Copáa, 2004(4), pp. 770-782

Mitochondrial DNA and Color Pattern Variation in Three Western Atlantic (Labridae), with the Revalidation of Two Species

Luiz A. ROCHA

Genetic surveys of widely distributed marine species often find previously unde- tected biodiversity. In the present study, populations of three species of Halichoeres were sampled across their entire geographical ranges: Halichoeres cyanocephalus and Halichoeres maculipinna were sampled on both sides of the Amazon freshwater out- flow, the main biogeographic barrier in the tropical western Atlantic; and Halichoeres garnoti was sampled in the Caribbean and . Genetic divergences between populations separated by the Amazon ranged from 2.3% in H cyanocephalus to 6.5% in H. maculipinna. There is inconsistency between color differences and genetic partitions in the species surveyed. The color differences between populations of H cyanocephalus and H maculipinna correspond to deep genetic partitions at the cy- tochrome b locus. However, genetic similarity at this same locus was observed be- tween populations of H. garnoti with striking color differences. Based on the com- bination of the observed genetic differences with diagnostic color differences, the BraziUan species Halichoeres dimidiatus (Agassiz) and Halichoeres penrosei Starks, 1913 are revalidated. In addition, a neotype is designated to H cyanocephalus (Bloch, 1791), to clarify its taxonomic status and type locaUty. All species analyzed have a similar larval dispersal potential, but varying degrees of genetic divergences were observed, indicating that benthic stage ecology may also play a role in speciation in this group.

THE presence of potentially widely dispers- populations that also show slight morphological ing planktonic larva in most marine fishes differences, those populations are usually ele- has led to the assumption that such species are vated to specific status (Randall, 1998). Howev- composed of genetically homogeneous popula- er, color alone is rarely used as a character to tions (Warner, 1997). However, recent studies define species, especially in groups with highly on larval behavior (Leis and McCormick, 2002) variable intraspecific color patterns such as and océanographie processes (Cowen, 2002) in- (Labridae). Despite the presence of dicate that larvae often disperse far less than bright color patterns and their wide variation their potential. Moreover, habitat selection by observed in reef fishes, little is known about its larvae can also overcome the effects of long-dis- evolutionary significance (McMillan et al., tance dispersal and give rise to genetic differ- 1999). Sexual selection related to coloration ap- ences in geographically close populations (Bier- pears to be present in damselfishes (Pomacen- ne et al., 2003). Consequently, genetic studies tridae; Thresher and Moyer, 1983), but no cor- of widely distributed taxa often reveal the pres- relation between mating success and color pat- ence of previously overlooked species (Knowl- tern was observed in the blue-head (La- ton, 2000). bridae; Warner and Schultz, 1992). In closely Early genetic surveys within the Caribbean re- related species, coloration has been proposed gion found no evidence of cryptic speciation in to be useful in mate recognition. The sharpnose reef fishes (Shulman and Bermingham, 1995). pufferfishes are an example of a group with lit- However, when broader geographic areas were tle or no diagnostic external morphological sampled, analyses of DNA variation demonstrat- characters, where color has been proposed as ed the presence of deeply divergent evolution- important in speciation and is routinely used as ary partitions in several species, such as bone- a species defining character (Allen and Randall, fishes (Colborn et al., 2001), blennies (Muss et 1977; Moura and Castro, 2002). al., 2001), surgeonfishes (Rocha et al., 2002), Wrasses of the genus Halichoeres in the west- groupers (Carlin et al., 2003), and wrasses (Ro- ern Atlantic provide an excellent system to ex- cha, 2003a). Some of these genetic divergences amine the relationship between genetic struc- are accompanied by slight color differences. ture and color differentiation. The populations In reef fish , color pattern is consis- surveyed in this study are present on both sides tently used as a useful diagnostic character. of recognized biogeographic barriers, have a When color differences are observed between similar dispersal potential (pelagic larval phase

© 2004 by the American Society of Ichthyologists and Herpetologists ROCHA•MITOCHONDRIAL DNA VARIATION IN HALICHOERES in of 25-30 days; Sponaugle and Cowen, 1997; go^W 75' Wellington and Robertson, 2001) and exhibit slight color differences. Populations of Halicho- 30' eres garnoti in the Caribbean and Bermuda are separated by at least 1400 km of open-ocean (Smith-Vaniz et al., 1999), and populations of Halichoeres cyanocephalus and Halichoeres maculi- 15°N pinna in and the Caribbean are separated by the Amazon barrier (Joyeux et al., 2001; Ro- cha, 2003b). These populations have been con- sidered to be the same species, but they show consistent color differences in three instances: (1) the juveniles of H. cyanocephalus in Brazil lack a blue spot on the midventral portion of IS'S the soft dorsal fin, which is present on juveniles of the Caribbean population; (2) terminal- phase males of H. garnoti at Bermuda have a red band on the upper posterior portion of the Fig. 1. Tropical western Atlantic. Numbers repre- body, whereas males from the Caribbean have a sent sampled locations: 1. Bermuda; 2. Bahamas; 3. black band at the same place; (3) the upper Florida Keys; 4. ; 5. St. Croix; 6. Venezuela; 7. body of individuals from the Brazilian popula- northeast Brazil; 8. southeast Brazil; 9. Trindade Is- land. Solid arrows indicate direction of mean surface tions of H. maculipinna is usually brownish-pink, oceanic currents; broken lines indicate subsurface whereas individuals from the Caribbean have a currents. yellowish-green upper body. Samples of the above-mentioned species were collected across their entire geographical range chrome b gene was amplified with the primers in the tropical western Atlantic, and an analysis L14725 (5' GTG ACT TGA AAA ACC ACC GTT of mitochondrial DNA sequences was carried G 3') and H15573 (5' AAT AGG AAG TAT CAT out aiming to assess the following questions: (1) TCG GGT TTC ATG 3'; Meyer, 1993). Based on Does genetic differentiation parallel the ob- initial sequences, the internal primers LI 4768 served patterns of color differentiation in west- (5' ACC CAC CCA CTC CTT AAA ATC 3'), and ern Atlantic Halichoeres? (2) Do biogeographic H15496 (5' TTC GAG ACC CAG ATA ATT TCA barriers (the Amazon freshwater outflow and C 3') were designed and worked consistently vast open-ocean distances) equally affect species among Halichoeres species, yielding a product of with similar dispersal potential? 690-709 base pairs. Thermal cycling in polymerase chain reac- MATERIALS AND METHODS tions (PCR) consisted of an initial denaturing step at 94 C for 1 min 20 sec, then 35 cycles of Specimens were collected with pole spears or amplification (40 sec of denaturation at 94 C, hand nets while scuba diving or snorkeling in 30 sec of annealing at 52 C, and 55 sec of ex- Bermuda, the Bahamas, Florida, Belize, St. tension at 72 C), and a final extension of 2 min, Croix (U.S. Virgin Islands), Venezuela, Paraiba 30 sec at 72 C. Excess primers were removed by (northeastern Brazil), Espirito Santo (south- incubation of PCR products with exonuclease I eastern Brazil) and Trindade Island (Fig. 1) be- and shrimp alkaline phosphatase (USB Corp., tween 1997 and 2002. Tissue (muscle and/or Cleveland OH). gill) was stored in a saturated salt-DMSO buffer Sequencing reactions with fluorescently la- (Amos and Hoelzel, 1991). beled dideoxy terminators (BigDye, Applied Total genomic DNA was extracted using QIA- Biosystems, Inc., Foster City, CA) were per- GEN DNeasy extraction kits following the man- formed according to manufacturer's recom- ufacturer's protocol. Extracted DNA was frozen mendations and analyzed with an ABI 377 au- in TE buffer (10 mM Tris-HCl, pH 7.5 and 1 tomated sequencer (Applied Biosystems, Inc., mM EDTA, pH 8.0, diluted in water) and ar- Foster City, CA). All samples were sequenced in chived at •20 C. Primer names indicate the the forward direction (with LI 4725 or LI 4768 DNA strand (H = heavy and L = light strand) primers), and rare or questionable haplotypes and the position of the 3' end of the oligonu- were sequenced in both directions to ensure ac- cleotide primer relative to the human mito- curacy of nucleotide designations. Haplotypes chondrial DNA sequence. A fragment of ap- of all species were deposited at GenBank (see proximately 800 base pairs of the mtDNA cyto- Material Examined section). 772 COPEIA, 2004, NO. 4

Sequences were aligned and edited with Se- quencher version 3.0 (Gene Codes Corp., Ann Arbor, MI). The computer program MODEL- TEST version 3.06 (Posada and Crandall, 1998) 100 Belize was used to determine the substitution model 100 and the gamma distribution shape parameter that best fit the data through hierarchical like- lihood ratio tests (hLRTs). The HKY (Hasegawa et al., 1985) substitution model was chosen for all datasets and the gamma distribution shape 95 Brazil parameter varied fi^om 0 in H. cyanocephalus to 96 L 0.1 in H. maculipinna to 0.4 in H. garnoti. The analyses were also run using the Tamura-Nei substitution model (Tamura and Nei, 1983) // -H. garnoti with no significant changes in tree topology. Re- • 0.005 substitutions/site lationships between unique haplotypes were re- Fig. 2. Phylogenetic tree of Halichoeres cyanocephal- constructed based on starting trees calculated us (Belize) and Halichoeres dimidiatus (Brazil) unique using the neighborjoining method (Nei, 1987) haplotypes. Neighbor-joining bootstrap support (> and ñarther searched by lO'^ tree bisection and 50%) indicated on nodes, maximum-parsimony sup- reconnection (TBR) iterations under the mini- port below nodes. Branch lengths are according to mum evolution criterion with the software indicated scale; the branch leading to the outgroup PAUP* version 4.0bl0 (D. L. Swofford, Sinauer, {Halichoeres garnoti) was reduced by 50%. Sunderland, MA, 2002, unpubl.). Additionally, maximum parsimony analyses were performed 17 individuals. A total of 26 polymorphic sites on datasets with deep phylogenetic breaks. To- distributed among 10 haplotypes were identi- pological confidence was evaluated with 1000 fied. Mean nucleotide frequencies were A = bootstrap replicates (Felsenstein, 1985). Equal 0.22, C = 0.31, G = 0.16, T = 0.31. The tran- weighting of all three codon positions was used. sitionitransversion ratio was 5.5:1. The phylo- Population structure and gene flow were as- genetic analysis (Fig. 2) assigned the haplotypes sessed with the program Arlequin v2.0 (S. Schneider, D. Roessli, and L. Excoffier, Univer- of H. cyanocephalus into two lineages (Brazil and sity of Geneva, Switzerland, 2000, unpubl.), Caribbean). Pairwise distances within lineages which generated Fg-r-values (a molecular analog ranged from 1.17-2.83 nucleotide substitutions of Fgy that takes into consideration sequence di- {d = 0.002-0.004). Although sample sizes are vergence among haplotypes). Genetic variation low, Brazilian and Caribbean populations are is described with nucleotide diversity (IT, equa- separated by at least 16 mutations {d = 0.023) tion 10.19; Nei, 1987) and haplotype diversity and the Y^ between Brazil and Belize was 0.91 (h, equation 8.5; Nei, 1987) within each loca- (P < 0.01). No morphological difference was tion. detected between the lineages; however, a slight In addition to the genetic analysis, morpho- color difference is present (Fig. 5C-D). logical comparisons were also carried out. Counts and measurements follow Randall and Halichoeres garnoti.•A 704 bp fragment from Lobel (2003). Measurements were taken with a the cytochrome b gene was analyzed from 116 digital caliper and recorded to the nearest tenth individuals. A total of 66 polymorphic sites were millimeter. Unless otherwise stated, institutional distributed among 55 haplotypes. Mean nucle- abbreviations follow Levitón et al. (1985). Spec- otide frequencies were A = 0.22, C = 0.30, G imens from the following museums were ex- = 0.16, T = 0.32. The transition:transversion ra- amined: MBML (Museu de Biología Meló Lei- tio was 10.16:1. No geographic signal was de- tâo. Espirito Santo, Brazil), MCZ, MZUSP, SU, tected in the phylogenetic analysis of H. garnoti UF, UFES (Coleçâo Ictiológica, Universidade (Fig. 3). Individuals with a red dorsum (indi- Federal do Espirito Santo, Brazil), UFPB (Co- cated by arrows and the letter B in Fig. 3) were leçâo Ictiológica, Universidade Federal da Pa- nested within the Caribbean lineage. Genetic raíba, Brazil), USNM and ZUEC (Museu de His- distances among populations ranged from 0.06- toria Natural, Universidade Estadual de Cam- 2.45 nucleotide substitutions (d = 0.001-0.004). pinas, Brazil). Haplotype diversity was relatively high and over- all nucleotide diversity (TT) was low in all pop- RESULTS ulations (Table 1). No significant population Halichoeres cyanocephalus.•A 701 bp fragment structure was observed across the species' entire from the cytochrome b gene was analyzed from range from Bermuda to Venezuela (Table 2). ROCHA•MITOCHONDRIAL DNA VARIATION IN HALICHOERES 773

sequence divergence). Genetic distances within lineages ranged from 0.01-2.05 nucleotide sub- stitutions (d = 0.001-0.003). Haplotype diversity was relatively high and overall nucleotide diver- sity (IT) was low in all populations of both line- ages (Table 1). The two lineages were allopatric, one north- ern (Florida, Bahamas, St. Croix, Belize, and Ve- nezuela) and the other southern (coastal Brazil and Trindade Island), being separated by highly significant FsT-values (Fgx = 0.958-0.984). No significant pairwise Fg^-values were observed within either lineage (Table 3). No morpholog- ical difference was detected between the line- ages; however, a slight color difference is pre- sent (Fig. 5E-F).

DISCUSSION Halichoeres cyanocephalus.•^The genetic diver- gence (d = 2.3%) between the Brazilian and Caribbean populations of H. cyanocephalus (Fig. 2) coincides with a subtle but consistent color difference: individuals in Belize, Florida, and the Virgin Islands have the color pattern shown in Figure 5D, whereas individuals in north and south Brazil have the pattern shown in Figure 5C. The Amazon is unlikely to be an effective barrier between Brazilian and Caribbean line- ages, because the Brazilian lineage occurs over deep sponge bottoms off Brazil (Rocha et al., 2000) and French Guiana (Uyeno et al., 1983), the latter under the Amazon plume. If not the Amazon barrier, what is causing ge- netic differentiation between these lineages? H. poeyi Rocha (2003b) suggests that it may be ecologi- • 0.005 substitutions/site cal speciation or speciation driven by "divergent Fig. 3. Phylogenetic tree of all (55) Halichoeres gar- selection on traits between populations or sub- noti unique haplotypes. No geographical signal was populations in contrasting environments" detected in the phylogeny. Arrows indicate haplotypes (Schlüter, 2001). Environmental differences be- shared between Bermuda and one or more Caribbean tween locations in the (charac- location; haplotypes only found in Bermuda are terized by clear waters all year round, relatively marked with a B. Bootstrap support (> 50%) is in- stable environmental conditions, and bottom dicated on nodes. Branch lengths are according to indicated scale; the branch leading to the outgroup sediments largely composed of calcium carbon- {Halichoeres poeyi) was reduced by 50%. ate) and the Brazilian coast (a typical continen- tal environment, with terrigenous substrates in- fluenced by run off from rivers, and high tur- Halichoeres maculipinna.•A 691 bp fragment bidity caused by wind driven suspension of bot- from the cytochrome b gene was analyzed from tom sediments) may generate divergent 86 individuals. A total of 71 polymorphic sites selection pressures in the insular versus conti- distributed among 44 haplotypes were identi- nental habitats, potentially promoting ecologi- fied. Mean nucleotide frequencies were A = cal speciation, even in the presence of small mi- 0.22, C = 0.29, G = 0.17, T = 0.32. The tran- gration between populations. sition:transversion ratio was 5.91:1. The phylo- genetic analysis (Fig. 4) assigned the haplotypes Halichoeres garnoti.•As pointed out by Smith- of Halichoeres maculipinna into two lineages (Bra- Vaniz et al. (1999, pi. 12, figs. 78-80), there are zil and the North Atlantic) separated by 45.2- noticeable color differences between Caribbean 49.0 nucleotide substitutions (d = 0.065-0.071 and Bermuda populations of this species (Fig. 774 COPEIA, 2004, NO. 4

TABLE 1. SAMPLE SIZE, NUMBER OF HAPLOTYPES, HAPLOTYPE DIVERSITY (h) AND NUCLEOTIDE DIVERSITY (TT) OF THE POPULATIONS SURVEYED.

N Haplotypes h 77 Halichoeres cyanocephalus Belize 12 6 0.68 •+- 0.15 0.002 ± 0.001 NE Brazil 5 4 0.90 ± 0.16 0.004 ± 0.003 Total 17 10 Halichoeres garnoti Bahamas 21 17 0.97 ± 0.03 0.008 ± 0.005 Belize 21 15 0.95 ± 0.03 0.009 ± 0.005 Bermuda 27 20 0.97 + 0.02 0.008 ± 0.004 Florida Keys 3 1 0.00 + 0.00 0.00 ± 0.00 St. Croix 23 18 0.98 + 0.02 0.009 ± 0.005 Venezuela 22 16 0.97 + 0.02 0.009 ± 0.005 Total 117 55 Halichoeres maculipinna Bahamas 19 12 0.90 + 0.06 0.003 ± 0.002 Belize 19 12 0.87 + 0.07 0.002 ± 0.001 NE Brazil 5 4 0.90 + 0.16 0.002 ± 0.001 Florida Keys 3 2 0.67 + 0.31 0.001 ± 0.001 St. Croix 18 9 0.80 + 0.09 0.002 ± 0.001 Trindade 5 3 0.70 + 0.21 0.001 ± 0.001 Venezuela 17 11 0.88 + 0.07 0.003 ± 0.002 Total 86 44

5A-B). Unlike what occurs in other Halichoeres notypically without an underlying genetic basis; species (color differences are matched by fixed (2) color differences are associated with differ- genetic differences in the pair Halichoeres radia- ences at loci other than the cytochrome tr, (3) tus/ brasiliensis and the Brazilian and Caribbean the population at Bermuda is young, possibly lineages of H. cyanocephalus and H. maculipinna; postdating the last glacial maximum, such that Rocha and Rosa, 2001b; Rocha, 2003a) color color differences that appeared as a result of differences are not accompanied by fixed ge- strong selection, drift or founder effect are not netic differences at the cytochrome h of H. gar- accompanied by fixed genetic differences at the noti. The population of H. garnoti in Bermuda cytochrome b. shares 12 of its 20 haplotypes with one or more The first two hypotheses remain untested; of the Caribbean locations (Fig. 3). Moreover, however, there is evidence favoring the third. haplotypes endemic to Bermuda are randomly During the last glacial maximum (25,000 to distributed across the tree on Figure 3. 15,000 yr B.R), temperatures at Bermuda were Three hypothesis may explain the absence of 3-5 C lower than today (Sachs and Lehman, genetic differentiation despite the presence of 1999), bringing mean annual temperatures be- color differences: (1) the population at Ber- low the 21 C (with much lower temperatures muda is exposed to some environmental con- during the winter) threshold necessary for reef dition that causes their color to change phe- coral survival and leading to the likely exter-

TABLE 2. POPULATION PAIRWISE «fcT FOR Halichoeres garnoti. No comparisons are significant at P = 0.05.

Locations 1 2 3 4 5 6 1. Bermuda 0 2. Florida 0.001 0 3. Bahamas -0.023 0.041 0 4. Belize 0.011 -0.026 -0.004 0 5. St. Croix -0.006 0.141 -0.024 -0.005 0 5. Venezuela -0.013 -0.078 -0.016 0.018 -0.017 0 ROCHA•MITOCHONDRIAL DNA VARIATION IN HALICHOERES 775

68 E showed the deepest genetic divergence ob- Brazil served in this study {d = 6.5%) between Brazil- 100 ^ ian and Caribbean populations, demonstrating the poor value of PLD as a predictor of genetic divergence. Despite the deep genetic split be- tween Brazil and the Caribbean, careful exami- nation of several individuals of H. maculipinna from locations at the two regions revealed no morphological differences. However, slight dif- ferences in coloration were detected (Fig. 5E- 82 : Caribbean F). 100 The divergence observed between Caribbean and Brazilian lineages of H maculipinna is the highest among all species examined. One pos- sible explanation is that this high divergence is a result of strict habitat preferences. There are no field survey data that indicates that H. ma- culipinna crosses the Amazon barrier; it is a shal- low water species (20-30 m maximum depth; -//- H. poey Randall and Böhlke, 1965; Humann and De- • 0.005 substitutions/site loach, 2002) and was not collected under the Fig. 4. Phylogenetic tree oí Halichoeres maculipinna Amazon plume (Collette and Rützler, 1977) or (Caribbean) and Halichoeres penrosei (Brazil) unique on deep sponge bottoms off northeastern Brazil haplotypes. Neighbor-joining bootstrap support (> (Rocha et al., 2000). Moreover, H. maculipinna 50%) indicated on nodes, maximum-parsimony sup- has the weakest jaw musculature, the smallest port below nodes. Branch lengths are according to mouth gap and the most specialized diet (fewer indicated scale; the branch leading to the outgroup and smaller species of benthic invertebrates (Halichoeres poeyi) was reduced by 50%. compared to congeners) among all Atlantic Halichoeres (Randall, 1967; Wainwright, 1988). As a specialized predator, H maculipinna may be mination of most of the tropical reef fish fauna more sensitive to habitat differences than the (Smith-Vaniz et al., 1999). Assuming that H. gar- other more generalist species surveyed here. In noti is a strictly tropical species (it does not oc- this case, stronger natural selection may lead to cur in the northern Gulf of Mexico or in any earlier and faster differentiation, possibly gen- coastal location north of Florida), it is probable erating the higher observed divergences in this that the Bermuda population is less than 15,000 species. years old. Thus, color differences may have a genetic basis but were not detected in neutral genes (such as the cytochrome b) because of Taxonomic implications.•The genetic divergence insufficient time for lineage sorting (Avise, between the Caribbean and Brazilian popula- 2000). tions observed in this study equals or exceeds divergences between recognized sister species Halichoeres maculipinna.•The pelagic larval du- pairs of other wrasses (Rocha, 2003a; Bernard! ration (PLD) of H. maculipinna (30 days; Spon- et al., 2004). In addition, there are diagnostic augle and Cowen, 1997) is the longest among color characters that distinguish the Brazilian the Atlantic Halichoeres. However, this species populations from their Caribbean counterparts

TABLE 3. POPULATION PAIRWISE

Locations 1 9 3 4 5 6 7 1. Florida 0 2. Bahamas -0.101 0 3. Belize -0.075 0.006 0 4. St. Croix -0.105 0.010 -0.014 0 5. Venezuela -0.100 -0.002 0.013 0.009 0 6. Coastal Brazil 0.973* 0.958* 0.969* 0.971* 0.959* 0 7. Trindade 0.984* 0.961* 0.972* 0.974* 0.963* 0.001 0 776 COPEIA, 2004, NO. 4

Fig. 5. Halichoeres garnoti at the Bahamas (A) and Bermuda (B). Halichoeres dimidiatus at northeastern Brazil (C) and Halichoeres cyanocephalus at Belize (D) Halichoeres penrosei at northeastern Brazil (E) and Halichoeres maculipirma the Bahamas (F). All photos taken by Luiz Rocha, except 5C (by Gerald Allen) and 5D (by Jack Randall).

(Fig. 5). Randall (1998) suggested that the im- Böhlke, 1965; Eschmeyer, 1998; Paepke, 1999; portance of diagnostic color characters in reef Parenti and Randall, 2000). The original de- fish systematics is dramatically increased when scription (Bloch, 1791:140) states [my transla- those characters are observed in combination tion]: "The residence of this fish is for me un- with (even minor) diagnostic counts or mea- known. The original drawing is kept in the cab- surements. Here I propose that this principle inet of Linkeschen (Heinrich Linck)." More- should be extended to genetics: the coinci- over, the original description is inaccurate and dence of diagnostic genetic and color charac- incomplete because it is based on a single, ap- ters should warrant elevation of evolutionary parently adult individual. Nonetheless, this lineages to specific status. Thus, the Brazilian name has been associated with the Caribbean populations of H. cyanocephalus and H. maculi- species since at least the end of the 19th century pinna should be considered valid species. (Jordan and Evermann, 1898). Bloch ( 1791 ) gave no type locality for H. cyan- To clarify the taxonomic status and the type ocephalus, and no types are known (Randall and locality of H. cyanocephalus (Bloch, 1791), and ROCHA•MITOCHONDRIAL DNA VARIATION IN HALICHOERES 777

Fig. 6. Neotype oí Halichoeres cyanocephalus (UF 224284). Photo by Luiz Rocha. in accordance with article 75(3) of the Inter- Distribution, habitat, and ecology.•Caribbean is- national Code of Zoological Nomenclature lands. North American coast from Florida to (ICZN, 1999), I designate UF 224284 (202.2 North Carolina and Central American coast mm SL) as a neotype. Detailed descriptions of from to Belize. Usually solitary, rare H. cyanocephalus are available in the scientific lit- in shallow waters, but juveniles and small adults erature (Randall and Böhlke, 1965) and in most frequently observed between 5 and 15 m at Be- Caribbean reef fish guides (Randall, 1996; Hu- lize; mostly observed at depths over 30 m at the mann and Deloach, 2002). remaining locations. The Brazilian species should now be consid- Halichoeres cyanocephalus (Bloch, 1791) ered valid as follows. Figures 5D, 6 Neotype.•\i¥ 224284, adult from Haulover Cut, Halichoeres dimidiatus (Agassiz, in Spix and North Miami Beach, Florida, collected on 8 Au- Agassiz, 1831) gust 1967 by W. Zeiller, at a depth between 18 Figures 5C, 7 and 27 m. This specimen was chosen because it has the typical characters described by Randall This species was originally described as Julis and Böhlke (1965) and because it is the best dimidiatus by Agassiz, in Spix and Agassiz, 1831, preserved specimen of this taxon at the Univer- and later placed in the synonymy of Halichoeres sity of Florida. cyanocephalus (Bloch, 1791) by (Jordan and Ev- ermann, 1898). Agassiz states that several spec- Description of the neotype.•^Dorsal IX, 12; anal III, imens were preserved, but only one (MHNN 12; pectoral ii,ll; total gill rakers on first arch 563, 153 mm SL) still exists (Kottelat, 1988). 19 (seven on the upper limb); lateral-line scales The type locality is given as Atlantic, off Brazil 27 (20 + 2 + 5); anterior lateral-line scales with (Spix and Agassiz, 1831). three pores; two pairs of enlarged canine teeth anteriorly in lower jaw, one pair on upper jaw. Diagnosis.•Dorsal IX, 12 (IX, 11 on remaining Standard length 202.2 mm; total length 227.5 Atlantic Halichoeres, except H. cyanocephalus); mm; body depth 29.1% SL; caudal peduncle anal III, 12; pectoral 13; total gill rakers 18 to depth 14.6% SL; head length (HL) 28.3% SL; 21; two pairs of enlarged canine teeth anteriorly snout length 37.8% HL; upper jaw length 22% in lower jaw. Juveniles and females blue with a HL; eye diameter 14.1% HL; interorbital width bright yellow region dorsally from mouth to 22.3% HL. posterior base of dorsal fin; a single dark spot on caudal (a small ocellated dark spot at rear Colcyr in alcohol.•Body light brown with a broad base of dorsal fin in addition to the caudal spot dark brown stripe on upper half, beginning at in H. cyanocephalus). Adults with a broad blue pectoral fin and extending into middle of cau- stripe on upper half of body ending at the be- dal fin; a narrower stripe of lighter brown be- ginning of the caudal fin (black and narrowing tween the broad stripe and the dorsal fin. Faint as it passes to end of caudal fin in H. cyanoce- indication of a dark bar on head, extending up- phalus); lower half of body light blue; a diagonal ward diagonally from upper posterior half of dark band from eye to nape. eye to upper portion of head. Dorsal fin dark brown on base, pale on margin; upper base of Distribution, habitat, and ecology.•From French pectoral fin with a dark spot. Guyana (Uyeno et al., 1983) to the State of Sao 778 COPEIA, 2004, NO. 4

Fig. 7. Adult Halichoeres dimidiatus, approximately 150 mm SL, Guarapari, southeast Brazil. Photo byjoao Luiz Gasparini.

Paulo (24°S, 46°W), in southeastern Brazil (Me- Diagnosis.•Dorsal IX, 11; anal 111, 11; pectoral nezes and Figueiredo, 1985). It is also present 14; gill rakers 13-15; one pair of enlarged ca- at Fernando de Noronha and Atol das Rocas nine teeth anteriorly in lower jaw (two pair of (20°30'S, 29°20'W) but apparently absent from canines on remaining Atlantic Halichoeres except the remaining Brazilian oceanic islands of Trin- H. macuUpinna). Juveniles and females with dade and St. Paul's Rocks (Gasparini and Floe- wide black stripe through eye to base of tail, ter, 2001; Feitoza et ah, 2003). Usually observed bordered above by prominent pinkish-brown solitary; juveniles relatively common in shallow line (bright yellow in H. macuUpinna). White waters (3-20 m), adults in deeper waters (30- scales with orange spots on lower half of body 60 m). (no spots on H. macuUpinna) Adult males are primarily green with scales bordered by pink Halichoeres penrosei Starks, 1913 and a black spot on anterior portion of the dor- Figures 5E, 8 sal fin. Narrow (< 0.5 mm width) orange stripes on lower half of head, from mouth to gill open- The Brazilian form of H. macuUpinna was de- ing (stripes are > 1 mm wide in H. macuUpin- scribed as Halichoeres penrosei by Starks (1913); na). the 55.7 mm SL holotype (SU 22211) is well preserved and was collected in tide pools at Na- Distribution, habitat, and ecology.•From Parcel tal, northeastern Brazil (5°S, 35°W), located at Manuel Luiz (0°52'S, 44°15'W; Rocha and Rosa, the middle of the range of the species. Digital 2001a) to the State of Sao Paulo (24°S, 46°W), photographs and a radiograph of the holotype in southeastern Brazil (Carvalho-Filho, 1999). It are available for download at the California is present at Trindade Island (20°30'S, 29°20'W; Academy of Sciences website (http://www. Gasparini and Floeter, 2001), but apparently ab- calacademy.org/research/ichthyology/Types/ sent from the remaining Brazilian oceanic is- index.html). lands (Atol das Rocas, Fernando de Noronha,

Fig.: Halichoeres penrosei, approximately 50 mm SL, Guarapari, southeast Brazil. Photo byjoao Luiz Gas- parini. ROCHA•MITOCHONDRIAL DNA VARIATION IN HALICHOERES 779 and St. Paul's Rocks; Feitoza et al., 2003). Com- be strongly influencing the biogeography of mon in coral and rocky reef tops to depths of western Atlantic reef fishes. 30 m and often observed solitary or in pairs.

MATERIAL EXAMINED CONCLUSIONS Voucher catalog numbers followed by num- Contrary to expectations of congruent pop- ber of specimens and GenBank accession num- ulation structure due to similar dispersal poten- bers in parentheses (when available). tial (25-30 days; Sponaugle and Cowen, 1997; Wellington and Robertson, 2001), the genetic Halichoeres cyanocephalus: : MCZ 14252 (3, effects of biogeographical barriers varied syntypes oí Julis internasalis Poey 1861); Flor- among species of Atlantic Halichoeres. Genetic ida: UF 65506 (1), UF 224284 (1, neotype of divergences between Brazil and the North At- H. cyanocephalus); North Carolina: UF 88359 lantic ranged from 2.3% in H. cyanocephalus/H. (2); Belize: UF 126324 (1, AY591376); St. dimidiatus to 6.5% in the pair H. maculipinna/ John, U.S. Virgin Islands: UF 206250 (1); Ja- H. penrosei. There is inconsistency between color maica: UF 229817 (1); Colombia: UF 231301 differences and genetic partitions in the species (1). Sequences without museum voucher: surveyed. The color differences between the AY591377-AY591379. species pairs H. cyanocephalus/H. difnidiatus and Halichoeres dimidiatus: (all from Brazil); Parcel H. maculipinna/H. penrosei correspond to deep Manoel Luiz, Maranhâo: MZUSP 53066 (2), genetic partitions at the cytochrome b gene. UFPB 3890 (1), UFPB 3977 (1); Fernando de However, genetic similarity at this same gene Noronha: MZUSP 14626 (1); Mamanguape, was observed between populations of H. garnoti Paraiba: UFPB 4312 (1, AY591380); Cabedelo, with striking color differences. Thus, caution Paraiba: UFPB 4325 (2); Joâo Pessoa: UFPB against the use of color differences (when not 3790 (1), UFPB 4354 (2, AY591381, supported by genetics and/or morphology) in AY591382); Recife: MZUSP 47482 (1); Salva- the alpha systematics of fishes is herein suggest- dor, Bahia: USNM 357718 (1); Guarapari, Es- ed, a conclusion also reached by others (Mc- pirito Santo: MZUSP 51543 (1), ZUEC 3184 Millan et al., 1999; Bernardi et al., 2004). (1), ZUEC 3196 (1), MBML 354 (28); Angra Genetic divergences ranged from 2.3-6.5% in dos Reis, Rio de Janeiro: MZUSP 66256 (2); lineages in Brazil and the Caribbean. A conven- Alcatrazes, Sao Paulo: MZUSP 47146 (1), tional molecular clock of 2% per million years USNM 357717 (1), ZUEC 2793 (1). Sequenc- for the cytochrome b gene (Avise, 2000) sug- es without museum voucher: AY591383. gests that this divergence is between 0.6 and Halichoeres garnoti: Bermuda: UF 119713 (11, 1.65 millions of years (myr) old, much younger AY591372, AY591374), UF 119725 (1, than the estimated age of the initial connection AY591375), USNM 348309 (1); Bahamas: UF of the Amazon with the Atlantic, 10-12 myr old 13485 (5); : UF 17620 (3); Co- (Hoorn, 1994; Nittrouer et al., 1996; Costa et lombia: UF 18830 (2); St. John U.S. Virgin al., 2001). Thus, vicariance alone cannot ex- Islands: UF 204900 (2); Belize: UF 209277 plain the observed pattern, and dispersal from (4); Cuba: USNM 343625; Tobago: USNM one area to the other after the formation of the 318896 (7). Sequences without museum Amazon is probably involved in speciation of voucher: AY591366-AY591371. western Atlantic Halichoeres. Halichoeres maculipinna: Bermuda: UF 119714 The species pair with the highest divergence (2), UF 119731 (9, AY591359); St. John, U.S. (6.5 %, H. maculipinna/H. penrosei) is also the Virgin Islands: UF 203752 (4); Bahamas: UF more reef specialized and the one with the nar- 206337 (5); Florida Keys: UF 219121 (12); Co- rowest diet width. Although apparently being lombia: UF 223079 (12); Belize: USNM able to cross the Amazon barrier, a deep phy- 329838 (1); Tobago: USNM 318864 (4). Se- logenetic break (2.3 %) between Brazil and the quences without museum voucher: Caribbean is present in the pair H. cyanocephal- AY591360-AY591365. us/H. dimidiatus. These observations indicate Halichoeres penrosei: (all from Brazil) ; Natal: SU that divergent environmental conditions (con- 22211 (1, holotype); Espirito Santo: MZUSP tinental sediment-rich Brazilian waters versus in- 52303 (1), MBML 223 (2), 318(1), 395(1, sular, clear Caribbean waters) can be as impor- AY591354), UFES 037 (1), 208(1); Ceará: tant as vast distances of unsuitable habitat (ei- MZUSP 65172 (1); Arembepe, Bahia: MZUSP ther open ocean waters or the area influenced 66259 (3); Salvador, Bahia: MZUSP 66268 (10 by the Amazon) in producing genetic partitions of 25). Sequences without museum voucher: and that benthic-stage habitat preferences may AY591355-AY591358. 780 COPEIA, 2004, NO. 4

ACKNOWLEDGMENTS preference and the marine-speciation paradox. Proc. R. Soc. Lond. B Biol. Sei. 270:1399-1406. I thank my Ph.D. advisor, B. Bowen, for pro- BLOCH, M. E. 1791. Naturgeschichte der ausländisch- viding scientific guidance, laboratory space, and en Fische. Naturg. Ausl. Fische v. 5, Berlin, Ger- financial support. I also thank A. L. Bass, A. Cas- many. tro, H. Choat, B. B. Collette, M. Coura, T. L. CARLIN, J. L., D. R. ROBERTSON, AND B. W. BOWEN. Dias, B. M. Feitoza, C. E. Ferreira, S. R. Floeter, 2003. Ancient divergences and recent connections J. L. Gasparini, Z. Hillis-Starr, S. Karl, B. Luck- in two tropical Atlantic reef fishes: Epinephelus ad- scensionis and Rypticus saponaceous (Percoidei: Ser- hurst, O. Luizjunior, D. Mûrie, D. Parkyn, B. ranidae). Mar. Biol. 143:1057-1069. Philips, J. Pitt, C. Rocha, D. R. Robertson, I. L. CARVALHO-FILHO, A. 1999. Peixes: costa brasileira. Rosa, R. S. Rosa, S. Sponaugle, B. Victor, and D. Marca D'Agua, Sao Paulo, Brazil. Weaver for help with sampling and laboratory CoLBORN, J., R. E. CRABTREE,J. B. SHAKLEE, E. PHILER, work. Discussions with C. Gilbert, S. A. Karl, N. AND B. W. BOWEN. 2001. The evolutionary enigma Knowlton, G. Paulay, and M. Westneat greatly of bonefishes (Albula spp.): cryptic species and an- improved the manuscript. S. Floeter, H. Lessios, cient separations in a globally-distributed shorefish. and W. F Smith-Vaniz critically read the manu- Evolution 55:807-820. script. G. Allen, J. L. Gasparini, and J. Randall COLLETTE, B. B., AND K. RùTZLER. 1977. Reef fishes provided excellent photographs. Specimen over sponge bottoms off the mouth of the Amazon loans, digital photographs of type specimens, River. Proc. 3rd Int. Symp. 1:305-310. COSTA, J. B. S., R. L. BEMERGUY, Y HASUI, AND M. S. and access to fish collections for morphological BORGES. 2001. Tectonics and paleogeography along analyses were made available by G. Burgess, D. the Amazon River. J. S. Am. EarÜi Sei. 14:335-347. Catania, W. Eschmeyer, K. Hartel, S. Jewett, N. CowEN, R. K. 2002. Larval dispersal and retention, Menezes, R. Robins, R. Rosa, I. Sazima, A. Var- and consequences for population connectivity, p. andas, and J. Williams. Collection permits were 149-170. In: Coral reef fishes. Dynamics and diver- provided by the Department of Agriculture and sity in a complex ecosystem. P. F. Sale (ed.). Aca- Fisheries of Bermuda (permit 01/12-2001), the demic Press, New York. Belize Fisheries Department (permit GEN/FIS/ ESCHMEYER, W. N. 1998. Catalog of fishes. California 15/04/2002), the National Park Service at St. Academy of Sciences, San Francisco. Croix, U.S. Virgin Islands (permit BUlS-2001- FEITOZA, B. M., L. A. ROCHA, O. J. LUIZ-JUNIOR, S. R. SCI-0004), and Instituto Brasileiro do Meio Am- FLOETER, AND J. L. GASPARINI. 2003. Reef fishes of biente e dos Recursos Naturais Renováveis, Bra- St. Pauls Rocks: new records and notes on biology and zoogeography. Aqua. J. Ichthyol. Aquat. Biol. zil (general collecting permit for nonprotected 7:61-82. areas). Fishes were collected following the FELSENSTEIN, J. 1985. Confidence limits on phyloge- guidelines of the University of Florida Institu- nies: an approach using the bootstrap. Evolution tional Care and Use Committee. This 39:783-791. work was financially supported by CAPES Bra- GASPARINI, J. L., AND S. R. FLOETER. 2001. The shore zilian Ministry of Education, the Brazilian Navy, fishes of Trindade Island, western South Atlantic. J. PADl Project Aware, the Smithsonian Tropical Nat. Hist. 35:1639-1656. Research Institute, the University of Florida, HASEGAWA, M., K. KISHINO, AND T YANO. 1985. Dating and the National Science Foundation through the human-ape splitting by a molecular clock of mi- a grant to B. Bowen (DEB 9727048). tochondrial DNA. J. Mol. Evol. 22:160-174. HooRN, C. 1994. An environmental reconstruction of the palaeo-Amazon River system (Middle-Late Mio- LITERATURE CITED cene, NW Amazonia). Palaeogeogr. Palaeoclimatol. ALLEN, G. R., ANDJ. E. RANDALL. 1977. Review of the Palaeoecol. 112:187-238. sharpnose pufferfishes (subfamily Canthigasteri- HuMANN, P., AND N. DELOACH. 2002. Reef fish iden- nae) of the Indo-Pacific. Rec. Aust. Mus. 30:475- tification. New World Publications, Jacksonville, FL. 517. IczN. 1999. International Code of Zoological Nomen- AMOS, B., AND A. R. HOELZEL. 1991. Long-term pres- clature. 4th ed., adopted by the International ervation of whale skin from DNA analysis. Rep. Int. Union of Biological Sciences. International Trust Whale Comm. Spec. Issue 13:99-103. for Zoological Nomenclature, Natural History Mu- AVISE, J. C. 2000. Phylogeography: the history and for- seum, London. mation of species. Harvard Univ. Press, Cambridge, JORDAN, D. S., AND B. W. EVERMANN. 1898. The fishes MA. of North and Middle America: a descriptive cata- BERNARDI, G., G. BUCCIARELLI, D. COSTAGLIOLA, D. R. logue of the species of fish-like vertebrates found ROBERTSON, AND J. B. HEISER. 2004. Evolution of in the waters of North America, north of the Isth- coral reef fish Thalassoma spp. (Labridae). 1. Mo- mus of Panama. Part II. Bull. U.S. Nat. Mus. 47:i- lecular phylogeny and biogeography. Mar. Biol. XXX -I- 1241-2183. 144:369-375. JOYEUX, J. C, S. R. FLOETER, C. E. L. FERREIRA, ANDJ. BiERNE, N., F. BONHOMME, AND P. DAVID. 2003. Habitat L. GASPARINI. 2001. Biogeography of tropical reef ROCHA•MITOCHONDRIAL DNA VARIATION IN HALICHOERES 781

fishes: the South Atlantic puzzle. J. Biogeogr. 28: , ANDJ. E. BöHLKE. 1965. Review of the Atiantic 831-841. labrid fishes of the genus Halichoeres. Proc. Acad. KNOWLTON, N. 2000. Molecular genetic analyses of Nat. Sei. Phila. 117:235-259. species boundaries in the sea. Hydrobiologia 420: , AND P. S. LoBEL. 2003. Halichoeres sodalis: a 73-90. new labrid fish from Belize. Copeia 2003:124-130. KoTTELAT, M. 1988. Authorship, dates of publication, ROCHA, L. A. 2003a. Ecology, the Amazon barrier, and status and types of Spix and Agassiz's Brazilian fish- speciation in western Atlantic Halichoeres (Labri- es. Spixiana 11:69-93. dae), p. 88. Unpubl. Ph.D. diss., Univ. of Florida, LEIS, J. M., AND M. I. McCoRMiCK. 2002. The biology, Gainseville. behavior, and ecology of the pelagic larval stage of . 2003b. Patterns of distribution and processes coral reef fishes, p. 171-199. In: Coral reef fishes. of speciation in Brazilian reef fishes. J. Biogeogr. 30: Dynamics and diversity on a complex ecosystem. P. 1161-1171. F. Sale (ed.). Academic Press, New York. , AND I. L. ROSA. 2001a. Baseline assessment of LEVITóN, A. E., R. H. GIBBS JR., E. HEAL, AND C. E. reef fish assemblages of Parcel Manuel Luiz Marine DAWSON. 1985. Standards in herpetology and ich- State Park, Maranhao, north-east Brazil. J. Fish Biol. thyology. Part I. Standard symbolic codes for insti- 58:985-998. tutional resource collections in herpetology and , AND R. S. ROSA. 2001b. Halichoeres brasiliensis ichthyology. Copeia 1985:802-832. (Bloch, 1791), a valid wrasse species (Teleostei: La- MCMILLAN, W. O., L. A. WEIGT, AND S. R. PALUMBI. bridae) from Brazil, with notes on the Caribbean 1999. Color pattern evolution, assortative mating, species Halichoeres radiatus (Linnaeus, 1758). aqua, and genetic differentiation in brightly colored but- J. Ichthyol. Aquat. Biol. 4:161-166. terflyfishes (Chaetodontidae). Evolution 53:247- , I. L. ROSA, AND B. M. FEITOZA. 2000. Sponge- 260. dwelling fishes of northeastern Brazil. Environ. MENEZES, N. A., AND J. L. FIGUEIREDO. 1985. Manual Biol. Fish. 59:453-458. de peixes marinhos do sudeste do Brasil, v. 4. Univ. , A. L. BASS, D. R. ROBERTSON, AND B. W. BOW- of Sao Paulo, Sao Paulo, Brazil. EN. 2002. Adult habitat preferences, larval dispersal, MEYER, A. 1993. Evolution of mitochondrial DNA in and the comparative phylogeography of three At- fishes, p. 1-38. In: Biochemistry and molecular bi- lantic surgeonfishes (Teleostei: Acanthuridae). ology of fishes. Vol. 2. P. W. Hochanchka and T P. Mol. Ecol. 11:243-252. Mommsen (eds.). Elsevier, New York. SACHS, J. P., AND S. J. LEHMAN. 1999. Subtropical MouRA, R. L., AND R. M. C. CASTRO. 2002. Revision of North Atlantic temperatures 60,000 to 30,000 years Atlantic sharpnose pufferfishes (Tetraodontifor- ago. Science 286:756-759. mes: Tetraodontidae: Canthigaster) with description SCHLüTER, D. 2001. Ecology and the origin of species. of three new species. Proc. Biol. Soc. Wash. 115:32- Trends Ecol. Evol. 16:372-380. 50. SHULMAN, M.J., AND E. BERMINGHAM. 1995. Early life Muss, A., D. R. ROBERTSON, C. A. STEHEN, P. WIRTZ, histories, ocean currents and the population ge- AND B. W. BowEN. 2001. Phylogeography of Ophioh- netics of Caribbean reef fishes. Evolution 49:1041- lennius: the role of ocean currents and geography 1061. in reef fish evolution. Evolution 55:561-572. SMITH-VANIZ, W. E, B. B. COLLETTE, AND B. E. LUCK- NEI, M. 1987. Molecular evolutionary genetics. Co- lumbia Univ. Press, New York. HURST. 1999. Fishes of Bermuda: history, zoogeog- NiTTROUER, C. A, S. A. KuEHL, A. G. FIGUEIREDO, M. raphy, annotated checklist, and identification keys. A. ALLISON, C. K. SOMMERFIELD, J. M. RINE, L. E. American Society of Ichthyologists and Herpetolo- FARIA, AND O. M. SILVEIRA. 1996. The geological re- gists. Spec. Pub. No. 4, Allen Press Inc., Lawrence, cord preserved by Amazon shelf sedimentation. KS. Cont. Shelf Res. 16:817-841. SPIX, J. B., AND L. AGASSIZ. 1831. Selecta genera et PAEPKE, H. J. 1999. Bloch's fish collection in the Mu- species piscium quos in itinere per Brasiliam annos seum für Naturkende der Humboldt Universität zu MDCCCXVII-MDCCCXX jussu et auspiciis Maxi- Berlin: an illustrated catalog and historical account. miliani Josephi I. Bavariae regis augustissmi peracto Theses Zool. 32:1-216, pis. 1-32. coUeget et pingendos curavit Dr J. B. de Spix [. . . ] PARENTI, P., AND J. E. RANDALL. 2000. An annotated Part 2, Monaco. checklist of the species of the labroid fish families SPONAUGLE, S., AND R. K. COWEN. 1997. Early life his- Labridae and Scaridae. Ichthyol. Bull. J. L. B. Smith tory traits and recruitment patterns of Caribbean Inst. Ichthyol. 68:1-97. wrasses (Labridae). Ecol. Monogr. 67:177-202. POSADA, D., AND K. A. CRANDALL. 1998. ModelTest: STARKS, E. C. 1913. The fishes of the Stanford expe- testing the model of DNA substitution. Bioinfor- dition to Brazil. Leland Stanford Junior Univ. Pub- matics 14:817-818. lications, Stanford, CA. RANDALL, J. E. 1967. Food habits of reef fishes of the TAMURA, K., AND M. NEI. 1983. Estimating the number West Indies. Stud. Trop. Oceanogr. Miami 5:665- of nucleotide substitutions in the control region of 847. mitochondrial DNA in humans and chimpanzees. . 1996. Caribbean reef fishes. TEH Publica- Mol. Biol. Evol. 15:512-526. tions, Neptune City, NJ. THRESHER, R. E., AND J. T MOVER. 1983. Male success, . 1998. Zoogeography of shore fishes of the courtship complexity, and patterns of sexual selec- Indo-Pacific region. Zool. Studies 37:227-268. tion in three damselfishes congeneric species of 782 COPEIA, 2004, NO. 4

sexually monochromatic and dichromatic (Pisces: Thalassoma bifasciatum: mating site acquisition, mat- Pomacentridae). Anim. Behav. 31:113-127. ing site defense, and female choice. Evolution 46: UYENO, T., K. MATSUURA, AND E. FUJII. 1983. Fishes 1421-1442. trawled off Suriname and French Guiana. Japan WELLINGTON, G. M., AND D. R. ROBERTSON. 2001. Var- Marine Fishery Resource Research Center, Tokyo. iation in larval life-history traits among reef fishes WAINWRIGHT, P. C. 1988. Morphology and ecology: across the Isthmus of Panama. Mar. Biol. 138:11- functional basis of feeding constraints in Caribbean 22. labrid fishes. Ecology 69:635-645. WARNER, R. R. 1997. Evolutionary ecology: how to rec- SMITHSONIAN TROPICAL RESEARCH INSTITUTE, oncile pelagic dispersal with local adaptation. Coral NAOS LABORATORY, UNIT 0948, APO AA Reefs 16:S115-S120. 34002-0949, USA. E-mail: [email protected]. , AND E. T. SCHULTZ. 1992. Sexual selection and Submitted: 5 April 2004. Accepted: 4 Aug. male characteristics in the Blueheaded Wrasse, 2004. Section editor: J. M. Quattro.