AMERICAN MUSEUM Novltates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2983, 57 pp., 50 figs., 1 table October 23, 1990
On Muller's and Cuvier's Concepts of Pharyngognath and Labyrinth Fishes and the Classification of Percomorph Fishes, with an Atlas of Percomorph Dorsal Gill Arches
DONN E. ROSEN' AND COLIN PATTERSON2
Foreword (Colin Patterson) euteleosts whose objective was "to lay a foun- dation for future analysis of the Acanthop- Donn Rosen's ambition, during the last de- terygii" (Rosen, 1973: 398); and its conclu- cade and more of his life, was to crack the sion was that acanthopterygians may be problem ofthe Perciformes, by far the largest defined by two features ofthe dorsal gill arch- and most diverse "order" offishes. Humphry es, but that Perciformes possess no known Greenwood and I first asked him to tackle synapomorphies, and so may not be mono- the job back in 1970, for the "Interrelation- phyletic. ships of Fishes" meeting held in London in During the rest of the 1970s, Donn was 1972. Donn took some persuading, but at occupied with other things, notably the rev- length agreed ("from now on, it's spines in olution in biogeography (Rosen, 1975, 1978, the chowder"-letter, 1971). The work for 1979; Nelson and Rosen, 1981), but by early that paper led him for the first time into the 1981 he was ready for another frontal assault anatomy of the dorsal part of the pharynx, on the perciforms, and had an NSF grant "to and of the dorsal gill arches in particular. As convert my office into a percomorph factory" might have been predicted, the problem of (letter, April 1981). Within a year, "the per- the Acanthopterygii, and the Perciformes in comorph problem has re-expanded into the particular, turned out to involve all or most neoteleost problem once again, so I'm look- of the Euteleostei. So Donn's paper on acan- ing at all paracanthopts, beryciforms, and thopterygians became a study of higher myctophids in addition" (letter, March 1982).
' Curator, deceased 1986, Department of Ichthyology, American Museum of Natural History. 2 Research Associate, Department of Ichthyology, American Museum ofNatural History; Department of Palaeon- tology, British Museum (Natural History), London SW7 5BD.
Copyright ©) American Museum of Natural History 1990 ISSN 0003-0082 / Price $7.00 2 AMERICAN MUSEUM NOVITATES NO. 2983
Out of this grew another review of higher pound, figures. There seemed to me to have euteleosteans (Rosen, 1985), a revision of been two principal impulses behind the work. paracanthopts (Patterson and Rosen, 1989), The first is an idea characteristic of the de- and one effort to tackle the perciforms by velopment of pattern cladism: that preevo- attrition rather than frontal assault (Rosen, lutionary systematists such as Muller and Cu- 1984). vier, unrestricted by theory, might have had In the spring of 1983, he underwent his a better insight into broad patterns of rela- first bout ofbrain surgery. His recovery from tionship than Donn's contemporaries. The that seemed excellent, and within a year he second is the belief that those contemporar- visited London to continue our joint work ies, particularly recent contributors on La- on paracanthopts. But he did not fully re- broidei ; Muller's Pharyngognathi acan- cover the use of his left hand, which meant thopterygii), tended to be too narrowly that he could no longer easily carry out the focused in their view of a problem and not microdissections ofcleared-and-stained fish- ready enough to cast the net of investigation es in the glycerin dish which for him "was more widely. But Donn would have been the crucible in which truth had to leave its ready to admit that his own approach to the residues" (Nelson et al., 1987). He was not problems was limited in different ways, prin- short of material, however, for in the "per- cipally by his physical handicap, which made comorph factory" in the two years before his not only dissections but his character and illness he had assembled a large collection of literature surveys more casual than they would pencil drawings, of dorsal gill arches in par- once have been. ticular, and as Nelson et al. (1987) said, Since Donn died, there have been a num- "Typically, he would accumulate the draw- ber of important contributions to the perci- ings and then write the manuscript to fit form problem, and to pharyngognaths in par- them." The following paper was begun in that ticular. Most notable is Stiassny and Jensen's way in summer 1985, after his recovery from (1987) review of labroid interrelationships, the first surgery, and was taken up again after which was dedicated to Donn's memory and the second operation, which occurred in the in which his manuscript was acknowledged fall of 1985, a year before his death. In those as a thought-provoking impetus. Sadly, I circumstances, it is not surprising that he felt found that this work had removed much of the need for the other characteristic of his the stuffing from Donn's manuscript, for one working method noted in Nelson and other's ofhis principal conclusions, in turn the step- obituary: "by nature he preferred collabora- ping-stone for further speculation, was that tion." He turned to two of his oldest collab- the Labroidei of Kaufman and Liem (1982) orators, Humphry Greenwood and myself, and Stiassny and Jensen (1987) (Cichlidae, and to one ofhis youngest, Melanie Stiassny, Embiotocidae, Pomacentridae, Labridae, now his successor at the AMNH; each of us Odacidae, Scaridae) might not be monophy- was shown drafts ofthe work in progress, and letic. In particular, Donn argued that certain was asked to join him. Working with Donn "percoid" groups (Girellidae, Kyphosidae, in those days, in late 1985 and 1986, was no Scorpididae) are more closely related to em- picnic, and perhaps we may each be forgiven biotocids than to other labroids, and he also for not accepting Donn's request for help with suspected that the Sparoidea and Haemuloi- the total commitment that he gave to this dea ofJohnson (1980), together with the Ger- project during what remained of his life. I reidae, might be more closely related to the spent two periods with him in early 1986, in Labridae, Odacidae, and Scaridae than are February and in April, and it was clear that cichlids, embiotocids, and pomacentrids. As he held on to the project as his lifeline back he would surely have acknowledged, his po- to scientific well-being during those last dif- tential apomorphies favoring those views are ficult months. thin on the ground and I was unable further The manuscript, as Donn left it, consisted to substantiate them, whereas Stiassny and of about 80 pages of typescript and almost Jensen's (1987) survey of labroid characters, 150 drawings, arranged in 47, mostly com- although it does not close the book on the 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 3 matter, leaves little room for Donn's hy- sen's (1987) comment, I have not added any potheses. With them removed, the substance reference to it. In a revision ofthis sort, there of his manuscript was much reduced, but I is a problem with the first person singular in have included his accounts of the character statements like "I have found," "I conclude," complexes which he thought bore on the or "my own studies": these are left as written, problem. The paper also included a general with the implication that the "I" or "me" cladogram covering the Acanthomorpha. This concerned in the main text is Donn Rosen. I incorporated a number of novel ideas, most have also left the bulk of the figures as they notably that myctophids, polymixiids, par- were, in the belief that the paper will have acanthopts, and atherinomorphs are more lasting value as an atlas of osteology, partic- closely related to higher perciforms than are ularly ofthe dorsal gill arches, in a wide range holocentrids and basal percoids. These ideas of perciforms. As it now stands, the paper is were not convincingly supported by charac- less the radical reorganization that Donn ters, whereas they are contradicted by char- planned, and more a critique of recent work acters cited by Johnson (1984), Stiassny on various perciform groups; it points out (1986), and others. I therefore omitted the incongruent or questionable characters, rath- cladogram and the text referring to it, but er than presenting a new synapomorphy have deposited it, together with a copy ofthe scheme. Nevertheless, I believe it deserves a manuscript as Donn left it, in the Dean Li- place in the literature. brary of the Department of Ichthyology in For help and hospitality during the work, the American Museum of Natural History. I am grateful to all the members of the De- In reworking the manuscript for publica- partment of Ichthyology in the American tion, I have incorporated references to more Museum of Natural History, in particular to recent work, have rewritten the historical sur- Gareth Nelson and Melanie Stiassny, also to veys, rewritten, reorganized or omitted major Stan Blum in the Department of Vertebrate parts ofthe main text, and added a few figures Paleontology, and above all to Donn's wid- and observations based on my work. In the ow, Mel Rosen, whose fortitude and dedi- original manuscript there was no mention of cation are beyond my praise. In London, Pholidichthys leucotaenia Bleeker, the only Humphry Greenwood and Gordon Howes member of the Pholidichthyidae. Pholidich- have been generous with time and advice. thys is a superficially blennylike fish which is Finally, I thank Dave Johnson for a most remarkably similar to the labroids in a num- thorough and sensitive review of the manu- ber offeatures ofgill-arch structure (Springer script. I regret that I have been unable to and Freihofer, 1976; Stiassny and Jensen, come up with rational responses to several 1987: 294); I do not know whether it was of his comments; yet I found myself unable omitted by intention or accident, but since I to respond by further excision, which would have no access to material and can add noth- have removed too much ofthe flavor ofwhat ing to Springer and Freihofer's (1976) de- Donn, with such effort, was striving toward. scription and discussion or Stiassny and Jen-
ABSTRACT A review of the anatomy of the buccal and pha- is supported to include the labroids, and the spar- ryngeal jaws, basicranial specializations associated oids and haemuloids, and also the gerreids, ky- with the pharyngeal jaws, and palatal musculature phosids, scorpidids, arnd girellids. attaching to the prootic and parasphenoid bones The problem of the Labyrinthici is also re- has led to a reevaluation of some percomorph re- viewed because certain aspects of their anatomy, lationships that extends and supports the recent especially in anabantoids, have been considered work of Johnson (1984), G. Nelson (1967, 1969), pharyngognathous. The group is considered cor- Stiassny (1986, Stiassny and Jensen, 1987), Lau- rectly to include the Anabantidae, Nandidae, Bad- der and Liem (1983), and some older taxonomic idae, Osphronemidae, Belontiidae, Helostomidae, concepts of Muller, Gunther, Gill, and Cuvier. Channidae, Synbranchidae and Luciocephalidae, The concept of a Pharyngognathi, in particular, and perhaps the Mastacembelidae. 4 AMERICAN MUSEUM NOVITATES NO. 2983
The Acanthomorpha as redefined by Stiassny Recent predictions that these large problems are (1986) are reduced to 10 branches with a few po- beyond the scope ofpresent investigation and res- lychotomies that need to be tied to each other and olution appear overly pessimistic. to the percoid residue discussed by Johnson (1984).
INTRODUCTION Merely corroborative detail, intended to give artistic verisimilitude to an otherwise bald and unconvincing narrative (Pooh-Bah to Pitti-Sing). [From the complete libretto of The Mikado, p. 152 of The Best Known Works of W. S. Gilbert; Arden Book Co., J. J. Little and Ives Co., New York.]
The above quotation captures the spirit of by subsequent investigations, many taxono- what has been written more recently as elab- mists continued to use Regan's system almost orations on Muller's (1844, 1845, 1846) and unchanged (e.g., Berg, 1940; Greenwood et Cuvier's (1828, 1831, in Cuvier and Valen- al., 1966; Gosline, 1971; McAllister, 1968). ciennes, 1828-1849) casually formulated In contrast, Jordan (1923) used a modified concepts of pharyngognath and labyrinth version of the Pharyngognathi, and such re- fishes, and captures also the spirit of what I cent studies as those by G. Nelson (1967), have to say below about those recent elabo- Stiassny (1981, 1982; Stiassny and Jensen, rations. It seems that, although Cuvier's and 1987), Liem and Greenwood (1981), and Muller's narratives were bald by modem Lauder and Liem (1983) have illustrated the standards, they were far from unconvincing general usefulness of Muller's conceptions. since ichthyologists have found these group The work reported here started with a study concepts apt, at least in part. ofthe dorsal gill arches and supports, in part, For more than a century percoid fishes have Muller's original notions. Muller's (1844, posed a taxonomic challenge to ichthyolo- 1845, 1846) Pharyngognathi comprised two gists concerned with producing a resolved hi- suborders, Ph. acanthopterygii, and Ph. mal- erarchical classification ofmonophyletic taxa. acopterygii: the first contained three families, As understood by Regan (1913), the percoids Labroidei cycloidei, Labroidei ctenoidei, and constitute a suborder that includes 12 form- Chromides; the second contained the family groups encompassing 82 families. One of Scomberesoces. The gill arch character cited Regan's form-groups is the division Perci- as uniting them was "the lower hyoid bones formes, recognized today as the suborder Per- united" (p. 533 in Griffith's 1846 transla- coidei, which contained 49 of his 82 percoid tion). Commenting on this arrangement, families. Regan's concept of this group was Gunther (1862: iii) wrote "I fully share the based only on symplesiomorphy since he re- opinion of those who do not consider the ferred to the percoids as a whole as "the most coalesced pharyngeal bones as a character of generalized suborder, defined by the absence sufficient importance to unite acanthopterous ofthe special peculiarities which characterize and malacopterous fishes into one Order, I the other suborders ofthe Percomorphi," thus have changed the name of Pharyngognathi appealing openly to the absence of informa- acanthopteri into Acanthopterygii pharyn- tion. Regan's percoid classification was a ret- gognathi." Within that order, Gunther placed rogressive step in our understanding of fish five families: Pomacentridae (= Muller's Ph. relationships, since some of the taxa in his ctenoidei), Labridae (= Muller's Ph. cycloi- basal Perciformes have long been grouped in dei), Embiotocidae, Gerr[e]idae (placed "near such subsequently corroborated assemblages the Embiotocidae"), and Chromides. Gill as the Pharyngognathi of Johannes Muller (1872: xiii) commented that Muller's pharyn- (1844, 1845, 1846), who was explicit about gognath character, united pharyngeal bones, his intent in producing a natural system. "would violently divorce forms from their Even though Muller's classification had natural allies, and ... was one liable to recur been supported, in part, and enlarged upon in very dissimilar groups, and not ... a tech- 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 5 nical expression of a natural group." Gill's 112) and his main text (p. 131): in the former Pharyngognathi (1872: 6) differed from Gun- they are called Embiotociformes and placed ther's in only two minor details; Chromides after the Labriformes, whereas in the latter were renamed Cichlidae, and Labridae were they are separated from Labriformes by the split into three families, Labridae, Scaridae, Pomacentriformes and are named Ditremi- and Gill's own (1863) monotypic Siphon- formes. Regan wrote (p. 131) that "The ognathidae. Gunther (1880) later modified 'pharyngognathous' fishes ... form three well- the content ofhis Pharyngognathi by remov- marked and probably not specially related ing Gerreidae (Gerres only) to the Percidae. groups, which resemble each other in the an- In Boulenger's classification (1 904a, kylosis of the lower pharyngeals.... Were it 1904b), the Pharyngognathi were no longer not for this all three would be included in the maintained as a separate group. Gunther's division Perciformes, since their external and and Gill's pharyngognath families were placed anatomical characters are those of typical in the division Perciformes, first of nine perciform fishes." Regan made a similar subgroups in the Acanthopterygii. Within the comment-"three distinct and apparently Perciformes, Boulenger separated the Po- unrelated groups" in his 1929 classification. macentridae, Labridae, and Scaridae from the Berg (1940) repeated Regan's classification, remainder by their possession of 31/2 gills in- ranking the three pharyngognath groups as stead of four. Among the four-gilled Perci- superfamilies. formes, he listed Embiotocidae and Cichlidae The next major classification of teleosts, next to the last three families, but put the Bertin and Arambourg's (1958), recognized Gerreidae close to the Percidae. Goodrich the Pharyngognathi as a suborder Labroidei, (1909) returned to a more explicit system in containing the four families Pomacentridae, retaining Pharyngognathi (as a "sub-tribe" Embiotocidae, Labridae, and Scaridae. The within the Perciformes). His five pharyngog- first of these families was called "un terme nath families were divided into three groups, de passage" between Cichlidae and the last Pomacentridae + Cichlidae, Embiotocidae two. Greenwood et al. (1966) wrote the first alone, and Labridae + Scaridae. Jordan's major classification entirely to omit any (1905, 1923) arrangement was akin to Good- pharyngognathous grouping: the suborder rich's: his Chromides included Pomacentri- Labroidei was maintained, but it contained dae and Cichlidae, with (1923) a third family only the Labridae, Odacidae, and Scaridae; for the Eocene tPriscacara; his Pharyngog- the pharyngognathous embiotocids and po- nathi contained the Labridae and Scaridae, macentrids were relegated to the Percoidei, with (1923) the tPhyllodontidae and five oth- where they are placed on either side of the er families split from among the labrids and Cichlidae. McAllister (1968) retained a more scarids of which only the Odacidae are still Regan-or Berg-like classification, with three recognized; and his Holconoti contained the pharyngognathous suborders, Embiotocoi- Embiotocidae alone. Jordan's Chromides, dei, Pomacentroidei, and Labroidei. Perhaps Pharyngognathi, and Holconoti were placed in recognition of Regan's comments on lack as three neighboring suborders in 1905, and of relationship between the three, the third raised to ordinal rank in 1923. is separated from the first two by several oth- Regan's (1913) influential classification of er groups. Gosline (1971) used a classification percoids resembled those of Goodrich and much like Regan's and Berg's, with adjacent Jordan in two respects. First, he isolated the superfamilies Embiotocoidae, Pomacentroi- Labridae, Scaridae, and Odacidae in a divi- dae, and Labroidae placed at the end of his sion Labriformes; and second, he isolated the Percoidei. J. Nelson (1984) adopted a scheme Embiotocidae in a division Embiotociformes. like that of Greenwood et al., recognizing a However, Regan separated the Pomacentri- suborder Labroidei (Labridae, Odacidae, dae from the Cichlidae, maintaining a divi- Scaridae), but leaving Embiotocidae and Po- sion Pomacentriformes for pomacentrids macentridae in the basal Percoidei, where they alone, and leaving cichlids in his basal di- are placed next to Cichlidae. vision Perciformes. Regan's treatment ofem- The modern works summarized in the last biotocids is inconsistent in his synopsis (p. paragraph express the precladistic phase of 6 AMERICAN MUSEUM NOVITATES NO. 2983
TABLE 1 Historical Summary of Classifications of Pharyngognath Fishes Muller, 1846 Suborder Pharyngognathi acanthopterygii Families Labroidei cycloidei, Labroidei ctenoidei, Chromides Gunther, 1862 Order Acanthopterygii pharyngognathi Families Pomacentridae (= Muller's Ph. ctenoidei), Labridae (= Muller's Ph. cycloidei), Embiotocidae, Gerreidae, Chrom- ides Gill, 1872 Superfamily Pharyngognathi Familes Pomacentridae, Embiotocidae, Gerreidae, Cichlidae (= Muller's Chromides), Labridae, Scaridae, Siphonognath- idae Goodrich, 1909 Sub-tribe Pharyngognathi Families Pomacentridae, Cichlidae, Embiotocidae, Labridae, Scaridae Jordan, 1923 Order Holconoti Families Embiotocidae, Hysterocarpidae Order Chromides Families Pomacentridae, tPriscacaridae, Cichlidae Order Pharyngognathi Families tPhyllodontidae, Labridae, Coridae, Neolabridae, Spar- isomidae, Scaridae, Odacidae, Siphonognathidae Regan, 1929 Division Embiotociformes Family Embiotocidae Division Pomacentriformes Family Pomacentridae Division Labriformes Families Labridae, Odacidae, Scaridae Berg, 1940 Superfamily Embiotocoidae Family Embiotocidae Superfamily Pomacentroidae Family Pomacentridae Superfamily Labroidae Families Labridae, Odacidae, Scaridae Bertin and Arambourg, 1958 Suborder Labroidei Families Pomacentridae, Embiotocidae, Labridae, Scaridae Greenwood et al., 1966, Suborder Percoidei J. Nelson, 1984 Families Embiotocidae, Cichlidae, Pomacentridae [and many others] Suborder Labroidei Familes Labridae, Odacidae, Scaridae fish systematics. The trend, in the 140 years Embiotocidae, and Gerreidae were not summarized in table 1, has been to play down grouped with them. the pharyngeal jaw character first recognized The first attempt to apply cladistic prin- by Muller, as convergence or parallelism, and ciples, or grouping by synapomorphy, among to dissociate the fishes that he and Gunther pharyngognaths was G. Nelson's (1967) study grouped by means of that character. The La- ofgill arches. He found no support for Tarp's broidei (Muller's Pharyngognathi cycloidei; (1952) proposal that the closest relatives of Gunther's Labridae) survived, but the Po- embiotocids are Girellidae, but did find dor- macentridae (Muller's Pharyngognathi cte- sal gill-arch features, additional to the fused noidei), Cichlidae (Muller's Chromides), fifth ceratobranchials, relating pomacentrids 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 7
and embiotocids (in that sequence) to la- kyphosids, sciaenids, gerreids, and poma- broids. Stiassny (1981) found further gill-arch dasyids, and certain anabantoids, centrar- apomorphies relating the other members of chids, and carangids. Muller's and Gunther's pharyngognaths- Stiassny and Jensen (1987) reexamined the Gerreidae and Cichlidae-to Nelson's as- implications of pharyngognathy in acantho- semblage. Stiassny found that these apo- morphs, and produced a better supported and morphic features showed a mosaic distribu- more carefully argued scheme than did their tion, with certain characters relating cichlids predecessors. A PAUP analysis of 15 char- more closely to embiotocids and labroids, acters in the four families comprising Kauf- others relating cichlids to embiotocids, and man and Liem's (1982) Labroidei (Cichlidae, at least one other (microbranchiospines) re- Embiotocidae, Pomacentridae, Labridae KL) lating gerreids and cichlids. She also found generated a single most parsimonious tree dif- indications that the Sparidae might be related fering from its predecessors in placing cich- to pharyngognaths. lids as the sister group of other labroids, and Liem and Greenwood (1981) provided the embiotocids as the sister of Pomacentridae first justified phylogenetic hypothesis of + Labridae KL. Eight out ofthe 15 characters pharyngognath fishes. They proposed that la- are homoplastic on this tree. The tree re- broids are monophyletic, that their sister quires 23 steps, whereas Liem and Green- group is Cichlidae + Embiotocidae, and that wood's (1981) and Kaufman and Liem's each ofthose families is monophyletic. Liem (1982) schemes require, respectively, 29 and and Greenwood considered pharyngognathy 26 steps with Stiassny and Jensen's charac- in pomacentrids, girellids, sciaenids, and also ters. A tree differing from Stiassny and Jen- in anabantoids, concluding (p. 98) "we reject sen's by reversing the position of pomacen- the hypothesis that any of the other phar- trids and embiotocids requires 24 steps. yngognathous acanthopterygians are related" Stiassny and Jensen (1987), like Liem and to their group. Kaufman and Liem (1982) Greenwood (1981) and Kaufman and Liem revised this conclusion and proposed a new (1982), were interested primarily in labroid hypothesis, summarized in a cladogram in monophyly and intrarelationships, but they which embiotocids are the sister group of la- mentioned previous conjectures of relation- broids (rather than of cichlids), cichlids are ship between various labroids and other tele- the sister group of embiotocids + labroids, osts. These include Tarp's (1952: 16) prop- and pomacentrids are the sister group of that osition that the Girellidae are "the closest assemblage. They named the whole group La- living relatives of the Embiotocidae"; Mor- broidei [it is in fact Gunther's (1862) Pharyn- ris's (1982) suggestion that pomacentrids and gognathi, minus gerreids], renaming the old embiotocids are sister groups with the Scor- Labroidei by lumping Labridae, Odacidae, pididae as possible ancestors (implying that and Scaridae in Labridae. Accepting that, in all or some scorpidids might be the sister the rest of his paper the expanded Labridae group of the two pharyngognath families); of Kaufman and Liem is called the Labridae Stiassny's (1981) note of labroidlike upper KL to distinguish it from the restricted Labri- pharyngeal features in Sparidae and Gerrei- dae of previous authors. Only two incon- dae; and Stiassny and Jensen's (1987) re- gruent characters are mentioned in Kaufman ported ofunited fifth ceratobranchials in some and Liem's paper: the unification ofthe lower leiognathids. pharyngeals and their dentition in pomacen- Thus recent cladistic studies have reem- trids, a feature that must be homoplastic in phasized the reality ofthe Pharyngognathi of their scheme since cichlids retain unfused Muller and Gunther, and the search for the lower pharyngeals without symphysial teeth; sister group ofthat assemblage, though hard- and pharyngocleithral joints in pomacentrids ly undertaken yet, might be directed toward and Labridae KL. As possible relatives of the nonpharyngognath families mentioned in their labroid assemblage ("one ofwhich may the work summarized above: Carangidae, represent the primitive sister group of the Centrarchidae, Gerreidae, Girellidae, Ky- Labroidei") Kaufman and Liem mentioned phosidae, Leiognathidae, Pomadasyidae, 8 AMERICAN MUSEUM NOVITATES NO. 2983
Sciaenidae, Scorpididae, Sparidae, and some parasphenoid teeth (Pristolepis, Nandus, anabantoids. The various percoid families Badis, Channa) in "muscular, neuronal and named bring in one of the great unsolved biomechanical" features so that "pharyngog- problems in systematic ichthyology, inter- nathy in the Anabantoidei represents a relationships among Regan's Percoidei; and uniquely specialized character complex both some of these percoids and the anabantoids functionally and morphologically." bring in two other ancient groups, Cuvier's Among workers previously concerned with Squamipinnes and Labyrinthici. Concerning the problem of what pharyngognathy might the first of these three groups, the Percoidei, mean taxonomically, there is significant dis- recent studies, notably those of Johnson agreement about which taxa should be in- (1980, 1984, 1986; Tyler et al., 1989), have cluded in a larger group with labrids and with begun to impose some order on chaos. John- anabantoids. Within a labroid clade, ques- son (1980) argued that the Sparidae are re- tions about including pomacentrids, girellids, lated to Centracanthidae, Lethrinidae, and sciaenids, sparids, and gerreids have been Nemipteridae in a monophyletic Sparoidea. raised. And, based on my own studies, I would In 1984 Johnson proposed (p. 496) that Ca- include haemulids on that list. Johnson's rangidae form a monophyletic Carangoidea (1980) concepts of Sparoidea and Haemu- with the Coryphaenidae, Echeneididae, Ne- loidea suggest that centracanthids, nemipter- matistiidae, and Rachycentridae; and (p. 469) ids, lethrinids, and inermiids might also be that the Scorpididae and Girellidae may be implicated in the pharyngognath problem. sister groups. Mok and Shen (1983) proposed So far as labyrinth fishes are concerned, that the Kyphosidae (in which they included previous work has dealt with comparisons Scorpididae and Girellidae) are the sister among such family-level groups as lucioceph- group of Toxotidae, and that those two fam- alids, channids (= ophicephalids), helosto- ilies are related successively to the Enoplosi- matids, osphronemids, belontiids, synbran- dae and Monodactylidae. For Mok and Shen chids, pristolepids, nandids, badids, these four families are members of Cuvier's anabantids, and mastacembelids. Travers (1828, in Cuvier and Valenciennes, 1828- (1984) has written extensively on the anat- 1849) Squamipinnes, and are the sister group omy and relationships ofthe last family, but, ofthe remainder ofthis group. However, Ty- in my view, the position of mastacembelids ler et al. (1989) have criticized both the de- is still problematical. scriptive work and the conclusions of Mok In addition to the 22 nominal families dis- and Shen (1983) (see further below on cussed above, I have examined representa- Squamipinnes). tives ofthe Leiognathidae and the two species Concerning the anabantoids, some ofwhich oftPriscacara from the Green River Eocene, might be nonpercoid relatives ofthe Pharyn- the latter genus primarily because it had been gognathi, Liem and Greenwood (1981) used included by Jordan (1923) in his Chromides the concept of pharyngognathy in a purely next to cichlids and pomacentrids [however, descriptive sense to include all cases in which Cavender (1986) referred Priscacara to the powerful pharyngeal jaws are present (e.g., Percichthyidae]. The position of tPriscacara sciaenids); they thus brought labyrinth fishes remains uncertain, but of the two known into the discussion because of the branchial- species, one (tP. serrata) has a well-devel- parasphenoid bite. Not all labyrinth fishes oped lower pharyngeal jaw (LPJ) with mo- have parasphenoid teeth or a single (median) lariform teeth, which is, perhaps, why Jordan lower pharyngeal jaw. But anabantoids form included it. The LPJ and UPJ of the othLer a monophyletic group defined, in part, on the species is not known. dorsal gill arch specializations discussed be- I do not know if some of the above 24 low. Liem and Greenwood (1981) presented higher taxa are monophyletic. In conse- an analysis of teleosts with a bite involving quence, I have tried to examine species be- the parasphenoid dentition, and concluded longing to type genera for the family group that the anabantoid condition is derived, dif- names. By so doing, I hope to minimize fu- fering from the primitive teleost condition ture taxonomic confusion resulting from the and from that in other acanthomorphs with cladistic arrangement of families that later 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 9 prove to be nonmonophyletic. For example, AP adductor arcus palatini process on para- assignment of the Cichlidae based on Cichla, sphenoid Pomacentridae on Pomacentrus, Haemuli- APP anterior part of autopalatine dae on Haemulon, etc. might be less trou- ARPR articular process on UPJ (PB3) BB1-3 (ossified) basibranchials blesome nomenclaturally ifthe family groups BH basihyal are shown to be unnatural. I have also as- BRR branchiostegal rays sumed that previous authors considered cer- Cl-5 ceratobranchials tain family-group taxa because they foresaw DHH dorsal hypohyal the possibility of a relationship among them. El -4 epibranchials Some ofthese hypotheses ofrelationship were FM foramen magnum advanced confidently, others tentatively, but FSB fenestra in wall of otic bulla contacting here they all were treated with equal regard swimbladder during a search for characters amongst the GR modified gill-raker entire AMNH collection of acanthomorph H 1-4 hypobranchials IAB interarcual bone (ossified IAC) skeletons. IAC interarcual cartilage LIG ligament ACKNOWLEDGMENTS MC mental cirrus MDC dorsal crest on maxilla I thank James W. Atz, Bruce Collette, G. MDP dorsal maxillary process David Johnson, Victor G. Springer, and NA nasal James Tyler for assorted advice; Barbara PAD pad of soft connective tissue Brown, Norma Feinberg, and Mary Rauch- PB 1-4 pharyngobranchials enberger for specimen preparation; Jeanne PCH posterior ceratohyal Forster for technical assistance; Lance PHA pharyngeal apophysis on basicranium PMPR combined articular and ascending pre- Grande, Sally Richardson, and Darrell Sie- maxillary processes bert for informative discussion; Humphry PSP parasphenoid Greenwood, Gareth Nelson, and Melanie PTF posttemporal fossa Stiassny for review and helpful comments; TP3,4 dermal toothplate of third or fourth Denise Takahashi, Christine Rossi, and Jan- pharyngobranchial ice Lillien for artwork; and especially my wife TPE2,3 dermal toothplate ofsecond or third epi- Mel for typing. branchial The work was supported by funds from VHH ventral hypohyal NSF grant BSR-8 100103. PHARYNGOGNATH RELATIONSHIPS ABBREVIATIONS For materials examined, see figure cap- LOWER AND UPPER PHARYNGEAL JAWS: tions. Pharyngognathy in its original usage by Mill- ler (1844, 1845) derived from the observa- Institutional tion that the fifth ceratobranchials in some AMNH American Museum of Natural History fishes were enlarged, strongly dentigerous, and ANSP Academy of Natural Sciences, Phila- joined in the midline to form a single trian- delphia gular bone (figs. 1-3) opposed by toothplates BMNH British Museum (Natural History) fused to the pharyngobranchials (figs. 4, 5). UMMZ Museum of Zoology, University of In later studies by Michigan Stiassny (1981, 1982; USNM United States Stiassny and Jensen, 1987), Liem and Green- National Museum wood (198 1), and Kaufman and Liem (1982) Illustrations these two features were referred to, respec- ACC accessory cartilage of unknown signifi- tively, as the lower pharyngealjaws (LPJ) and cance in dorsal gill arches upper pharyngeal jaws (UPJ). These workers ACH anterior ceratohyal noted also that the UPJ bears a dorsal artic- ACL area on LPJ for articulation with clei- ulating process (ARPR) on the third pharyn- thrum gobranchial (PB3) which directly contacts a 10 AMERICAN MUSEUM NOVITATES NO. 2983
A
Adductor 5
Obliquus dorsalis posterior
B
Fig. 1. Lower pharyngeal jaws (united fifth ceratobranchials) of two gerreids; A, Gerres cinereus (Walbaum), AMNH 21732, B, Gerres lineatus (Cuvier and Valenciennes), AMNH 12020, in dorsal view. corresponding basicranial apophysis on the from the basicranium by modified dorsal gill parasphenoid or parasphenoid and basioc- arch muscles (variably developed in poma- cipital (Greenwood, 1978); this condition is centrids: Stiassny and Jensen, 1987: 284), and defined as a diarthrosis. In addition the LPJ has a characteristic median ventral keel that is functionally a single structure that is slung is continuous with a pair of lateral struts or 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES I1I
A
BH
C>A :4 BB1
PCH
Fig. 2. Lower pharyngeal jaws (fused fifth ceratobranchials) and hyoid skeleton in the labrid Hali- choeres bivittatus (Bloch), AMNH 23598. A, B, LPJ in dorsal view, anterior up, and left lateral view; C, ventral hyoid skeleton in dorsolateral view, anterior up. "horns" (figs. 1, 3; Stiassny and Jensen, 1987: G. Nelson (1967), it now is clear that the figs. 1, 3). As a consequence of the studies by characteristic features of the LPJ and UPJ Stiassny and Liem and their co-workers, and mentioned above define a group including 12 AMERICAN MUSEUM NOVITATES NO. 2983
B
4.",'
ACL
F
Fig. 3. Lower pharyngeal jaws (LPJ) in a pomacentrid (A-C) and two haemulids (D-1I). A-C, Po- macentrus planifrons Cuvier, AMNH 23632, A, dorsal view of fused fifth ceratobranchials, anterior up, B, left lateral view of same, C, ventral hyoid skeleton in dorsolateral view, anterior up. D-F, LPJ of Haemulon carbonarium Poey, AMNH 30825, in D, dorsal, E, ventral, and F, left lateral view, anterior up in each. G-I, LPJ of Haemulon album Cuvier and Valenciennes, AMNH 30827, views as in D-F. 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 13
Fig. 4. Left dorsal gill arches ofthe gerreid Eucinostomus gula (Quoy and Gaimard), AMNH 17764. Dorsal view (left) and medial view of PB2-4 (right). Note the platelike flanges along the anterior margin of E 1 and E2, a possible precursor of the anterior extension of these bones in cichlids and pomacentrids (fig. 18), and the typical pharyngognath teeth with concave tips and constricted bases.
Labridae KL, embiotocids, pomacentrids, a suture between them is visible. The LPJ of and cichlids. Other groups of teleosts also other labroids shows no such signs of a bi- have LPJ ofpharyngognath type (e.g., Belon- lateral origin, and suggests, therefore, that de- iformes, which Muller included in his Pha- rived characteristics of LPJ are a synapo- ryngognathi as Pharyngognathi malacopte- morphy uniting embiotocids, pomacentrids, rygii). In general, these lack the muscular sling and Labridae KL, as Stiassny and Jensen and characteristic median ventral keel and (1987) concluded. lateral struts mentioned above, but among In the dorsal gill arches a separate, cartila- beloniforms Stiassny and Jensen (1987) re- ginous fourth pharyngobranchial (PB4) is ab- ported a median keel in exocoetoids, and in sent in all labroids and in gerreids (fig. 4), but hemiramphids and exocoetids a muscular is (primitively) present in some sparoids (fig. sling "simulating the labroid configuration in 6B), all haemuloids (fig. 7A, C), and in ky- remarkable detail" (p. 284). Not all cichlid phosids (fig. 6A). A separate toothplate in the fishes show these ventral LPJ features. Ma- PB4 position is present in all but Labridae chado-Allison (1973) and Stiassny (1982) il- KL. Pharyngeal dentition, when represented lustrated the LPJ of Cichla, showing that a by separate teeth, is distinctive in labroids, single median keel is not present as a simple sparoids, haemuloids, and some squamipin- consequence of the failure of the right and neans. Each tooth is constricted proximally, left ceratobranchials to suture posteriorly, and distally is capped by a conical, decurved where the two bones are not in contact. tip of brown or umber colored enameloid Among adult labroid fishes only cichlids re- (acrodin) (figs. 4-10). This characterizes ju- tain anatomical evidence ofthe paired origin venile teeth, and teeth along the margins of of the LPJ, either because the two halves are the pharyngeal jaws. Medially, and especially incompletelyjoined (as in Cichla), or because in larger individuals, these teeth may be mo- 14 AMERICAN MUSEUM NOVITATES NO. 2983
O°°Oo°--o-o-a+^0 lariform or mammiform distally, retaining A /oa ° aK=F>So °° ° YJ the proximal constriction. Cichlids are unusual among pharyngog- naths in having the oral margin ofthe second //S',-C\a(,epibranchial greatly enlarged, extending for- ward under the first epibranchial, and tipped /\°;--<. with a platelike cartilaginous extension (fig. 5A; Stiassny, 1981). Pomacentrids and em- biotocids also possess an expansion of the z<>3' 4I second epibranchial in the same position, with '-//k~e^
A
B
Fig. 6. Left dorsal gill arches ofa kyphosid (A) and a lethrinid (B). A, Kyphosus sectatrix (L.), AMNH 20855, dorsal view, B, Lethrinus ornatus Cuvier and Valenciennes, AMNH 14916, dorsal (left) and medial views. Both have a primitive interarcual cartilage and a well-developed cartilaginous PB4. Only B shows a well-developed articular surface for the pharyngeal apophysis on the parasphenoid (fig. 14), but both show the pharyngeal teeth characteristic of pharyngognaths: large, decurved teeth, posteriorly concave at the tip and slightly constricted at the base. noidea (Lutjanidae, Caesionidae), Sparoidea betically. J. Nelson (1984: 298-303) refers to (Sparidae, Centracanthidae, Lethrinidae, Johnson in associating the eight families in Nemipteridae), and Haemuloidea (Haemu- a similar though not identical way (reversing lidae, Inermiidae). Johnson was careful to the order ofthe haemuloids and sparoids and point out that he found no satisfactory evi- inserting the gerreids before them). I find in- dence to relate lutjanoids, sparoids, and hae- dications in Johnson's paper that his Spar- muloids (or even the fishes contained within oidea and Haemuloidea may be related to the his Lutjanoidea; see also Johnson, 1984: 491), Pharyngognathi as construed here (including although an implication of such relationship Gerreidae). might be drawn from his classification (pp. The maxilla of pharyngognath fishes (figs. 9-12) in which the three superfamilies are 12-15) has a high posterolateral crest (MDC) listed in the above order, rather than alpha- just below and lateral to a large dorsal process 16 AMERICAN MUSEUM NOVITATES NO. 2983
Fig. 8. Left dorsal gill arches of a labrid (A) and an embiotocid (B). A, Halichoeres bivittatus (Bloch), AMNH 23598, dorsal view, B, Cymato- gaster aggregata Gibbons, AMNH 1973, dorsal (left) and medial views. The articular process for the pharyngeal apophysis on the parasphenoid is well-developed in both and the marginal teeth on PB3 have the shape noted in figure 6. A also shows platelike expansions on the anterior margins ofthe first and second epibranchials, as in cichlids and some pomacentrids (fig. 5). We agree with Stiassny and Jensen (1987: 292) that the anterolateral ex- pansion of the articular head of EB2 in embioto- cids (B) is nonhomologous with the cartilage- capped anterior expansion of EB2 in cichlids.
Fig. 7. Left dorsal gill arches of three hae- muloids, dorsal views. A, the inermiid Emme- These taxa have some relatively primitive traits: lichthyops atlanticus Schultz, USNM 188198, B, poorly developed PB3 articular surfaces, primitive C, the haemulids Anisotremus virginicus (L.), interarcual cartilages and a cartilaginous PB4 in A AMNH 21910, and Pomadasys sp., AMNH 14388. and C. 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 17
A E2,2 C4
ARPR
B El B E1'P~~B2P/
E2 E3 ~ B
Fig. 10. Left dorsal gill arches in a leiognathid (A) and a cichlid (B). A, Leiognathus equulus (For- skAl), AMNH 27027, dorsal (left) and dorsomedial Fig. 9. Left dorsal gill arches ofthe scarid Sca- views, B, Cichlasoma centrarchus (Gill and Brans- rus bowersi (Snyder), AMNH 38114, showing the ford), AMNH 22096, dorsal view. Pharyngogna- maximum development of the articular process thous PB3 tooth form can be seen in A. The ar- on PB3 for contact with pharyngeal apophysis on ticular process on PB3 is particularly obvious in the parasphenoid. A, lateral view, anterior end up, B, but present in a reduced state in A. B, dorsal view. The characteristic pharyngognath PB3 teeth are small and just visible in a reduced form in A. mal ethmoid and roofed by the long, flat nasal bones which are sutured or tightly attached to the frontals and lateral ethmoid. Rognes (MDP); the posterolateral crest is named the (1973: 100) commented that the labrid nasal dorsal wing, with its anterior border called "is a relatively long bone movably connected the shank ridge, in cichlids (Barel et al., 1976). at its posterior end with the anterior edge of Between the crest and the dorsal process is a the frontal by strong connective tissue and at deep, oblique groove in which the anterior its anterior end to the dorsal process of the process of the autopalatine (APP) fits and is maxillary by a ligament, the nasal-maxillary attached by connective tissue. The ascending ligament." In some cichlids (e.g., Stiassny and processes of the premaxillae slide in and out Jensen, 1987: fig. 18C), pomacentrids (e.g., of a rostral fossa (Rognes, 1973: figs. 6-10, Emery, 1980: fig. 12), and embiotocids (e.g., "ascending process fossa") floored by the der- fig. 12B; Morris, 1982: fig. 29), the maxillary 18 AMERICAN MUSEUM NOVITATES NO. 2983
A figs. 7-12) the maxillary crest and palatine groove are both well developed except in the haemuloids Xenocysjessiae and Inermia vit- tata (Johnson's figs. 16, 17). The dorsal pro- cesses ofthe maxillae in gerreids form an arch over the ascending premaxillary processes when the maxilla is rotated outward during opening of the mouth (Schaeffer and Rosen, 1961: fig. 7, although that illustration of the maxilla fails to show the posterolateral crest). The more primitive percomorph maxilla has a straight dorsal margin and the posterolater- al crest is represented by a knoblike process lateral to a dorsal process which clasps the articular process on the premaxilla. That primitive maxillary anatomy occurs in ser- ranids (Morone; Stiassny, 1986: fig. 10), the centrarchid Micropterus, and in the lutjanids, caesionids, and inermiids illustrated by Johnson (1980: figs. 2-6, 17, 18). The maxillary anatomy described above is also very well developed in the squamipin- neans Chaetodon xanthurus (fig. 14E), Pomacanthus paru, AMNH 38130, Scato- phagus sp., AMNH 20329, and among ephippidids, in Drepane punctata (fig. 14C), but not in Chaetodipterus or Platax which have a simple, primitive maxilla, as in Mon- odactylus and lutjanids. The derived maxil- lary structures appear to be present in some zeoids and plectognaths (Rosen, 1984: figs. 27-29; Antigonia, triacanthoids), but not in Capros or acanthuroids. Tyler et al. (1989) concluded that acanthuroids (including si- ganids, luvarids, zanclids, and acanthurids) are most related to Fig. 11. Left dorsal gill arches in an odacid (A) closely successively sca- and a labrid (B), dorsal views. A, Olisthops cy- tophagids and ephippidids. anomelas Richardson, AMNH 31319, B, Coris Another feature of the upper jaw is also of gaimard (Quoy and Gaimard), AMNH 38142. The some interest. In Labridae KL (fig. 13B, C) articular surfaces for the pharyngeal apophysis on the articular process is coalesced with the as- the parasphenoid are very well developed. B shows cending process of the premaxilla (Rognes, an anterior expansion of E2, as in cichlids and 1973); the long, slender articulo-ascending pomacentrids (fig. 5). process is able to slide in and out under the intermaxillary ligament within the rostral crest and dorsal process (and therefore, the fossa and fills the fossa when the mouth is intervening groove for the autopalatine) are closed. The alveolar process ofthe premaxilla relatively small, but still differ in these fea- is only about half the length of the articulo- tures from serranids, centropomids, and cen- ascending process. Coalescence and elonga- trarchids, among more generalized percoids. tion of the articular and ascending processes In the gerreids (fig. 13D) and the taxa in- is also present in gerreids (fig. 1 3D), and many cluded by Johnson (1980) in the Haemuloi- cichlids (fig. 1 2D), haemuloids (fig. 14), and dea (fig. 14A, B, D; Johnson, 1980: figs. 13- sparoids (fig. 15) but not in pomacentrids (fig. 18) and Sparoidea (fig. 15; Johnson, 1980: 1 3A) or embiotocids (fig. 1 2A-C). These two 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 19
A
MDC - APP
D
Fig. 12. Buccal jaws of embiotocids (A-C) and a cichlid (D) showing the long nasals sutured to frontals covering a rostral fossa for the premaxillary ascending processes. A, Cymatogaster aggregata Gibbons, AMNH 2739; B, Hyperprosopon argenteum Gibbons, AMNH 27947; C, Rhachochilus sp., AMNH 27943; D, Cichlasomafriedrichsthali Steindachner, AMNH 27822. features contrast with the more primitive ter also have ascending and alveolar pre- conditions in serranids and lutjanids (John- maxillary segments that are subequal in length son, 1980), among other percoids, in which [except Drepane (fig. 1 4C) which has a rather the relatively short ascending process is sep- long ascending process]. arated from the articular process by a deep Both pomacentrids (fig. 13A) and embi- notch. In this primitive configuration the ar- otocids (fig. 1 2A-C) have relatively short pre- ticular process also is set offfrom a relatively maxillary ascending processes. Among po- long alveolar process (twice or more the length macentrids this process is most elongate and ofthe ascending process) which bears a broad, most closely coalesced with the articular pro- triangular, postmaxillary process on its dor- cess in Chromis (Emery, 1973: fig. 38). In sal margin. Various sparoids and haemuloids embiotocids (Morris, 1982) the ascending are figured by Johnson (1980) as having a process is short and separated from the ar- thumblike postmaxillary process on a short ticular process by a notch in Amphistichus alveolar arm, a feature not seen in any ofthe and Hyperprosopon, but is about equal in squamipinnean material examined. The lat- length to the alveolar process and coalesced 20 AMERICAN MUSEUM NOVITATES NO. 2983
B
C
Fig. 13. Buccal jaws of a pomacentrid (A), two labrids (B, C), and a gerreid (D). A, the pomacentrid Hypsipops rubicundus (Girard), AMNH 42156; B, Labrus bimaculatus L., from Rognes (1973, fig. 80); C, the labrid Bodianus rufus (L.), AMNH 30875; D, Gerres cinereus (Walbaum), AMNH 21732.
with the articular process in other genera. ryngeal apophysis, a labridlike maxilla, but a Among cichlids, premaxillae specialized as primitive premaxilla. discussed above are characteristic of certain PHARYNGOGNATH BRAINCASE: Features of taxa with lengthy protrusile mouths (e.g., Pe- the neurocranium, as well as those ofthejaws tenia splendida). But it appears that neither and palate discussed elsewhere, suggest a re- cichlids, pomacentrids, nor embiotocids lationship among gerreids, Labridae KL, primitively have the advanced maxillary or sparoids, and haemuloids. premaxillary features of labrids. Upper jaw Derived features ofthe parasphenoid bone evidence aligns labrids with gerreids, spar- appear to be alternative expressions of some oids, and haemuloids. And each of the last specializations of the dorsal gill arches dis- three groups shows some development of a cussed above. The parasphenoid of labroids pharyngeal apophysis on the parasphenoid shows three features that do not occur in more (see below) and an articulatory facet on PB3. generalized acanthomorphs such as Holocen- Gerreids in addition have a labroidlike ba- trus (fig. 16), Centropomus, Morone, Epi- sibranchial series (below, and fig. 27). Lastly, nephelus, or Micropterus. These are: certain squamipinnean taxa have a weak pha- 1. A strong ventral keel from the sides of 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 21
D E
Fig. 14. Buccal jaws of three haemuloids (A, B, D, Haemulidae) and two Squamipinnes (C, Dre- panidae, E, Chaetodontidae). A, Haemulon carbonarium Poey, AMNH 30825; B, H. plumieri (Lacepede), AMNH 21747; C, Drepane punctata (L.), AMNH 13922; D, Anisotremus sp., AMNH 21569; E, Chae- todon xanthurus Bleeker, AMNH 38110. which an adductor arcus palatini muscle aris- the posterior fibers of the adductor muscle es that fills the floor of the orbit. The more arise. The adductor process takes various primitive condition appears to be the con- forms (pointed, squared off, or rounded) and finement ofthis muscle to the rear ofthe orbit is variable in size. It is quite long and squared with a parasphenoidal attachment of rugose off in labrids (fig. 17; Rognes, 1973: figs. 11- bone in the form ofa narrow ellipse. The shaft 15), or lower and rounded in pomacentrids; of the parasphenoid anterior to the site of it is least developed in the embiotocids (Mor- muscle attachment is primitively smooth and ris, 1982: figs. 1-4) and cichlids, where often convex ventrally. Morris (1982: 143) noted the adductor process is all that remains ofthe that a parasphenoid keel equal in size to that ventral keel (Stiassny and Jensen, 1987: fig. ofmany embiotocids occurs in scorpidids and 6). girellids, and that leiognathids also have such 3. A pharyngeal apophysis (PHA) more a keel. posteriorly on the otico-occipital region of 2. A ventral extension of the median keel the parasphenoid. PHA consists either of an in the form ofa triangular process posteriorly elevated transverse shelf or of a raised pair just in front of the prootic to form an ad- ofoval knobs oriented anteroposteriorly. The ductor process (AP, figs. 17-26) from which articular surface of PHA has a basioccipital 22 AMERICAN MUSEUM NOVITATES NO. 2983
A B
MDC
C
Fig. 15. Buccal jaws of sparoids (A-C, Sparidae; D, Nemipteridae) showing the long nasals sutured to frontals covering a rostral fossa that houses the premaxillary processes. A, Pagrus pagrus (L.), AMNH 56321; B, Calamus calamus (Cuvier and Valenciennes), AMNH 35480; C, Lagodon rhomboides (L.), AMNH 55947; D, Nemipterus virgatus (Cuvier and Valenciennes), AMNH 56390.
component in some cichlids (fig. 18; Green- wood, 1978), but is parasphenoidal in po- macentrids, Labridae KL, embiotocids (Stiassny and Jensen, 1987: 282, were mis- taken in mentioning a basioccipital contri- bution to the apophysial surface in embioto- >,OKA'cids),PT and gerreids (figs. 19-21). PHA
Fig. 16. Posterior view ofthe braincase ofHol- ocentrus rufus (Walbaum), AMNH 35496, to il- lustrate primitive absence of basicranial apophys- es associated with dorsal or ventral gill arch specializations. Compare with labrid, sparoid, ger- reid, cichlid, lethrinid, and haemuloid neurocrania in figures 17-26. 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 23
Fig. 17. Pharyngognath basicranial features. The braincase ofthe labrid Bodianus rufus (L.), AMNH 35434, in A, posterior, B, lateral, and C, ventral view.
articulates indirectly (in gerreids) or directly tween PHA and PB3 are discussed at length (in labroids) with an articular facet of the and figured by Stiassny (198 1, 1 982; Stiassny third pharyngobranchial. The relations be- and Jensen, 1987). 24 AMERICAN MUSEUM NOVITATES NO. 2983 A
Fig. 18. Pharyngognath basicranial features in cichlids. A-C, Braincase ofPetenia splendida Gunther, AMNH 27833, in A, posterior, B, lateral, and C, ventral view. D, Braincase of Cichla ocellaris Bloch and Schneider, AMNH 40023, in posterior view. 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 25
Fig. 19. Pharyngognath basicranial features in the gerreid Gerres lineatus (Cuvier and Valenciennes), AMNH 21949. Braincase in A, posterior, B, lateral, and C, ventral view. 26 AMERICAN MUSEUM NOVITATES NO. 2983
A
Fig. 20. As figure 19, but Gerres cinereus (Walbaum), AMNH 21732.
Basicranial pharyngeal apophyses occur in 1989: 49). As a character, the pharyngeal a variety of teleosts: beloniforms, some cen- apophysis appears ambiguous. Its relation to trarchids, gerreids (figs. 19-21), sciaenids, PB3 in labroids has been described as a di- sparids (figs. 23-25), lethrinids (fig. 22), and arthrosis (direct bone-to-bone contact), as in some of the Squamipinnes (Tyler et al., opposed to an amphiarthrosis (or mixed ar- 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 27
PHA
C
Fig. 21. As figure 19, but the gerreid Diapterus mexicanus Steindachner, AMNH 28060. ticulation in which dorsal gill-arch muscle haemuloids, sparoids, and gerreids. The lat- and connective tissue intervene between PB3 ter three taxa include species with a definite and the bony pharyngeal apophysis), as in platform (? articular facet; ARPR, figs. 4, 6B) 28 AMERICAN MUSEUM NOVITATES NO. 2983
B
NA Fig. 22. Braincase of a lethrinid, Lethrinus sp., AMNH 30872, in A, lateral, and B, ventral view, to show the pharyngeal apophysis and the long nasal sutured to the frontal to cover the rostral fossa. dorsally on PB3, and they and the labroids and roofing the fossa for the premaxillary possess very similar pharyngobranchial den- ascending processes (figs. 12-15, 22). tition with decurved acrodin tips and a con- 3. Maxilla with a dorsal crest and process stricted base (see above, and figs. 5-10). forming a groove for the autopalatine (figs. AN EXPANDED PHARYNGOGNATHI? Char- 12-15). acters reviewed above that might be used to One character of the Labroidei is ventral align sparoids, haemuloids, and the Pharyn- displacement of the first basibranchial (BB 1) gognathi as understood by Gunther 1862 as described by G. Nelson (1967) and Stiass- [1859-1870] and Gill (1872) [Labroidei of ny and Jensen (1987). Young embiotocids Stiassny and Jensen (1987) plus Gerreidae] (e.g., Cymatogaster aggregatus, AMNH 1973, include: standard length 33 mm; and embryos ofAm- 1. Two parasphenoidal processes; an ante- phistichus argenteus, Morris and Gaudin, rior one to which the adductor arcus pal- 1982: fig. 8) do not show this feature. BB I in atini muscle attaches and a posterior these small individuals is in line with the apophysis to which attaches an articular basihyal and BB2. Similarly this feature is process of PB3 ofthe UPJ (AP, PHA, figs. best seen in cichlids of larger sizes. Such ob- 17-26). servations suggest that the ventral position 2. Long nasals firmly united with the frontals of BB1 results from an ontogenetic process 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 29
A
AP
(I N>sPHA
Fig. 23. Pharyngognath basicranial features in the sparoid Lagodon rhomboides (L.), AMNH 55947. Braincase in A, posterior, B, lateral, and C, ventral view. 30 AMERICAN MUSEUM NOVITATES NO. 2983
A
PH,
AP'
Fig. 24. As figure 23, but the sparoid Calamus proridens Jordan and Gilbert, AMNH 21673. 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 31
A
AP'
Fig. 25. As figure 23, but the sparoid Calamus bajonado (Schneider), AMNH 35442. 32 AMERICAN MUSEUM NOVITATES NO. 2983
A
Fig. 26. Pharyngognath basicranial features in the haemuloid Haemulon carbonarium Poey, AMNH 30839. Braincase in A, posterior, B, lateral, and C, ventral view. 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 33
B
BB2
D
Fig. 27. Basibranchial series of young and adult gerreids in left lateral view. A, Gerres cinereus (Walbaum), UMMZ 172787, ca. 35 mm standard length (SL); B, G. abbreviatus Bleeker, BMNH 1974.5.25.2336, ca. 65 mm SL; C, G. cinereus, BMNH unreg. dry skeleton, ca. 150 mm SL; D, G. cinereus, AMNH 21949, ca. 200 mm SL. Scale bar at lower right of each = 1 mm. in which backward growth of the basihyal ulation of the basihyal (BH) with BB2. In excludes BB1 from the plane of the basi- Stiassny and Jensen's (1987: fig. 1OE) illus- branchial series by forcing it downward, or tration of Embiotoca lateralis there is no di- that BB 1 pulls itselfinto ventral disalignment rect articulation between BH and BB2, but by growth of a posteroventral extension be- BB1 is almost entirely below BH. The po- low BB2. macentrid illustrated by Nelson (1967: fig. 1, The first basibranchial of young gerreids Chromis ovalis), is intermediate, with BB1 (e.g., Gerres cinereus, G. abbreviatus, fig. 27A, anteriorly below the proximal end ofBH, but B) is in line with the basibranchial series, but articulating posteriorly with BB2 and having is very short and deep, with a broad postero- a long narrow process that extends back be- ventral process or flange. In adult Gerres (fig. low BB2 as far as the anterior end of BB3. 27D) the posteroventral process extends be- Stiassny and Jensen's (1987: fig. lOC) illus- low BB2 much as Nelson (1967: fig. 3B) il- tration of Pomacentrus shows virtually the lustrated for a labrid. In the embiotocids il- same condition. This pomacentrid confor- lustrated by Nelson (1967: fig. 2, Embiotoca mation is almost identical to the relations jacksoni) and Morris (1982: fig. 38A, Hys- between BB 1, BB2, and BH in some large terocarpus traski) BB1 is entirely excluded cichlids such as Petenia splendida, AMNH from the basibranchial series by direct artic- 27845 (cf. Stiassny and Jensen, 1987: fig. lOB, 34 AMERICAN MUSEUM NOVITATES NO. 2983
C). From all of this I conclude that the ba- epibranchial of the sparoids and haemuloids sibranchial configuration described by Nel- (fig. 7) studied is not modified in this way. son and by Stiassny and Jensen properly de- Embiotocids have the oral border ofE2 with- fines some group of pharyngognathous out a keellike expansion, although there is an "percoids," including pomacentrids, cich- enlarged uncinate process distally that merges lids, embiotocids, gerreids, labrids, and with the oral border (fig. 8B; Stiassny and possibly also kyphosids, since Stiassny and Jensen, 1987: fig. 14B). In the labrid Coris Jensen (1987: 288) reported the same config- (fig. 1 iB) and the scarid Scarus (fig. 9), the uration in Girella, and in Kyphosus BB1 is oral border of E2 is expanded respectively in almost completely overgrown by BH (which the manner of pomacentrids and cichlids, bears paired toothplates), and so lies beneath making embiotocids odd-man-out in this the BH-BB axis. Among labroids, embioto- comparison. cids are linked to Labridae KL by the reduc- Thus a combination of dorsal and ventral tion and specialization of PB2 as a small, gill arch features corroborates a group La- edentulous cylinder bent along its middle so broidei, including cichlids, pomacentrids, that the concave surface faces PB3. I note embiotocids, and Labridae KL, and within that PB2 in gerreids (fig. 4) shows the same that group features of the dorsal gill arches curvature as in labrids and embiotocids (fig. corroborate a subgroup including embioto- 8), but is toothed, as it is in cichlids and cids, pomacentrids, and Labridae KL. Two pomacentrids. features of the ventral gill arches suggest a Other "percoid" families show some ofthe relationship between gerreids and labroids. gill-arch features found in all or some la- Each of the most recent studies of phar- broids, as summarized in the following list: yngognath interrelationships begins with a premise that there exists a natural group of families of which the Labridae KL are the 1. absence of PB2 dentition, most derived. Each has added supporting 2. first basibranchial ventral to axis of ba- evidence of the group's naturalness. But the sibranchial series, evidence offered pertains entirely to the pha- 3. posterior orientation of PB2 and anterior ryngobranchial apparatus. Since the time of epibranchials, Muller (1844), one feature has been judged 4. a functionally median LPJ, to be crucial; the fused fifth ceratobranchials 5. teeth constricted proximally and capped that form the lower pharyngeal jaw (LPJ). by conical, decurved tip of acrodin, Later studies by Liem and Greenwood (1981), 6. PB3 with a dorsal diarthrodial articula- Kaufman and Liem (1982), Stiassny (1981, tion with a parasphenoidal apophysis, 1982), Liem (1986), Liem and Sanderson 7. No PB4, (1986), and especially Stiassny and Jensen 8. No IAC, (1987) have emphasized additional features: 9. Oral margin of first and/or second epi- diarthrosis between PB3 and a ventrally pro- branchial expanded. jecting, rounded parasphenoid apophysis; a muscular sling, formed from the fourth In gerreids (fig. 1) the LPJ consists of two external levator and obliquus posterior mus- separate elements, but, as in cichlids and hae- cles, that suspends the LPJ from the neuro- mulids (fig. 3D-I), the right and left parts are cranium; presence of a transversus epibran- so closely united that, functionally, they con- chialis division 2 of the anterior transversus stitute a median structure. The similarity to dorsalis muscle; an undivided sphincter oes- the derived LPJ of labroids is heightened by ophagi muscle; and a first basibranchial lying the modification in some taxa of the inner partially below the longitudinal axis of the teeth (figs. 1, 3) into a pavement, molariform basihyal-basibranchial series. dentition. The work ofTarp (1952) and Morris (1982) Gerreids also show a slight expansion of appears unique in proposing that some phar- the oral margin of the second epibranchial yngognaths (embiotocids and pomacentrids) (E2) (fig. 4), but less than that in the poma- are more closely related to some "percoids" centrids examined (fig. SB, C). The second (kyphosids, here including scorpidids and gi- 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 35
rellids) than they are to other pharyngog- underlying V-shaped foundation is still clear- naths. In fact, although other authors have ly visible (e.g., in Petenia splendida, AMNH refrained from making specific taxonomic 28054). The fundamental feature of this proposals, the kyphosids have been men- apophysis, therefore, appears to be that tioned repeatedly in this connection (e.g., V-shaped wedge which can be formed en- Nelson, 1967; Stiassny, 1981; Liem and tirely from the parasphenoid or from a com- Greenwood, 1981; Kaufman and Liem, bination of the parasphenoid and basioccip- 1982). Also mentioned as possibilities for ital. study are sciaenids, gerreids, and anaban- A V-shaped pharyngeal apophysis that toids. But no other "percoid" family has been merges anteriorly with a hyomandibular pro- considered seriously for inclusion in a group cess on a midventral parasphenoidal ridge that contains labroids. For Nelson (1967) the occurs in gerreids (figs. 19-21), as noted kyphosid Girella was simply the outgroup for above, as well as in sparoids (figs. 23-25), deciding the polarity oflabroid characters, as kyphosids, sciaenids, some centrarchids, and were the anabantoids and other mentioned haemuloids. It is associated with hypertro- "percoids" for Liem and Greenwood (1981) phied LPJs, as noted in Kaufman and Liem in deciding the polarity of cichlid and em- (1982), and these authors write of "articu- biotocid characters, and as were the above- lations between the [UPJ] and the basicra- mentioned "percoids" for Stiassny (1981) in nium" in gerreids and certain sciaenids, but making decisions about cichlid, pomacen- Stiassny (1981; Stiassny and Jensen, 1987) trid, and embiotocid character polarity. did not regard these as diarthrodial because Hence, in proposing that pomacentrids and of intervening muscle and connective tissues embiotocids are sister taxa whose nearest rel- (? a plesiomorphous state of the UPJ-basi- atives lie with certain nonpharyngognath cranial character). Kaufman and Liem did "percoids," Morris' (1982) and Tarp's (1952) not exclude the possibility of a relationship theories stand apart, since they amount to between some of those taxa and labrids, but proposals that the Labroidei ofKaufman and interpreted all nonlabroid conditions as more Liem are nonmonophyletic. primitive. Stiassny (1981) also allowed for Ontogenetic data concerning the pharyn- the possibility ofsome linkage among all these geal apophysis ofthe parasphenoid show that taxa. the now well-described labroid condition be- The problem ofthe gill arch data -the un- gins as a V-shaped wedge with the point of certainty ofits significance as evidence ofsys- the V continuous with an elevated part ofthe tematic relationship- may be resolved, in median parasphenoidal ridge from which the part, by (1) treating some of the features as posterior or hyomandibular section ofthe ad- parts ofa complex, multistate transformation ductor arcus palatini originates. The arms of series, and (2) being aware of the evidence the V become increasingly thickened by con- that some pharyngognath features can be in- tinuing ossification, and eventually are duced ontogenetically by diet (i.e., are eco- bridged near their tips to form a transverse phenotypic). Evidence for ecophenotypy was arthrodial surface heavier at the ends than discussed by Greenwood (1964) for the hy- the middle. This is the pharyngeal apophysis pertrophy of the pharyngeal mill of cichlids, referred to by Liem, Stiassny and others; it whereas Ismail et al. (1982) discussed reasons extends downward and backward away from for believing that the formation of a pharyn- the parasphenoid on a bony column or neck geal apophysis on the parasphenoid, or at that may include bone from the basioccipital least, its size, is influenced by "epigenetic fac- (e.g., in certain cichlids). Even in some ofthe tors, due to movements of the UPJ's against fishes considered previously to be labroids the skull floor." In support oftheir argument, (sensu lato), the pharyngeal apophysis retains Ismail et al. cited the developmental studies its original V-shaped structure as a result of of Bassett (1964, 1971) and Hall (1975). failure of the transverse bridging bone to de- Returning to the first, or transformational, velop fully, as in the pomacentrid Hypsipops resolution of the comparative anatomical rubicunda, AMNH 42156. And even in some problem, I envisage the following sequence large cichlids with the transverse bridge, the ofcharacter states, from primitive to derived: 36 AMERICAN MUSEUM NOVITATES NO. 2983
1. hypertrophy of the pharyngeal jaws, 2. development of a V-shaped pharyngeal A apophysis on the parasphenoid, 3. development ofa transverse bony bridge over the arms of the V-shaped apophy- SiS, 4. reduction of connective tissue and mus- cle over the coalesced toothpatches of the third and fourth pharyngobranchials and the articulation of PB3 with the pharyngeal apophysis on the basicrani- um, 5. suspension of LPJ from modified fourth external levator and posterior levator muscles, 6. direct bony contact between PB3 and the pharyngeal apophysis, 7. reduction and posterior reorientation of PB2, 8. loss of teeth on PB2, B 9. coalescence of the right and left LPJs, 10. development of an articulation between the coalesced LPJs and a notch in the cleithrum. Item (10) was used by Kaufman and Liem (1982) to define Labridae KL, but Stiassny and Jensen (1987) reported that the phar- yngocleithral joint appears to be developed there only in scarids and odacids, whereas there is a very similar joint in certain po- macentrids. Item (9) characterizes embioto- cids, pomacentrids, and Labridae KL. Items (7) and (8) characterize embiotocids and La- Fig. 28. Left dorsal gill arches in two labyrinth bridae KL. Items (5) and (6) characterize fishes, an anabantid (A) and a belontiid (B). A, cichlids, pomacentrids, embiotocids, and La- Ctenopoma acutirostrum Pellegrin, AMNH 42125, bridae KL. Those four and dorsal view, showing the highly derived El asso- groups gerreids ciated with a suprabranchial organ for aerial res- are characterized by (4), and those five and piration; B, Betta splendens Regan, AMNH 45082, sparoids and most haemuloids (sensu John- dorsal view illustrating the highly modified El with son, 1980) by (1-3). (right) an anterodorsal view showing the relation A number of squamipinnean taxa have a between PB2 and IAC. well-developed pharyngeal apophysis of a shape, position, and orientation similar to that in the pomacentrid Hypsipops, as well their general discussion of pharyngognathy as an upper jaw anatomy very like that in because of the powerful LPJ and its bite many pharyngognaths. I do not regard ana- against a tooth-bearing basicranial apoph- bantoids as especially relevant to this general ysis; these were described and illustrated in taxonomic problem, because oftheir distinc- detail by Liem (1963). Liem and Greenwood tive gill arch anatomy. identified the labyrinth pharyngeal jaws as functionally different from those of labrids LABYRINTH FISHES and other fishes traditionally combined with them in a Pharyngognathi (cichlids, poma- For functional reasons Liem and Green- centrids and embiotocids). wood (1981) included the labyrinth fishes in The term Labyrinthici seems to have orig- 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 37
Fig. 29. Left dorsal gill arches of a belontiid, the dwarf gourami Colisa lalia (Hamilton Buchanan), AMNH 221 10, to show the modified El (cf. fig. 28), dorsal view, anterior to left.
inated with Cuvier in vols. 1 (1828) and 7 ers referring mastacembelids to the Synbran- (1831) ofCuvier and Valenciennes (1828-49) chiformes (an arrangement originated by as "acanthopterygiens a branchies labyrinth- McAllister, 1968). Gosline (1968, 1971) ar- iques" or "poissons a pharyngiens labyrinthi- gued for relationship between the anaban- formes." Muller (1845) placed Cuvier's toids (in which he included Channa and Lu- Labyrinthici (p. 532) or Labyrinthiformes ciocephalus) and the Badidae, Nandidae, and (p. 537) as a family within his order Acan- Pristolepidae. He did not transfer these three thopteri. Gunther (1861) [1859-1879] also "percoid" families to the Anabantoidei, but treated Labyrinthici as a family, and its con- wrote that "they perhaps should be" (1971: tents are equivalent to the Anabantoidei of 162); he also hinted (p. 161) that synbran- most current usage (e.g., Liem, 1963; J. Nel- chids might be involved. Barlow et al. (1968) son, 1984: Anabantidae, Belontiidae, Helo- removed Badis from the Nandidae as a new stomatidae, Osphronemidae). Gunther placed family Badidae, and argued (in effect) that it his Labyrinthici between the nandids and Lu- is the sister group ofthe anabantoids. G. Nel- ciocephalus. Gill (1872) used the Labyrin- son (1969) found some evidence for relation- thici as a subdivision of his Acanthopteri, ship between mastacembelids and a nandid, and listed three families (Helostomidae, An- pristolepid, channid, and anabantoid assem- abantidae, Osphronemidae), with an appar- blage. ent disclaimer in adopting the group only On the question of relationship between provisionally as one offive of "very dubious badids, nandids, pristolepids, and labyrinth value ... kept in prominence to attract future fishes, Liem and Greenwood (1981; also Lau- examination" (Gill, 1872: xliv). Some sub- der and Liem, 1983) found further evidence sequent history of ideas on the group, and that Badis is the sister group of anabantoids, opinions on relations with Luciocephalidae but concluded that their data from functional and Channidae, is in Liem (1963), McAllister morphology gave no support to relationship (1968), Lauder and Liem (1983), and J. Nel- between nandids, Pristolepis, and Channa, or son (1984). Berg (1940) suggested a relation- between any of them and anabantoids. One ship between Channidae (Ophicephalidae) character relevant to this (fig. 30A, B) is re- and Synbranchidae, an idea that was given ported below. substance by Lauder and Liem (1983). Gos- Among the eight families that have been line (1983) and Travers (1984) proposed a associated with the Labyrinthici [Anabanti- sister-group relationship between the Mas- dae, Belontiidae (= Polyacanthidae), Helo- tacembelidae and the Synbranchidae, Trav- stomatidae, Osphronemidae, Channidae (= 38 AMERICAN MUSEUM NOVITATES NO. 2983
A Ophicephalidae), Luciocephalidae, Synbran- chidae, Mastacembelidae], all but the last two are characterized by gill-arch specializations associated with aerial respiration, discussed below. Parasphenoid teeth are present in some (Anabantidae, Belontiidae, Helostomatidae, Osphronemidae, Ophicephalidae); such teeth are not known among synbranchids, lucio- cephalids [except for Regan's (1909: 768) re- port], or mastacembelids, but occur in ba- dids, nandids, and pristolepids. LABYRINTH GILL ARCHES (figs. 28-32): A derived feature that has been taken to char- acterize labyrinths is a suprabranchial organ associated with a modified first epibranchial (El). This modified El (figs. 28-31) is present in all five anabantoid families (Anabantidae, Belontiidae, Helostomatidae, Osphronemi- dae, Luciocephalidae) and in channids, but not in synbranchids (fig. 31 B) or mastacem- belids (fig. 32). The complex El and supra- branchial organ are related to adventitious or obligate aerial respiration. The aerial respi- ration ascribed to mastacembelids, which was one justification for aligning them with syn- branchids (McAllister, 1968), is actually in- ferred from their habit of aestivating, but mastacembelids are not known to have any modification ofthe dorsal gill arches (Fig. 32) or ofanything else (Sufi, 1956) in connection with it. Mastacembelids have neither of the anatomical features (parasphenoid teeth or a modified first epibranchial) characteristic of other labyrinth fishes, and aerial respiration is widespread among teleosts (more than 300 species: Hughes, 1976; J. W. Atz, personal commun.).
Fig. 30. Left dorsal gill arches ofa channid (A) and two nandids (B, C), dorsal views. A, Channa ocellata (Lacepede), AMNH 10652, B, Monocir- rhus polyacanthus Heckel, AMNH 22648, C, C Polycentropsis abbreviata Boulenger, AMNH 43454, with (at top left) a posterior view of El showing the absence of an uncinate process as- sociated with IAC, and a platelike ventral expan- sion. In A and B, note the large, strongly offset comblike gill-raker (GR) associated with the distal end of El, possibly associated with the develop- ment of a suprabranchial organ and a possible synapomorphy between these two taxa. Monocir- rhus (B) appears to have a primitive (small) IAC, comparable with that in Kurtus (fig. 46A), there- fore a homoplasy as compared with the long IAC in such nandids as Polycentropsis (C). 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 39
A PB2
B
Fig. 31. Comparison ofleft dorsal gill arches in A, the channid Channa arga (Cantor), AMNH 10648, and B, the primitive synbranchid Ophisternon aenigmaticum Rosen and Greenwood, BMNH 1846.2.16- 129, dorsal views, anterior up. Note that they share a first epibranchial that is displaced away from the midline.
Gosline's (1968, 1971) and J. Nelson's might be present in polycentrine nandids. (1969) alignment of the Nandidae with the That proposal is inconsistent with Lauder and labyrinths is supported by the common pos- Liem's (1983) hypothesis of a sister-group session in the Polycentrini and Channa of a relationship between Channa and synbran- large, offset, comblike gill raker associated chids, in which the dorsal gill arches (fig. 30) with the first epibranchial (fig. 30A, B; Lau- share little beyond a first epibranchial that is der and Liem, 1983: fig. 59, Channa), sug- displaced from the midline. Lauder and gesting that a suprabranchial organ of sorts Liem's hypothesis is also inconsistent with 40 AMERICAN MUSEUM NOVITATES NO. 2983
A
B
Fig. 32. Left dorsal gill arches of Rhynchob- della sinensis Bleeker (= Mastacembelus sinensis), AMNH 11078, in dorsal view. Note unmodified El and see text for discussion of aerial respiration in mastacembelids. The gill arches yield no ana- tomic justification for including mastacembelids with the labyrinth fishes.
Gosline's (1983) and Travers' (1984) argu- ments for a sister-group relationship between synbranchids and mastacembeloids. COMMENT ON BLENNIOIDS Lauder and Liem (1983), McAllister (1968), and Gosline (1968) were the last investigators that I am aware of to comment broadly on the general classification ofpercomorph fish- Fig. 33. Left dorsal gill arches of a clinid blen- es and of blennioids in particular. Gosline nioid (A, B) and a callionymid (C). A, B, Gibbonsia elegans (Cooper), USNM 200386, in (A) antero- (1968) followed Regan (1929) and recognized dorsal and (B) posterodorsal view; the different a Percoidei (minus the families Gadopsi- views obscure the axhead shape of PB3. C, Syn- dae, Nototheniidae, Callionymidae, Draco- chiropus ? ocellatus (Pallas), USNM aquarium nettidae, and most trachinoids). McAllister specimen, dorsal view. (1968) recognized the trachinoids as a sub- order distinct from the blennioids. Like Gos- line, Lauder and Liem (1983: 172) included "distinguished by the firm attachment of the some trachinoid families along with the blen- fin rays of the pectoral fin to hypertrophied nioids, implying that what defines the latter pterygiophores" ("hypertrophied" is merely applies also to the former. Lauder and Liem "well-developed" in Bertin and Arambourg's cited only Bertin and Arambourg (1958) on text). the defining characters of blennioids: fishes On the subject of the Blennioidei, I have 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 41
A =
B 3
E2 -TP3 B E3 E4
Fig. 35. Left dorsal gill arches of gobiesocids, dorsal view. A, Pherallodiscus funebris (Gilbert), AMNH 5557; B, Gobiesox meandricus (Girard), AMNH 18282; C, Tomicodon fasciatus (Peters), AMNH 33191. E3 and E4 are directly apposed at their midpoints, and apparently ankylosed. PB3 appears to be represented by the dermal toothplate only, with no endoskeleton. received much instruction from Victor Springer, who pointed out that blennioids fall into two groups, and gill-arch anatomy is rel- evant to that distinction. The tropical or "true" blennies lack all but the third phar- yngobranchial (PB3), which has the shape of the head of an ax or halberd when viewed Fig. 34. Left dorsal gill arches ofcottoids, dor- from above (figs. 33A, B, 37). This configu- sal view. A, Cottidae, Myoxocephalus aenaeus (Mitchill), AMNH 20787; B, Cottidae, Oligocottus ration of PB3 occurs in gobiesocids (fig. 35), maculosus Girard, USNM 188372, with (below) which also lack PB 1, 2 and 4 (as do some sketch of ventral gill-arch structure; C, Cyclopter- cottoids, figs. 34A, C, 47C). It was for these idae, Liparis coheni Able, USNM 213822, with and other reasons that Springer (personal (below) sketch ofventral gill-arch structure -there commun.) objected to the alignment of go- are no ossified basibranchials. biesocids with gadiforms and lophiiforms in ,42 AMERICAN MUSEUM NOVITATES NO. 2983
A
- ' cr H
C5 C4 H 05
Fig.36. Left dorsal gill arches ofuranoscopids, dorsal view. A,Astroscopusygraecum (Cuvier), USNM 185669, with (right) diagram of ventral gill-arch elements converging on an unossified copula; B, Kath- etostoma albigutta (Bean), USNM 185666, with (below right) sketch of ventral gill-arch elements; C, Uranoscopus scaber L., USNM 198087, with (below) diagram of ventral gill-arch elements, as in A. the Paracanthopterygii (Greenwood et al., paracanthopterygians (Patterson and Rosen, 1966; Rosen and Patterson, 1969) instead of 1989), neither do they belong with dragonets. with blennioids (Springer's view) or with cal- Clingfishes have the four hypobranchials lionymoids (as Gosline, 1970, had suggest- converging on a short basibranchial copula ed). Callionymoids (dragonets) have a similar without ossifications, as do some uranoscop- halberd-shaped PB3, but have a sizable PB2 ids (fig. 36), but dragonets have a primitive tucked under the anterior margin of PB3 (fig. ventral gill arch configuration with well- 33C; Nakabo, 1983: figs. 25, 26). Although developed ossified basibranchials (Nakabo, clingfishes (gobiesocids) might not belong with 1983: figs. 25, 26). 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND~BB1LABYRINTH FISHES 43
4 < BB2
C BH
BB1
Fig. 37. Left dorsal gill arches of blennioids, dorsal views, with (right) sketches of ventral gill-arch elements. A, Dactyloscopidae, Gillelus rubrocinctus Longley, ANSP uncataloged (BB 1-2 are sketched in lateral view at extreme right); B, Tripterygiidae, Gilloblennius tripinnis (Forster), USNM 214271; C, Blennius trigloides Cuvier and Valenciennes, USNM 259168-Fl9 (BH and BB1-3 of this specimen are sketched in lateral view, with a sketch of the same bones in Blennius normani Poll, USNM 199533, at extreme right, to show variability in the posteroventral process on BB 1, which is short in B. normani).
BLENNY GnuL ARCHES (figs. 33A,B, 37, 38B, unusual process on BB 1. Springer's views on C): Springer also pointed out to me that in blennioid classification, in George and blennioids, basibranchial 1 (BB1) has a pos- Springer (1980: 3), are summarized to restrict teroventral process extending below BB2. But true or tropical blennies to the Clinidae (in- it does not develop to the same extent in all cluding Ophiclinidae and Peronedysidae), species (fig. 37C). It appears, therefore, that Labrisomidae, Tripterygiidae, Dactyloscop- blennioids are characterized by the absence idae, Chaenopsidae, and Blenniidae. The re- of PB1, 2, and 4, by the shape of PB3, and duced dorsal gill-arch anatomy of a single at least in some taxa, by the presence of an pharyngobranchial (PB3) in the shape of an 44 AMERICAN MUSEUM NOVITATES NO. 2983
B
C .. /~~~~~~~~~~~ui~~ BB2 ,~H2 C3 ~~H3 C52
Fig. 38. Left dorsal gill arches of a stichaeid (A) and two blennioids, dorsal views. A, Stichaeidae, Ulvaria subbifurcata (Storer), USNM 201949; B, Chaenopsidae, Chaenopsis alepidota (Gilbert), USNM 200391, with (below right) a sketch of the ventral gill-arch elements; C, Tripterygiidae, Lepidoblennius marmoratus (Bennett), USNM 201625. ax head is found in all these fishes, except that some chaenopsids (fig. 38B) and labri- MONOPHYLETIC GROUPS AND somids (Springer, personal commun.) retain MATERIALS FOR A PB1. The blennioids of Gosline (1968) in- PERCOMORPH CLASSIFICATION cluded notothenioids, trachinoids, congro- Ifthere is to be a Perciformes it would have gadoids, and zoarceoids as well as (p. 68) a to be defined by having a dorsal gill arch distinct group of tropical blennioids (whose anatomy that includes a long, rodlike inter- relationship to other fishes "is by no means arcual cartilage (IAC) between El and PB2. clear"). The main trouble with this is that the Merely having IAC is not enough, because nontropical "blennioids" are rife with prim- there are at least six states of IAC in cteno- itive features, at least in their dorsal gill arch squamates: (1) the nodule in some mycto- anatomy. phids; (2) the short plug in anoplogasterid 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 45
IAC A ,PB1
,PB1
B
Fig. 40. Left dorsal gill arches of perciforms with unresolved relationships in dorsal view. A, Centrarchidae, A mbloplites rupestris (Rafinesque), AMNH 39695; B,? Elassomatidae, Elassoma zon- atum Jordan, AMNH 21849. The gill arches pro- vide no good reason to regard B as a centrarchid (cf. Johnson, 1984: 465), but according to Chang Chang-Hwa (personal commun., 1988), it appears to be immediately related to centrarchids.
beryciforms and kurtids (fig. 46A); (3) the Fig. 39. Left dorsal gill arches of echeneoids rodlike form in more normal percoid fishes (of Johnson, 1984) in dorsal view. A, Coryphaen- (e.g., figs. 4, 6); (4) the ossified rod present in idae, Coryphaena equisetis L., AMNH 22117; B, Rachycentridae, Rachycentron canadum (L.), synbranchids (fig. 31 B) and most carapids USNM 260175; C, Echeneididae, Remora remora (L.), USNM 159685. Johnson (1984) has shown that these three families form a monophyletic group of toothplates on E2 and E3 are shown by broken whose sister is the Carangidae; and that within lines; these toothplates are not fused to the bones that group, A and B are sister groups. The outlines in A, whereas in C TPE2 is free and TPE3 is fused. 46 AMERICAN MUSEUM NOVITATES NO. 2983
Fig. 41. Left dorsal arches ofperciforms with unresolved relationships in dorsal view. A, Cirrhitidae, Oxycirrhites typus Bleeker, AMNH 45081; B, Apogonidae, Apogon townsendi (Breder), AMNH 23513; C, Mullidae, Mullus barbatus L., AMNH 22228 (to right of C is a ventral view, anterior down, of the skeleton of the bases of the mental barbels, with the right branchiostegals).
(Travers, 1981; Patterson and Rosen, 1989: primitive euteleostean condition is direct 25); and (5a) a short, stout ligament in the contact between the cartilaginous tips of the interarcual position, as in Percopsis (Patter- uncinate processes of EB1 and PB2 (Rosen, son and Rosen, 1989: fig. 13) or (5b) a more 1973: figs. 3, 5, 58; Travers, 1981: fig. 14), I elongate ligament as in Raniceps (Markle, presume that the varieties of condition (5) 1989: fig. 2) and Trichodon (fig. 48B), which are derived relative to conditions (1) to (3), may contain a small IAC close to its attach- and that among those three, condition (1) is ment to PB2, as in certain gadiforms (Pat- primitive, (2) is somewhat more derived, (3) terson and Rosen, 1989: fig. 12B; Markle, is the derived "percoid" condition, and (4) 1989: fig. 3). On the assumption that the is surely derived from (3). This means that 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 47
LAC S /pharyngognaths and labyrinths, atherino- A \ J g morphs, and paracanthopts as newly defined PB2 (Patterson and Rosen, 1989) are linked with perciform acanthopterygians since some ple- siomorphous members of all these groups have condition (3) [e.g., labroids (fig. 5) and gerreids (fig. 4), some anabantoids (fig. 28), / "atherinoids" (Rosen and Parenti, 1981: figs. 6-9), and ophidioid ophidiiforms (Patterson and Rosen, 1989: fig. 1 3F, G, H)], unless some PB3 of the IACs are homoplastic. The absence of P84 IAC when El lacks an uncinate process, or PB2 is small or absent, is certainly likely to be secondary, as in some gadiforms and in bythitoid ophidiiforms among paracanthopts ACC/ (Patterson and Rosen, 1989; Markle, 1989), some atherinomorphs (Rosen and Parenti, 1981: figs. 14-17), and various acanthopts B illustrated here (e.g., figs. 33-38, 44A, B, 47C). One view of the problem of percomorph relationships seems to boil down to discov- ering how to link the Pharyngognathi, Squamipinnes, Labyrinthici, Plectognathi, Scombroidei, atherinomorphs, and paracan- thopts with a monophyletic Percoidei and with some but not all "beryciforms," and, when assembled in some corroborated hier- archy, all ofthose with polymixiids and myc- tophids. Previously I have suggested how the occipital region (Rosen, 1985) and the caudal skeleton (Rosen, 1984) might be used to clus- ter some taxa to these major groups. For reasons perfectly illustrated by Regan's C >\ (1913) conclusion that about 60 percent of his 95 percomorph families had to be left ...... unresolved in his equivalent ofour Percoidei, a nongroup defined by symplesiomorphies ("the absence ofthe special peculiarities which characterize the other suborders," p. 112), the intrarelationships of percomorphs have justly been considered one of ichthyology's major unsolved problems. Some contribu- tions on this problem (e.g., Gosline, 1968) accepted Regan's solution by treating the Per- coidei as a natural group with defining char- acters.Johnson(1975, 1980, 1984, 1986) and
22125; B, Toxotidae, Toxotes jaculatrix (Pallas), Fig. 42. Left dorsal gill arches of perciforms AMNH 1969-79 (TPE2 shown by broken lines); with unresolved relationships in dorsal view. A, C, Malacanthidae, Hoplolatilus starcki Randall and Lobotidae, Lobotes surinamensis (Bloch), AMNH Dooley, AMNH 38129. 48 AMERICAN MUSEUM NOVITATES NO. 2983
A~~~~~~~A
B2 A 9 ~ ~~~E1X P/D
*PB3~
PB4 ~ ~ ~ ~ 0
B82~ ~ ~ ~ ~ ~ ~ B
H3~~ ~ ~ ~ ~ ~~H
Fig. 43. Left dorsal gill arches of perciforms with unresolved relationships in dorsal view. A, Bathy- clupeidae, Bathyclupea hoskynEi Alcock, AMNH 22067; B, Leptoscopidae, Leptoscopus macropygus (Richardson), USNM 213490, with sketch of ventral gill-arch elements; C, Percophidae, Hemerocoetes monopterygius (Schneider), USNM 214077, with sketch of ventral gill-arch elements; D, Polynemidae, Polydactylus sexfilis Cuvier and Valenciennes, USNM 214123. A
B
Fig. 45. Left dorsal gill arches of perciforms with unresolved relationships in dorsal view. A, Fig. 44. Left dorsal gill arches of acantho- Priacanthidae, Pristigenys alta (Gill), AMNH morphs with unresolved relationships in dorsal 29378; B, Grammidae, Gramma loreto Poey, view. A, Gasterosteidae, Culaea inconstans Kirt- AMNH 23811; C, Sciaenidae, Isopisthus parvi- land, AMNH 40320; B, Pegasidae, Pegasus voli- pinnis (Cuvier and Valenciennes), AMNH 20764. tans L., AMNH 13876 (10 large, clublike teeth on TP3 are shown in transparency); C, Ammodyti- dae, A mmodytes americanus DeKay, AMNH 21897. 50 AMERICAN MUSEUM NOVITATES NO. 2983
PB1 A
B
Fig. 46. Left dorsal gill arches of perciforms Fig. 47. Left dorsal gill arches of scorpaeni- with unresolved relationships in dorsal view. A, forms. A, B, Scorpaenidae, Ectreposebastes imus Kurtidae, Kurtus gulliveri Castelnau, AMNH 1969- Garman, AMNH 27991 (B, in posterior view with 105 (toothplate fused to E3 shown by broken line); E4 removed); C, Icelidae, Icelus spiniger Gilbert, B, Percidae, Perca flavescens L., AMNH 22340; USNM 208352. C, Serranidae, Rypticus bistrispinus (Mitchill), AMNH 14890. 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 51
A
B
TP4 Fig. 48. Left dorsal gill arches of perciforms with unresolved relationships in dorsal view. A, Mu- giloididae, Parapercis ommatura Jordan and Snyder, AMNH 26921; B, Trichodontidae, Trichodon trichodon (Tilesius), USNM 213532.
Rosen (1984, 1985) attempted a simplifica- Tyler et al., 1989), atherinomorphs (Rosen tion by assembling the families of perco- and Parenti, 1981), scorpaeniforms, and a morphs (J. Nelson, 1984, listed 149 families revised Paracanthopterygii (Patterson and in Perciformes) into cladistically defined nat- Rosen, 1989). In this scheme there is left a ural groups to reduce the scope to the more smaller residue ofpercoid families about each manageable one of interrelationships among of which questions of taxonomic affinity can ten groups: the scombroids (Collette et al., be asked in relation to one or more of those 1984; Johnson, 1986), gobioids (Hoese, 1984; 10 groups listed above. The same reasoning Springer, 1983, 1988), Pharyngognathi, La- applies to the various "beryciform" taxa. Al- byrinthici, plectognaths (Rosen, 1984), blen- ternatively, some of the 10 core groups and nioids (Rosenblatt, 1984; Springer, in pro- a few of the leftover "percoids" might be gress), Squamipinnes (Mok and Shen, 1983; linked on the basis of characters of general 52 AMERICAN MUSEUM NOVITATES NO. 2983
PB2
PB4
Fig. 49. Left dorsal gill arches ofstromateoids and pomatomids, dorsal views. A, Nomeidae, Nomeus gronovii (Gmelin), AMNH 22249 (broken line beneath head ofE3 is outline ofa membranous extension ofTP4 along lateral face ofPB3); B, Pomatomidae, Pomatomus saltatrix (L.), AMNH 18804 (toothplates on E2, free, and E3, fused to bone, are shown by broken lines); C, Stromateidae, Peprilus triacanthus (Peck), AMNH 1808, with (right) medial view of PB3, 4, and head of E4.
applicability such as Johnson's (1975) caudal and perhaps features of the dorsal gill arches, spur, occipital anatomy (Rosen, 1985), cau- as sampled in figures 39-50. There will al- dal skeletal features outlined in Rosen (1984), ways be a residue of taxa that are difficult to 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 53
C
TP4
Fig. 50. Left dorsal gill arches ofscombroids, dorsal views. A, Scomber australasicus Cuvier, AMNH 26894; B, Trichiurus sp., AMNH 17566; C, Xiphias gladius L., AMNH 15657. assign, but the problem of percomorph clas- rich (1909), Regan (1913, 1929), Jordan sification appears to be somewhat simpler (1923), Berg (1940), McAllister (1968), than it was for Gunther (1859-1870), Good- Greenwood et al. (1966), and J. Nelson (1976, 54 AMERICAN MUSEUM NOVITATES NO. 2983
1984), although simplification carries no im- Collette, B. B., T. Potthoff, W. J. Richards, S. Ue- plication ofcorrectness and I am not so naive yanagi, J. L. Russo, and Y. Nishikawa as to believe that, after reading Gill (1872), 1984. Scombroidei: development and rela- I might only be reinventing the wheel. But tionships. In H. G. Moser et al. (eds.), we need to invent a new Ontogeny and systematics of fishes, pp. perhaps wheel before 591-620. Special Publication No. 1, some of the larger questions of relationship American Society of Ichthyologists and can be dealt with. The narrative is still un- Herpetologists. convincing, adorned as it is with significant Cuvier, G. L. C. F. D., and A. Valenciennes bald spots. 1828-49. Histoire naturelle des poissons, 22 vols. Paris. REFERENCES Emery, A. R. 1973. Comparative ecology and functional os- Barel, C. D. N., F. Witte, and M. J. P. van Oijen teology offourteen species ofdamselfish 1976. The shape of the skeletal elements in (Pisces: Pomacentridae) at Alligator the head ofa generalized Haplochromis Reef, Florida Keys. Bull. Mar. Sci. 23: species: H. elegans Trewavas 1933 649-770. (Pisces, Cichlidae). Netherlands J. Zool. 1980. The osteology of Lepidozygus tapeino- 26: 163-265. soma (Pisces: Pomacentridae). Ibid., 30: Barlow, G. W., K. F. Liem, and W. Wickler 213-236. 1968. Badidae, a new fish family-behaviour- George, A., and V. G. Springer al, osteological, and developmental evi- 1980. Revision of the clinid fish tribe Ophi- dence. Proc. Zool. Soc. London 156: clinini, including five new species, and 415-447. definition ofthe family Clinidae. Smith- Bassett, C. son. Contrib. Zool. 307: 1-31. 1964. Environmental and cellular factors reg- Gill, T. ulating osteogenesis. In H. Frost (ed.), 1863. On a new genus of fishes allied to Au- Bone biodynamics, 233-244. Boston: lorhynchus and on the affinities of the Little Brown. family Aulorhynchoidae, to which it be- 1971. Effect of force on skeletal tissues. In J. longs. Proc. Acad. Nat. Sci. Philadel- A. Downey and R. C. Darling (eds.), phia 1862: 233-235. Physiological basis of rehabilitation 1872. Arrangement of the families of fishes. Medicine, pp. 283-316. Philadelphia: Smithson. Misc. Collect. 247: i-xlvi, 1- Saunders. 47. Berg, L. S. Goodrich, E. S. 1940. Classification offishes, both Recent and 1909. Vertebrata Craniata: first fascicle, cy- fossil. Trav. Inst. Zool. Acad, Sci. clostomes and fishes. In E. R. Lankester U.S.S.R. 5(2): 517 pp. [Also reprint, J. (ed.), A treatise on zoology, vol. 9. Lon- Edwards, 1947, Ann Arbor, Michigan] don: A. and C. Black. Bertin, L., and C. Arambourg Gosline, W. A. 1958. Superordre des Teleost6ens (Teleostei). 1968. The suborders ofperciform fishes. Proc. In P. P. Grasse (ed.), Traite de Zoologie, U.S. Natl. Mus. 124(3647): 1-78. 13(3): 2204-2500. Paris: Masson et Cie. 1970. A reinterpretation of the teleostean fish Boulenger, G. A. order Gobiesociformes. Proc. Califor- 1904a. A synopsis ofthe suborders and families nia Acad. Sci. 38: 363-382. of teleostean fishes. Ann. Mag. Nat. 1971. Functional morphology and classifica- Hist., ser. 7, 13: 161-190. tion ofteleostean fishes. Honolulu: Univ. 1904b. Fishes, systematic account ofTeleostei. Hawaii Press. In S. F. Harmer and A. E. Shipley (eds.), 1983. The relationships ofthe mastacembelid The Cambridge natural history 7: 541- and synbranchid fishes. Japanese J. Ich- 727. London: Macmillan. thyol. 29: 323-328. Cavender, T. M. Greenwood, P. H. 1986. Review of the fossil history of North 1964. Environmental effects on the pharyn- American freshwater fishes. In C. H. geal mill of a cichlid fish, Astatoreo- Hocutt and E. 0. Wiley (eds.), The zoo- chromis alluaudi, and their taxonomic geography of North American fresh- implications. Proc. Linn. Soc. London water fishes, pp. 699-724. New York: 176: 1-10. Wiley. 1978. A review of the pharyngeal apophysis 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 55
and its significance in the classification ilies and genera as far as known. Stan- of African cichlid fishes. Bull. Br. Mus. ford Univ. Publ., Biol. Sci. 3: 77-243. (Nat. Hist.), Zool. 33: 297-323. Kaufman, L. S., and K. F. Liem 1985. Notes on the anatomy and phyletic re- 1982. Fishes ofthe suborder Labroidei (Pisces: lationships of Hemichromis Peters, Perciformes): phylogeny, ecology, and 1858. Bull. Br. Mus. (Nat. Hist.), Zool. evolutionary significance. Breviora 472: 48: 131-171. 1-19. Greenwood, P. H., D. E. Rosen, S. H. Weitzman, Lauder, G. V., and K. F. Liem and G. S. Myers 1983. The evolution and interrelationships of 1966. Phyletic studies ofteleostean fishes, with the actinopterygian fishes. Bull. Mus. a provisional classification of living Comp. Zool. 150: 95-197. forms. Bull. Am. Mus. Nat. Hist. 131: Liem, K. F. 339-456. 1963. The comparative osteology and phylog- Gunther, A. C. L. G. eny of the Anabantoidei (Teleostei, 1859-70. Catalogue offishes in the British Mu- Pisces). Illinois Biol. Monogr. 30: 1-149. seum, vols. 1-8. London: British Mu- 1986. The pharyngeal jaw apparatus of the seum. Embiotocidae (Teleostei): a functional 1880. An introduction to the study of fishes. and evolutionary perspective. Copeia Edinburgh: A. and C. Black. 1986: 311-323. Hall, B. K. Liem, K. F., and P. H. Greenwood 1975. Evolutionary consequences of skeletal 1981. A functional approach to the phylogeny differentiation. Am. Zool. 15: 329-350. ofpharyngognath teleosts. Am. Zool. 21: Hoese, D. F. 83-101. 1984. Gobioidei: relationships. In H. G. Mo- Liem, K. F., and S. L. Sanderson ser et al. (eds.), Ontogeny and system- 1986. The pharyngeal jaw apparatus of labrid atics of fishes, pp. 588-591. Special fishes: a functional morphological per- Publication No. 1, American Society of spective. J. Morphol. 187: 143-158. Ichthyologists and Herpetologists. McAllister, D. E. Hughes, G. F. (ed.) 1968. Evolution of branchiostegals and clas- 1976. Respiration of amphibious vertebrates. sification ofteleostome fishes. Natl. Mus. London: Academic Press. Canada, Bull. 221: 1-239. Machado-Allison, A. Ismail, M. H., W. Verraes, and A. Huysseune 1973. Contribucion al conocimiento de la 1982. Developmental aspects of the pharyn- tax- in onomia del genero Cichia (Perciformes: geal jaws Astatotilapia elegans (Tre- Cichlidae) en Venezuela. Acta Biol. Ve- wavas, 1933) (Teleostei: Cichlidae). nezuelica 8: 155-205. Netherlands J. Zool. 32: 513-543. Markle, D. F. Johnson, G. D. 1989. Aspects ofcharacter homology and phy- 1975. The procurrent spur: an undescribed logeny ofthe Gadiformes. Sci. Ser., Nat. perciform caudal character and its phy- Hist. Mus. Los Angeles Co. 32: 59-88. logenetic implications. Occas. Pap. Cal- Mok, H.-K., and S. C. Shen ifornia Acad. Sci. 121: 1-23. 1983. Osteology and phylogeny of Squami- 1980. The limits and relationships of the pinnes. Taiwan Mus. Spec. Publ. Ser., Lutjanidae and associated families. Bull. Zool. 1: 1-87. Scripps Inst. Oceanogr., Univ. Califor- Morris, S. L. nia, San Diego 24: 1-114. 1982. The osteology and relationships of the 1984. Percoidei: development and relation- Embiotocidae (Pisces). Ph.D. diss., Or- ships. In H. G. Moser et al. (eds.), On- egon State Univ., Corvallis. togeny and systematics of fishes, pp. Morris, S. L., and A. J. Gaudin 464-498. Special Publication no. 1, 1982. Osteocranial development in the vivi- American Society of Ichthyologists and parous surfperch Amphistichus argen- Herpetologists. teus (Pisces: Embiotocidae). J. Mor- 1986. Scombroid phylogeny: an alternative phol. 174: 95-120. hypothesis. Bull. Mar. Sci. 39: 1-41. Muller, J. Jordan, D. S. 1844. Uber den Bau und die Grenzen der 1905. A guide to the study of fishes, two vols. Ganoiden und uiber das naturliche sys- New York: Henry Holt. tem der Fische. Ber. Akad. Wiss. Berlin, 1923. A classification of fishes including fam- 1844: 416-422. 56 AMERICAN MUSEUM NOVITATES NO. 2983
1845. Uber den Bau und die Grenzen der lationships of fishes, pp. 397-513. Lon- Ganoiden und uiber das naturliche sys- don: Academic Press. tem der Fische. Arch. Naturgesch. 11: 1975. A vicariance model of Caribbean bio- 91-141. geography. Syst. Zool. 24: 431-464. 1846. Uber den Bau und die Grenzen der 1978. Vicariant patterns and historical expla- Ganoiden und iuber das naturliche sys- nation in biogeography. Syst. Zool. 27: tem der Fische. Phys. Math. Abh. K. 159-188. Akad. Wiss. Berlin, 1846: 117-216. 1979. Fishes from the uplands and intermon- [English translation by J. W. Griffith, tane basins of Guatemala: revisionary 1846, Scient. Mem. 4: 499-558] studies and comparative geography. Nakabo, T. Bull. Am. Mus. Nat. Hist. 162: 267- 1983. Comparative osteology and phyloge- 376. netic relationships of the dragonets 1984. Zeiforms as primitive plectognath fish- (Pisces: Callionymidae) with some es. Am. Mus. Novitates 2782: 45 pp. thoughts of their evolutionary history. 1985. An essay on euteleostean classification. Publ. Seto Mar. Biol. Lab. 28: 1-73. Am. Mus. Novitates 2827: 57 pp. Nelson, G. J. Rosen, D. E., and L. R. Parenti 1967. Gill arches of some teleostean fishes of 1981. Relationships of Oryzias, and the groups the families Girellidae, Pomacentridae, ofatheriniform fishes. Am. Mus. Novi- Embiotocidae, Labridae and Scaridae. tates 2719: 25 pp. J. Nat. Hist. 1: 289-293. Rosen, D. E., and C. Patterson 1969. Gill arches and the phylogeny of fishes, 1969. The structure and relationships of the with notes on the classification of ver- paracanthopterygian fishes. Bull. Am. tebrates. Bull. Am. Mus. Nat. Hist. 141: Mus. Nat. Hist. 141: 357-474. 475-552. Rosenblatt, R. H. Nelson, G. J., J. W. Atz, K. D. Kallman, and C. 1984. Blennioidei: introduction. In H. G. L. Smith Moser et al. (eds.), Ontogeny and sys- 1987. Donn Eric Rosen, 1929-1986. Copeia tematics of fishes, 551-552. Special 1987: 541-547. Publication no. 1, American Society of Nelson, G. J., and D. E. Rosen (eds.) Ichthyologists and Herpetologists. 1981. Vicariance biogeography: a critique. Schaeffer, B., and D. E. Rosen New York: Columbia Univ. Press. 1961. Major adaptive levels in the evolution Nelson, J. S. of the actinopterygian feeding mecha- 1976. Fishes of the world. New York: Wiley. nism. Am. Zool. 1: 187-204. 1984. Fishes of the world [second ed.]. New York: Wiley. Springer, V. G. Patterson, C., and D. E. Rosen 1983. Tyson belos, new genus and species of 1989. The Paracanthopterygii revisited: order western Pacific fish, with discussions of and disorder. Sci. Ser., Nat. Hist. Mus. gobioid osteology and classification. Los Angeles Co. 32: 5-36. Smithson. Contrib. Zool. 390: 1-40. Regan, C. T. 1988. Rotuma lewisi, new genus and species 1909. Asiatic fishes of the family Anabanti- offish from the southwest Pacific. Proc. dae. Proc. Zool. Soc. London 1909: 767- Biol. Soc. Washington 101: 530-539. 787. Springer, V. G., and W. C. Freihofer 1913. The classification of the percoid fishes. 1976. Study of the monotypic fish family Ann. Mag. Nat. Hist., ser. 8, 12: 111- Pholidichthyidae (Perciformes). Smith- 145. son. Contrib. Zool. 216: 1-43. 1929. Fishes. Encyclopaedia Britannica, 14th Stiassny, M. L. J. ed., 9: 305-328. 1981. The phyletic status of the family Cich- Rognes, K. lidae (Pisces: Perciformes): a compara- 1973. Head skeleton and jaw mechanism in tive anatomical investigation. Nether- Labrinae (Teleostei: Labridae) from lands J. Zool. 31: 275-314. Norwegian waters. Acta Univ. Bergen- 1982. The relationships of the neotropical ge- sis, Ser. Math. Rer. Nat. 4: 1-149. nus Cichla (Perciformes: Cichlidae): a Rosen, D. E. phyletic analysis including some func- 1973. Interrelationships of higher euteleos- tional considerations. J. Zool., London tean fishes. In P. H. Greenwood, R. S. 197: 427-453. Miles, and C. Patterson (eds.), Interre- 1986. The limits and relationships ofthe acan- 1990 ROSEN AND PATTERSON: PHARYNGOGNATH AND LABYRINTH FISHES 57
thomorph teleosts. J. Zool., London, ser. Travers, R. A. B, 1: 411-460. 1981. The interarcual cartilage; a review ofits Stiassny, M. L. J., and J. S. Jensen development, distribution and value as 1987. Labroid interrelationships revisited: an indicator of phyletic relationships in morphological complexity, key inno- euteleostean fishes. J. Nat. Hist. 15: 853- vations, and the study of comparative 871. diversity. Bull. Mus. Comp. Zool. 151: 1984. A review of the Mastacembeloidei, a 269-319. suborder ofsynbranchiform teleost fish- Sufi, S. M. K. es. Part II: phylogenetic analysis. Bull. 1956. Revision of the Oriental fishes of the Br. Mus. (Nat. Hist.), Zool. 47: 83-150. family Mastacembelidae. Bull. Raffles Tyler, J. C., G. D. Johnson, I. Nakamura, and B. Mus. 27: 93-146. B. Collette Tarp, F. H. 1989. Morphology of Luvarus imperialis (Lu- 1952. A revision of the family Embiotocidae varidae), with a phylogenetic analysis of (the surfperches). California Dep. Fish the Acanthuroidei (Pisces). Smithson. Game, Fish Bull. 88: 1-99. Contrib. Zool. 485: vi + 78 pp.
Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, New York 10024.
THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER.