423 Abstract...Jfwo cosmopolitan spe­ Revision of the frigate tunas cies of frigate tunas, Auxis thazard (Lacepede) and A. rochei (Risso), are (Scombridae, AuxisJ, with descriptions differentiated primarily by the width of the corselet under the origin of the of two new SUbspecies from the second dorsal fin and by the anterior extentofthe dorsal scaleless areaabove eastern Pacific the pectoral fin. Auxis thazard has five orfewer scales inthe corselet under the second dorsal fin, and the dorsal scale­ Bruce B. Collette less area extends anterior to the tip of National Marine Fisheries Service Systematics Laboratory the pectoral fin; A. rochei has six or more scales and the dorsal scaleless National Museum of Natural History. Washington, D.C. 20560 area does not reach the tip ofthe pec­ toral fin. Of nine morphometric char­ Christopher R. Aadland acters examined with ANCOVA, four body depth measurements were signifi­ National Marine Fisheries Service Systematics Laboratory cantly different between the species. National Museum of Natural History. Washington, D.C. 20560 Auxis rochei is a slender species with Present address: AsCI Inc.. c/o USAE Waterways Experiment Station each of the four body depth measure­ Trotters Shoals Research Facility ments approximately equal, whereas P.O. Box 533. Calhoun Falls. South Carolina 29628 A. thazard is more tuna-shaped, with a deeper, more robust body and with more unequal body depth measure­ ments. Forboth species, eastern Pacific populations differ from Atlantic and Indo-West Pacific populations at what we consider the subspecific level. The Itis not clearwhetherthere are one that only one species occurred in the eastern PacificA. thazard brachydorax or two species offrigate tunas ofthe Mediterranean Sea and Atlantic new subspecies has more gill rakers. genus Auxis. In the Indo-West Pa­ Ocean and that rochei was the ear­ usually 43-48 vs. usually 38-42 in cific region, several investigators liest name for it. Richards and A. t. thazard. Body depth atanal-fin ori­ have distinguished a narrow-cor­ Randall (1967) have shown that gin, pectoral-fin length, and corselet width were also significantly different seleted species (N) from a wide­ both species occur in the Atlantic, between A. thazard brachydorax and corseleted species (W): Bleeker reopeningthe question ofthe proper A. t. thazard. The eastern Pacific (1854, 1855) from the East Indies name for Atlantic Auxis. A. rochei eudorax new subspecies has as A. tapeinosoma (N) andA. thyn­ Frigate tunas (Auxis) are the an even wider corselet than that of noides (W); Kishinouye (1915, 1923) smallest members of the tribe A. rochei rochei, usually more than 20 scales wide compared with 6--19 scales. from Japan as A. hira (N) and A. Thunnini, the true tunas. Auxis maru (W); Wade (1949) from the rochei reaches a fork length (FL) of Philippines and Jones (1958) from 600 mm andAuxis thazard, at least India as A. thazard (N) and A. 420 mm. To date, frigate tunashave tapeinosoma (W); and Matsumoto not been greatly exploited, partly (1960a) from Hawaii as A. thazard because of their small size. Of the (N) andA. thynnoides (W). However, total world catch oftunas, bonitos, most Atlantic-based workers have and billfishes in 1993 (4,654,750 considered the frigate tunas to be a metric tons (t), FAO, 1995), only single worldwide species which they 4.5% (208,305 t) was Auxis. The have identified as Auxis thazard largest landings are byAsian coun­ (Dresslar and Fesler, 1889; Jordan tries (Philippines 110,357 t; Thai­ and Evermann, 1896; Fraser-Brun­ land 36,300 t; Japan 27,896 t; and ner, 1950; Rivas, 1951; Zavala­ India 15,684 t). Auxis has soft meat Camin, 1983). Fitch and Roedel and is not suitable for sashimi (1963) have shown that there are (slices of raw fish) because of its two sympatric species offrigate tu­ strong astringent taste, probably nas in the Indo-West Pacific: the due to the high level offree amino narrow-corseletedA. thazard (Lace­ acids in the meat (Murata et al., pede) and the wide-corseleted A. 1994). At least as larvae, Auxis are rochei (Risso) (Fig. 1). Theyused the widespread and very abundant Manuscript accepted 23 January 1996. name A. rochei for the wide-cor­ (Yabe et al., 1963, fig. 15;Ahlstrom, Fishery Bulletin 94:423--441 (1996), seleted species on the supposition 1971; Ramiriz-Estevez and Ornelas- 424 Fishery Bulletin 94(3), )996 A. thazard 10 em A. roche; Figure 1 Auxis thazard (topl and A. rochei (bottom) showing anterior extent of dorsal scaleless area in relation to distal end ofpectoral fin lIeft line) and width ofcorselet under origin of second dorsal fin (right line). Roa, 1991). Richards (1984:28) concluded that Auxis names for these species. We evaluate the systematic "... is the most widely distributed and abundant status ofeastern Pacific populations ofthe two spe­ scombrid in tropical seas." With stocks ofother tuna cies and describe them as new subspecies, Auxis species continually diminishing, there is increased thazard brachydorax and Auxis rochei eudorax. potential for greater exploitation of Auxis. Before exploitation increases, itis necessary to establish how many species exist and what their correct names are, Materials and methods to determine how to identify them, andto obtain some idea ofthe major worldwide populations. Methods follow those used by Collette and Gillis Fitch and Roedel (1963) also showed that the east­ (1992), in a revision of the double-lined mackerels ern Pacific populations of both species differed sig­ Grammatorcynus, and by Collette and Russo (1985) nificantlyfrom the Indo-West Pacific populations but in a revision of the Spanish mackerels Scomber­ did not name them pending further work. We con­ omorus. Morphometric characters used in this study firm Fitch and Roedel's findings that the eastern are as follows: FL (fork length); HdL <head length); Pacific population ofA. thazard has more gill rakers PIL (pectoral-fin length); PIT (distance from pecto­ (usually 43-48) than other populations (usually 38­ ral-fin tip to anterior edge of dorsal scaleless area 42 in Atlantic and Indo-West Pacific populations), above lateral line); and CW (corselet width under similar to the counts for A. rochei. The eastern Pa­ second dorsal-fin origin). Owing to the different cific population ofA. rochei has an even wider corse­ shapes of A. rochei and A. thazard, we postulated let (usually more than 20 scales wide under the sec­ that four body depth measurements would illustrate ond dorsal fin) than other populations (8-19 scales these differences: IDor (body depth at origin offirst in populations from the Indo-West Pacific). dorsall; IDend (body depth at posterior end of first This paper distinguishes the two species. We have dorsaD; 2Dor (body depth at origin ofsecond dorsaD; located the type specimen ofrochei and designate a andAor (body depth at anal-fin origin). Auxis rochei neotype for thazard to stabilize current use ofthese is a more slender species and the four body depth Collette and Aadland: Revision of the frigate tunas rScombridae. Auxisj 425 measurements are approximately equal, whereas thazard and 291 specimens ofA. rochei. Abbrevia­ A. thazard is more tuna-shaped, having narrow en­ tions ofinstitutions housing the material examined tering and departing wedges and more unequal body follow Leviton et a1. (1985). CMFRI is the Central depths. Marine Fisheries Research Institute, Cochin, India. Morphometric data were analyzed with regression ZRC is the current institutional code for the Zoologi­ analysis, analysis ofcovariance (ANCOVAl, and prin­ cal Reference Collection at the National University cipal component analysis (PCA). Regression analy­ of Singapore. sis was run on the combined morphometric data for all populations ofA. rochei vs. the combined popula­ tions of A. thazard to determine whether morpho­ Systematics metric differences exist between species. Regression analyses were also run on morphometric data for Auxis Cuvier, 1829 major populations ofeach species to look for intraspe­ cific differences. The following combinations of geo­ Auxis Cuvier, 1829:199. Type-species: Scomber rochei graphic populations were tested: forA. rochei-west­ Risso, 1810, by subsequent designation of Gill, em Atlantic (n=36) vs. Mediterranean (n=34) plus 1862 eastern Atlantic (n=64), western Pacific (n=82) vs. eastern Pacific (n=37l; forA. thazard-GulfofGuinea Diagnosis (Modified from Collette, 1979.) Auxis (n=20) vs. Indian Ocean (n=38), western Pacific lacks the prominent paired fronto-parietal fenestra (n=75) vs. central Pacific (n=17), and westcentral in the top of the skull found in all the .other tunas, Pacific (n=92) vs. eastern Pacific (n=69). tribe Thunnini (exceptAllothunnus). The dorsal and PCA was performed with the character variables ventral branches of the cutaneous artery originate FL, 1Dor, 1Dend, 2Dor, andAor to determine whether separately from the dorsal aorta in Auxis (Godsil, these characters supported species level separation 1954); a common cutaneous artery divides into dor­ ofA. rochei from A. thazard and separation between sal and ventral branches laterally from the dorsal the eastern Pacific population and the other popula­ aorta in the other Thunnini. The ventral branch is tions for both species. Very little separation was ob­ very poorly developed, less so than inotherThunnini. served with PCA because 97.9% of the variation is The dorsal cutaneous artery lies v'entral to the cor­ explained by principal component 1. This means that responding vein inAuxis; itlies dorsal inEuthynnus almost 98% ofthe variation can be explained by dif­ (Godsil, 1954). The vertebral column differs from all ferences in size, whereas only a little more than 2% otherscombrids inhaving a long pedicel orepihaemal ofthe variation can be explained by character differ­ process (1.5-2 times the centrum depth) that forces ences among the different populations. the dorsal aorta to run much further ventrally from ANCOVA was performed on the variables 1Dor, the vertebral column than in other tunas.