BULLETIN OF MARINE SCIENCE, 25(3): 393-421, ]975 A REVISION OF THE INDO-PACIFIC ANGELFISH GENUS GENICA NTH US, WITH DESCRIPTIONS OF THREE NEW SPECIES John E. Randall ABSTRACT The Indo-Pacific pomacanthid genus Gellicallthus containing nine species is dis- tinguished chiefly by its emarginate to lunate caudal fin (often with produced lobes), relatively short teeth, and sexual dichromatism. Males have either bars or stripes on the body (black on all species except G. bel/us which has a yellow stripe) and lack black on the lobes of the caudal fin, whereas females (except the striped G. lamarck and the diagonally banded G. bel/us) lack dark markings on the body and all except G. spillus have a prominent black band in each lobe of the caudal fin. G. semicillctu.l' (Waite), known only from Lord Howe Island, is resurrected from the synonymy of G. melallospilos (Bleeker) of the western Pacific. G. caudibicolor (Lienard, in Sauvage) and G. zebra (Lienard, in Sauvage) are regarded as synonyms of G. caudovillatus (GUnther) of the western Indian Ocean; the latter, however, is not a synonym of G. melallospilos, as some authors have contended. G. macclesfieldiellsis Chan is referred to the synonymy of G. melanospilos. G. fucosus Yasuda and Tominaga is the probable female form of G. semi- fasciatus (Kamohara), which ranges from the Philippines to Japan. The distribution of G. melllllOspilos is extended to the New Hebrides, Loyalty Islands, Solomon Islands, Palau Islands, Fiji Islands, New Guinea, and the Ryukyu Islands; that of G. lamarck to the Solomon Islands; and that of G. watanabei to Pitcairn Island, Austral Islands, New Cale- donia, and Osprey Reef (Coral Sea). The following three new species are described: G. spillus (a relative of G. semicillctus) from the Pitcairn Group and Austral Islands; G. bel/us from Tahiti and Cocos-Keeling Islands; and G. persollatus from Hawaii. The male of the latter has not yet been collected. Swainson (1839) established the poma- re-established as a genus of Pomacanthidae. canthid genus Genicanthus, noting that the In addition to the concave caudal fin, he body was more elongate than Holacanthus distinguished Genicanthus from Holacanthus and Pomacanthus and the caudal fin "deeply as follows: "Preorbital notched mesially, its lunated, the ends extended into filaments." hind margin free, serrated. Teeth in both He added "operculum spined as in Hola- jaws short. Scales on operculum in six to canthus." He mistakenly included Hola- eight rows." (Holacanthus listed as having canthus tricolor (Bloch) with H. lamarck about nine rows). Though many authors Lacepede in Genicanthus; tricolor is the type such as Fowler (1934), Weber and de Beau- specics of Holacanthus. fort (1936), Smith (1955), Norman (1957), Genicanthus went unrecognized as a valid Chan (1965), who named G. macclesfieldi- genus for many years. Bleeker (1857) ensis from the South China Sea, Munro described Holacanthus melanospilos from (1967), Araga (1972), and Randall (1973) Ambon, East Indies; Gunther (1860) named have followed Fraser-Brunner, others such the related H. caudovittatus from Mauritius; as Kamohara (1934, 1961), who described and Waite (1900) described H. semicinctus Holacanthus semi/asdatus from southern from Lord Howe Island. All of these species Japan, Watanabe (1949), Herre (1953), should have been placed in Genicanthus. and Yasuda and Tominaga (1970), have Not until the revision of the angelfishes reverted to Holacanthus. by Fraser-Brunner (1933) was Genicanthus In their excellent paper on "long-tailed" 393 394 BULLETIN OF MARINE SCIENCE, VOL. 25, NO.3, 1975 angelfishes, Yasuda and Tominaga described Holacanthus as Genicanthus, one or two two new species, G. fucosus from Japan and more rows may be counted on the former. G. watanabei from Okinawa. They wrote It is with the emarginate to lunate caudal that the pre orbital of Genicanthus is not fin, however, that Genicanthus is most readily always notched mesially and its hind mar- differentiated from all other angelfishes. gin is not always free, that the teeth are not Fraser-Brunner separated Genicanthus and always short, and they doubted the value of Holacanthus from Heteropyge (= Euxiphi- the number of vertical scale rows on the pops) in part as follows: "Lateral line ter- operde as a distinguishing character because minating at end of soft dorsal, sometimes a the count "may differ by one or two depend- separate portion on caudal peduncle." The ing on the investigators' counting method." lateral line was described as complete for They preferred to retain the long-tailed spe- Euxiphipops. Although the lateral line of cies in Holacanthus until an osteological Genicanthus and Holacanthus may be poorly study of the world representatives of the developed in the descending portion beneath family can be undertaken. the hind part of the soft portion of the dorsal While osteological research is certainly fin and one or two scales of the series in this needed to resolve confusion in the generic sector may lack pores, I would not describe it classification of pomacanthids, there does as incomplete for these genera. seem to be enough justification from external Genicanthus is unique among the poma- characters, the mode of life, and the striking canthid genera in the striking sexual di- sexual dichromatism to warrant the retention chromatism exhibited by all the species (ex- of Genicanthus as a genus. cept G. personatus for which the male is not I find Fraser-Brunner's separation of yet known). I first suspected that the color Genicanthus from Holacanthus largely cor- differences in Genicanthus were sexual when rect. The anteroventral notch in the pre- I speared four specimens of G. watanabei in orbital is present on the various species of Tahiti in 1969 and found a striped male and Genicanthus though not always as deep as three females with unmarked bodies. I ob- illustrated by Fraser-Brunner (1933: text- served that the two color forms occurred fig. 16), and it may be partially obscured by together in the same deep reef habitat; also, spines. Furthermore, the hind margin of the no small individuals in the striped pattern preorbital is free for about one-third to one- were seen. Similar observations were made half the distance from the lowermost edge to of G. melanospilos in the Palau Islands in the eye. However, a similar notch can be 1970. More convincing was the observation found on the preorbital of some specimens of of courtship of the barred male and plain- Holacanthus, and the posterior margin of bodied female of G. caudovittatus in the Gulf of Aqaba, Red Sea, in 1972, as well as this bone can be as free as that of Geni- noting an adult individual just developing canthus. the barred color pattern of the male. No Compared to Holacanthus, the teeth of small individuals of any species of Geni- Genicanthus are indeed relatively short. canthus have been found with the male color Their length is contained about five times in pattern, thus suggesting that sex reversal the diameter of the eye, whereas the length takes place, as has been demonstrated for of the teeth of Holacanthus is contained from various Serranidae, and Labridae. less than two to three times in the eye. The species of Genicanthus are poorly Admittedly, the vertical scales on the represented in the collections of the museums opercle are difficult to count because they of the world. In fact, some major museums do not lie in regular rows, and there may be have no specimens of the genus. This is small scales at either end. But if the same partly due to the occurrence of the species method is used to count the larger rows on in relatively deep water. Although one or RANDALL: REVISION OF THE ANGELFISH GENUS GENICANTHUS 395 two species may occasionally be seen in as marek, G. melanospilos, G. watanabei, und little as 15 m, these fishes are not commonly G. semifasciatus. It is possible that G. bellus found in less than about 30 m. Also, they will ultimately be found there also, for Indo- arc associated with reefs or rocky bottom, nesia is contained within the present range oftcn ncar vertical discontinuities, hence are of Tahiti and Cocos-Keeling Islands. not apt to be taken in trawls. They have Only two species are definitely known from small mouths, thus are rarely caught on the Indian Ocean: G. caudovittatus in the hook and line. western part, ranging from Mauritius to the They swim higher above the bottom, Red Sea, a:Id G. bellus from Cocos-Keeling. gcncrally, than other angelfishes, probably The one record of a juvenile G. lamarck from reflecting their feeding to a significant degree Kenya cannot be confirmed due to loss of the on the larger animals of the zooplankton (the specimen. fcw stomach content analyses suggest that In southern Oceania there are three spe- pelagic tunicates are the dominant plank- cies: G. watanabei, G. semicinctus, and tonic animals consumed; benthic material G. spin us (the last two are an allopatric such as polychaetes, bryozoans, and algae pair) . The only record from northern have also been found in some stomachs, Oceania is G. personatus from Hawaii. however) . In contrast, the species of The apparent absence of the genus from Holacanthlls, Pomacanthus, Pygoplites, and the broad central region of the tropical Euxiphipops fecd mainly on sponges, and Pacific is somewhat puzzling. Some of the Centropyge on algae and detritus (Randall, island groups such as the Marshall Islands, 1967; Randall and Hartman, 1968; Randall, Mariana Islands, and Samoa Islands have MS). been well collected. At Enewetak in the In view of the morphological similarity, it Marshalls the author and associates have is plausible that Genicanthus has been de- made numerous deep dives, but no Geni- rived from Holacanthlls.