OSTEOLOGY OF THE MALAYSIAN PHALLOSTETHOID FISH CfRATOSTETHUS BICORNIS, WITH A DISCUSSION OF THE EVOLUTION OF REMARKABLE STRUCTURAL NOVELTIES IN ITS JAWS AND EXTERNAL GENITALIA^
TYSON R. Roberts-
Abstract. The osteology of the phallostethoid an externalized pelvic bone. Phallostetliidae ap- Ccratostethtis bicornis (Regan) is described and parently arose from Neostcthus. The toxactinium, figured. Comparative observations on osteology of the main externalized bony element in the atherinoids, cyprinodontoids, and other phallo- priapium of Phallostethidae, is derived from the stethoids are also given. Phallostethoids apparently inner pulvinular bone, which is the anteriormost originated from atherinids. The most closely internal liony element in tlie priapium of related atherinids are Taeniomembrasinae. The Neostetliidae. The inner pulvinular bone of Neo- osteological observations tend to support the idea stethus bears a small lateral projection, the that atherinoids and cyprinodontoids are related, pulvinular spine, which may be a rudimentary as in toxactinium. structure of the postulated ])y Rosen ( 1964 ) proposing the The papillary ])one, order Atheriniformes. Atherinifomis exhil^it a intimately associated with the genital pore, is widespread tendency to develop teeth with two relatively simple in Gulaphallinae and Phallo- and three cusps, especially on the pharyngeal stethidae, ])ut in Neostethinae it divides into bones. The trend is pronounced in cyprinodontoids, numerous thin processes each bearing a booklet at exocoetoids, and scomberesocids, practically absent its tip. A comprehensive definition is given for in atherinoids, and completely absent in phallo- the superfamily Phallostethoidea. All taxa used stethoids and belonids. in this paper were proposed l^y previous authors. The Phallostethoidea can be divided into two Oviparous Atheriniformes with internal fertiliza- families, Phallostethidae and Neostethidae. Neo- tion have external genitalia far more complicated stetliidae comprises two subfamilies, Neostethinae than the gonopodium of any of the viviparous and Gulaphallinae. These groups are distinguish- .\theriniformes. The explanation of this difference able on the basis of morphological differences in is sought in terms of selection pressures for and the jaws and external genitalia. The highly pro- against the evolution of liighly complicated ex- ternal in forms with internal fertilization. tractile jaws of Neostethinae ( as exemplified" ])y genitalia Ceratostethus and Neostethtis) are remarkalile in Neotenic characters probably played an important having several pairs of bones without homologues role in the origin of phallostethoids from Atherin- in other fishes. These new bones, invohed mainly idae. in protrusion of the jaws, evolved in soft structures already present in the jaws of atherinids. The INTRODUCTION functional anatomy of the jaws of phallostethoids is discussed. Phallostethidae and Gula- briefly Ichthyologists have marvelled at phallo- phallinae lack the neomorphic jaw bones found in stethoids since thev \\'ere first made known Neostethinae. Two ctenactinia formed from pelvic C. T. Males of fin spines or rays occur only in CeratostetJius. The by Regan (1913; 1916). so-called "second ctenactinium" of Gulaphallus is these delicate little fishes from Southeast Asia can be distinguished in an instant by ^ Published a from tlie Wetmore Colles by grant their strange subcephalic copulatory organ, Fund. - or as designated it. The Museum of Comparative Zoology, Harvard Uni- priapium, Regan have been divided versity, Cambridge, Massachusetts 02138 19 species now known
Bull. Mus. Comp. Zool., 142(4): 393-418, December, 1971 393 394 Bulletin Museum of Comparative Zoology, Vol. 142, No. 4
into two families and ten genera, largely based on differences in the external arma- ture of the priapium (see Herre, 1942). H. M. Smith (1927) made the surprising an- nouncement that phallostethoids he ob- served in Thailand are oviparous. This discovery was confirmed in species from the Philippines by Villadolid and Manacop (1934) and Woltereck (1942a). Despite their outstanding interest, phallostethoids have remained virtually unknown to the
a general zoological public. The present paper gives a relatively complete, illustrated account of the osteol- > Cerotostethu.s ])icornis o ogy of (Regan) (Fig. 1), with special attention to the hitherto undescribed jaws and their func- ,-K a c tional Previous work D anatomy. osteological .cU on Phallostethoidea (with the exception of brief observations by Regan [1913; 1916], skeletal who lacked adequate preparations ) has been almost totally restricted to the J3 O priapium (Bailey, 1936; Aurich, 1937; The is U Woltereck, 1942a, b). priapium derived mainly from pelvic bones and fin with contributions from the anterior- o rays, 3 most ribs and pectoral girdle. The pelvic 'K elements are so excessively modified that their homologies remain unresolved. This copulatory organ is perhaps even more on specialized than the copulatory organs derived from the anal fin in the U cyprin- odontoids Horaichthi/s and Tomeunis. — and o Interestingly enough, Iloraichthys ,^ E Tomeunis are also oviparous, and they exhibit numerous morphological and eco- logical similarities to phallostethoids. Tlie E do not E morphological similarities, however, CO extend to the jaws of Ceratostethtis: al- though basically similar to the jaws of atherinoids, they are even more highly t;r: protractile and possess two major and two 3 minor paired bones found so far in no other o fishes except the closely related Neostethus also in 0) (probably present SolenophaUm and Plectrostethus and possibly Phallo-
0) but absent in PJienacostethus and k_ stethus, Guhphallus). No observations have evei Ceratostethus—Osteology and Structural Novelties • Roberts 395
been made on the feeding movements of toids. The most closely related forms ap- Ceratostethus or any other phallostethoids; pear to be Stenatherina and its Indo-Pacific I have tried to detemiine the main features allies (placed by Schultz, 1948, in the of their functional atherinid anatomy by manipula- subfamily Taeniomembrasinae ) . tion of alizarin-stained specimens macerated If one considers only zoogeographic dis- in potassium hydroxide and cleared in 50 tribution and priapial morphology (which percent glycerine. has been worked out in considerable detail My initial objectives in studying the for almost all of the genera by Bailey [1936] osteology of Ceratostethus were to provide and especially by Aurich [1937]), a rel- information that might lead to a better atively simple picture of phyletic relation- of 1 of Phallo- \\'ithin understanding ) relationships ships Phallostethoidea emerges. stethoidea to other fish groups, and 2) There are two families: Phallostethidae and relationships among phallostethoid genera. Neostethidae. Phallostethidae, presumably I chose Ceratostethus because, judging most highly modified from the primitive from priapial stiiicture, it seemed to repre- or ancestral type of phallostethoid, com- sent a relatively primitive phallostethoid, prise two genera, Phallostethus and Fhena- and I had an abundant supply of fresh costethus, confined to the Malay Peninsula material from Thailand. Regan (1913; 1916) and adjacent parts of the mainland of regarded the phallostethoids as an aberrant Southeast Asia. Neostethidae comprise six subfamily of Cyprinodontidae, and noted to eight genera (Herre, 1942, probably that their osteology was "typically cyprin- went too far in splitting genera) belonging odontid." He did not indicate to which to two subfamilies, Neostethinae and Gula- cyprinodontids they might be most closely phallinae. Gulaphallinae are restricted to related. Subsequent to the discovery by the Philippines. Neostethinae, which in- Herre 1925 that clude ( ) phallostethoids possess Ceratostethus, SoJenophaUus, Plectro- a spiny (al]:)eit minute) first dorsal fin, stethus, and Neostethus, are more widely most followed 1928 distributed. Neostethus ichthyologists Myers ( ) and Ceratostethus in relating them to the Atherinidae. Myers occur in the Philippines, Borneo, and intimated that Atherinidae and Cyprin- both sides of the Malay Peninsula.^ The odontidae might actually be fairly closely priapium of Neostethinae, in which the related, a suggestion generally viewed with only externalized elements are derised from disfavor (cf. Hubbs, 1944) until Rosen pelvic spines and rays, is evidently more (1964) pointed out a large number of primitive than are the priapia of Phallo- stethidae and anatomical, morphological, and reproduc- Gulaphallinae. In phallo- tive characters and trends commoj! to stethid priapia the main externalized atherinoids, cyprinodontoids, and allied element is homologous witli the neostethid forms outer (including Phallostethoidea ) and pulvinular (Bailey, 1936; Roberts, united them in a new order, Atherinifonnes. 1971), which is the anteriormost internal While placing the superfamily Phallo- element in the priapium of Neostethidae. stethoidea in the suborder Atherinoidei Neostethus, alone among Neostediidae, has {ibid.: 261), he suggested that it might be a single spur near the base of its cten- more closely related to cyprinodontoids actinium, resembling the series of spurs on than to atherinoids after all {ibid.: 242). the ctenactinium of PhaUostethus (the During the course of the present study ctenactinium of Phenacostethus is absent or numerous supplemental observations were ^ made on cyprinodontoids and atherinoids. There are two records of indeterminalile phallostethoids: from Sabang, northwestern Su- Phallostethoids are more closely apparently matra (Aurich, 1937: 282-284) and from related to atherinoids than to cyprinodon- Zamboanga (Banjagan, 1966: 46). 396 Bulletin Museum of Comparative Zoology, Vol. 142, No. 4
Phenacostethus greatly reduced). In addition, the outer bone of Neostethus has a small puK'inular Phallo stethus but distinct lateral projection, immediately anterior to the pulvinulus (absent in other Manacopus neostethids), which might be an incipient Plectrostethus Gulaphallus or toxactinium. Solenophallus rudimentary Regan (1916: / as the Neostethus 6, fig. 3) referred to this projection pulvinular spine. It would be of great interest to compare the osteology of Neo- stethus and PhaUostethus. Unfortunately, Phallostethus is represented in collections by only four type specimens in poor con- dition. The priapial characteristics of As previously indicated (Roberts, 1971), Phenacostcthus have been treated in detail I am inclined to believe that the order elsewhere 1971). Both Gulo- (Roberts, Atheriniformes, as conceived by Rosen })luillus and Cerato.stethus have been de- (1964), constitutes a natural assemblage. scribed as two ctenactinia. The having Thus far my researches have not uncovered second ctenactinium of Cerato.stethus, any facts that cast serious doubt on this however, is evidently a modified pelvic concept. Although Rosen (1964: 255) or whereas that of spine ray, Gulaphallus stated that the circumorbital series in is an externalized pelvic bone. In addition, Atherinifonnes is represented only by ill (Tulaphallinae {Gulaphallus and Miro- lacrimal and dermosphenotic (lacrimal phallus) the anterior end of the cten- with separate ventral piece in a few cases), actinium fits into a fleshy sheath in the in some phallostethoids and atherinids anterior end of the priapium, whereas in there is a large, trough-shaped infraorbital Ncostethinae it remains uncovered (see bone immediately beneath the lacrimal, Aurich, 1937: 266, figs. 1 and 2). and in some atherinids {Melanotaenia and The discovery of neomorphic jaw bones Telmatherina) there are two separate, in Ceratostethus and Neostethus adds a troughlike or laminar infraorbitals below new dimension to the above sketch. Their the lacrimal. Rosen (p. 288) found in in these genera and their absence presence melano'taeniids a small, spatulate element in indicate the distinctness of CAilaphallus broadiv and firmlv joined to the ventral the subfamilies Ncostethinae and Gula- surface of the lacrimal, and noted that these and make direct derivation of phallinae two bones in melanotaeniids together re- from Ncostethinae harder to Gulaphallinae semble the single elongate lacrimal of I conceive. Absence of the neomorphic ele- Xenopoecilus. The second infraorbital bone ments is perhaps to be expected in the (considering the lacrimal as the first of minute species Phcnacostethtis; they may infraorbital) in phallostethoids and ather- in Their well be present PhaUo.stethus. inids observed by me is quite separate from absence in Gulaphallus, in which the jaws the lacrimal. A first pharyngobranchial is are distinctly less protractile than in Neo- present in Cerato.stethus, Melanotaenia, andi stethus, Ceratostethus, and Phenacostethus, Allanetta, although Rosen (p. 237) stated! indicates that more diversity exists among that the first pharyngobranchial is lacking' Phallostcthoidea than might otherwise have in Atheriniformes (Melanotaenia indicatedl been thought. as a possible exception). The Atherini- The intergeneric relationships of Phallo- formes are diverse, and it is understandable stcthoidea, as currentlv understood, mav that as additional representatives are be represented in a diagram as follows: studied more thoroughly, definitions will Ceratostethus—Osteology and Structural Novelties • Rohctis 397
have to be revised and expanded. It is un- Prajit Wongrat, and Mrs. Supap Monkol- deniable that certain evolutionarv trends, prasit, of the College of Fisheries, Kasetsart for instance those concerning reproduction, University; by Miss Prachuab Sukcharean seem to be more readily understandable if and Mr. Sopon Chantarat, of the Marine the groups placed in Atheriniformes are Fisheries Station of Songkhla; and by Dr. indeed related. Thus some observations on Vagn Hansen, of the Phuket Marine Bio- the widespread occurrence of bi- and tri- logical Center, has already been acknowl- cuspid teeth in Atherinifonnes, which seem edged (Roberts, 1971): without it the to confirm their relationship, are reported observations and collections of phallo- in this paper even though the dentition of stethoids that provide the basis for my phallostethoids is entirely conical. My own studies could not have been made. My feeling, however, is that the concept of fieldwork was approved by the National Atherinifonnes should be verified by Research Council of Thailand. thorough osteological comparisons among Miss Deborah White and Dr. Elizabeth the primitive or annectant representatives Deichmann kindly translated the papers by of the suborders assigned to it. Aurich and Woltereck. Much of the in- Villadolid and Manacop (1934) analyzed spiration for the discussion section of this stomach contents throughout one year in paper came from an essay on "The Emer- wild-caught GiilaphalJiis mirabilis, and gence of Evolutionary Novelties," presented observed courtship, copulation, and egg- by Prof. Ernst Mayr at the Centennial laying of this species in aquaria. These Celebration of Darwin's Origin of Species aspects of behavior have not been observed in London (Mayr, 1960). Acknowledg- in any other phallostethoids. No new spe- ment is gratefully made to Prof. George S. cies of Neostethidae have been described Myers. Prof. M>'ers acquainted me with since the genera and species were reviewed phallostethoids when I was an undergradu- by Herre (1942). I reviewed the Phallo- ate at Stanford University, and we have stethidae, described a new species of discussed them on numerous occasions. Phenocostethus from the Indian Ocean coast of Thailand, and remarked upon OSTEOLOGY OF CERATOSTETHUS the ecology of Phallostethidae and Neo- (Figures 2-12) stethidae in Thailand. I also discussed the The account is selective advantage of internal fertilization following osteological based on 14 in Phallostethoidea as well as trends in the specimens (ten males and four females) 17.0-26.8 mm in standard reproductive biology of atheriniforms that of a series collected in a might be conducive to the evolution of length, part large roadside ditch shaded by nipa about internal fertiUzation (Roberts, 1971). The palm two miles from Tha Chalap on the road to discussion section of the present paper Chantaburi Citv, Chantaburi Province, deals with evolutionary topics such as the Tliailand (MCZ' 47304). The only osteo- origin of the neomorphic jaw bones found logical differences detected between males in Ceratostethus and Neostethus, neotenic and females involve genitalia, and the characters in phallostethoids, and the anteriormost ribs and the cleithrum nature of selection for and pressures against in to ( modified males form a suspensorium the development of highly complicated for the priapium). Tlie figures are based external in atherinifomis. genitalia on four specimens, 24.8-26.8 mm. A few observations are included on the ACKNOWLEDGMENTS osteology of Phenacostethus, Neostethus, and CaiIq- The help rendered during my stay in phalhis. Comments and comparati\e obser- Thailand by Dean Jinda Thiemeedh, Mr. vations on cyprinodontoids and atherinoids 398 Bulletin Museum of Comparative Zoology, Vol. 142, No. 4
sphenotic pterotic frontal maxillary pararostral rostral ^ submaxillary OS max. -mand. palatine dentary
exoccipital
first vertebr;
supraoccipita
epiotic
dermosphenotic
Figure 2. Ceratosfethui bicornis. Dorsal view of skull and first vertebra; jaws sligfitly protruded. are inserted in numerous places. I have Foster) and atherinoids (including alizarii examined alizarin preparations of a con- preparations in the Department of Ichthyol siderable variety of cyprinodonts (a num- ogy of the American Museum of Natura of ber them kindly provided by Neal History). The osteology of atherinids ii
dermosphenotic pterosphenoid sphenotic pterotic parasphenoid
prefrontal
nasal exoccipital
basioccipital
first vertebra
posttemporal
supracleithrum prtmaxillary prootic hyomandibular maxillary
Figure 3. Cero/osfefhus b/corn/s. Ventral view of cranium and first vertebra; witfi portions of jaws, suspensorium, one sfioulder girdle of right side. Ceratostethus—Osteology and Structural Novelties • Roberts 399
dermosplienotic pterosphenoid supraoccipital maxillary frontal
rostral i „;isal epiotic exoccipital lacrimal
paradentary
dentary
premaxillary
OS max. -mand.
maxillary
articular
infraorbital angular quadrate
interopercle
preopercle subopercle
symplectic opercle interhyal
Figure 4. Ceratostethus bicornis. Lateral view of si Figure 5. Cera/os/efhus bicornis. Dorsal view of anterior part of skull witfi jaws greatly protracted. Compare Fig. 6C. stethus are perhaps even more specialized, (their length almost equal to a third of for they include a number of hitherto un- cranial length). reported ossified elements, some of which Posteriorly directed, broad-based, roundec 1 are without knowai osseous homologues in processes ("articular processes") arising atherinoids or cyprinodontoids or even at midlength of prcniaxillaries; sucl analogues in other teleosts. These evidently processes (either rounded or pointed) neomorphic structures, apparently present characteristic of many atherinoids, are usu only in phallostethoids, are functionally in- ally (invariably ?) absent in cyprin volved in the extreme protrusibility of the odontoids. Several atherinoids have slender jaws (evidently including closure of the elongate ascending processes as in Cerate jaws in protruded position). The new stcthiis; ascending processes in cyprin- bones, discussed below, are paired ele- odontoids usually (invariably ?) relativel) ments, named as follows (in order of oc- short, broad-based, and wide. Maxillar) currence : with curved anteroposteriorly ) paradentary, well-developed, strongly in-j maxillomandibular, pararostral, and sub- ternal hooks. As in atherinoids (but not iri maxillary bones. cyprinodontoids) the maxillary has i\ Bones in upper jaw: paired premaxil- laminar projection dorsal to the interna' to th( laries, maxillomandibularies, maxillaries, hook, part of which projects dorsally pararostrals, and submaxillaries; in lower ascending premaxillary processes (wherj mouth is Lower limb jaw: dentaries, paradentaries, articulars, abducted). (distal) of slender. A angulars, and coronomeckelians. Premaxil- maxillary uniformly swelling (cranial condyle) at junction of interna laries and dentaries with a single row of hook with the main body of maxillar) conical teeth; other jaw bones and palate articular surface for provides submaxillar}^ toothless. Proximal third of premaxillary bone. Maxillary without ligamentous ati with eight to ten relatively small teeth, tachment to palatine. In atherinoids and ii middle third with about ten teeth, enlarged some (all ?) cyprinodontoids maxillary distal third toothless. Dentaries with three articulates directly to prevomer or to the to five very small (vestigial ?) teeth near ethmoid by a submaxillary meniscus. Ii symphysis; otherwise toothless. Ascending atherinoids and cyprinodontoids there i premaxillary processes slender, elongate usually (always ?) a direct ligamentou Ceratostethus—Osteology and Structural Novelties • Roberts 401 connection between palatine and maxillary In the following account I attempt to at or near cranial condyle. Head of maxil- give an idea of functional (mechanical) re- lary bound in place, probably by liga- lations of the jaw bones even though their mentous connections with nasal, lacrimal, movement was studied only in macerated and perhaps mesethmoid which, however, specimens. I have assumed that in live permit swinging and screwing movements, specimens the mouth is opened mainly by Lower (free) limb of maxillary unifonnly depression of the lower jaw; that return of slender. Internal hooks attached by dense the jaws to normal "resting" position in- connective tissue extending ventrally to volves two distinct, consecutive stages, rostral bone. Rostral tissue 1 of it is connective namely ) closing the mouth while (ethmopremaxillary ligaments ?) with a still protracted, accomplished by retraction pair of very small pararostral bones dorso- on the lower limb of the maxillary, drawing lateral to anterior part of rostral bone. Near free ends and premaxillaries back (while tip of lower limb of premaxillary, in the position of lower jaw is unaffected), and maxillomandibulary ligament lies the small, 2) inoving the entire assemblage back into rounded, dorsoventrally compressed maxil- the nearly vertical position it occupied be- lomandibulary bone. The maxillomandibu- fore, probably accomplished by retraction lary ligament connects the ends of the lower on the coronoid process of the dentary and or free limbs of the premaxillary and maxil- on the "articular process" of the premaxil- to the is From the work of Alexander lary dentary. The mouth opened lary. ( 1966, b it seems that in fishes when movement originated by depressing 1967a, ) likely many the lower jaw is transferred by this liga- with protractile jaws, the mouth is closed ment to the upper jaw. The attachment of while the jaws are protracted. It is the ligament to the dentary extends from unclear whether the mouth can be closed middle of dentary to symphysis of lower when it is protracted in atherinoids, but it in jaws. Near symphysis main body of liga- can be some cyprinodontoids, as well as in and Alex- ment very tough and round in cross sec- cyprinoids, acanthopterygians ( tion. In shape and extent of attachment the ander, 1967a, b). ligament resembles strongly that in certain Wlien mouth is in resting position, as- atherinoids, particularly members of the cending premaxillary processes fit snugly subfamily Taeniomembrasinae (for defini- into rostral concavity, their distal (free) tion and discussion of this taxon see ends lying between depressed anterioiTnost Schultz, 1948). In other atherinids and in portion of frontal bones. When jaws are as cyprinodontoids the attachment of the liga- maximally protracted ( estimated by ment does not extend much anterior to the manipulation of KOH-macerated speci- middle of the and is out or dentary, spread mens) premaxillaries have moved forward sheetlike. The anterior of the part liga- ^bout 125 percent of their length, or about ment in Ceratostethus differs strikingly 35 percent of the headlength. • • rrom that in Taeniomembrasinae in that \t ^^ c . n .1 i r . 1 ,. 1 . , . . . . Maxillanes of typically athermoid tonn, it is lymg entirely witnm a very conspicuous 1 i i r 1 ..1 n i 1 • • ^„ r r- . r- . .1 excluded from gape and with well-de- and .11 the tunctionally significant ossification, 1 i 1 ^'^1^?^^ internal hooks. With mouth com- paradentary bone. Distal end of paraden- closed, slender ventral tary lies free in the ligament; its proximal P^^tely straight, limb of maxiUaries Hes at an end foi-ms the ball of a ball and socket joint ( distal) angle of about 10 to a with the dentary at anterionnost point of degrees posterior vertical; with mouth ventral limb attachment of ligament, i.e., near symphysis fully protracted, of at an of about 45 de- of lower jaw. Dentary with a high coronoid maxillary angle process, as in many atherinids; coronoid grees anterior to a vertical (see Fig. 6, A-C). processes usually absent in cyprinodontoids. Screw movement is slight or nonexistent 402 BuUetin Museum of Comparative Zoology, Vol. 142, No. 4 cyprinodontids examined by me failed to exhibit this movement. The screw move- ment evidently also causes dilation of the lower limbs of the premaxillary, causing the lateral series of enlarged premaxillary teeth (which are directed sideways when the is in to mouth resting position ) point straight ahead. As stated by Alexander (1967b: 241) protrusion of the jaws must be accompanied by screw movement of the maxillary, sc that the internal hooks and consequenth rostral bone are moved anteriorly. In Ceratostethiis, however, at least some pro- trusion occurs before the screw movemeni begins. Furthermore, it is possible to causf considerably more protrusion (about 21 percent more) after the rostral has ceasec to move forward by simply continuing t( depress the lower jaw. If this is done, the premaxillary tilts upwards posteriorly sc that the mouth opening is slightly down- wards instead of vertical. Wliile it i; doubtful that the jaws are normally pro truded so far, this observation indicate; that depression of the jaws can act un accompanied by screw movements of tht maxillary to bring the premaxillaries for ward. In cyprinodontoids the screw movemen - of the maxillary causes not only anterio: movement of the internal hooks, but alsc Figure 6. Cerofosfefhus bicornh. Lateral view of jaws in lateral movement (Alexander, 1967b: 239 varying stages of protrusion (for explanation see text). fig. 5). I have observed this in mam cyprinodontoids. In Ftindiihis the rostra until after more than half of the swinging cartilage is Y-shaped, with the interna movement is completed and the premaxil- hooks attached finnly to the ends of th( laries are considerably protracted. Up to arms of the Y; as the internal hooks mov(! this of the is point movement premaxillaries laterally, the amis of the Y spread apart to that of the rostral; from this equal point In addition, in many (but not all) cyprin, the move rel- on, however, premaxillaries odontoids, the internal hooks are so fimib atively further forward than do the rostrals. bound to the ascending premaxillary tha^ Thus not only does the rostral slide forward when the hooks move laterally, the ascend relative to the cranium, but the premaxil- ing processes diverge posteriorly (see Alexi lary processes slide forward relative to the ander, 1967b, fig. 6e of Futululus). h\ rostral; Alexander ( 1967b ) found a similar Ceratostethus there is either no latera movement present in Atherina but absent lateral movemen in the cyprinodontids Aplocheilus and movement or very slight Fundulus. Alizarin preparations of various of the internal hooks; the ascending pre Ceratostethus—Osteology and Structural Novelties • Roberts 403 maxillary processes never diverge from each other. Submaxillary process of maxillary articu- lated to prevomer by submaxillary bone (developed in place of the usual submaxil- lary meniscus). Submaxillary bone forms ja meniscus with submaxillary process of .maxillary and with anterior end of pre- v^omer. When mouth is closed, submaxillary bone lies with its upper end at an angle of ibout 15 or 20 degrees anterior to a vertical; when mouth is protracted, rotation Figure 7. Ceratostethus bicornis. Hypothetic closure of Df end of sub- maxillary depresses upper protracted jaws (for explanation see text). naxillary, which finally lies at an angle of ibout 45 degrees from vertical with mouth Ceratostethus are two bones: the ully protracted. The curved, ventrally di- paired rected internal hooks of the maxillaries are small maxillomandibulary bones, lying free in that of the igamentously attached to a median ossified part ligament connecting and and the elon- element or rostral bone lying ventrally to premaxillary maxillary, of gate paradentaries, which are attached by iscending processes premaxillaries. ( This a sort of ball and socket to the slement evidently is represented by a joint dentaries near ramus of lower Move- •QStral cartilage in atherinoids; it is similar jaw. ment of the bones is n position, but not in movement, to the maxillomandibulary dependent on movement of the premaxil- 3yprinoid kinethmoid. ) With mouth closed, their with to lower interior tip of rostral is horizontal with the lary; position respect limb of remains interior tips of the nasal bones and an- premaxillary unchanged. bone erior margins of maxillaries; when mouth Maxillomandibulary forming conspic- uous rounded in of s fully protracted, it moves forward ap- prominence angle rictus of when mouth proximately 50 percent of its length. Above jaws fully opened with :he rostral, in thickened connective tissue (Fig. 6c). Paradentaries, longitudinal axis to that of lie attaching it to ascending premaxillary proc- parallel dentary, snugly against dentaries in base of esses and internal hooks of maxillary, lies a grooves; groove fonned dorsal surface of canal for lair of small bones or pararostrals. Tliese by bony mandibular of latero- ire evidently equivalent to accessory rostral segment cephalic sensory system. With mouth about half :;artilages of some atherinoids. Movement remain )f pararostrals equal to movement of open, paradentaries snugly against as their ostral. dentary; mouth opens further, distal end and Lower jaws similar in form to those in (free) swings sideways up- wards until, with mouth itherinoids: dentaries with very large cor- fully protracted, they are at right angles to )nary process, and with ventrally opened dentary (Fig. When mouth closes in lange forming a trough for mandib- 6c). protracted position (closing accomplished in macerated ilar segment of cephalic laterosensory specimens back on lower limb iystem. Maxillomandibulary ligament, link- by pushing of the return ng tips of the ventral limb of maxillary and maxillary), paradentaries to lie flush with side of while ,)remaxillary to dentary, is similar to that dentaries, the lower limb of the is In atherinoids (but not to that in cyprin- only premaxillary )dontoids in its retracted; the ) that connection to dentary ascending premaxillary proc- extends anteriorly to ramus of lower jaws, esses remaining as far or almost as far n the maxillomandibulary ligaments of forward as they are when mouth is fully 404 BuUetin Museum of Comparative Zoology, Vol. 142, No. 4 opened (Fig. 7). Tims closure is ac- in the neostethid Guhphalhis in which the almost complished entirely by movement upper jaws do not protract very far, the of the upper jaw. The lower jaw remains mandible must be depressed further before the depressed, lower jaw bones, excepting the upper jaw moves fonvard at all than the little or paradentary, undergoing no when the upper jaw is fully protracted in movement. Ceratostethm. Second, the paradentaries The paradentaries evidently are involved do not rotate outward until the upper jaw in increased of the is protrusibility upper jaw. considerably protracted, and only slightly In in \\hich the specimens maxillomandibu- more forward movement of the upper jaw lary ligaments have been cut immediately is involved for the paradentaries to rotate to the posterior distal end of the para- as far outward as in Figure 5. It seems tc dentaries, depression of the lower jaw still me highly likely that the paradentaries dc causes premaxillary protraction, but the in fact rotate this much. Tliird, when one prcmaxillaries do not project as far an- depresses the lower jaw of a macerated and teriorly they project upwards (with specimen, the entire jaw assemblage move^ leading margin considerably elevated) in- readily and smoothly as far forward as ir stead of straight forward. The paradentaries Figure 5. In particular, the motion of the also help spread the gape sidewards as the maxillary is unencumbered. This contrast.' mouth is opened. with GuIaphoUus and with various ather Comments. inoids in Two objections that might which movement of the maxillar) be raised to this description of jaw function is relatively restricted, are 1 whether it is ) normal for the jaws to Whether closure of the jaws occurs ir be so the greatly protracted, and 2) whether manner indicated is much less sure. II closure actually occurs in the manner sug- seems likely that the jaws are closed ir There gested. are some fishes (including protracted position. Granted this happens Morwcirrlms among Nandidae [Liem, perhaps stages 1 and 2 envisioned in re- 1970], Epibiihis- among Labridae, several turning the jaws to resting position act ir with concert atherinoids) protractile upper jaws rather than consecutively. It ma>! and elongate ascending premaxillary proc- be that the mouth never closes with the esses in which the ascending premaxillary upper jaw fully protracted (the positior are indicated in processes advanced no more than a Fig. 7). It should be noted; fraction half in that the (a Epibitlns: about a third however, jaws of macerated speci- or less in Monoc/r;7jt/5 and the mens are thi5 atherinoids ) readily manipulated into of their and length when the jaws are fully position, that from this position the> In can be protracted. Monocirrhus and Epihulus readily manipulated back to resting which the (in lower jaws as well as the position, i.e., closed and completely re- are tracted. upper protractile) the ascending pre- Liem maxillary— processes are exceedingly elon- (1970: 106), considering fishes gate as long or longer than the total ^^ith protractile jaws, stated that the salieni cranial features of length. The ascending premaxillary feeding (and respiratory) of movements are processes Ceratostethm are relatively 1) opening and closing ol short compared to those of Monocirrhus the jaws, 2) of the protrusion jaws, 3)| and Epihidus: their length is about one- volume and pressure variations in the' third of cranial length. Three facts indicate buccopharyngeal cavity, 4) abduction and' that the upper jaw is protracted as far or adduction of the opercular apparatus, and' nearly as far forward as in shown Figure 5. 5) complex movements of the gill arches ^ irst, protraction of the upper jaws this far He pointed out that these functions an is achieved by only moderate depression of mutually interdependent and should b( the lower In various jaw. atherinoids and considered in conjunction. Assessment ol 1 Ceratostethus—Osteology and Structural Novelties • Roberts 405 the movements of the opercular apparatus epibranchial i;ill rakers and gill arches and their relationships in liypobrunchial Cerotostethti.s has not been since attempted, basibranchial "natural" movements of these parts are less glossohyal ceratobranchial readily achieved (or achieved with less confidence) by manipulation of macerated interhyal specimens. The main, or at least a major, advantage of protractile jaws may lie in increasing the buccal component of the liypohyal buccopharyngeal pumping mechanism; one ceratohyal of the advantages in closing the mouth with the lies premaxillaries protracted probably brancliioste't^al rays in increasing the volume of water that can be sucked into the mouth without being Figure 8. Ceratostethu% bicornis. Lateral view of hyoid forced back out as the mouth closes (see arch, urohyal, and first brancfiial arcfi. Hyoid and branchial more detailed discussion in Alexander, arches slightly separated. 1967a: 59-62). It seems likely that these factors operate in Ceratostethus. laterosensoiy canal. Dermosphenotic simi- Suspensorhim (Fig. 4). Bones in larly troughlike, its attachment anterior to suspensorivim : palatine, ectopterygoid, sphenotic (attachment of dermosphenotic entopterygoid, quadrate, symplectic, and posterior to sphenotic in Horaichthys and hyomandibular. Anterodorsal part of pala- Onjzias). Preopercle with a ventrally open tine slender, tubular, apparently connected flange fonning a trough for a segment of by ligaments to prefrontal, but free of at- cephalic laterosensory canal. Opercle with- tachment to nasal, lacrimal, and maxillary. out spiny projections, its posterior margin Movement of fails to induce move- jaws in in all concave as some atherinids; many ( ment in palatine. Anterior margin of ?) cyx^rinodontoids posterior margin of suspensorium (formed by palatine, ecto- opercle broadly rounded. Posterior margin pterygoid and quadrate) almost vertically of gill cover formed by opercle, not by inclined. Ventral portion of hyomandibular subopercle. Anterodorsal comer of opercle with processes contacting and symplectic where it attaches to hyomandibular rel- preopercle. Symplectic with a ventrally atively unmodified, without a strongly directed laminar portion. reinforced socket to receive hyomandibular Facial bones (Figs. 2-4). Nasal well process. Interopercle and subopercle developed, almost as long as orbit, a dor- weakly ossified. Anterodorsal projection of sally projecting flange on its medial margin subopercle weakly developed; posteriorly immediately lateral to depressed anterior- subopercle projects considerably beyond most portion of frontal, and a ventrally ventral margin of opercle. directed flange at its anterior tip that is Hyoid and gill arches (Figs. 8 and intimately associated with anterodorsal 9). Hyoid arch of typical atherinifonn comer of lacrimal bone. Circumorbital structure, with a single liypohyal; five series comprised of lacrimal, a single infra- branchiostegal rays on each side, first four orbital directly beneath it, and denno- branchiostegal rays attached to ceratohyal, sphenotic. Lacrimal a large lamellar plate last one to epihyal; dorsal half of joint with a flange in its dorsoposterior comer between and forming a gutter for a segment of cephalic ceratohyal epihyal strongly with a concave antero- laterosensory canal. Ventral to lacrimal a ossified; ceratohyal Gill arches with three single, concave infraorbital, forming a ventral margin. bony trough for a segment of cephalic median basibranchials, four pairs of h\^o- 406 BuUetin Museum of Compnrotwc Zoologij, Vol. 142, No. 4 glossohyal postteniporal hypohyal distal ceratohyal radials hypobranchials epihyal interhyal basibranchials ccratobranchials epibranchials Figure 10. Cera/ostethus bicornis. Lateral view of pectorc girdle. inlrapharyngeal moderately expanded where it meeft Figure 9. Ceratostethus bicornis. Dorsal view of hyoid scapular to form scapular foramen, with arch (minus branchiostegal rays) and branchial arches. Up- out an expanded portion extending poste permost branchial elements of right side removed. riorly to scapular. In mature males anterio limb of cleithrum elongated anteriorly, t branchials, ccratobranchials, epibranchials, almost double relative length of cleithrur in immatures and pharyngobranchials, and a pair of and females, approximatin; , infrapharyngeals. Third epibranchial T- pulvinular bone of priapium. Cleithrum o shaped when viewed from side (figured one side elongated slightly more than tha of other side only in dorsal view). Second, third, and and ligamentously attached t fourth pharyngobranchials and anterior half puK'inular; at point of attachment sHghtl of infrapharyngeals covered with moder- expanded to form a concave lamella fittin; over a or in ately sized, irregularly arranged, conical snugly convexity bump dorss surface of teeth. Infrapharyngeals separate. Gill rak- pulvinular; anterior tip of op cleithrum ers only present on leading margin of first posite ending in a fine poinl Relative of anterior gill arch, 12 or 13 rakers on each side. First development prolon of cleithra epibranchial bearing a single smaller raker, gation variable. In some sped mens of cleithrum t i.e., all rakers except this one are borne tip unattached falls far short of in other on lower half of gill arch. Anteriormost pulvinular it, spea mens in cleithrum t eight gill rakers on lower limb with distal equal length attached of cleithrum ends directed laterally and posteriormost pulvinular. Laterality attachei to four or five rakers with distal ends directed pulvinular independent of laterality c increasingly medially, arrangement of these priapium itself. Apart from this differenc in uppermost rakers thus simulating nonnal development of the cleithrum, the pec arrangement of rakers on upper limb of toral girdle is similar in males and female; Anterior arch found in many fishes. end of coracoid not prolonged i males. Posterior end of Pectoral fin (Fig. 10). Posttemporal coracoid fused wit of forked; upper fork tightly bound to epiotic, ventrally expanded portion scapular, an^ not lower fork short and not reaching base of associated with radials of pectoral fii cranium. Supracleithiiim (as in other Postcleithra absent. Proximal radials two i atheriniforms a small lowermost ) scalelike element number, considerably enlargec completely interposed between posttempo- their proximal ends fitting snugly into ral and cleithrum. Clcithiiun slender, onlv concavity in posterior border of scapula; Ceratostethus—Osteology and Structural Novelties • Roberts 407 fa/corn/s. Axial skeleton. igure 11. Cerofosfefhus \ series of about four small distal radials. interposed between haemal spines of \n unpaired splint attached to proximal antepenultimate and penultimate vertebrae iorsal surface of uppermost pectoral ray. (such an element present in Dermogenijs, Pectoral rays 10 or 11. Onjzias, and Xenopoecihis: see Rosen, 1964, Median fins (Fig. 11). First dorsal fin fig. 21). Caudal fin with 7 upper and 7 tvith two very short, slender, refractile or 8 lower procurrent rays, and 7 upper ipines; second spine even shorter and slen- and 7 lower principal rays, the outermost derer than first, attached to a pterygium principal ray in each lobe unbranched. :onsisting of a single element. Second Posterior margins of hypurals straight and iorsal with five segmented rays. First ray vertically aligned. imbranched; last ray divided to its base, Vertebral column (Fig. 11). Vertebrae interior division branched, posterior divi- 35 or 36 excluding hy^Dural; precaudal plus sion simple. Second dorsal fin pterygials caudal either 18 + 17, 17 + 18 or 18 + 18. 3ach consisting of a single element, with Distal ends of all except a few of the slender proximal and expanded distal por- anteriormost and posteriormost ribs (which tions. Anal fin with 14 or 15 elements, first are associated with anterionnost pterygial element simple, short, and unsegmented, elements of anal fin) meet opposite mem- second long and segmented; third through bers in ventral midline of body. In females, penultimate elements branched, segmented anteriormost ribs are borne on third verte- rays; last element divided to base, both bra: first pair short and slender, without divisions simple. Form of anal fin pteryg- modified parapophyses; first three or four ials as in second dorsal fin, except for pairs of ribs gradually increasing in length. enlarged anterionnost pterygial, which is In males, first ribs borne on fourth vertebra. prolonged anteriorly in ventral midline for These ribs greatly expanded, their distal a distance equal to eye diameter in front ends entering priapium just anterior to vertebra of anal fin origin. Caudal skeleton (Fig. pulvinulus; parapophyses of fourth 11) with two hypurals, both fused to greatly enlarged, projecting anteriorly be- hypural centrum. No epineural or urostyle. neath \ertebral column to a point below Two slender epurals. Penultimate vertebra first and second vertebrae (Fig. 11). First with well-developed haemal spine but three vertebrae without parapophyses. Ribs neural spine reduced to expanded basal of fifth vertebra short and slender, their portion only. Neural and haemal spines of parapophyses unenlarged and posteriorly vertebrae preceding penultimate vertebra directed. Ribs of sixth vertebra of full equally slender and elongate. No accessory length, reaching ventral midline of abdo- haemal spine or separate slender element men. Haemal arches not expanded (swim- 408 BuUcfin Museum of Comparative Zoology, Vol. 142, No. 4 Figure 12. Cerafosfefhus bicorn/s. Laferal view of shoulder girdle, urohyal, and priapial skeleton of dextrol male; ctenot tinio abducted. bladder fitting snugly against posterior wall from this sac, providing its posterior mar of abdominal cavity, without posterior pro- gin with a "fringe" or "comb." jection into haemal arches). Intermuscular As in Solenophalhis, the main element bones absent. First four vertebrae with in the axial skeleton of the priapium ar( distal ends of neural spines longitudinally the two pelvic bones. The proctal pclvi< expanded. Supraneurals absent. bone (i.e., the one opposite the ctenactina" Priapium (Fig. 12). The priapial skeleton is considerably enlarged and acts as th( of Ceratosteihus is very similar to that of suspensorimn of the entire priapium. It i: Solenophalhis (Aurich, 1937) and Neo- suspended anteriorly by the outer pulvin stethus. The first and second ctenactinia of ular and inner pulvinular ("Pulvinulus tun Ceratostethtis evidently are homologous, trager" of Aurich), which are in respectively, with the ctenactinium and suspended by the attachment of the pulvin short, hooklike "Priapklaue" of Soleno- ulus to the cleithrum and perhaps also th{ phalhis. The anterior bundle of tliree seg- urohyal. Tlie inner pulvinular is absent ir mented branched rays and posterior bundle Giihphalhis and Mirophallus (of. Bailey sus of three segmented branched rays also are 1936; Aurich, 1937). Posteriorly it is present. These rays lie in the wall of a sac pended from the vertebral column by it: that becomes filled with spennatophores. ligamentous attachment to the antcpleura Tlie branched ray tips protrude slightly cartilage and anterionnost modified rib; Ceratostethus—Osteology and Structural Novelties • Roberts 409 (see Aurich, 1937, fig. 3 of SolenophaUus) . spine-bearing process of the priapium and rhe proximal ends of rays in posterior ray in the wing-like margin of the cten- bundle firmly articulated to side of pos- actinium." The ctenactinium has a broad teriormost part of proctal pelvic bone. membranous margin along the lower side \proctal pelvic bone movably articulated of its proximal half. At the base of the to proctal pelvic bone by a series of ctenactinium is a flat fleshy process armed complicated bones, the morphology of on its upper and posterior border with a vvhich is difficult to make out in intact row of nine or ten short sharp recurved Driapia (see also Aurich, 1937). Two spines and on its anterior border by two :-tenactinia movably articulated to aproctal longer spines directed forward (Myers, Delvic bone. Lying between anterior end 1935: 5-6). The flat fleshy process is )f aproctal pelvic bone and accessory presumably homologous with the "ventral xilvinular is a small bone, the pulvinular Penisbone" (= papiilaiy bone ?) of Cerato- >sselet, evidently homologous with the stethus and Neostethus. Tlie only known pulvinular cartilage in Solenophalhis. At specimens of Plectrostethus are 12 type he posterior end of the priapium, in the specimens of P. palawanensis Myers (U. S. ixis of the ctenactinia, is a series of ex- National Museum nos. 93421-93423). pressively modified bones surrounding or partly surrounding the genital pore. These CHARACTERIZATION OF THE elements represent the "dorsal," "middle," SUPERFAMILY PHALLOSTETHOIDEA md "ventral penisbones" of Aurich. In This characterization of Phallostethoidea 'Jeratostethiis and Neostethus the "ventral is based on a survey of the literature and penisbone" gives rise to a large number of my own observations on Phenacostethus straight, thin bony projections, constituting (both species), Ceratostethus hicornis, 1 sort of flap over the genital pore. Many, Neostethus siamensis (probably equals N. f not all, of these projections are doubled lankesteri) and GulaphaUus mirahilis. My back at the tip, and end as a booklet. In observations have been relatively complete Neostethtis the projections, about 80 in only for Ceratostethus. Many, if not most, lumber, are parallel to one another. In of the characters cited below can be found Ceratostethus the projections are much scattered in the literature; virtually all of nore numerous, some are considerably en- these characters have been verified by me, larged, and at each end of the genital pore and erroneous statements in the literature 1 large number of the projections are rectified (e.g., concerning the pectoral oriented away from the main body of pro- girdle and first pleural ribs). Tlie char- jections. The "ventral penisbone" of Solenp- acters are presented in the following order: nhaUus ctenophortis is similarly modified, 1. general body features, size, habitat; [t evidently is homologous with the greatly 2. reproduction; 3. sensory organs; 4. squa- enlarged "ventraler Penisknochen" of Gula- mation; 5. fins; 6. skull and visceral arches; ohalhis (called papillary by Bailey, 1936), 7. dentition; 8. 9. vertebral vvith the slender "ventraler Penisknochen" pectoral girdle; column (including ribs and intramuscular >f MirophoUus, and perhaps with the en- bones . larged papillary bone in the base of the ) 1. translucent, com- oenis in Phenacostethiis. In GulaphaUus, Largely bilaterally and or MirophaUus, and Fhenacos-tethus, however, pressed moderately elongate, tiny atherinifonn adults from "iie parallel projections and booklets are very small, fishes, absent. 14 {Phenacostethus smithi) to 37 mm in standard Plectrostethus Myers (1935) is said to be (GulaphaUus eximus) length closely related to Neostethus, "from which (females usually slightly larger than t differs trenchantly in the presence of the males). Inhabiting fresh to brackish, usu- 410 Bulletin Museum of Comporativc Zoology, Vol. 142, No. 4 ally turbid, coastal streams of mainland and and perhaps absent in Fhallostethus), con- insular Southeast Asia, usually within sisting of one or two tiny, weak (nonerectile range of the tides. ?) spines movably articulated to an elon- 2. Oviparous; gonad single; fertilization gate pterygium. The second spine, when internal, males with a clasping and intro- present, usually even thinner and shorter mittant organ, die priapium, derived than the first. First dorsal completely mainly from the pelvic fins; pelvics evanes- separated from second dorsal, but much cent, vestigial or absent in females. In closer to it than in atherinids, its origin males anus opens on side of priapium op- posterior to a vertical through anal fin posite genital pore, in females anal and origin. Second dorsal with one or twc genital openings close together on throat. simple, unbranched, segmented rays (nc Eggs demersal, chorionated, with adhesive spines), and three to nine branched rays filament verified in Phallostethidae of second dorsal well behind ana' (not ) ; Origin sperm transferred in large masses held to- fin origin, or even posterior to anal fin gether by a mucoid substance (sper- Anal fin with a short flexible spine, fol- mozeugma), at least in Neostethidae. lowed by an unbranched, segmented ray Newly hatched young are miniatures of and ten to 28 branched rays. Initial pteryg- adults, probably immediately capable of iophore of anal fin an enlarged backward.' active feeding and swimming. L-shaped bone (Fig. 11). A translucent 3. Eyes large, laterally directed, free median, abdominal, membranous keel in- from orbital rim. Nostrils, if present, minute variably present. Caudal fin forked; in at (a single opening which may be a nostril least some species the upper lobe is slightly lies on the side of the snout near the upper longer than the lower lobe and is pointed, jaw in Neostethiis but is apparently absent while the lower lobe is rounded (Fig. 1: in Ceratostethus; olfactorv lamellae are not Roberts, 1971, figs. 2 and 3 of Phenaco- evident tlie stethus caudal 5 7 beneath the skin of snout). A ) ; principal rays from + large pore midway between tip of snout or 6 + 7 to 7 + 8 or 8 + 8. and anterior margin of eye in Neostethidae 6. Upper jaw usually highly protractile. is evidently a pore opening into supra- Mesethmoid a single, irregularly hexagona orbital canal of cephalic laterosensory lamina; intercalar absent; parietals absent; system. Cephalic laterosensory system well infraorbital series represented by lacrimal, developed, supraorbital, preorbital, post- second infraorbital and dermosphenotic; orbital, preopercular, and mandibular dermopalatine absent. Lower arm of canals with large pores in Neostethidae. In maxilla separated from mandible by ex- Phallostethidae laterosensory organs on top panded lower arm of premaxilla; maxillo- of head housed in a membranous dome, mandibulary ligament round in cross sec- evidendy without pores (see Roberts, 1971). tion where it attaches to dentary, the Lateral line canal absent on body. Otoliths attachment near symphysis of lower jaws. absent (dissolved by fonnahn ?). Contact Hind border of opercle concave; opercle organs absent. and preopercle without spines or serrations; 4. Scales cycloid; body except "neck" interoperculum not extending posterior to completely scaled; scales in lateral series vertical from preopercle or overlapping 31-58. Head scaleless or sparsely scaled, subopercle. Distal half of ceratohyal scales on head confined to posterior border abruptly expanded, a dorsal bridge of bone of skull roof and preopercle. joining it to epihyal; branchiostegal rays 5. Pectoral fin set high on sides of body, four or five; gill rakers present only on its shape slightly falcate, with ten or eleven leading edge of first gill arch, other arches rays. First dorsal fin usually present (ab- rakerless. sent in Mirophalhi.^, SolenophaUus thessa 7. Teeth strictly conical, confined to Ceratostethus—Osteology and Structural Novelties • Roberts 411 jremaxillary, dentary, second through atherinoids, and within atherinoids, to the ourth pharyngobranchials, and infra- subfamily Taeniomembrasinae. The sug- )haryngeals. Teeth on premaxillary and gestion by Rosen (1964: 242) that phallo- ientary in a single row (Ceratostethus, stethoids might be more closely related to Seostethus, Phenocostethiis) or in two to cyprinodontoids than to atherinoids was 'our rows (Giila phallus). Dentition of based solely on the erroneous idea that the ower jaw well developed in Gulaphalhts, first pair of pleural ribs in phallostethoids )oorly developed (vestigial ?) in Cerato- originated on the second vertebra. In ;tethus and Neostethus. phallostethoids, as in many (all ?) ather- 8. Supracleithrum a small bone sand- inoids, the first pair of ribs is borne on viched between posttemporal and dorsal either the third or the fourth vertebra, ip of cleithrum; cleithrum without wing- whereas in cyprinodonts the first pair of ike dorsal expansion; scapula and coracoid ribs is usually (invariably ?) borne on the used together; only two proximal pectoral second vertebra. It is noteworthy that the adials or actinosts, the lowermost en- lowest vertebral count in atherinoids is 31, arged; cleithra extended anteriorly in only three less than the lowest count in nales, one or both of them attached to phallostethoids, whereas a number of )ulvinular bone of priapium. cyprinodontoids have as few as 26. Cyprin- 9. Vertebrae 34 to 38. In Neostethidae odontoids usually (invariably ?) have he numbers of precaudal and caudal verte- intramuscular bones, while at least some brae are about equal. In Phallostethidae (but not all) atherinids agree with phallo- ?audal vertebrae more numerous than stethoids in lacking them. Various other )recaudal: Phenacostethus with 14 to 16 points in which phallostethoids agree with Drecaudal and 19 or 20 caudal, Phallo- atherinoids more than with cyjDrinodontoids itetlius with about 10 precaudal and 28 have been pointed out in my osteological ?audal vertebrae. First pair of ribs borne description of Ceratostethus. Among the 3n third vertebra in females and on fourth most important of these are the basically /ertebra in males of Ceratostethus, Neo- similar morphology of the jaws and at- itetlius, and PlienocostetJius. In GuhiphaUus tachment of the maxillomandibulary liga- mirahiUs (MCZ 33904) a 23.1-mm. immature ment to the dentary. nale and a 22.0-mm specimen that appears While phallostethoids are the only :o be an immature female both have the Atherinifonnes k^^o^^^l to me in which the first ribs on the fourth vertebra. First pair first pair of ribs is ligamentously attached 3f ribs in males greatly enlarged and at- to the pelvic girdle, a ligamentous connec- tached to axial support of priapium; par- tion between the pelvic girdle and the xpophyses of aproctal priapial rib enlarged distal ends of a pair of ribs evidently is a ind directed anteriorly so that its distal primitive atherinifonn character. Which i^nd, to which the rib is attached, lies be- pair of ribs is attached depends mainly on aeath second vertebra. Intramuscular bones the position of the pelvic girdle. Thus the ibsent. ribs attached to the pelvic girdle are the Comments. The Phallostethoidea con- third pair in Melanotaenia, the fourth in stitute a distinctive and well-defined taxon. Telmatherina, the fifth in Fhiciphyhx, the Vlmost all of their characters are in reason- sixth in Menidia, etc. ible agreement with Rosen's concept of According to Bailey (1936: 464), in the Atheriniformes, with the (probably GuhphaUus mirahilis the distal ends of the minor) exception of number of pectoral first pair of ribs are "embedded in a heavy radials and relationship of the scapula and triangular mass of fibrocartilage, the ante- coracoid. Within the Atheriniformes it pleural cartilage, which is attached by appears to be most closely allied to the fibers to the posterior crest of the axial 412 Bulletin Museum of Comparative Zoology, Vol. 142, No. 4 bone." In The nacoste thus (ibid.: 472) the forms have conical jaw teeth, bi- and tri- tips of the priapial ribs are embedded in cuspid teeth occur in the jaws of diverst the antepleiual cartilage, "which is ossified representatives; this is especially tiiie ir throughout most of its length and fonns a cyprinodontoids but is not confined tt distinct rod-like antepleural bone." An them. Miller (1956: 8-9) hypothesized thai ossified antepleural is present in Cerato- tricuspid teeth evolved independently ai stethus and Neostethus. In atherinoids and least three times in New World cyprin cyprinodontoids a "secondary postclei- odontids, and suggested that the Old Worlc thrum" is intimately associated with the cyprinodontid genera with tricuspid teetl distal end of each of the first pair of ribs. also originated independently. Tricuspic There is no trace of this element in female jaw teeth occur in some genera of Poc phallostethoids, and none in males either, ciliidae; the outer row of jaw teeth o unless it corresponds with the antepleural. Jenynsiidae has tricuspid teeth. Bicuspid jav The antepleural cartilage or bone of phallo- teeth characterize the atherinid genu: stethoids is perhaps neomorphic. The inner Atherinops, and tricuspid jaw teeth occu puK'inular bone of Ccratostethus and 'Neo- in some genera of Exocoetidae anc stethus, which apparently gave rise to the Hemiramphidae. The significance of thes( phallostethid toxactinium, probably is neo- facts has not been fully appreciated, be moq^hic. It apparently has no homologue cause the remarkably widespread occur in atherinoids. For a discussion of the rence of bi- and tricuspid pharyngeal teetl homologies of the priapial elements see in atheriniforms has passed virtualh' un Woltereck (1942b: 343-347). noticed. Gannan (1895, pis. I-V) showec The haemal spines of the anterionnost that practically all cyprinodontoid group: exhibit caudal vertebrae are expanded to perniit a considerable variety of tootl extension of the swimbladder posterior to forms in the jaws and pharyngeals; in ; the abdominal cavity in various exocoetoids number of instances the pharyngeal teetl exhibit foni (only in the family Exocoetidae ?), cyprin- considerably more variation in odontoids, and atherinoids. In phallo- than the jaw teeth. Anahleps, for example stethoids the swimbladder extends to, but has conical jaw teeth, but the pharyngea teeth in not beyond, the posterior wall of the ab- exliibit a \\'ide variety of fonns in wit) dominal cavity (as taeniomembrasines ) , eluding tricuspid. Many cyprinodonts and the haemal spines are unexpanded. conical teeth in the jaws nevertheless ha\'( some teeth of bi- or The single gonad of Phallostethoidea pharyngeal tricuspic (observed in Phallostethus and Neostethus fonn. Pantanodon podo.xys, with no trace by Regan [1916], in Phenacostethus and of jaw teeth, has tricuspid teeth on tli( Guhphallus by TeWinkel [1939]) is prob- pharyngeals, even in specimens only 17 mn ably a primitive character widespread in in standard length (Whitehead, 1962, fig atheriniforms. The gonads are usually (in- 8). As one might expect, however, the variably ?) single in cyprinodonts, includ- smallest cyprinodontids and poeciliids usu ing Oryzias, Rivulus, Fundiihis, and others. ally have simple conical teeth in both jaw; The teeth of phallostethoids, both on the and pharynx. This does not seriously dc jaws and in the pharynx, are evidently tract from the generalization that most, i invariably simple conical teeth, as one not all, groups of cyprinodontoids have th( would expect in fishes of their tiny size. tendency to develop bi- and tricuspid teeth Hence, they fail to exhibit one of the most and that this is seldom completely sup' persistent trends of the atheriniforms, pressed. This generalization can be ex namely possession of tricuspid teeth. tended to include most groups of atherini Wide.s^jiead occurrence of tricuspid teeth forms. Thus tricuspid teeth occur on th( in atheriniforms. While most atherini- pharyngeals in exocoetids and hemiram Ceratostethus—Osteology and Structural Novelties • Roheiis 413 )hicls (some of which also have tricuspid attachments and movements of the aw teeth) and in scomberesocoids (jaw acanthopterygian rostral and cyprinoid eeth usually or invariably conical; verified rostral or kmethmoid are radicallv dif- n Scomberesox saiirus). Belonidae ex- ferent. As fishes with protractile jaws imined by me have conical pharyngeal are studied more closely, additional evo- eeth; the conical pharyngeal teeth of lutionary novelties are bound to be dis- everal belonids are figured by Collette covered. For example, in Epihidus 1966, figs. 2 and 3). The pharyngeal teeth insidiator (family Labridae) the lower jaw )f adrianichthyoids are conical (verified in is just as protractile as the upper jaw. As (.enopoecihis poptae and X. samsinoriim). might be expected, this extraordinary rhroughout the atheriniforms, the smallest adaptation involved major changes in the onns are likely to have simple conical suspensorium. The outstanding jaw mech- eeth (e.g., phallostethoids, On/zias); again, anisms of many teleosts remain equally his does not seriously detract from the uninvestigated. Some of the most peculiar ;eneralization that atheriniforms have a jaws occur in tropical inshore fishes for emarkable predisposition to develop bi- which no observations on feeding have nd tricuspid teeth. been recorded, and suggestions as to how the jaws are used would necessarily be )ISCUSSION fairly speculative. To no other fishes have TJw protractile jaws of phallostethoids. my knowledge, stiTictures to the "he diversity of jaw mechanisms in teleosts bony comparable paraden- taries and submaxillaries of Neostethidae. ^ becoming ever more apparent. Probably The rostral bone of neostethids is nore neomorphic joints and bones have probably to the rostral of irisen in the jaws than in any other of the homologous cartilage and ele- najor teleostean functional anatomical acanthopterygians, cartilaginous ments similar to or with the mits (as defined by Liem, 1967). A great homologous occur in atherinoids nany of these structural innovations are parardstrals ( accessory rostral cartilages). Analogous rostral car- ssociated with mobility of the upper jaw. or bones and rostral t has been estimated that the jaws are tilages accessory cartilages occur in cyprinoids. Bones anal- )rotractile in about half of the living to the submaxillaries occur in eleosts. While the majority of fonns with ogous Cobitidae. The bones )rotractile upper jaws are acanthopteryg- maxillomandibulary are small bones, and similar elements in ans, paracanthopterygians, or cyprinoids, other fishes, if they occur, are likely to )rotractile upper jaws also occur in have been overlooked. A small bone Vtherinifonnes, in the gonorhynchoid similar to the maxillomandibulary bone, Vnactolaemus, and the characoid Bivi- but lying dorsal to the lower limb of the ^ranchia. Tlie functional mechanism of maxillary instead of ventral to it, occurs in irotractility is completely different in the characoid Chilodus. Such "supramaxil- .canthopterygians, cyprinoids, Phracto- lary" bones probably have evolved inde- aemus and Bivihranchia (for acanthopte- pendently in numerous lines. The presence ygians, see Alexander, 1967a; Liem, 1970; of such an element in CdnJodiis cannot be )sse, 1969; Schaeffer and Rosen, 1961; for construed as a primitive character. yprinoids, Matthes, 1963; Alexander, 1966; The paradentaries and submaxillaries are or Phractohiemtts and Bivihranchia, Gery, large elements, obviously of considerable 963). The functional mechanisms in functional significance. They are evidently itheriniforms are relatively similar to those involved in increasing the extent to which n acanthopterygians. Although both per- the ja\\'s can be protracted. The paraden- nit protraction of the upper jaws, the taries probably are also important in keep- 414 Bulletin Museum of Comparative Zoology, Vol. 142, No. 4 ing the mouth wide open when it is fully the submaxillary bone and maxillary in neo protracted and perhaps in pennitting the stethids is similar but of greater magnitude upper jaw to close dowTJ over the lower Tlie condition of the prevomer should b( jaw while the lower jaw remains fully noted. In almost all fishes this dermal bon( depressed. The paradentaries are joined by is a single, median element. In Cerate a ball and socket joint to the dentaries stethus, however, it sometimes ossifies ii near the symphysis of the lower jaw, and two pieces that remain separate, each witl are capable of rotating outward 90 degrees its anterior end greatly expanded where i and upward 45 degrees from their resting is joined by a meniscus to the submaxillar position in a trough on the side of the bone. While dermal bones usually maintaii dentary. The submaxillaries are tubular their integrity, it is not impossible for then bones with a meniscus at either end : the to give rise to separately ossifying elements anterior meniscus with the maxillary bone, Frazzetta (1970) described and discusse( the posterior meniscus with the prevomer. the subdivision of the maxillary bone b Although novel as bony elements, the form two maxillary bones united by a: paradentaries and submaxillaries arise in intramaxillary joint in bolyerine snakes, structures that are characteristic of atherini- Neotenic characters of plmllostetJwich forms, and particularly of atherinoids, the In teleosts it is extremely unusual for th group to which they seem most closely anus and urogenital openings to be placci related. Furthermore, the paradentary anterior to the pelvic fins. In most teleost bones and submaxillary bones can be seen in which these openings lie at the throii as the ultimate consequences of clearly the pelvics are either absent or are eve defined evolutionary trends exhibited in farther forward than the openings. In thi atherinoids. Thus the anterionnost part of respect phallostethoids are exceptional. I the maxillomandibulary ligament in some males and in females in which the pelvic atherinoids (notably in the Taenio- are vestigial or evanescent, anal am membrasinae) has the peculiarities of the urogenital openings are morphologicall anterior part of the maxillomandibulary anterior to pelvic fins. In adult atheriij ligament in phallostethoids except that no oids the position of the openings i paradentary bone develops in it. That is, variable, but they always lie posterior t in these atherinids the attachment of the the pelvic origin. It may be immediatel ligament to the dentary is extremely far in front of the anal fin, distinctly in front c fins. forward, near the symphysis; the body of it, or even between the pelvic Woltt this portion of the ligament is tough and reck (1942a: 256) made the importan round in cross section, and when the mouth discovery that the openings lie anterior t is closed it lies in a trough on the side of the pelvic fins in larval Atherina. This sm the dentary identical with the trough in gests that the condition in phallostethoid which the paradentary bone lies in neo- may be neotenic. stethids. Concerning the submaxillary bone. It may well be that the anterior positio it evidently serves much the same function of the pelvic fins, and perhaps even thei as the submaxillary meniscus in atherinids. attachment to the shoulder girdle, ar Alexander (1967b: 234) found the sub- neotenic, at least in part. In atherinoid maxillary meniscus much thicker in thought to be primitive, Rheocles ani Atherina than in Melanotaenia; in Atherina Bedotia, the pelvics are relatively far foi "the submaxillary meniscus can swing an- ward. The ancestors of atherinoids ma teriorly and somewhat medially about its have had the acanthopterygian charactei articulation with the cranium, carrying the istic of thoracic pelvic fins connected wit cranial condyle of the maxilla anteriorly the shoulder girdle, and ventrally" (ibid.: 236). Movement of The subcephalic position of the priapiur Ceratostethus—Osteology and Structural Novelties • Roberts 415 cannot, however, be ascribed entirely to Poeciliidae, Goodeidae, and Jenynsiidae- leoteny. No atherinoids or atheriniforms Anablepidae—it has led to viviparity. >f any kind have the pelvics anywhere near Gonopodial morphology in these viviparous IS far fonvard as they are in phallo- fonns ranges from relatively simple in itethoids. The anteriad prolongation of the Goodeidae and in the poeciliid tribe Poe- ?leithrum and of the parapophyses of the ciliini to remarkably complex in the ourth vertebra in male phallostethoids poeciliid tribe Cnesterodontini (see Rosen ndicates that the priapium has "migrated and Bailey, 1963, for figures of the poe- ' or\\'ard ( in an . ciliid It is a evolutionary sense ) The gonopodia). striking fact that elective advantage of an anteriorly located, in the three groups that evolved internal .e., subcephalic, priapium may lie in fertilization and are oviparous—Tomeur- greater visual coordination during mating. idae, Horaichthyidae, and Phallostethoidea The phallostethoid "neck" may also be —the primary copulatory organ is far more onsidcrcd a neotenic character, for it ap- complex than in any of the \'iviparous ^ )ears to result in large part from absorj^tion forms. The gonopodium of Tomeunis can if the yolk. In larval phallostethoids the be compared directly \^'ith that in Poe- ompact yolk sac lies immediately posterior ciliidae, since Tomeunis is evidentlv related 3 the head; it occupies considerable space to the poeciliids and its gonopodium is /here the neck eventually forms. derived mainly from the same rays in the Almost certainly neotenic is the phallo- anal fin as the poeciliid gonopodium. Per- tethoid abdominal keel, derived from the haps the most complicated gonopodium in ledian embryonic fin fold. In larval phallo- any poeciliid is that of the cnesterodontin tethoids the fin fold extends continuously Pliallotoiynus, which bears an enlarged rom its at at its origin a point on the dorsum scooplike element tip ( ibid., fig. 31a, bout midway between snout and caudal b). The gonopodium of Tomeunis is much lase (where the first dorsal fin arises), more complex: its tip bears a pair of greatly round the caudal lobe and then on the enlarged antlerlike processes, a far more entrum uninterruptedly until it ends at complicated scooplike arrangement, and he posterior margin of the yolk sac (see the distal end of the fifth anal ray, which ^illadolid is modified in a the and Manacop, 1934, pi. 2, fig. manner recalling even -2, pi. 3, figs. 1 and 2 of larval GuhphaUus more greatly modified papillary bone of vmhilis). In adult phallostethoids the neostethids {ibid., figs. 7, 13-16). The bdominal keel consists of actinotrichia gonopodimn of Horoichthijs (Kulkarni, the fibrous rays characteristic of embry- 1940), which evolved independently of the . of Poeciliidae and nic fins ) gonopodium Tomeur- The abdominal keel presumably plays idae, is as complicated as that of Tomeunis. le most important of the hydrodynamic The neostethid priapium is moq:)hologically as as or even more than oles formerly played by the pelvic fins. If, complex complex the of Tomeunis or Is may have been, acquisition of the ab- gonopodia Horaichthys, and far than the ominal keel permitted phallostethoids to more complicated gono- of of the forms. tilize the pelvic fins for an entirely new podium any viviparous The be are these jnction, it was a key step in their evo- question may now raised, \vhy ition. structures in oviparous fonns more com- have Evolution of complicated external ^cni- plex? Conversely, why viviparous ilia. Internal fertilization is a regular ^ The situation is the 'ature in the reproductive biology of admittedly complicated liy development in \ iviparous forms of modified pelvic lany groups of atheriniforms (excepting or pectoral fin structures that fimction as acces- hallostethoidea, these to groups belong sory copulatory structures, but this does not le Cyprinodontoidea). In three groups— seriouslv alter the main theme under consideration. 416 Bulletin Museum of Comparative Zoologij, Vol. 142, No. 4 forms failed to evolve gonopodia as Tomeurus apparently occur in comparable complicated as those in oviparous forms? habitats, Horaiclithys along the Bombay- The puzzle increases \\'hen we recall that Kerala coast of India, Tomeurus along the the gonopodium of Hoiaichthys is not an northeastern coast of South America, in- intromittant organ but serves only to con- cluding the mouth of the Amazon. The vey spennatophores with dartlike attach- selective advantage of internal fertilization ments that are implanted externally near in these fishes may lie in pennitting tem- the female's vent. poral separation of mating and egg-laying It would seem that perfection of the corresponding, respectively, with periodj fertilization of organs of would be just as im- low water and high water ( Roberts 1971 . It that in portant to viviparous forms as to oviparous ) seems likely mating sue! ones; and yet nature has favored the more habitats is difficult, because of current anc complicated structures in oviparous forms, low visibility. The waters in which these and selected against their development in fishes live are probably especially turbic the viviparous forms. during the rainy season, when mating prob The explanation of this seemingly con- ably reaches its peak. In this connection tradictory state of affairs perhaps lies in it is noteworthy that copulation in Gula two considerations: first, effects on mating phallus mirahilis is prolonged, the mak behavior of differences in the habitats clasping the female for one to two minute occupied by oviparous and viviparous (Villadolid and Manacop, 1934: 200). atheriniforms, and second, deleterious ef- In contrast to phallostethoids, poeciliid fects of excessively complicated external tend to occiu- in habitats in which mating genitalia. Evidence concerning these fac- can occur much of the time and under rel tors was gathered during my fieldwork on atively favorable conditions. Poeciliid gono phallostethoids in Thailand. podial thrusts are relatively short in dur In Tliailand neostethids and phallo- ation, some almost instantaneous; they ma; stethids occur only a short distance inland, be repeated several times without spem in waters strongly influenced by tides. transfer and often strike wide of the mark Usually the current is strong and the water perhaps even more in some forms wit] muddy. They are swift swimmers, generally less specialized gonopodia. Yet such male stay in midwater and can maintain them- are undoubtedly quite efficient at fertiliz selves in the current. During flood waters ing females; the number of females ii they are likely to be widely dispersed. Thus populations of poeciliids is general! several hours of collecting at Khlong Langu greatly in excess of that of males. in Satul Province during high tides com- In all my collections of neostethids, male bined with heavy rain yielded only a half- outnumbered females. In places where the; dozen immature Neostethus, all taken were particularly abundant, it was some singly. Whenever either Ceratostethus or times possible to catch 25 or 50 at once, anc Neostethus were found during low water, in these instances the ratio of males to fe they were usually aggregated and some- males was sometimes about equal. Fron times exceedingly numerous (as in the some places where individuals were reli estuary of Chantaburi River, where Neo- atively scarce or collecting was particular! stethus and Ceratostethus were found to- difficult, collections of half-a-dozen or in . dozen of gether large numbers ) When the tide is specimens Ceratostethus are en out, Ceratostethus are likely to be confined tirely males. During fieldwork it appeared to that 1 well-defined creeks and channels; when ) males of neostethids are at least a the tide is in, they are probably dispersed abundant, if not more abundant, tha over a wide area of partially submerged females, and 2) males are more readil mangroves and nipa palm. Horaichthys and caught because the priapium prevents theri Ceratostethus—Osteology and Structural Novelties • Roberts 417 from swimming away as quickly as females. numbers of eggs that are expelled in small This suggests that males are more suscep- batches or even singly, and in which em- or be tible to predation than females. It seems bryonic development hatching can be to the evo- probable that the exceptionally large deferred, may preadapted gonopodia of Horaichthys and Tomeurus lution of internal fertilization (Roberts, would encumber their swimming. 1971). If internal fertilization is favored, modification of or anal Another disadvantage of highly compli- then every pelvic fins matter how that increases cated genitalia may lie in an increased (no slight) the of the ovi- proportion of developmental anomalies. probability sperm entering Thus Developmental anomalies in the priapium duct will confer a selective advantage. the is set for intensification of func- of phallostethoids are probably relatively stage tion which can frequent. Woltereck (1942a) mentioned, (Mayr, 1960), eventually lead to structures as as the but did not describe, developmental anom- complicated and bone of Cerato- alies in the priapia of neostethids from the priapium papillary stethus bicornis. Philippines. I find a number of anomalies in my material, including a Neostethiis with a peculiarly bowed ctenactinium LITERATURE CITED could not be into (which properly swung Alexander, R. McN. 1966. The functions and beneath the and resting position head) mechanisms of the protrusible upper jaws of additional externalized elements or "super- two species of cyprinid fish. J. Zool., London, numerary ctenactinia" and a Ceratostethus 149: 288-296. . 1967a. The functions and mechanisms in which the priapium is equally developed of tlie protrusible upper jaws of some (but not fully foiTned) on both sides. acanthopterygian fish. J. Zool., London, ex- Finally, the development of a large 151: 13-64, 6 pis. 1967b. Mechanisms of the of some ternal genitalium reqviires a considerable jaws atheriniform fish. ]. London, 151; diversion of energy and materials, and such Zool., 233-255, 1 pi. "expenditures" will always be selected AuRiCH, H. 1937. Die Phallostethiden ( Unterord- unless confer a selective ad- against they nung Phallostethoidea Myers). Intnatl. Rev. vantage. The highly complicated copula- Ges. Hydrobiol. Hydrogr., 34: 263-286. 1936. The and relation- tory organs presumably ensure successful Bailey, R. J. osteology of the fishes. J. Morph., sperm transfer with each copulatory epi- ships phallostethoid 59(3): 453-483, 4 pis. sode. In fishes in which copulatory episodes Banjagan, M. C. 1966. [Untitled letter]. Dopeia are rare this will have selective ad- great (Miami), June, 1966: 46. vantage. CoLLETTE, B. B. 1966. Bclonion, a new genus of Provided environmental conditions are fresh-water needlefishes from South America. Mus. No. 2274: 1-22. such that internal fertilization has a selec- American Novitates, Frazzetta, T. H. 1970. From hopeful monsters tive advantage, there is little to prevent its to Bolyerine snakes? Amer. Nat., 104(935): are numerous development in fishes. There 5.5-72. records of internal fertilization occurring in Garman, S. 1895. The cyprinodonts. Mem. Mus. 12 fishes, such as trout, which normally have Comp. Zool., 19(1): 1-179, pis. 1963. buccal de external fertilization; sperm simply enter Gery, J. L'appareil protracteur Bivibmnchia (Characoidei) (1) avec une the oviduct and fertilize eggs within the note SUV Pliractolacmiis (Chanoidei) (Pisces). female. Such occurrences do not lead to Vie et Milieu, 13(4): 729-740. the evolution of internal fertilization, but Herre, a. W. 1925. Two strange new fishes from are selected against, when the life history Luzon. Philippine J. Sci., 27(4): 507-514, 2 strategy of the fishes involved overwhelm- pis. little . 1942. New and known phallo- ingly favors external fertilization. Ovi- stethids, with keys to the genera and Philip- on the other parous atheriniform fishes, pine .species. Stanford Ichth. Bull., 2(5): hand, which tend to have relatively small 137-156. 418 Bulletin Museum of Comparative Zoology, Vol. 142, No. 4 HuBBS, C. L. 1944. Fin structure and relation- Roberts, T. R. 1971. The fishes of the Malaysian ships of phallostethid fishes. Copeia, 1944 family Phallostethidae ( Atheriniformes ) . Bre- (2): 69-79. viora. No. 374: 1-27. KuLKARNi, C. V. 1940. On the systematic position, RosEX, D. E. 1964. The relationships and taxo- de- structural modifications, bionomics and nomic position of the halflseaks, killifishes, \elopnient of a remarkable new family of silversides and their relatives. Bull. American fishes the of Bom- cyprinodont from province Mus. Nat. Hist, 127(art. 5): 217-267, pls.j bay. Rec. Ind. Mus. Calcutta, 42: 379-423. 14 and 15. K. F. 1967. Functional of the LiEM, moiphology Rosen, D. E., and R. M. Bailey. 1963. The poe- teleost Helostoma head of the anabantoid fish, ciliid fishes (Cyprinodontiformes), their struc- 135-158. tcmmincki. J- Morph., 121: ture, zoogeography, and systematics. Bull. . 1970. fimctional of Comparative anatomy American Mus. Nat. Hist., 126(art. 1): the Nandidae (Pisces: Teleostei). Fieldiana 1-176. 56: 156. (Zool.) iii-f Schaeffer, B., and D. E. Rosen. 1961. Major H. 1963. of tlie Matthes, A comparative study adaptive levels in the evolution of the of some African feeding mechanisms Cyprin- actinopterygian feeding mechanism. Amer- idae. 33 : 1-36. Bijdr. Dierk., ican Zool., 1(2): 187-204. E. 1960. The of Mayr, emergence evolutionary ScHULTZ, L. P. 1948. A revision of six subfamilies novelties. In S. Tax (ed. ), The Evolution of of atherine fishes, with descriptions of new Life. Chicago, University of Chicago Press, genera and species. Proc. U. S. Nat. Mus., pp. 349-380. 98(3220): 1-48, 2 pis. R. R. 1956. A new and Miller, genus species Smith, H. M. 1927. The fish Neostethiis in Siam. of cyprinodontid fish from San Luis Potosi, Science (n. s.), 6S: 353-355. Mexico, with remarks on the subfamily TeWixkle, L. E. 1939. The internal anatomy of Cyprinodontinae. Occ. Papers Mus. Zool. two phallostethid fishes. Biol. Bull. Woods Univ. Michigan, No. 581, 17 pp., 2 pis. Hole, 76(1): 59-69. Myers, G. S. 1928. The systematic position of ViLLADOLiD, D. v., AND P. R. Manacop. 1934. The the phallostethid fishes, witli diagnosis of a Philippine Phallostethidae, a of a new genus from Siam. American Mus. Novi- description new species, and a report on the biology of tates, No. 295, 12 pp. GithiphaUus mimhilis Herre. Philippine J. Sci., . 1935. A new phallostethid fish from 5 Palawan. Proc. Biol. Soc. Washington, 48: 55(3): 193-220, pis. 5-6. Whitehead, P. J. P. 1962. The Pantanodontinae, edentulous toothcarps from East Africa. Bull. OssE, J. W. M. 1969. Functional moiphology of British Mus. (Nat. ) : Hist.), Zool., 9(3): the head of tlie perch ( Perca fluviatilis L. 103-137. an electromyographic study. Neth. J. Zoology, 19: 289-392. WoLTERECK, R. 1942a. Stufen der Ontogenese Regan, C. T. 1913. Phallostethus dunckeri, a und der E\olution von Kopulationsorganen bei Neostethiden remarkable new cyprinodont fish from Johore. (Percesoces, Teleostei). Internatl. Rev. Ges. 42: Ann. Mag. Nat. Hist., 12: 548-555. Hydrobiol. Hvdrogr., 253-268. . 1916. The morphology of the cyprinodont fishes of the subfamily Phallostethinae, with : 19421). Neue Organe, durch postembryo- descriptions of a new genus and two new nale Umkonstruction aus Fischflossen entste- species. Proc. London Zool. Soc, 1916: hend. Internatl. Rev. Ges. Hydrobiol. Hydrogr., 1-26, pis. 1-4. 42: 317-355.