Marine Biology (2004) 144: 369-375 DOI 10.1007/S00227-003-1199-0 RESEARCH ARTICLE G. Bernardi ? G. Bucciarelli ? D. Costagliola D. R. Robertson ? J. B. Heiser Evolution of coral reef fish Thalassoma spp. (Labridae). 1. Molecular phylogeny and biogeography Received: 12 November 2002/ Accepted: 6 August 2003 / Published online: 1 October 2003 ? Springer-Verlag 2003 Abstract Wrasses in the genus Thalassoma comprise 27 recognized species that occur predominantly on coral reefs and subtropical rocky reefs worldwide. The phy- logenetic relationships for 26 species were examined based on two mitochondrial genes (cytochrome b and 16S rRNA) and one nuclear intron (the first intron of the ribosomal protein S7). Two closely related species, the bird-wrasses (Gomphosus varius Lacep?de, 1801 and G. caerulaeus Lacep?de, 1801), were also included in the analysis. These species grouped within the genus Thalassoma. Thalassoma newtoni (Os?rio, 1891) from Sao Tome, which is generally synonymized with the Atlantic/Mediterranean Thalassoma pavo (Linnaeus, 1758) appears to be a vaHd species. Using a molecular clock, the genus was estimated to have originally di- verged 8-13 million years ago, with Thalassoma hallieui (Vaillant and Sauvage, 1875) from Hawaii and Thalas- soma septemfasciata Scott, 1959 from Western Australia as the ancestral species. Approximately 5-10 million years ago, a sudden burst of speciation resulted in seven clades, which were resolved with the sequence data. The terminal Tethyan event and the closing of the Communicated by J.P. Grassle, New Brunswick G. Bernardi (E3) ? G. Bucciarelli ? D. Costagliola Department of Ecology and Evolutionary Biology, University of California, 100 Shaffer Road, Santa Cruz, CA 95060, USA E-mail: bernardi@biology.ucsc.edu Fax: +1-831-4594882 D. Costagliola Dipartimento di Scienze della Vita, Seconda Universit? di Napoli, Via Vivaldi 43, 81100 Caserta, Italy D. R. Robertson Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Rep. de Panama J. B. Heiser Shoals Marine Laboratory and Department of Ecology and Evo- lutionary Biology, Cornell University, Ithaca, NY 14853, USA Present address: G. Bucciarelli Stazione Zool?gica A. Dohrn, Naples, Italy Isthmus of Panama were probably the major historical factors shaping the evolution of species in the genus Thalassoma. These data on the spatio-temporal pattern of speciation in the Indo-Pacific indicate that peripheral species have been generated at various times throughout the history of the genus, and that none of the wide- spread species are relatively young. Thus, there is no clear support for centrifugal (youngest at the periphery) versus centripetal (oldest at the periphery) modes of generation of species, two theories which have been used to account for geographic gradients in species diversity. Electronic Supplementary Material Supplementary ma- terial is available in the online version of this article at http://dx.doi.org/10.1007/s00227-003-1199-0. Introduction Wrasses in the genus Thalassoma form a distinctive group (family Labridae, subfamily Julidini; Gomon 1997). They are abundant worldwide in circumtropical and subtropical coral and rocky substrates at shallow depths, generally to 25 m. Several species have been intensively studied from a variety of perspectives (e.g. Warner 1982; Kramer and Imbriano 1997; Barry and Hawryshyn 1999; Swearer et al. 1999), making com- parative analyses especially interesting. Of the 24 species discussed by Allen (1995), all have a generalized body shape typical of julidin wrasses. Al- though adult sizes range from 135 mm standard length (SL) for the smallest species {Thalassoma noronhanum) to >460 mm SL in the largest {T. purpureum), gener- ating some allometric differences, the morphology of all the species is exceptionally homogeneous (Heiser 1981; Rocha et al. 2001). However, Hke other julidins, Thalassoma spp. vary widely in color pattern and hue, both inter- and intra-specifically. The ontogenetic and adult color patterns and hues (as well as locality data) are most useful in distinguishing species and 370 understanding their evolutionary relationships (Heiser 1981). Indeed, those few phylogenetic relationships among Thalassoma spp. that can be inferred from studies aimed chiefly at identifying sister species groups are based primarily on color pattern similarities (Randall and Edwards 1984; Allen and Robertson 1994; Randan 1994, 1995; Allen 1995; Randall et al. 1996). A reliable phylogeny is a prerequisite for a critical analysis of such relationships. In other taxa, where morphological data (including color pattern and hue) shed little light on relationships, molecular approaches have proven useful (Avise 1994; Seehausen et al. 1999; Crochet et al. 2000; Knowlton 2000). Here, we generate a molecular phylogeny based on two mitochondrial genes (cytochrome b and 16S rRNA) and one nuclear intron (ribosomal protein S7) for 26 (out of 27) recog- nized taxa. In addition to the 26 Thalassoma species in this study, we included both species of bird-wrasse, genus Gomphosus. The juveniles of the bird-wrasse Gomphosus varius have been described as a distinct species of Thalassoma (T. stuckiae), due to their mor- phological similarities to many species of Thalassoma (Whitley 1959). Furthermore, courtship behavior is vir- tually identical in Gomphosus spp. and Thalassoma spp. (Heiser 1981). and 0.3 mM of each primer, and used a cycling profile of 45 s at 94?C, 45 s at 50?C, and 1 min at 72?C, for 35 cycles. Automated sequencing was performed in both directions with the primers used in the amplification using an ABI 3100 automated sequencer (Applied Biosystems, Foster City, Calif). Sequence analysis We used the computer program Clustal V implemented by Se- quence Navigator (Apphed Biosystems) to align the mitochondrial and nuclear sequences. Phylogenetic relationships were assessed using the neighbor-joining (Kimura-2-distances) method (Nei 1987) and the maximum-parsimony method implemented by the software package PAUP (phylogenetic analyses using parsimony, ver- sion 4.0, Swofford 1998). Phylogenetic relationships were obtained for the three separate datasets (cylh, 16S, S7), and each is presented in the Electronic Supplementary Material. Topological confidence was evaluated with 1000 bootstrap rephcates (Felsenstein 1985). In both neighbor-joining and maximum-parsimony methods, boot- strapping analysis was performed with equal weighting of transi- tions and transversions. Alternative tree topologies were tested using the Kishino and Hasegawa (1989) method and the topology- dependent tail permutation test (Faith 1991). Consistency among the three markers was determined using permutation tests. All statistical tests were performed using the PAUP package. Testing for the actual presence of polytomies (as opposed to data artifacts) was done using the method of Walsh et al. (1999). Results Materials and methods Tissue samples and DNA extraction The sampling localities and geographic range of the sequenced specimens are provided in Table 1 and Fig. 1. The species Hali- choeres semicinctus and Coris juHs, which belong to the same wrasse subfamily (Julidinae) as the Thalassoma spp., were used as out- groups (Gomon 1997). Two individuals of each species were ana- lyzed, except for T. duperreyi for which a single specimen was available. Thalassoma heiseri (endemic to the Pitcairn Island group) was the only Thalassoma species that was not included in our study. Liver or muscle tissue of adult fish was extracted immedi- ately upon capture of the fish and preserved in 95% ethanol, then stored at 4?C in the laboratory. Tissues were digested overnight at 55?C in 500 \i\ of extraction buffer (NaCl 400 mM, Tris 10 mM, EDTA 2 mM, SDS 1%). We purified the DNA by standard chloroform extraction and isopropanol precipitation (Sambrook et al. 1989). Polymerase chain reaction (PCR) amphfication Amphfication of the mitochondrial cytochrome h (cylh) and 16S ribosomal gene (16S) regions was accomplished using the primers GLUDG-L, CB3H and 16SAR, 16SBR, respectively, following Kocher et al. (1989). In the case of cyt?, after amplifying several individuals, a specific primer was designed to replace GLUDG-L, namely CYTTHAL-5' AAC GGA GCA TCN TTC TTC TTT 3'. These primer sets amphfy a 560-bp region of the cylh and a 503-bp region of thel6S gene. The nuclear marker used in this study was the first intron of the ribosomal protein S7 (S7), a single locus gene (Chow and Hazama 1998). This region was amphfied using the universal fish primers of Chow and Hazama (1998). All amplifi- cations (25 nl) contained 10-100 ng of DNA, 10 mM Tris HCL (pH 8.3), 50 mM KCl, 1.5 mM MgCb, 2.5 U of Taq DNA po- lymerase (Perkin-Elmer, Norwalk, Conn.), 150 mM of each dNTP, Sequences and molecular phylogeny Out of the 1839 aligned base pairs (560 for the cyt? region, 503 for the 16S region, 776 for the S7 region), 232 could not be aligned unambiguously and were therefore removed from the analysis (GenBank acces- sion numbers: cyt?: AY328856-AY328885; 16S; AY328983-AY329012; S7: AY329640-AY329669). Of the remaining 1607 bp, 881 were variable, and 598 were phylogenetically informative. The two individuals se- quenced for each species were identical, except for Thalassoma purpureum for which the Easter Island and Rangiroa individuals differed by a single substitution. One previous molecular study based on cyt? sequences focused on the relationships of three Thalassoma species (Mikami and Machida 1999), but because there were only 30 bp in common between our study and theirs, the sequences could not be directly compared. In addition, the Thalassoma spp. sequences presented by Mikami and Machida appeared to be contaminated (possibly with human DNA following a GenBank BLAST search). Phylogenies obtained with the mitochondrial cyt?, 16S, and the S7 gene were almost identical. Data from the three markers were found to be statistically indis- tinguishable (permutation tests), and the topologies obtained independently for each marker were found not to be statistically different either by a topology- dependent tail permutation test (T-PTP) (see Electronic Supplementary Material). We therefore combined the three datasets (1607 bp). Seven most-parsimonious trees were obtained with the maximum-parsimony method, one of them being identical to the neighbor-joining tree. Table 1 Species names, common names (most widely used, des- criptive), distribution, and collection localities of Thalassoma spp. and outgroup species used in the present study. Specimens were 371 collected by: G.R. Allen (G?), E. Azzurro (EA), G. Bernardi (GBe), G. Bucciarelli {GBu), J.B. Heiser {JH), N.L. Crane {NQ, S. Planes {SP), and D.R. Robertson {DRR) Species (author) Common name Distribution Sampling locality Date of collection Ingroup T. amblycephalum (Bleeker, 1856) Twotone wrasse Indo-Central Pacific Moorea, Fr. Polynesia (GBe) Sep 2000 T. ascensionis (Quoy & Gaimard, 1834) Ascension wrasse Ascension Island Ascenci?n Island (DRR) Jun 1997 T. ballieui (Vaillant & Sauvage, 1875) Blacktail wrasse Hawaii Hawaii (DRR) Sep 1998 T. bifasciatum (Bloch, 1791) Bluehead wrasse Caribbean San Bias, Panama (DRR) Oct 1996 T. cupido (Temminck & Schlegel, 1845) Nishikibera Japan to Taiwan Akajima, Japan (DRR) May 1997 T. duperreyi (Quoy & Gaimard, 1824) Saddle wrasse Hawaii aquarium trade (GBe) Apr 2003 T. genivittatum (Valenciennes, 1839) Mascarene wrasse South-East Africa Reunion Island (DRR) Mar 1996 T. grammaticum Gilbert, 1890 Green sunset wrasse Tropical eastern Pacific ; Clipperton Island (DRR) Jun 1998 T. hardwicke (Bennett, 1830) Sixbar wrasse Indo-Central Pacific Moorea, Fr. Polynesia (GBe) Sep 2000 T. hebraicum (Lacep?de, 1801) Goldbar wrasse East Africa Zanzibar, Tanzania (NC) Aug 1999 T. heiseri Randall & Edwards, 1984 Heiser's wrasse Pitcairn Islands group Not sampled T. jansenii (Q\e.?kn?, 1856) Jansen's wrasse Central Indian Ocean to W. Pacific Lizard Island, Australia (DRR) Sep 1997 T. loxum Randall & Mee, 1994 Slantband wrasse Oman Masirah Island, Oman (SP) Dec 1998 T. lucasanum (Gill, 1862) Cortez rainbow Tropical eastern Pacific ; Guaymas, Sea of Cortez (GBe) Jul 1998 T. tunare (Linnaeus, 1758) Moon wrasse Indo-Central Pacific Rangiroa, Fr. Polynesia (GBe) Sep 2000 T. lutescens (Lay & Bennett, 1839) Sunset wrasse Indo-Central Pacific Moorea, Fr. Polynesia (GBe); Easter I. (DRR) Sep 2000, Jun 1997 T. newtoni (Os?r?o, 1891) Newton's wrasse West Africa Sao Tome (DRR) Apr 1996 T. noronhanum (Boulenger, 1890) Noronha wrasse F. de Noronha, Brazil Fernando de Noronha (DRR) May 1994 T. pavo (Linnaeus, 1758) Turkish wrasse Mediterranean, E. Atlantic Ustica, Italy (EA) Feb 1999 T. purpureum (Forssk?l, 1775) Surge wrasse Indo-Pacific Rangiroa, Fr. Polynesia (NC); Easter I. (DRR) Jun 1997 T. quinquevittatum (Lay & Bennett, 1839; 1 Five-stripe wrasse Indo-Central Pacific Moorea, Fr. Polynesia (GBe) Sep 2000 T. robertsoni Allen, 1995 Clipperton wrasse Clipperton Island Clipperton Island (DRR) Jun 1998 T. rueppellii (Klunzinger, 1871) Rueppell's wrasse Red Sea Aquarium trade (JH) Jan 2000 T. sanctaehelenae (Valenciennes, 1839) Saint Helena wrasse Saint Helena Island Saint Helena Island (DRR) Jun 1997 T. septemfasciata Scott, 1959 Seven banded wrasse ; Western Australia Perth, W. Australia (GA) Jul 1999 T. trilobatum (Lacep?de, 1801) Christmas wrasse Indo-Central Pacific Moorea, Fr. Polynesia (NC) Sep 2000 T. virens Gilbert 1890 Green wrasse Revillagigedo, Clipperton Chpperton Island (DRR) Jun 1998 Gomphosus varius Lacep?de, 1801 Pacific Bird wrasse Pacific Ocean Moorea, French Polynesia (GBe) Sep 2000 Gomphosus caerulaeus Lacep?de, 1801 Indian Bird wrasse Indian Ocean Zanzibar, Tanzania (GBe) Aug 1999 Outgroup Con.? ji/fc (Linnaeus, 1758) Rainbow wrasse Mediterranean, E. Atlantic Naples, Italy (GBu) Feb 1998 Halichoeres semicinctus (Ayres, 1859) Rock wrasse California, Sea of Cortez Baja California (GBe) Jun 1997 As expected, differences among the seven trees were all located in regions that were weakly supported (see de- tails below). The seven most-parsimonious trees were 2252 steps long (consistency index = 0.586). One of these most-parsimonious trees and the bootstrap supports for both maximum-parsimony and neighbor-joining meth- ods are shown (Fig. 2). Phylogenetic relationships DNA sequences resolved our samples into eight well- supported clades. Thalassoma ballieui and T. septem- fasciata formed a strongly supported clade (clade 1, Fig. 2), which was the sister clade to all other studied taxa. The monophyly of the remaining Thalassoma/ Gomphosus species was also strongly supported (Fig. 2). Within this group, samples could be divided into seven major clades that were strongly supported, although their interrelationships were not. Those seven clades grouped the following species: clade 2 included the two bird-wrasse species Gomphosus varius and G. caerulaeus. Clade 3 was fully resolved and well supported (Fig. 2). It included several smaller subclades; T. amblycephalum, T. robertsoni, and T. lucasanum formed one subclade; T. cupido, T. loxum, and T. trilobatum formed another subclade, which grouped with the T. purpureum/T. virens subclade. T. lunare was also included in clade 3 and was ancestral to all the species named above. Clade 4 included T. duperreyi, T. genivittatum, T. lutescens, and T. rueppellii, which formed a subclade. T. hebraicum was included in clade 4 in an ancestral position. Clade 5 included the two widespread species, T. hardwicki and T. jansenii. Clade 6 included only the five-stripe wrasse, T. quinquevittatum. Clade 7 included the western Atlantic species T. bifasciatum and T. noronhanum. 372 Fig. 1 Map of sampling locations mentioned in Table 1 160W 120W 80 W 40 W E dc Nor?nha Atol das Rocas Fig. 2 Molecular phylogeny of Thalassoma?Gomphosus species based on mitochondrial (cyt? and 16S) and nuclear (S7) markers. One most- parsimonious tree is shown with bootstrap support (>50%) indicated on nodes, maximum- parsimony helow the nodes, neighbor-joining above the nodes. Coris julis and Halichoeres semicinctus were used as outgroups -//- -//- -//- Coris julis ? Halichoeres semicinctus . T. septemfasciata - Gomphosi4s varius ?? r. amblycephaluin T. roberlsoni rC -T. lucasanum T. cupido ? T. loxum ^~ T. trjlnbatum ? T. purpureum T. duperreyi " T. genivUtutum ? T. grammaiiaun ? T. rueppeUii - T. hehraicuin T. ?lardwicki T.jansenii - T. quinqueviUalum T. noronbanum j%C - 7". bifascicititm T. newloni J. sancTaehelenae - T. ascensioids - T. pavo Clade 1 Clade 2 Clade 3 Clade 4 Clade 5 3 Clade 6 Clade 7 Clade 8 Indo-Pacif?c Ocean Atlantic Ocean Clade 8 included the mid-Atlantic to eastern Atlantic group T. ascensionis, T. sanctaehelenae, T. newtoni, and T. pavo. Clades 7 and 8 from the Atlantic were grouped together, but this grouping was weakly supported (61% of neighbor-joining bootstraps, < 50% for maximum parsimony). Relationships among the seven clades were weakly supported as they corresponded to very short internal branches (thus generating more than one most-parsi- monious tree). These polytomies did not result from saturation effects (plots transitions/transversions vs. divergence, not shown), and statistical tests showed that these polytomies were not due to data artifacts (Walsh et al. 1999). Polytomies were more likely to correspond to a group of divergence events occurring within a brief period of time, whose ordering was not resolved. As shown in Fig. 2, the bird-wrasse genus Gomphosus nests within the genus Thalassoma. To test for the monophyletic status of the genus Thalassoma, we con- strained our tree to place the genus Gomphosus outside the genus Thalassoma and re-ran the analysis to produce the shortest tree consistent with that topology. This new 373 topology was 98 steps longer than the most-parsimoni- ous topology. Both Kishino-Hasegawa and T-PTP tests showed that these two topologies were significantly dif- ferent, and that our data rejected (P< 0.0001) placing the genus Gomphosus outside Thalassoma. Molecular clocks and divergence times In order to estimate Thalassoma spp. divergence times, we restricted our analysis to the cyt? marker, as it is the most widely used for fish divergence time estimates (Meyer 1993; McCune and Lovejoy 1998). Considering the generally used molecular clock of 1.5-2.5% sequence divergence per million years, we found that the first splitting event in the genus, i.e. the divergence between the T. ballieuijT. septemfasciata clade and the remaining Thalassoma spp. occurred ~8 to 13 million years ago (Mya). The burst of divergence that produced polyto- mies between the seven clades described above is esti- mated to have occurred ~5 to 10 Mya. Discussion As mentioned in the "Introduction", several coral reef fish groups, including parrotfishes and wrasses, are very difficult to identify using morphological characters alone. Colors are heavily relied upon for proper identi- fication (Bernardi et al. 2002). The genus Thalassoma is probably the most striking example, with most species sharing overlapping meristic and morphometric char- acters (Heiser 1981). Occasionally, authors have sug- gested possible relationships between a subset of species, again primarily based on coloration patterns. Here, we review those suggestions in Hght of our findings. Randall and Edwards (1984) considered a closely related group of six Indo-Pacific species exhibiting sim- ilar color patterns: Thalassoma cupido, T. rueppellii (then called T. klunzingeri), T. purpureum, T. trilobatum, T. quinquevittatum, and T. heiseri. They considered T. purpureum and T. trilobatum to be closely related. Similarly they considered T. heiseri and T. cupido to be closely related. Our phylogeny is mostly concordant with these findings, with T. cupido, T. trilobatum, and T. purpureum being in the same clade. Interestingly, initial phases of T. trilobatum and T. purpureum are al- most indistinguishable, yet they are not closest relatives. In contrast with that study, we did not find T. rueppelli in the same clade as the previous taxa, and our data could not group T. quinquevittatum with any other spe- cies with great confidence. Allen and Robertson (1994) placed T. virens close to T. purpureum (confirmed by Randall 1995), and T. grammaticum close to T. lutescens. Both placements are in agreement with our findings. Allen (1995) considered T. robertsoni, T. lucasanum, and T. amblycephalum to be closely related, which is also what we found. Randah (1994) suggested that T. loxum was close to T. cupido, and this is also well supported by the molecular phylogeny. Overall there is a remarkable match between the predictions based on coloration patterns and the molecular phylogeny presented here. Classification work involves a comprehensive ap- proach based on morphological, behavioral, biogeo- graphical, and genetic methods. Therefore, we are not proposing new interpretations here to the systematics of the genus Thalassoma. Our data, however, shed light on some relationships that warrant further work. 1. The species collected in Sao Tome is usually referred to as Thalassoma pavo (Gomon and Forsyth 1990; Seret 1990). This species, however, was originally described as Thalassoma newtoni (Os?rio 1891). Our data show T. pavo and T. newtoni to be genetically as distant as several other pairs of species, and, in addition, T. newtoni is closer to the mid-Atlantic island endemic species {T. sanctaehelenae and T. ascensionis) than it is to T. pavo (Fig. 2). Further work on this subclade (T. pavo, T. newtoni, T. ascensionis, and T. sanctaehelenae) is presented in an accompanying paper (Costaghola et al. 2003). 2. The two bird-wrasse species in the genus Gomphosus appear to be included in the genus Thalassoma. Bird- wrasses are fishes with very elongated snouts. Their unique adaptation sets them apart from the remain- ing Thalassoma spp., yet, as larvae and early juve- niles, Gomphosus spp. do not have an elongated snout and can be very difficult to distinguish from Thalas- soma spp. Gomphosus spp. should therefore be included in the genus Thalassoma as a specialized morphological variant. 3. One Thalassoma sp. individual was obtained from a pet store without geographic origin information (not included in this study). Its coloration did not match any described coloration for a Thalassoma species, and its sequence did not match any of the sequences described here. The individual looked most Hke T. duperreyi and T. lutescens. Its sequence was sim- ilar, but not identical to the sequence of T. duperreyi, ruling out the possibility of being a hybrid. Thus, it may correspond to an undescribed species. Hybridization in fishes has been seen as a possible indicator of genetic similarity (Craig et al. 2000), and some Thalassoma spp. are known to hybridize. In the Red Sea, T. rueppellii was shown to hybridize with T. lunare (Randall and Miroz 2001). In the Hawaiian archipelago, T. duperreyi and T. lutescens are also known to hybridize frequently (Hoover 1993; Witte and Mahaney 2001; P. Lobel, personal communication). Finally, a hybrid initial phase Thalassoma/Gomphosus specimen was collected by G.R. Allen in September 1994 at Cassini Island, Western Australia. The Thalassoma sp. parent of the hybrid was later assigned to T. lunare by J.E. Randall (G.R. Allen, personal communication). The molecular phylogeny presented here suggests that T. duperreyi and T. lutescens are genetically closely 374 related, and that T. lunare/T. rueppellii, and T. lunarej Gomphosus spp. are related but are not closest relatives. Thus, while hybridizations indicate that some parental species can interbreed, supporting the conclusion that the genus Gomphosus should be included in the genus Thalassoma, hybridization may not indicate that hybridizing species are always closest relatives. Species in the genus Thalassoma provide an ideal sys- tem to study biogeographic patterns as they are distri- buted over a vast geographic range. The molecular phylogeny presented here raises some interesting ques- tions that can be tested. Our data indicate that the genus Thalassoma originated approximately 8-13 Mya. It has been postulated that several groups of coral reef fishes evolved during the Terminal Tethyan Event (approxi- mately 15 Mya) (Bellwood and Wainwright 2002). The genus Thalassoma may be an example of such an event. Furthermore, our data also indicate that 5-10 Mya, a speciation explosion occurred in the genus Thalassoma. This was accompanied by a large variety of coloration patterns, behavior differences, and vast geographic expansion. The separation of Indo-Pacific from Atlantic clades occurred after the first major divergence event (clade 1 divergence), but at about the same time as the major divergence of the seven more derived clades. Fur- thermore, within the Atlantic, western (e.g. T. bifascia- tum) and eastern (e.g. T. pavo) clades seem to have separated at approximately the same time as the Atlantic and Indo-Pacific split. This may suggest that a common factor was involved in the two differentiating events. The rise of the Isthmus of Panama, which occurred approxi- mately 3.5 Mya, has been proposed as the major historical event responsible for the separation of Atlantic and Indo- Pacific faunas. The eastern Pacific T. lucasanum was ori- ginally thought to have been separated by the rise of the Isthmus of Panama from its Caribbean geminate species T. bifasciatum (e.g. Bermingham et al. 1997), although Heiser (1981) pointed out that T. amblycephalumi and T. lucasanum were closer to each other than either was to T. bifasciatum. Our data show that eastern Pacific groups such as T. lucasanum, T. robertsoni, or T. grammaticum are not closely related to Atlantic taxa. In contrast, T. lucasanum and T. robertsoni were found to be more closely related to the Indo-Pacific T. amblycephalum. Thalassoma spp. may be widespread (e.g. T. lunare), have a restricted range (e.g. T. duperreyi), found at the periphery of the Indo-Pacific, or are more abundant in its central regions. We found that Indo-Pacific periph- eral species (T. ballieui, T. septemfasciata, T. robertsoni, T. lucasanum, T. cupido, T. loxum, T. virens, T. duper- reyi, T. genivitattum, T. grammaticum, and T. rueppelli) were derived from splits occurring across a range of times from ancient to quite recent. Thus, our data did not provide evidence that peripheral species are recent additions to ancestral widespread or central species (Jokiel and MartinelH 1992; Briggs 1999a, 1999b, 2003). The production of such peripheral species has occur- red in at least four Indo-Pacific clades, with existing widespread species. Interestingly, the spHts involving currently widespread species, T. amblycephalum, T. trilobatum, T. purpureum, T. lunare, on one hand, and T. lutescens, T. hardwicki, T.jansenii, T. quinquevittatum, on the other hand, are all fairly ancient. Widespread species are ancestral in some (e.g. T. trilobatum and T. lunare), but not all cases (e.g. T. lutescens). Taken together, these patterns suggest a mixture of centripetal (youngest in center) and centrifugal (youngest at edges) generation of species in the Indo-Pacific, with no recent generation of widespread species. A unique situation is found in Hawaii, where the two endemic species, T. ballieui and T. duperreyi, are found early and late in the evolution of the genus. The use of more molecular markers may provide the appropriate tools to determine the precise succession of events that led to the unique radiation of Thalassoma spp. Notes added in proof while this manuscript was in press, two studies came to our attention: 1. A phylogeny of the closely related genus Halichoeres was done by P. Barber and D. Bellwood. This study determined that the Caribbean species Halichoeres maculi- pinna was more closely related to Thalassoma spp. that to order Halichoeres species. When including H. maculipinna to our dataset, we found that it is close to Thalassoma but not within the genus. It is however, an ideal outgroup to be used. When H. maculipinna was used as an outgroup, our results were unchanged (we would like to acknowledge P. Barber and D. Bellwood for permission to use their data before publication). 2. A new species of Thalassoma nigro- fasciaium, a new species of labrid fish from the south-west Pacific, Aqia 7(1), (1-8). This species is closely related to T.jansenii and is found in the Great Barrier Reef and adjacent areas. As such, our T. jansenii sample should be labeled T. nigrofasciatum. We sequenced the 12S rRNA, 16S rRNA and the cytochrome b regions for 2 individuals of bona fide T.jansenii (collected in June 1998 by DRR in Ishigaki, Japan). The two individuals had identical sequences. Their 16S sequence was identical to the Australian T. nigrofascia- tum and differed by one nucleotide at the cytochrome b locus. We conclude that T. nigrofasciatum and T. jansenii are either very closely related species with no detectable genetic divergence at these two loci, or that T. nigrofasciatum is a color variant of T. jansenii. Acknowledgements We would like to thank the people who provi- ded samples and/or comments for this study: G. Allen, E. Azzurro, N. Crane, A. Edwards, and S. Planes. This research was partly supported by faculty research funds granted by the University of California Santa Cruz to G.B. References Allen GR (1995) Thalassoma robertsoni, a new species of wrasse (Labridae) from Clipperton Island, tropical eastern Pacific Ocean. Rev Fr Aquariol 22:3-4 Allen GR, Robertson DR (1994) Fishes of the tropical eastern Pacific. University of Hawaii Press, Honolulu Avise JC (1994) Molecular markers, natural history and evolution. Chapman and Hall, New York Barry KL, Hawryshyn CW (1999) Spectral sensitivity of the Hawaiian saddle wrasse, Thalassoma duperrey, and implications for visually mediated behaviour on coral reefs. Environ Biol Fishes 56:429^42 Bellwood DR, Wainwright PC (2002) The history and biogeogra- phy of fishes on coral reefs. In: Sale PF (ed) Coral reef fishes. Dynamics and diversity in a complex ecosystem.Academic, San Diego, pp 3-15 375 Bermingham E, McCafferty SS, Martin AP (1997) Fish biogeog- raphy and molecular clocks: perspectives from the Panamanian Isthmus. In: Kocher T, Stepien C (eds) Molecular systematics of fishes. Academic, New York, pp 113-126 Bernardi G, Holbrook SJ, Schmitt RJ, Crane NL, DeMartini E (2002) Species boundaries, populations and colour morphs in the coral reef three-spot damselfish (Dascyllus trimaculatus) species complex. Proc R Soc Lond B Biol Sei 269:599-605 Briggs JC (1999a) Coincident biogeographic patterns: Indo-West Pacific Ocean. Evolution 53:326-335 Briggs JC (1999b) Modes of speciation: marine Indo-West Pacific. Bull Mar Sei 65:645-656 Briggs JC (2003) Marine centers of origin as evolutionary engines. J Biogeogr 30:1-18 Chow S, Hazama K (1998) Universal PCR primers for S7 ribo- somal protein gene introns in fish. Mol Ecol 7:1255-1256 Costaghola D, Robertson DR, Guidetti P, Stefanni S, Wirtz P, Heiser JB, Bernardi G (2003) Evolution of coral reef fish Thalassoma spp. (Labridae). 2. Evolution of the eastern Atlantic species. Mar Biol (in press). DOI 10.1007/s00227-003- 1200-y Craig MT, Pondella DJ, Franck JPC, Hafnert JC (2000) On the status of the serranid fish genus Epinephelus: evidence for paraphyly based upon 16 s rDNA sequence. Mol Phylogenet Evol 19:121-130 Crochet PA, Bonhomme F, Lebreton JD (2000) Molecular phy- logeny and plumage evolution in gulls (Larini). J Evol Biol 13:47-57 Faith DP (1991) Cladistic permutation tests for monophyly and nonmonophyly. Syst Zool 40:366-375 Felsenstein J (1985) Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783-791 Gomon MF (1997) Relationships of fishes of the labrid tribe Hypsigenyini. Bull Mar Sei 60:789-871 Gomon MF, Forsyth P (1990) Labridae. In: Quero JC, Hureau JC, Karrer C, Post A, Saldanha L (eds) Check-list of the fishes of the eastern tropical Atlantic. UNESCO, Paris, pp 919-942 Heiser JB (1981) Review of the labrid genus Thalassoma (Pisces: Teleostei). PhD thesis, Cornell University, Ithaca, N.Y. Hoover JP (1993) Hawaii's fishes: a guide for snorkelers, divers, and aquarists. Mutual, Honolulu, Hawaii Jokiel P, Martinelh FJ (1992) The vortex model of coral reef bio- geography. J Biogeogr 19:449-458 Kishino H, Hasegawa M (1989) Evaluation of the maximum like- hhood estimate of the evolutionary tree topologies from DNA sequence data. J Mol Evol 29:170-179 Knowlton N (2000) Molecular genetic analyses of species bound- aries in the sea. Hydrobiologia 420:73-90 Kocher TD, Thomas WK, Meyer A, Edwards SV, Paabo S, Villablanca FX, Wilson AC (1989) Dynamics of mitocho- ndrial DNA evolution in animals: amphfication and sequen- cing with conserved primers. Proc Nati Acad Sei USA 86:6196-6200 Kramer CR, Imbriano MA (1997) Neuropeptide Y (NPY) indu- ces gonad reversal in the protogynous bluehead wrasse, Thalassoma bifasciatum (Teleostei: Labridae). J Exp Zool 279:133-144 McCune AR, Lovejoy NR (1998) The relative rate of sympatric and allopatric speciation in fishes: tests using DNA sequence divergences between sister species and among clades. In: Howard DJ, Berlocher SH (eds) Endless forms: species and speciation. Oxford University Press, New York, pp 132-153 Meyer A (1993) Evolution of mitochondrial DNA in fishes. In: Hochachka PW, Mommsen TP (eds) Biochemistry and molecular biology of fishes, vol 2. Molecular biology frontiers. Elsevier, Amsterdam, pp 1-38 Mikami Y, Machida Y (1999) External and internal morphology and nucleotide sequence of mitochondrial cytochrome h gene in three Thalassoma species (Perciformes: Labridae). Mem Fac Sei Kochi Univ Ser D Biol 20:35-46 Nei M (1987) Molecular evolutionary genetics. Columbia Univer- sity Press, New York Os?rio B (1891) 3a Nota. Peixes mar?timos das ilhas de S. Thom?, do Principe e ilheo das Rolas. J Sei Math Phys Nat Lisboa 2:97- 139 Randall JE (1994) Coastal fishes of Oman. University of Hawai'i Press, Honolulu Randall JE (1995) On the vahdity of the eastern Pacific labrid fishes Thalassoma grammaticum Gilbert and T. virens Gilbert. Bull Mar Sei 56:670-675 Randall JE, Edwards A (1984) A new labrid fish of the genus Thalassoma from the Pitcairn Group, with a review of related Indo-Pacific species. J Agrie Aquat Sei 4:13-32 Randall JE, Miroz A (2001) Thalassoma lunarexThalassoma ruep- pellii, a hybrid labrid fish from the Red Sea. Aquat J Ichthyol 4:131-134 Randall JE, Allen GR, Steene RC (1996) Fishes of the Great Barrier Reef and Coral Sea, revised and expanded edition. Crawford House, Bathurst, N.S.W., Australia Rocha LA, Guimaraes RZ, Gasparini JL (2001) Redescription of the Brazilian wrasse Thalassoma noronhanum (Boulenger, 1890) (Teleostei: Labridae). Aquat J Ichthyol 4:105-108 Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: a laboratory manual, 2nd edn. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Seehausen O, Mayhew PJ, Van Alphen JJM (1999) Evolution of colour patterns in East African cichlid fish. J Evol Biol 12:514- 534 Seret B (1990) Poissons de mer de l'Ouest Africain tropical. OR- STOM, Par?s Swearer SE, Caselle JE, Lea DW, Warner RR (1999) Larval retention and recruitment in an island population of a coral-reef fish. Nature 402:799-802 Swofford DL (1998) PAUP*: phylogenetic analysis using parsi- mony (and other methods). Sinauer, Sunderland, Mass. Walsh HE, Kidd MG, Moum T, Friesen VL (1999) Polytomies and the power of phylogenetic inference. Evolution 53:932-937 Warner RR (1982) Mating systems, sex change and sexual demo- graphy in the rainbow wrasse, Thalassoma lucasanum. Copeia 1982:653-661 Whitley GP (1959) More ichthyological snippets. Proc R Zool Soc NSW 58:11-26 Witte A, Mahaney C (2001) Hawaiian reef fish. Island Heritage, Honolulu