC 2 (Fig. 3B).

Differences in neurocranial morphology are likely to translate into differences in head shape. The few pictures available (some of which are presented in Fig. 1) suggest that Lethrinus sp. B has a more slender snout than Lethrinus sp. A.

Cytochrome b gene sequences from 27 L. olivaceus individuals sampled across the species range allowed the reconstruction of the phylogeny of mitochondrial lineages distribution-wide (Fig. 4). Two major clades were observed: one clade comprised mitochondrial haplotypes sampled from the Pacific Ocean, including two locations from the eastern Coral Triangle (Taiwan and West Papua); the other clade consisted exclusively of haplotypes sampled in the Indian Ocean (forming an exclusive sub-clade) and in the Coral Triangle (forming another sub-clade). The specimen from West Papua that was used for the morphometric analysis (Gutila 12: Appendices 1–3) was found to belong to the “Coral Triangle” sub-clade. The net nucleotide distance (Kimura-2 parameter distance [19]) between the “Pacific Ocean” clade and the (“Indian Ocean” + “Coral Triangle”’) clade of L. olivaceus was 0.095. The net distance between the “Indian-Ocean’ and “Coral-Triangle” subclades was 0.017. Nucleotide sites diagnostic between the “Pacific Ocean” clade vs. the (“Indian Ocean” + “Coral Triangle”) clade are highlighted in Fig. 5.

In the only molecular phylogeny of species of the genus Lethrinus published to date [16], the nucleotide divergence at the cytochrome b locus between sister species at the extremity of the tree was comprised between 7% and 14%. The lowest value in this comparison (7%, between L. lentjan and L. ornatus) separates lethrinid species which are morphologically distinct by current taxonomic standards [2]. Using this as a yardstick, the deep branching in the phylogenetic tree (Fig. 4) and the associated nucleotide divergence between the “Indian Ocean-Coral Triangle” and “Pacific Ocean” forms of L. olivaceus (9.5%; greater than 7%) indicate distinct species. These two species are here provisionally designated as Lethrinus sp. A and Lethrinus sp. B, respectively. The distinction of two species is further supported by morphometric differences.

The level of divergence between the “Indian-Ocean” and the ‘Coral-Triangle’ forms of Lethrinus sp. A (1.7%), although much lower than morphologically recognizable species pairs in Lethrinidae, may nevertheless be high enough to compare with Indo-West Pacific species pairs in other phylogenetically well-documented fish families, e.g. Acanthuridae [21], Chaetodontidae [22], and Siganidae [23,24]. Pending further investigation, we here adopt a conservative approach by considering the latter two forms to represent allopatric or parapatric populations of a single species (namely, Lethrinus sp. A). The Indo-West Pacific barrier, which runs from the Indo-Malay Sunda Shelf in the west to the Papuan-Australian Sahul shelf in the east, separates the marine biota of the Indian Ocean and of the Pacific Ocean [25]. Both shelves were exposed to form uninterrupted continental masses west and east of the Indo-Malay-Papua archipelago when the sea level fell during glacial periods, thus restricting the water flow between the two oceans [26]. Pleistocene vicariance on either side of the Indo-West Pacific barrier may account for the genetic differences observed between the ‘Indian Ocean’ lineage and the ‘Coral Triangle’ lineage of Lethrinus sp. A. Nuclear genotyping will be helpful to assess the degree of reproductive isolation between these two forms. Further examination based on time-calibrated molecular phylogenetic analysis will certainly be helpful to address more specifically the evolutionary history of this species.

It has been argued here that the two main clades observed within L. olivaceus correspond to distinct species, Lethrinus sp. A and Lethrinus sp. B. It is now necessary to assign to either of the latter the type specimens of L. olivaceus (MNHN 0000-5741) and L. rostratus (RMNH D443). Depending on the results, it might further be necessary to do so with L. waigiensis (MNHN 0000-9065), and it might also be necessary to collect specimens from the Caroline Islands, the type locality of L. xanthopterus (a nomen nudum) and assign them similarly. Finally, it might also be necessary to examine the type specimens of L. longirostris (BMNH 1865.9.21.6), the last in the list of nominal species synonymised with L. olivaceus [2]. To base the assignment on type locality would be hazardous: the type-localities of both L. olivaceus (Sunda Strait, Indonesia) and L. waigiensis (Waigeo Island, West Papua) lie within the Coral Triangle where both species may occur, according to our genetic evidence. Also, the type locality of L. rostratus has not been clearly indicated by A. Valenciennes, who only mentioned it was “envoyé de Batavia par MM. Kuhl et van Hasselt” [1]. Because of this, and because of possible overlap in the geographic distribution of the two species, nucleotide sequencing the type specimens at the cytochrome b locus is warranted. In Table 1, we propose a molecular-based strategy to rationalize the actions needed for the purpose of revising the taxonomy and nomenclature of species currently under L. olivaceus. Additional nomenclatural problems might arise in case the “Coral-Triangle” form and the ‘Indian Ocean’ form also proved upon further research to be distinct species within Lethrinus sp. A.

The authors declare that they have no conflicts of interest concerning this article.