AMERICAN MUSEUM

NorntatesPUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORYCENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024

Number 2779, pp. 1-12, figs. 1-7, table 1 February 8, 1984

On the Relationships of Phallostethid Fishes(Atherinomorpha), With Notes on the Anatomy of

Phallostethus dunckeri Regan, 1913

LYNNE R. PARENTI'

ABSTRACT

The Phallostethidae (including Neostethidae) isa family comprised of approximately 20 speciesof small, fresh, brackish, and occasionally salt-water atherinomorph fishes of Indo-Australia.Phallostethids have variously been suggested asclosest relatives of the atherinoid or cyprinodon-tiform fishes among the atherinomorphs, or ofthepolynemids or gobioids among the percomorphs.Phallostethids uniquely share several derivedcharacters ofthe jaws and the anal fin with a groupof Indo-Australian and Pacific atherinoids. The

westem Pacific Dentatherina Patten and Ivantsoffis proposed as the sister group of the Phallosteth-idae.The anatomy of Phallostethus Regan, the type

genus, is poorly known because ofthe scarcity andunsatisfactory condition of available material. Areport on the anatomy of Phallostethus dunckeriRegan, the sole species in the genus, based on ex-amination of the syntypes and on unpublishednotes and sketches is also included.

INTRODUCTION

The phallostethids (Atherinomorpha,Phallostethidae) are a little-known group ofcoastal fishes distributed throughout the Phil-ippines, Borneo, Java, Malay Peninsula, andSoutheast Asian mainland.2 They are definedas monophyletic by a complex subcephaliccopulatory organ in males, termed the pria-

pium (Regan, 1913, 1916), and among oth-ers, a series of derived characters related toreproduction as, for example, the anteriorplacement of the urogenital opening and re-duction and/or modification ofthe pelvic finsand fin girdles (see Roberts, 1971b).The primary purpose of this report is to

' Research Associate, Department of Ichthyology, American Museum of Natural History.2 Report ofa collection ofphallostethids from northwestern Sumatra (Aurich, 1937) is considered to be unconfirmed.

Copyright © American Museum of Natural History 1984 ISSN 0003-0082 / Price $1.25

AMERICAN MUSEUM NOVITATES

present the hypothesis of close relationshipamong the phallostethids and several atheri-noid genera ofIndo-Australia and the Pacific.The definition and relationships of phallo-stethid species and genera are the subjects ofan ongoing study (Parenti, in prep.). How-ever, such a study could not be carried outwithout a detailed, well-corroborated hy-pothesis of the relationship of phallostethidsto other fishes.When phallostethids were first discovered,

it was assumed that they were viviparous cy-prinodontiforms (then Microcyprini) be-cause the priapium superficially resemblesintromittent organs found among somemembers of that group, for example, poecili-ids (Regan, 1913, 1916). Smith (1927) re-ported that phallostethids which he observedin Thailand were oviparous. The priapium isapparently used by males for passing spermbundles to females who subsequently lay fer-tilized eggs, although details of priapial func-tion are unknown.

Both Herre (1925) and Myers (1928) point-ed out that some phallostethids have a spi-nous first dorsal fin which is lacking in thecyprinodontiforms. On this basis, and be-cause of the overall resemblance of phallo-stethids to atherinid fishes (the silversides orhardyheads), Myers transferred the phallo-stethids to the order Percesoces, which thencontained the silversides, mullets, barracu-das, and threadfins. Bailey (1936) proposedthat phallostethids were close relatives of thethreadfins (the polynemids) based on a su-perficially similar association of the pectoraland pelvic fins.Hubbs (1944) concurred with Myers that

the phallostethids were more closely relatedto the atherinids than to the cyprinodonti-forms; however, whereas Myers consideredthe cyprinodontiforms and the atherinids tobe closely related, Hubbs considered thepercesocans (including the phallostethids andatherinids) to have a relatively more derivedfin structure than that of the cyprinodonti-forms. Thus, Hubbs supported an alignmentof the percesocans closer to some of the per-ciform fishes.Rosen (1964) placed the phallostethids,

atherinids, cyprinodontiforms, along with theadrianichthyoids (the ricefishes), and the exo-

coetoids (flying fishes, halfbeaks, sauries, andneedlefishes) into the newly named seriesAtherinomorpha, which he considered to bemost closely related to the series Percomor-pha. Rosen and Parenti (1981) formally de-fined the atherinomorph fishes, giving as sev-eral of their derived characters specializationsof the egg, embryo, sperm formation, rostralcartilage and association of the premaxillaryascending processes, and dorsal gill arches.Phallostethids remained in the Atherino-morpha based on their possession of severalof these derived characters. Rosen and Par-enti (1981) concluded that the atherinoidfishes are not currently definable as mono-phyletic and simply listed the families ofath-erinoids in their Division I of the Atherino-morpha, abandoning the use of the termatherinoid in formal classification to empha-size uncertainty in our knowledge of rela-tionships of these fishes (table 1).The phallostethids have undergone eleva-

tions and reductions in taxonomic rank sincetheir discovery; however, no formal state-ment concerning their relationship to anothergroup of fishes has been made previously.Myers (1935) created a new suborder withinthe Percesoces, what he termed Phallosteth-oidea, which served to emphasize further theunique characters ofthese fishes. Rosen (1964)placed the phallostethids in the superfamilyPhallostethoidea, suborder Atherinoidei, butsuggested that they may be more closely re-lated to the cyprinodontiforms. Roberts(1971b) speculated that phallostethids areclosely related to the atherinid subfamilyTaeniomembradinae of Schultz (1948, 1950).However, the Taeniomembradinae was notdefined by Schultz as monophyletic. He stat-ed only that these atherinids possess theprimitive state of the swimbladder characterused to define a relatively more derivedsubfamily. Even the idea that phallostethidsare atherinomorph fishes has been ques-tioned recently by V. G. Springer (personalcommun.) who suggests that they may beclosely related to the gobioid fishes.

This report was prompted by: (1) the recentdiscovery and diagnosis of a new genus andspecies ofatherinid, Dentatherina merceri byPatten and Ivantsoff(1983), which I proposeas the closest living relative of the phallo-

2 NO. 2779

PARENTI: PHALLOSTETHID FISHES

stethid fishes; (2) the opportunity to examineRegan's syntypes of Phallostethus dunckeriat the British Museum (Natural History); and,(3) most important, by the gift of notes andsketches prepared by Dr. Ethelwynn Trewa-vas in the 1930s as part of her planned re-vision of phallostethid fishes. Both becauseof the scarcity of material and present poorcondition, which precludes a formal rede-scription, Dr. Trewavas's notes and sketchesof P. dunckeri contain some of the only dataavailable on the anatomy of this uniquespecies.

ACKNOWLEDGMENTS

Dr. Donn E. Rosen, American Museum ofNatural History, New York, has frequentlydiscussed aspects ofatherinomorph anatomyand systematics with me. Dr. Tyson R. Rob-erts, California Academy of Sciences, SanFrancisco has, on numerous occasions, freelygiven information on the anatomy and dis-tribution of phallostethids. Drs. WalterIvantsoff, Macquarie University, Australia,and Brian White, Los Angeles County Mu-seum ofNatural History, Los Angeles, kindlyshared information on atherinoids. The re-port on the anatomy of Phallostethus dunck-eri Regan would not have been possible with-out the gift of notes and sketches made byDr. Ethelwynn Trewavas, Curator Emerita,British Museum (Natural History) as well asher kind permission to use them here.

I thank Drs. P. Humphry Greenwood, Brit-ish Museum (Natural History), London(BMNH), Gareth Nelson, American Mu-seum ofNatural History, New York (AMNH),and Richard P. Vari and Ms. Susan L. Jewett,National Museum ofNatural History, Smith-sonian Institution, Washington, D.C.(USNM), for lending or allowing me to ex-amine specimens in their care. Dr. Clyde Bar-bour, Wright State University, Dayton, andDr. Roberts, kindly donated specimens to theAmerican Museum ofNatural History whichwere used in this study. This project was ini-tiated during the tenure of a NATO Post-doctoral Fellowship at the British Museumand completed while I was a Research As-sociate at the American Museum of NaturalHistory.

TABLE 1Classification of Atherinomorph Fishes

(From Rosen and Parenti, 1981)

Series AtherinomorphaDivision I

Family AtherinidaeFamily BedotiidaeFamily IsonidaeFamily MelanotaeniidaeFamily PhallostethidaeFamily Telmatherinidae

Division IIOrder CyprinodontiformesOrder Beloniformes

Drs. Ivantsoff, John Patten, and Rosen readand commented on the manuscript.

Phallostethus dunckeri ReganPhallostethid anatomy has been the focus

ofnumerous studies published since Regan's(1913) description of Phallostethus dunckeri(e.g., Regan, 1916; Myers, 1928; Bailey, 1936;Aurich, 1937; TeWinkle, 1939; Herre, 1942;Woltereck, 1942a, 1942b; Hubbs, 1944).Roberts (1971 a, 1971 b) presented the mostrecent reviews of some problems in phallo-stethid anatomy and systematics. He statedthat basic comparative data of Phallostethusare unknown because Regan (1913) did notreport states ofcharacters which we now con-sider to be useful in defining phallostethidrelationships, and because P. dunckeri isknown today only from syntypes (Roberts,197 ib). For example, Roberts stated that thenumber of branchiostegal rays, and whetheror not P. dunckeri has a first dorsal fin, wereunknown.Regan (1913) based his description of

Phallostethus dunckeri on seven specimenscollected from Johore, on the Malay Penin-sula. The description centered on the struc-ture ofthe priapium and included other char-acters that are still considered to distinguishP. dunckeri from all other phallostethidspecies. One such character is a high numberof anal fin rays, ranging from 26 to 28 asopposed to 14 to 15 in Phenacostethus, thepresumed closest relative (Roberts, 1971 a).Regan (1913, pp. 548-555) stated that his

1 984 3

AMERICAN MUSEUM NOVITATES

s9mplectic

FIG. 1. Lateral view of neurocranium, jaws and jaw suspensorium, and opercular series, Phallostethusdunckeri Regan, Syntype, BMNH 1913.5.24:22. From a sketch prepared by Dr. Trewavas. Bone stippled;cartilage open circles.

description was based on seven specimens,two of which were sectioned for study of in-ternal anatomy. Five specimens, four alcohol(BMNH 1913.5.24:18-20, BMNH 1913.5.24:21) and one cleared and stained (BMNH1913.5.24:22), are present in the Recent fishcollection of the British Museum (NaturalHistory) and hence, are treated as the re-maining syntypes. The syntypes, augmentedby notes and sketches ofTrewavas, constitutethe study material on which the followinganatomical description is based.NEUROCRANIUM (figs. 1, 2): Supraoccipital

overlapped anteriorly by frontals. Frontalsforming convex roof of orbits, a lateral limbposterior to nasal capsule, articulating withpreorbital, and anterior limb reaching an-terolateral surface of lateral ethmoid. Pari-etals absent. Temporal region concave. Os-sified epioccipital and pterotic present.Intercalar absent. Pterosphenoid small, notmeeting sphenotic and just meeting frontal.

Basisphenoid absent. Exoccipital and basioc-cipital (not shown in fig. 1) with condyles welldeveloped. Infraorbital series represented bya dermosphenotic and a preorbital bone.Mesethmoid an ossified triangular plate, thebase anterior, with a notch for passage of ol-factory nerve. Vomer, with small toothplateand patch of teeth, ventral to ethmoid car-tilage.JAWS AND JAW SUSPENSORIUM (figs. 1, 2):

Upper jaw represented by premaxilla withascending process long and narrow; maxillawith a process dorsal and a process ventralto premaxilla. Lowerjaw represented by den-tary, paradentary, articular and retroarticularbones, the last two bones not necessarily dis-tinct. Premaxilla, dentary, and paradentarywith small, unicuspid teeth. Submaxillarybone between maxilla and vomer. Rostralcartilage long and of moderate width. Hyo-mandibula with a large, single dorsal headarticulating with sphenotic anteriorly and

4 NO. 2779

PARENTI: PHALLOSTETHID FISHES

iibranchial 2

FIG. 2. Dorsal view ofupperjaw, ethmoid regionand anterior portion of neurocranium, Phalloste-thus dunckeri Regan, Syntype, BMNH 1913.5.24:22. From a sketch prepared by Dr. Trewavas. Bonestippled; cartilage open circles.

pterotic posteriorly. Symplectic a long, nar-row bone. Metapterygoid present as a smallbone at hyomandibular-symplectic junction.Quadrate with slender posterior ramus. En-dopterygoid narrow and elongate; ectopter-ygoid minute. Autopalatine reaching maxillaanteriorly.OPERCULAR SERIES (fig. 1): Opercle oval

with posteroventral indentation, lacking ser-rations. Preopercle, subopercle and inter-opercle narrow.HYOBRANCHIAL APPARATUS (fig. 3): Hyoid

bar represented by a single hypohyal, and an-terior and posterior ceratohyal. Four bran-chiostegal rays. Interhyal ossified. Ossifiedportion of basihyal narrow and elongate.Three ossified basibranchials.3 Hypobran-chials, if present, not ossified. First cerato-branchial with 12 slender gill rakers, secondceratobranchial without gill rakers, the thirdand fourth ceratobranchials each with atoothplate and patch of teeth, fifth cerato-branchial expanded medially and with atoothplate bearing a patch ofcurved, pointedteeth. Four epibranchial bones, the first withgill rakers. Two upper pharyngeal bones, the

3In her notes, Dr. Trewavas indicated that Phallo-stethus dunckeri has three ossified basibranchials. In heroriginal sketch, there are just two ossified basibranchialsposterior to an elongate ossified basihyal, however. Iinterpret the elongate basihyal as a basihyal and the firstbasibranchial.

ceratobranchial 1

ceratobranchial:2ceratobranchial

ceratobranchial 4

ceratobranchial 5

-I-basibranchial 3

- haryngobranchial 2

I epibranchial

epibranchial 2

pibranchial 2pibranchial 3

pharyngobranchial 3

FIG. 3. Dorsal view ofgill arches, with left dorsalportion removed, Phallostethus dunckeri Regan,Syntype, BMNH 1913.5.24:22. From a sketch pre-pared by Dr. Trewavas. See text for discussion onidentification of structures.

anterior (pharyngobranchial 2) articulatingwith the second epibranchial, the posterior(pharyngobrancial 3) with the third and fourthepibranchials.VERTEBRAL COLUMN: Forty vertebrae, 13

abdominal, 27 caudal. First pleural rib onfourth vertebra. No epineurals or epipleurals.CAUDAL SKELETON (fig. 4): Last vertebra

consisting ofa half-centrum (PU2) with whichare united a dorsal and a ventral hypural plate.Parhypural autogenous. Two epurals.

procurrent rays

parhypural

procurrent rays

FIG. 4. Lateral view of caudal skeleton, Phal-lostethus dunckeri Regan, Syntype, BMNH1913.5.24:22. From a sketch prepared by Dr. Tre-wavas. Epurals are blackened.

1 984 5

AMERICAN MUSEUM NOVITATES

N

-, I.1::f- .. - - /I

/

papillary

toxactinium

1xial

Nctenactini umFIG. 5. Schematic diagram of internal priapial structure of Phenacostethus smithi Myers, BMNH

1927.12.29:1-10. From a sketch prepared by Dr. Trewavas.

FINS: No spinous first dorsal fin; secondsoft-rayed dorsal fin with eight to 10 rays.Caudal fin truncate, seven procurrent raysdorsally, 16 branched caudal rays, and 11procurrent caudal rays ventrally. Minute pel-vic fins and fin girdles in females; males withpelvic fins and fin girdles modified into pria-pium (see below).

PRIAPIUM: The present report does not re-quire a comprehensive review of priapialanatomy, which will be considered in detailin a revision ofthe Phallostethidae sensu lato(Parenti, in prep.). However, certain detailsare pertinent to this discussion.The priapium primitively has two con-

spicuous externalized bones, the ctenactin-ium and pulvinulus, both of which are de-rived from pelvic fin structures (Regan, 1913,1916). Phallostethus Regan and Phenacos-tethus Regan are distinguished from otherphallostethids by the type of externalizedpriapial bones; they have a ctenactinium and

a toxactinium, rather than one or two cten-actinia and a pulvinulus. The currently ac-cepted homologies of priapial parts requiresthat the toxactinium of Phallostethus andPhenacostethus be viewed as a modified pul-vinulus (Roberts, 197 1b). The ctenactiniumofPhenacostethus (fig. 5) is rudimentary andnot as well developed as that of Phallostethusin which it is serrated (see also Regan, 1913;Bailey, 1936; Roberts, 1971a).

RELATIONSHIPS OF THEPHALLOSTETHIDS

Characters in addition to the priapium andassociated reproductive traits have been pro-posed as unique to phallostethids: a submax-illary bone (fig. 2), and a paradentary bone(fig. 1) (Roberts, 197 lb). Rosen and Parenti(1981) treated the two families (Phallosteth-idae and Neostethidae) of phallostethoidssensu Myers as one, the Phallostethidae. Thedivision of phallostethids into two families

6 NO. 2779

1%

PARENTI: PHALLOSTETHID FISHES

was based on differences in priapial mor-phology, and by the identification of sub-maxillary and paradentary bones in generathat had been assigned to the Neostethidae(Roberts, 197 ib), such as Ceratostethus Myersand Neostethus Regan. Roberts (197 ib) calledthese bones "neomorphs."The recent discovery and diagnosis by Pat-

ten and Ivantsoff (1983) of Dentatherina, awestern Pacific marine atherinid, allows fora reinterpretation of these as well as othercharacters used to hypothesize the relation-ships of phallostethids and other Indo-Aus-tralian and Pacific atherinoids.SUBMAXILLARY BONES:4 Submaxillary

bones are prominent endochondral bones thatlie between the medial ramus of the maxillaand the anterolateral portion of the vomer(figs. 2, 6). Bony elements in this position arefound in Dentatherina (Patten and Ivantsoff,1983, figs. 4, 5) and phallostethids in boththe Phallostethidae and Neostethidae sensustricto, contra Roberts (1971b), who statedthey occur only in the latter. He did not knowthe condition in Phallostethus. In some phal-lostethids (e.g., Gulaphallus mirabilis, BMNH1933.3.11:179-186) and taeniomembradineatherinids (e.g., Craterocephalus cuneiceps,AMNH 43184SW, fig. 7C) a submaxillaryelement is present as a large cartilage ratherthan a bone. Roberts (197 lb) postulated thatthe submaxillary bones of neostethids sensustricto contributed toward the extremely pro-tractile mouths of these fishes.

It is not my purpose here to present a hy-pothesis of the relationships of all atherinidfishes, or even atherinid fishes of the sub-family Taeniomembradinae (comprising thegenera Taeniomembras, Craterocephalus,Stenatherina, Alepidomus, Hypoatherina,Atherinomorus, and tentatively Quirichthys).Therefore, I have not surveyed all the generaof any such group to determine whether ornot each has a submaxillary bone or cartilage.However, from the limited survey of Indo-Pacific atherinoids and the subfamily Tae-

4This element should not be confused with one be-tween the maxilla and autopalatine, called a subauto-palatine cartilage by Parenti (1981, p. 406, fig. 35A).Such an element, found in some cyprinodontiforms, isapparently primitive for acanthopterygian fishes.

niomembradinae, I have determined thatmost, but not all, members of these groupshave submaxillary cartilages or bones. Forexample, the Western Australian isonid, Isorhothophilus (AMNH 55027SW), as well asthe New World taeniomembradine Atherino-morus stipes (AMNH 52025SW), have nosuch cartilages or bones. The maxilla andvomer are separated by a small connectivetissue meniscus, which is a primitive char-acter for acanthopterygian fishes.

Furthermore, other atherinoids, such asQuirichthys stramineus (AMNH 20571 SW),Telmatherina ladigesi (AMNH 35378SW),and Pseudomugil tenellus (AMNH 36598SW)have prominent submaxillary cartilages. Eachof these genera was placed in a different fam-ily of atherinoid fishes by Allen (1980). Ac-cessory cartilages and bones in the ethmoidregion are not uncommon among euteleosts,nor are elements in the position of the sub-maxillary bones. I judge these to be derivedat one level within the atherinoid fishes, dis-cussed below.PARADENTARY BONES: A separate, slender

bone, termed the paradentary, lies lateral tothe dentary in both Dentatherina and thephallostethids. Patten and Ivantsoff (1983)noted what they called "calcified nodules" inthe lower jaw ligaments, preferring not toconsider them homologues of the paraden-tary bones of phallostethids. However, fortwo reasons I take the view that the structuresare homologous.

First, the suggestion that they are fortui-tous ossifications of jaw ligaments is con-tradicted by the presence of a paradentarybone with a single row of small, unicuspidteeth in Phallostethus dunckeri (fig. 1). As faras known, these bones are unique to Denta-therina and phallostethids. In some closelyrelated taeniomembradine atherinids, suchas Craterocephalus cuneiceps (AMNH43184SW), there is a mass of connective tis-sue in the position of the paradentary.

Second, the homology ofparadentary bonesin Dentatherina and phallostethids is sup-ported by parsimony. Reasons for their in-clusion in yet a larger group of atherinoids,is discussed below. The same ontogenetic se-quences of paradentary formation in bothgroups would support this homology, how-ever, no such data are available. Histological

1 984 7

AMERICAN MUSEUM NOVITATES

A

pleural rib ofvertebra 13'

pleural rib ofvertebra ; C

first proximal anal radial

anal fin r ys

FIG. 6. Internal view of anal fin, anteriorto left, of A. Gulaphallus mirabilis, BMNH1933.3.11:179-186; B. Dentatherina merceri,USNM 230374; C. Pseudomugil tenellus, AMNH36598SW. Bone stippled; cartilage open circles.

study is also needed to determine whetherthese are dermal or endochondral bones.PREMAXILLA AND ROSTRAL CARTILAGE: In

phallostethids (fig. 2), Dentatherina and sometaeniomembradine atherinids, such as Cra-terocephalus cuneiceps (fig. 7C), the ascend-ing processes of the premaxillae, as well asthe rostral cartilage, are thin and elongate, asopposed to being short in many other ath-erinids. There are exceptions among NewWorld menidiine atherinids such as Melan-orhinus microps (AMNH 25878SW) in whichthere are elongate premaxillary ascendingprocesses; however, I view these as indepen-dently derived in Melanorhinus because oth-er characters indicate that it is distantly re-lated to the phallostethids.ANAL FIN: In phallostethids, Dentatherina,

taeniomembradine atherinids (of the generaCraterocephalus, Atherinomorus and Quir-ichthys), telmatherinids (the genus Telma-

therina) and Pseudomugil, the first proximalanal radial is expanded anteriorly. In phal-lostethids (fig. 6A) and Dentatherina (fig. 6B)the first proximal anal radial is a long, blade-like element. The main shaft of the radial isoriented dorsally. That is, the radial is ex-panded anteriorly, rather than being recum-bent.

In the other taxa listed above, the firstproximal anal radial is expanded anteriorlyand may be bladelike, as in Atherinomorusstipes (AMNH 52025SW), or may be ex-panded just slightly, as in Pseudomugil te-nellus (fig. 6C). In each case, the main shaftof the radial is evident. In other atherinoids(e.g., Bedotia geayi, AMNH 28132SW) thefirst proximal anal radial has no anterior ex-pansion; it is represented solely by the main,dorsally directed shaft.ADDITIONAL CHARACTERS: Two other de-

rived characters, the absence of parietal bonesfrom the skull, and the absence of a dorsalor first postcleithral bone from the pectoralskeleton support a close relationship betweenDentatherina and phallostethids.

RELATIONSHIPS OFATHERINOMORPH FISHES

It is hypothesized that the Phallostethidaeand Dentatherina are sister groups, and thatthey in turn are members of a group of Indo-Australian and Pacific, and some New World,atherinoids that include the taeniomembra-dines as far as I have examined them. Thus,Roberts's (1971b) speculation that phallo-stethids and taeniomembradines are closelyrelated is corroborated. Patten and Ivantsoff(1983) named a new subfamily, the Denta-therininae, for their new genus, because theycould not place it in any other known subfam-ily of the Atherinidae. The definitions andrelationships of the six families of DivisionI, the atherinoids, of the Atherinomorpha(table 1), are poorly known. As stated above,genera that I believe to be closely related tophallostethids and Dentatherina have beenplaced in several different families by recentworkers. One such genus, Quirichthys, en-demic to three river systems in northern Aus-tralia, has been placed most recently in theAtherinidae by Allen (1980, p. 483); how-ever, previously it has been considered a close

8 NO. 2779

PARENTI: PHALLOSTETHID FISHES

relative of the Telmatherinidae, which nowcontains a single genus, Telmatherina. Allen(1980) also placed Pseudomugil in the Me-lanotaeniidae. The Isonidae (or Notocheiri-dae) has been considered by Patten (1978, inPatten and Ivantsoff, 1983) to constitute asubfamily ofthe Atherinidae. These changes,or suggested changes, in classification serveto point out where additional work is needed.For example, Allen (1980, p. 465) claimedthat melanotaeniids, as he defined them, aredistinguished from atherinids in having maleswith elongate dorsal, anal, and pelvic fin rays,and more brightly colored than females.However, not only does Pseudomugil sharethis characteristic, it is found also in otheratherinids such as, for example, Telma-therina. No definition of the Atherinidae interms ofunambiguous, derived characters hasever been proposed. It is inherent in the pres-ent argument concerning the relationship ofthe phallostethids, that one does not exist.That is, the Atherinidae is not monophyletic.More important, I believe that there is ad-

ditional evidence to support Rosen and Par-enti's (1981) claim that the atherinoids can-not be defined as monophyletic. Two of thecharacters for the monophyly ofthe atherino-morph fishes as a group, of the 10 listed byRosen and Parenti (1981, pp. 20-21) are thederived ethmoid region (their character 9),and the decoupling of the rostral cartilagefrom the ascending processes of the premax-illae (their character 7). With these charac-ters, Rosen and Parenti (198 1) supported themonophyly ofthe Atherinomorpha, but theycould not make a firm statement concerningits relationship to other acanthopterygians.They stated only that it was the sister groupof the Percomorpha, a group that is itself notdefinable as monophyletic. To hypothesizethe polarity of particular characters withinthe Atherinomorpha, therefore, particulargroups of percomorphs may be chosen foroutgroup comparison. The holocentrid be-ryciforms, which in some characters areprimitive and in others derived relative toatherinomorphs, comprise such a group cho-sen to hypothesize polarity of characters ofthe upper jaw and ethmoid region.

Primitively for acanthopterygian fishes, asin the holocentrid Holocentrus rufus (fig. 7A),the ethmoid region consists of a well-devel-

ascending process

articular process

rostral cartiagev lalveolar process

autopalatine-- - maxilla

vomer A eehodv pramaxilla

ascending process

rostral cartilaged/ articular process

alveolar process_

autopalatine-

vomer mesethmoid mala

_maxil Iarostral cartilage-

submaxillary cartilage,thmoid3process

C

FIG. 7. Dorsal view of upper jaw and ethmoidregion ofA. Holocentrus rufus, AMNH 27118SW;B. Bedotia geayi AMNH 28132SW; C. Cratero-cephalus cuneiceps, AMNH 43184SW. Bone stip-pled; cartilage open circles.

oped mesethmoid bone. This is not the casein atherinomorphs (with the exception of theapomorphic Iso) in which the mesethmoid isrepresented by a small wedge or disc of bone(fig. 7B-C) or may be cartilaginous as in someaplocheilichthyine and orestiine killifishes(Parenti, 1981).

Furthermore, acanthopterygian fishesprimitively have distinct ascending and ar-ticular processes of the premaxillae, with theascending processes intimately in contact withthe rostral cartilage, and often wrappedaround the cartilage so that the processes andthe cartilage move together as one unit uponopening and closing of the mouth (fig. 7A)(see Alexander, 1967).

Primitively for atherinomorphs, as in thebedotiid Bedotia geayi (fig. 7B), the rostralcartilage is not as firmly attached to the as-cending processes ofthe premaxillae, and yet

1984 9

AMERICAN MUSEUM NOVITATES

there is what I identify here as a distinct ar-ticular process that comes in contact with themaxilla.

Generally, the ascending is identified as thatprocess of the premaxilla in contact with therostral cartilage or connected to it via liga-ments, and the articular as that process incontact with the maxilla, either directly orvia a connective tissue meniscus. Greenwoodet al. (1966) claimed that atherinomorphs donot have true ascending processes of the pre-maxillae. Alternatively, Alexander (1967)claimed that atherinomorphs have no artic-ular processes of the premaxillae, only as-cending processes. Rosen and Parenti (1981)called the processes in atherinomorphs as-cending. I claim that there is a distinct artic-ular as well as ascending process in someprimitive atherinomorphs and that the artic-ular process is reduced or lost in relativelymore derived members. In most atherino-morphs of table 1, minus the Bedotiidae andthe Melanotaeniidae sensu stricto, there is nodistinct articular process of the premaxilla,only a distinct ascending process, as in Cra-terocephalus cuneiceps (fig. 7C). Further-more, there is less contact between the rostralcartilage and the ascending processes, and thetwo often move rather independently duringjaw movement (Alexander, 1967).

DISCUSSION

Rosen and Parenti (1981, p. 14) suggestedthat the Bedotiidae and Melanotaeniidaesensu stricto, may be primitive relative to allother atherinomorphs (that is, the atherinids,telmatherinids, pseudomugilids sensu stric-to, phallostethids, cyprinodontiforms, andbeloniforms). Characters suggested to sup-port this alignment, although just briefly ex-plained, were conditions of the state of thedorsal fins, pelvics, extent of spine develop-ment, and number of vertebrae.The two characters proposed here, reduc-

tion ofthe articular process on the premaxillaand the further decoupling of the rostral car-tilage from the ascending processes ofthe pre-maxillae, support the hypothesis that all oth-er atherinomorphs are derived relative to thebedotiids and melanotaeniids (minus Pseu-domugil.The two characters proposed here are in

conflict with two proposed by White, Lav-enberg, and McGowen (in press) who claimthe atherinoid fishes can be defined as mono-phyletic by a short preanal length at flexure,and a unique dorsal pigmentation pattern.They propose using the ordinal term Athe-riniformes for the atherinoids, the fishes ofDivision I (table 1). Parsimony does not al-low us to choose between the hypothesis ofWhite, Lavenberg, and McGowen and thatproposed here. However, I feel it is perhapspremature to treat the fishes of Division I asmonophyletic. Much comparative anatomi-cal work needs to be done, particularly withregard to fin spine development, to hypoth-esize the polarity of characters that exhibitdifferent states among the atherinoids. It hasbeen stated repeatedly (e.g., Myers, 1928; Ro-sen, 1964; Rosen and Parenti, 1981) that ath-erinomorph fishes are acanthopterygians inwhich the spinous first dorsal fin is reducedphylogenetically, from the strongly devel-oped spinous dorsal of the bedotiids, to thereduction or loss in phallostethids and nu-merous atherinids, and to its eventual loss(absence) in cyprinodontiforms and beloni-forms. Yet, the ontogenetic sequence of thisreduction and loss, as well as other characterswith which it may be correlated, have neverbeen stated clearly.

Therefore, I propose the following classi-fication of atherinomorphs to reflect some ofthe findings of this paper:

Series AtherinomorphaDivision I

Family AtherinidaeFamily BedotiidaeFamily IsonidaeFamily MelanotaeniidaeFamily Telmatherinidae

Superfamily PhallostethoideaFamily Phallostethidae (includ-

ing Neostethidae)Family Dentatherinidae

Division IIOrder CyprinodontiformesOrder Beloniformes

Patten and Ivantsoff (1983) placed theirnew genus in its own subfamily, the Den-tatherininae; I raise the rank to family. Theorder of the taxa is arbitrary, as in the clas-sification of Rosen and Parenti (1981). Any

10 NO. 2779

PARENTI: PHALLOSTETHID FISHES

classification of atherinomorphs must, in myopinion, reflect the sister group relationshipofthe phallostethids and Dentatherina. I havechosen not to place the Dentatherininae inthe Phallostethidae solely for reasons of tra-dition. However, it is critical to recognize thephallostethids and Dentatherina as belongingto a group distinct from other members ofDivision I. To list all the families withoutthis indication would represent a loss of in-formation in the printed classification. Shouldwe wish to include some of the taeniomem-bradine atherinids in the group includingphallostethids and Dentatherina, they maybe included in the Phallostethidae, Denta-therinidae, or a third family to be placed inthe superfamily Phallostethoidea.On the other hand, I cannot support the

monophyly of the fishes of Division I, anddo not use the ordinal term Atheriniformesfor them.

I tentatively accept the definition of theMelanotaeniidae of Allen (1980) to includePseudomugil and its presumed close relativePopondetta Allen; however, the evidenceherein suggests that Pseudomugil (and per-haps Popondetta) is not closely related to oth-er melanotaeniids, but is rather more closelyrelated to a group that includes phallosteth-ids, Dentatherina, some taeniomembradinesand Telmatherina.The primary goal of the present paper is

the clear definition of the relationship ofphallostethids to other atherinomorphs forthe purpose of carrying out a taxonomic re-vision ofthe included species. The definitionsand relationships of the families not treatedin detail here are currently under study byother workers.

SUMMARY1. The Indo-Australian fresh, brackish, and

occasionally saltwater fishes of the familyPhallostethidae and the western Pacific ma-rine atherinid Dentatherina are hypothesizedto be sister groups that share three derivedcharacters: a paradentary bone, absence ofparietals, and absence of a dorsal (first) post-cleithrum from the pectoral skeleton.

2. A submaxillary bone or cartilage be-tween the maxilla and vomer is present inphallostethids, Dentatherina, the Australiantaeniomembradine atherinids Crateroce-

phalus and Quirichthys, the Australian me-lanotaeniid or pseudomugilid Pseudomugil,and Telmatherina, endemic to Sulawesi.Other taeniomembradines examined, for ex-ample the New World Atherinomorus, havethe primitive state ofa connective tissue me-niscus.

3. An enlarged first proximal of the analradial fin is present in phallostethids, Denta-therina, Craterocephalus, Quirichthys, Pseu-domugil, Telmatherina, and Atherinomorus,the taxa listed in 2, above. It has not beenfound in bedotiids, melanotaeniids, and inother nontaeniomembradine atherinids ex-amined.

4. The atherinoid fishes, those of DivisionI of Rosen and Parenti (1981), are not cur-rently definable as monophyletic.

5. The classification ofatherinomorph fish-es proposed above represents the finding ofthis paper that phallostethids are most closelyrelated to the western Pacific atherinoid Den-tatherina. Other families listed in the clas-sification, such as the Atherinidae, cannotcurrently be defined as monophyletic. Theyare currently under revision by other work-ers.

6. At least some of the taeniomembradineatherinids appear to be most closely relatedto the phallostethids and Dentatherina. Inparticular, the western Australian Cratero-cephalus shares the derived character ofelon-gate premaxillary ascending processes androstral cartilage with these two groups. Othertaeniomembradines most likely share this aswell as other characters with the phallosteth-ids and Dentatherina. A clearer statement ofthe relationship of all taeniomembradinesawaits a revision of the subfamily, as well asthe rest of the Atherinidae.

LITERATURE CITED

Alexander, R. McN.1967. Mechanisms of the jaws of some ath-

eriniform fish. Jour. Zool. London, vol.151, pp. 233-255.

Allen, G. R.1980. A generic classification of the rainbow-

fishes (family Melanotaeniidae). Rec.Western Australian Mus., vol. 8, no. 3,pp. 449-490.

Aurich, H.1937. Die Phallostethiden (Unterordnung

1984 1 1

AMERICAN MUSEUM NOVITATES

Phallostethoidea Myers). Internatl. Rev.Ges. Hydrobiol. Hydrogr., vol. 34, pp.263-286.

Bailey, R. G.1936. The osteology and relationships of the

phallostethoid fishes. Jour. Morph., vol.59, no. 3, pp. 453-483.

Greenwood, P. H., D. E. Rosen, S. H. Weitzman,and G. S. Myers1966. Phyletic studies of teleostean fishes, with

a provisional classification of livingforms. Bull. Amer. Mus. Nat. Hist., vol.131, art. 4, pp. 339-456.

Herre, A. W.1925. Two strange new fishes from Luzon.

Philippine Jour. Sci., vol. 27, no. 4, pp.507-514.

1942. New and little known phallostethids withkeys to the genera and Philippine species.Stanford Ichth. Bull., vol. 2, no. 5, pp.137-156.

Hubbs, C. L.1944. Fin structure and relationships of phal-

lostethid fishes. Copeia, pp. 69-79.Myers, G. S.

1928. The systematic position of the phallo-stethid fishes, with diagnosis of a newgenus from Siam. Amer. Mus. Novi-tates, no. 195, 12 pp.

1935. A new phallostethid fish from Palawan.Proc. Biol. Soc. Washington, vol. 48,pp. 5-6.

Parenti, L. R.1981. A phylogenetic and biogeographic anal-

ysis of cyprinodontiform fishes (Te-leostei, Atherinomorpha). Bull. Amer.Mus. Nat. Hist., vol. 168, art. 4, pp.335-557.

Patten, J. M., and W. Ivantsoff1983. A new genus and species of atherinid

fish, Dentatherina merceri from theWestern Pacific. Japanese Jour. Ichthy.,vol. 29, no. 4, pp. 329-339.

Regan, C. T.1913. Phallostethus dunckeri, a remarkable

new cyprinodont fish from Johore. Ann.Mag. Nat. Hist., vol. 12, pp. 548-555.

1916. The morphology of the cyprinodontfishes of the subfamily Phallostethinae,with description ofa new genus and twonew species. Proc. London Zool. Soc.,26 pp.

Roberts, T. R.197 la. The fishes ofthe Malaysian family Phal-

lostethidae (Atheriniformes). Breviora,no. 374, 27 pp.

197 lb. Osteology ofthe Malaysian phallosteth-oid fish Ceratostethus bicornis, with adiscussion of the evolution of remark-able structural novelties in its jaws andexternal genitalia. Bull. Mus. Comp.Zool., vol. 142, no. 4, pp. 393-418.

Rosen, D. E.1964. The relationships and taxonomic posi-

tion of the halfbeaks, killifishes, silver-sides and their relatives. Bull. Amer.Mus. Nat. Hist., vol. 127, art. 5, pp.217-268.

Rosen, D. E., and L. R. Parenti1981. Relationships of Oryzias, and the groups

of atherinomorph fishes. Amer. Mus.Novitates, no. 2719, 25 pp.

Schultz, L. P.1948. A revision of six subfamilies ofatherine

fishes, with descriptions of new generaand species. Proc. U.S. Natl. Mus., vol.98, no. 3220, 48 pp.

1950. Correction for "A revision of sixsubfamilies of atherine fishes, with de-scriptions of new genera and species."Copeia, no. 2, p. 150.

Smith, H. M.1927. The fish Neostethus in Siam. Science,

vol. 65, pp. 353-355.TeWinkle, L. E.

1939. The internal anatomy of two phallo-stethid fishes. Biol. Bull. Woods Hole,vol. 76, no. 1, pp. 59-69.

White, B. N., R. J. Lavenberg, and G. E. Mc-Gowen

[In press] Ontogeny and Systematics ofthe Ath-eriniformes. In, Ontogeny and system-atics offishes. E. H. Ahlstrom MemorialSymposium, American Society of Ich-thyologists and Herpetologists.

Woltereck, R.1942a. Stufen der Ontogenese und der Evolu-

tion von Kopulationsorganen bei Neo-stethiden (Percesoces, Teleostei). Inter-natl. Rev. Ges. Hydrobiol. Hydrogr.,vol. 42, pp. 253-268.

1942b. Neue Organe, durch postembryonaleUmkonstruction aus Fischflossen en-tstehend. Ibid., vol. 42, pp. 317-355.

12 NO. 2779