ZOOLOGICA Iff SCIENTIFIC CONTRIBUTIONS OF THE NEW YORK ZOOLOGICAL SOCIETY VOLUME 44 • 1959 • NUMBERS 1-13 PUBLISHED BY THE SOCIETY The ZOOLOGICAL PARK, New York NEW YORK ZOOLOGICAL SOCIETY GENERAL OFFICE 30 East Fortieth Street, New York 16, N. Y. PUBLICATION OFFICE The Zoological Park, New York 60, N. Y. OFFICERS PRESIDENT VICE-PRESIDENT SECRETARY TREASURER Fairfield Osborn Laurance S. Rockefeller George Wall Merck David H. McAlpin SCIENTIFIC STAFF: JohnTee-Van General Director Christopher W. Coates. .Director, Aquarium ZOOLOGICAL PARK Joseph A. Davis, Jr.. .Assistant Curator, Mammals Grace Davall Assistant Curator, Mammals and Birds William G. Conway. .Curator, Birds Herndon G. Dowling. Associate Curator, Reptiles Charles P. Gandal . . Veterinarian Lee S. Crandall General Curator Emeritus William Beebe Honorary Curator, Birds AQUARIUM James W. Atz Associate Curator Carleton Ray Assistant to the Director Ross F. Nigrelli Pathologist & Chair- man of Department of Marine Biochem- istry & Ecology Sophie Jakowska Research Associate in Experimental Biology C. M. Breder, Jr Research Associate in Ichthyology Harry A. Charipper. . .Research Associate in Histology Homer W. Smith Research Associate in Physiology GENERAL William Bridges . . Editor & Curator, Publications Sam Dunton Photographer Henry M. Lester. .Photographic Consultant DEPARTMENT OF TROPICAL RESEARCH William Beebe .... Jocelyn Crane. . . . David W. Snow . . . Henry Fleming . . . John Tee-Van .... William K. Gregory AFFILIATES L. Floyd Clarke Director, Jackson Hole Biological Research Station SCIENTIFIC ADVISORY COUNCIL A. Raymond Dochez Caryl P. Haskins Alfred E. Emerson John S. Nicholas W. A. Hagan EDITORIAL COMMITTEE Fairfield Osborn, Chairman James W. Atz William G. Conway William Beebe Lee S. Crandall William Bridges John Tee-Van Christopher W. Coates .... Director Emeritus . . . . Assistant Director . . . . Resident Naturalist . . . . Entomologist Associate . . . . Associate Contents 510. Sli A A ■n-n Part 1. March 31, 1959 PAGE 1. Middle- American Poeciliid Fishes of the Genera Carlhubbsia and Phal- lichthys, with Descriptions of Two New Species. By Donn Eric Rosen & Reeve M. Bailey. Plates I- VI; Text-figures 1-10; Maps 1 & 2 1 Part 2. July 20, 1959 2. Studies on the Histology and Histopathology of the Rainbow Trout, Salmo gairdneri irideus. II Effects of Induced Inflammation and Corti- sone Treatment on the Digestive Organs. By Eva Lurie Weinreb. Plates I & II 45 3. A Study of the Structure and Development of Certain Reproductive Tis- sues of Mugil cephalus Linnaeus. By Albert H. Stenger. Plates I-VIII. 53 4. The Production of Underwater Sound by Opsanus sp., a New Toadfish from Bimini, Bahamas. By Marie Poland Fish & Willam H. Mowbray. Plates I-III; Text-figures 1-5 71 5. Some Aspects of the Behavior of the Blennioid Fish Chaenopsis ocellata Poey. By C. Richard Robins, Craig Phillips & Fanny Phillips. Plates I-III; Text-figure 1 77 Part 3. December 11, 1959 6. The Ctenuchidae (Moths) of Trinidad, B.W.I. Part II. Ctenuchinae. By Henry Fleming. Plates I-III 85 7. Eastern Pacific Expeditions of the New York Zoological Society. XLIV. Non-intertidal Brachygnathous Crabs from the West Coast of Tropical America. Part 1 : Brachygnatha Oxyrhyncha. By John S. Garth. Plate I; Text-figures 1 & 2 105 8. Stomach Contents and Organ Weights of Some Bluefin Tuna, Thunnus thynnus (Linnaeus), near Bimini, Bahamas. By Louis A. Krumholz 127 SMITHSONIAN o 2 HMSTiTUTION * 4 Part 4. December 31, 1959 PAGE 9. Heterotopic Thyroid Tissues in Fishes. III. Extrapharyngeal Thyroid Tis- sue in Montezuma Swordtails, a Guppy and a Cherry Barb. By K. France Baker. Plate I; Text-figure 1 133 10. Observations on the Spawning Behavior and Egg Development of Strongy- lura notata (Poey). By C. M. Breder, Jr. Plates I & II; Text-figure 1.. . 141 1 1 . Effects of Four Combinations of Temperature and Daylength on the Ovo- genetic Cycle of a Low-latitude Fish, Fundulus conjluentus Goode & Bean. By Robert Whiting Harrington, Jr. Text-figures 1-4 149 1 2. A Study of Lipids and Water of Liver and Muscle in Fundulus heteroclitus (Linnaeus) and Stenotomus versicolor (Mitchill) . By X. J. Musacchia. . . 169 13. Ectyonin, an Antimicrobial Agent from the Sponge, Microciona prolifera Verrill. By Ross F. Nigrelli, Sophie Jakowska & Idelisa Calventi. Plate 1 173 Index to Volume 44 177 ZOOLOGICA SCIENTIFIC CONTRIBUTIONS OF THE NEW YORK ZOOLOGICAL SOCIETY VOLUME 44 • PART 1 • MARCH 31, 1959 • NUMBER 1 PUBLISHED BY THE SOCIETY The ZOOLOGICAL PARK, New York Contents PAGE 1. Middle-American Poeciliid Fishes of the Genera Carlhubbsia and Phal- lichthys, with Descriptions of Two New Species. By Donn Eric Rosen & Reeve M. Bailey. Plates I-VI; Text-figures 1-10; Maps 1 & 2 1 1 Middle-American Poeciliid Fishes of the Genera Carlhubbsia and Phallichthys, with Descriptions of Two New Species1’ ' Donn Eric Rosen Genetics Laboratory, New York Zoological Society, and New York University, New York & Reeve M. Bailey Museum of Zoology, University of Michigan, Ann Arbor, Michigan (Plates I- VI; Text-figures 1-10; Maps 1 & 2) Table of Contents Introduction 1 Materials and Methods 2 Systematic Account 3 Genus Carlhubbsia Whitley 3 Carlhubbsia stuarti, new species 5 Carlhubbsia kidderi (Hubbs) 8 Genus Phallichthys Hubbs 16 Phallichthys amates (Miller) 18 P. amates pit fieri (Meek) 19 P. amates amates (Miller) 22 Phallichthys fairweatheri, new species 24 Assessment of Taxonomic Criteria 29 Gonopodium 29 Gonopodial Suspensorium 30 Head Skeleton and Dentition 32 Sensory Canals 32 Relationships of Phallichthys and Status of the Poeciliopsinae 33 Carlhubbsia and the Cuban Endemic Poeciliids. 35 Girardinus and Quintana 36 Characters Indicating Relationship of Carl- hubbsia with Quintana and Girardinus . . . 36 Resume of Morphological Analysis and Con- clusions 37 Zoogeographic Considerations 39 Summary 40 Acknowledgments 41 Literature Cited 41 1 Much of the material in this paper was included in a thesis presented by Rosen in partial fulfillment of the degree of Master of Science in the Department of Biology, New York University. Introduction THE fishes of the family Poeciliidae (Or- der Cyprinodontiformes, also known as Cyprinodontida, Cyprinodontes and Mi- crocyprini) are all from the New World and most of the species bear living young. They abound in the fresh and brackish waters of Mexico, the West Indies and northern and eastern South America, but attain their maximum diversity in Central America (Rosen & Gordon, 1953: 1-6). In Middle America three nominal species classi- fied in two genera, Carlhubbsia (formerly Allo- phallus) and Phallichthys, have heretofore been regarded as closely related. Their many super- ficial resemblances include the asymmetric twist- ing or folding, either sinistrally or dextrally, of the external genitalium (gonopodium) of the adult male. This modification is shared by several other genera (Poeciliopsis, Aulophallus, Phallop- tychus and Xenophallus). On the basis of this one feature, all of them have been grouped as the subfamily Poeciliopsinae (Hubbs, 1926; 1936). We now find evidence, however, that asym- metry of the gonopodium is not in itself an adequate criterion of the implied phylogenetic relationships of the fishes in the Poeciliopsinae. As in other poeciliids, it is the fine details of the 2 This work was supported by a grant from the Na- tional Science Foundation to the New York Zoological Society for the project, “A Biological Synthesis of Poeciliid Fishes,” Dr. Myron Gordon, New York Zoological Society, Principal Investigator. flSS8RSS*w 1 5 1959 2 Zoologica: New York Zoological Society [44: 1 gonopodia and characters in the gonopodial sus- pensoria that provide the most useful criteria in determining relationship. On the basis of new data, Carlhubbsia shows affinities to Quintana and Girardinus (including Toxus, Glaridichthys, Allodontium and Dactylophallus); these are en- demic Cuban fishes with symmetrical gonopodia. Phallichthys is closely allied to Poeciliopsis (in- cluding Poecilistes) and Aulophallus, but the remaining genera previously associated with them in the Poeciliopsinae are not intimately related. For these reasons the dissolution of the Poeciliopsinae is now recommended. Materials and Methods Material.— Most of the specimens used in this study are preserved in the Museum of Zoology of the University of Michigan (UMMZ). Addi- tional specimens are from the collections of the Chicago Natural History Museum (CNHM), Genetics Laboratory of the New York Zoological Society (NYZS-GAF) and the United States National Museum (USNM) . In addition to most of the material of Carlhubbsia and Phallichthys previously recorded, we have had access to nu- merous and extensive series of the four known species belonging to these genera from Guate- mala and Honduras. These were collected largely by the Rev. Gerald Fairweather, Drs. Myron Gordon, Carl L. Hubbs, Laurence C. Stuart and their field associates. Counts and Measurements.— The methods of counting and measuring are those described by Miller (1948: 8-14) for certain cyprinodontid fishes. Fin ray counts include small anterior rays, but in dorsal and anal fins the last ray as counted consists of two elements that are separate to their bases. In the genera studied either one or two anterior dorsal rays are simple, the rest are branched distally. Careful examination under transmitted light is necessary to establish the number of simple rays, and since branching may be delayed, the definitive presence of a single simple ray can be reliably determined only in adults. Caudal ray counts include branched rays plus two. Scales in lateral series are counted from the upper angle of the gill aperture to the caudal base at midside. Body-circumference scales are counted from about two scale rows before the dorsal origin to an equal distance in front of the pelvic fins. The vertebral count includes the uro- stylar vertebra. Head angle is measured with an arm protractor: one arm is placed parallel to the predorsal contour (if flat) or tangential to it (if gently curved); the other is adjusted to coincide with the straight line along the lower surface of the head and the anterior part of the breast (the oblique upward slant of the lower jaw is disregarded) . Standard length is measured from snout tip to caudal base. Head length is taken to the opercular margin. Diameter of orbit is established by slipping caliper points within the orbital rim and spreading them gently. All measurements were recorded to tenths of milli- meters. Skeletal Material— For clearing and staining, alcoholic (ethyl alcohol) specimens were washed briefly in tap water and transferred to 3 % solu- tion of KOH in tap water. Ten fish averaging IV2 inches in standard length were placed in 300 cc. of this macerating solution. When the fish became translucent in alkali (4 to 6 weeks at room temperature) enough Alizarin Red S was added to color the solution deep wine red. Within a week the bone and scales became in- tensely stained, and then all scales were removed. The fish were then placed directly into 50 cc. of glycerine where they cleared in about three days. In disarticulating a skeleton for study a section of the cleared and stained specimen was first placed in warm 50% KOH solution. The rapid progress of the maceration was watched with a dissecting microscope; when minute bubbles formed, muscle and connective tissues were freed readily from the bone with a dissecting needle. As the bones separated they were trans- ferred to cool tap water to which a few drops of acetic acid had been added to check further disarticulation. Skeletons or parts thereof were drawn with the aid of a camera lucida. The material (in water or glycerine) was placed in a petri dish and intense reflected light was directed on it from illuminators on each side of the microscope stage. A manipulator fitted with solid glass needles was used to orient specimens. The cam- era lucida was adjusted to give a distortion-free image on the recording surface by sighting down a hollow tube placed at the geometrical center of the proposed drawing. The specimen was brought into line with the image of the hollow tube that was resting on the recording surface. Relative positions and dimensions of different structures were determined by utilizing the co- ordinate system of the graph paper employed as a drawing surface. Preliminary drawings were refined and details added in free-hand from direct observations. The few cleared and stained skeletons were checked for accuracy by comparison with X-rays of series of specimens which were prepared as outlined by Miller (1957). The distal part of the gonopodium is equipped with a variety of spine-like processes and other 1959] Rosen & Bailey: Middle-American Poeciliid Fishes: Carlhubbsia & Phallichthys 3 specializations. The terminology here adopted for these structures is that proposed and de- scribed by Rosen & Gordon (1953: 18-23). For example, spines are specialized bony structures that arise on the ventral (anterior) surface of the distal third of ray 3; serrae, if present, are located on the posterior edges of rays 4 or 5, those found proximally on ray 4p function as an anchor for the collagenous tissue sheets that envelop the rays, those found near the distal end of the rays serve as holdfast structures dur- ing copulation; the hook is a terminal bony segment on ray 3 or ray 4a. Systematic Account Genus Carlhubbsia Whitley Allophallus.— Hubbs, 1936: 232 (original de- scription; type species Allophallus kidderi Hubbs). Carlhubbsia— Whitley, 1951: 67 (replacement synonym of Allophallus Hubbs, which name is preoccupied by Allophallus Dziedzicki, 1923, in Diptera). Description— Body moderately deep and com- pressed, with distinctly or moderately angulated dorsal and ventral margins, covered with large cycloid scales. Dorsal fin typically falcate, often pointed, with 8 to 10, usually 9, rays; the first two rays simple (rarely only the first) , other rays bifurcate one or more times (in adult). Pelvic fin without fleshy appendages, constantly with 6 rays, the second and third barely prolonged in adult males. Anal rays 9 to 11, usually 10. Gonopodium permanently folded to form a broad groove on the right side; single series of flat, irregular unpaired serrae on right half of ray 5p extending 8 to 20 segments beyond tip of ray 6; single series of unpaired serrae distally on the lateral margin of left half of ray 5a; sin- gle series of unpaired serrae originating distally on left half of ray 4p, tightly grouped into a clus- ter; distal and subdistal elements of ray 4p that lack serrae extremely slender, reduced or obso- lescent; terminal segments of ray 4a widened transversely; ray 3 terminated by minute bony or membranous hook, without consolidated termi- nal or subterminal segments, right half of ray with row of unpaired broad and flat spines form- ing ventral wall of groove, left half with minute denticles on subdistal elements; segments of dis- tal half of ray 6 swollen, transversely thickened, those of basal half asymmetrical, the paired ele- ments not side by side; rays 7 and 8 simple, dis- tinctly separated, not converging or in contact along middle of their lengths. Gonopodial sus- pensorium with three gonapophyses; uncini on first two gonapophyses emerging near base of spine, not curved, moderately slender; uncini on third gonapophysis, if present, emerging midway along spine, usually closer to vertebral axis than tip of spine, not curved downward, moderately thickened, rarely slender; uncini of all gonapo- physes overlappping one another, forming un- cinar plane extending downward and backward at angle of approximately 30 to 45 degrees with horizontal. Dorsal half of primary gonactinostal complex greatly dilated antero-posteriorly, upper edge of complex slightly notched or uniform and unbroken. Vertebrae 28 or 29, rarely 30. Pectoral girdle somewhat triangular in outline, its longest dimension vertical; four discrete ac- tinosts recessed within posterior margins of scapula and coracoid, not approximating lower margin of coracoid; upper part of cleithrum pro- duced backward above scapula as large spatu- late process; posterior edge of coracoid below actinosts produced backward as flat process similar in outline to cleithral process but smaller and variously developed. Skull deep and wedge- shaped, with well-developed supraoccipital pro- cesses and variously developed epiotic processes; jaws weak, consisting of slender elements with delicate articulations; preorbital (lacrymal) heavily sculptured, with a well-developed process extending backward toward lateral ethmoid; pre- maxillae and dentaries flattened in front, the paired elements not joined at midline and sepa- rated by a distinct tissue space, each with an outer series of movable, compressed or narrow incisor-like teeth in a single, largely transverse row that is weakly indented near midline, and an inner series of minute slender and pointed teeth in a narrow band from 1 to 3 teeth wide. Intestine long, coiled, lying largely on right side of coelom. Peritoneum dark. Cephalic canals rather well developed; supraorbital canal typical- ly with a developed tube connecting pores 2, 3, and 4a, and a postorbital section connecting pores 6 and 7 (Gosline, 1949), sometimes with a third section connecting pores 4b and 5; pre- opercular canal typically with 7 pores; mandibu- lar canal undeveloped; preorbital canal with 3 or 4 pores in adult. (See Table 1.) Status— Carlhubbsia consists of two distinct species, C. kidderi (Hubbs) and C. stuarti, n. sp. (pages 5-16, and Tables 1-5, 13-19), and prob- ably a third, as yet undescribed because of inade- quate material. The genus is confined to the Atlantic drainage of Middle America from the Isthmus of Tehuantepec to southern Guatemala. Allophallus [= Carlhubbsia ] kidderi was grouped by Hubbs (1936) with Phallichthys amates and P. pittieri in a subfamily Poeciliop- sinae that Hubbs (1924) erected to include all 4 Zoologica: New York Zoological Society [44: 1 Table 1. Comparison of Two Species of Carlhubbsia Measurements are based on adults and are expressed as percent, of standard length Character C. stuarti C. kidderi Pectoral rays 13, rarely 14 9 or 10, rarely 11 Preorbital pores (adults) Usually 4; if 3 with an open groove at ventral end Usually 3; without groove Vertebrae Usually 28 Usually 29 Gonapophyses With uncini on I and II only With uncini on I, II and III Gonopodium: Left half of ray 5a With 15 to 20 terminal serrae With about 6 terminal serrae Left half of ray 5p Continuous with terminal serrae on ray 5 a Obsolescent distally, not continuous with terminal serrae on ray 5 a Tip of ray 4p With cluster of about 8 serrae With cluster of 5 serrae Body circumference scales 23 to 26; infrequently 23 22 to 24; infrequently 24 Dorsal fin With dusky marginal band With jet black spot near posterior margin Vertical bars on body Well defined Faint Dark scale borders Not well marked Well defined Greatest body depth Males 39 to 42 27 to 30 Females 37 to 42 29 to 33 Least depth Males 22 to 25 14 to 18 Females 21 to 24 16 to 18 Predorsal length Males 54 to 58 50 to 54 Females 56 to 59 50 to 53 Preanal length Males 54 to 58 48 to 52 Females 64 to 67 57 to 62 Dorsal origin to caudal base Males 49 to 54 51 to 55 Females 48 to 51 50 to 54 Anal origin to caudal base Males 49 to 55 52 to 54 Females 40 to 42 42 to 45 Head length 29 to 34 25 to 30 Head width 18 to 20 14 to 18 Snout length 8 to 10 5 to 9 Postorbital length of head 11 to 13 9 to 12 Interorbital, bony width Males 13 to 15 9 to 10 Females 13 to 14 11 to 13 Dorsal fin, depressed length Males 31 to 34 27 to 32 Females 27 to 31 31 to 34 Anal fin, depressed length Males 36 to 41 49 to 54 Females 22 to 26 23 to 27 Size (standard length) Males Usually 30-40 mm., largest 45 mm. Usually 16-21 mm., largest 23 mm. Females Usually 35-50 mm., largest 55 mm. Usually 25-40 mm., largest 5 1 mm. Head angle 47° to 51° 37° to 42° 1959] Rosen & Bailey: Middle-American Poeciliid Fishes: Carlhubbsia & Phallichthys 5 poeciliid genera in which the gonopodium of the adult male is asymmetrical. In his key to this group Hubbs (1936) utilized only the most general features of body form and tooth struc- ture to associate Carlhubbsia and Phallichthys. He regarded the differences in their gonopodia as being sufficient for generic distinction. Gonopodial Characters in Carlhubbsia.— The dextral folding of gonopodial rays 3, 4 and 5 in Carlhubbsia is accompanied by asymmetric modifications of many individual elements. Each type of segment ornamentation, e.g., spines, hooks and serrae (Rosen & Gordon, 1953), re- flects the over-all asymmetry either by serial or unilateral reduction or by fusion and consolida- tion with adjacent or underlying structures. Thus, the basic architecture of the gonopodium in Carlhubbsia, as in other poeciliids with asym- metric gonopodia, is often masked by the super- imposed concomitants of folding. In view of the proved value of specialized terminal features of the gonopodium to poeciliid taxonomy, it is of phylogenetic importance to distinguish between the basic pattern of terminal modifications and the asymmetric distortions to which the structure has been subjected. Ray 3, for example, is bilaterally asymmetri- cal; the segments of the left half of this ray are simple oblong elements with minute denticles on the subdistal elements, whereas those of the right half are transversely widened with long, flat processes arising from their outer, ventral margins. The latter segments form a continuous ridge, that is, the ventral wall of the gonopodial groove. If this ridge is mechanically flattened, or viewed from above, the proximal section re- sembles the eccentric groove and the distal section the series of gonopodial spines that commonly occur in bilaterally symmetrical gonopodia in numerous poeciliid genera. The unilateral reduction of an entire series of seg- ments and the distortion of the remaining series on ray 3 tends to obscure the character of the individual segment types. Nevertheless, these structures appear to have arisen in a manner similar to, if not identical with, the eccentric grooves and spines in forms with symmetrical gonopodia. For this reason they are so treated in our taxonomic reassessment. Consequent to their reduction and consolida- tion in the evolution of asymmetrical gono- podia, some segments have become isolated from their parent rays and/or subsequently fused to others. The isolated cluster of serrae near the tip of the gonopodium in Carlhubbsia is known to arise from ray 4p because early develop- mental stages show these serrae to be continuous with the proximal elements of this ray. Subse- quently the proximal elements are partly ob- literated by fusion with the underlying segments of 4a. The serrae on ray 5 in Carlhubbsia kidderi are isolated from the elements of 5p and are consolidated with the underlying segments of 5a; thus, in this species, they might appear as a de novo feature of ray 5a. In C. stuarti, how- ever, these same serrae, though equally well consolidated with the segments of 5a, are con- tinuous with the elements of 5p, which in this species are persistent. It seems certain, therefore, that these serrae in Carlhubbsia originated from a developmental field associated with ray 5p. Carlhubbsia stuarti, n. sp. PI. I; Text-figs. 1, 3, 5 Material.— The holotype (UMMZ 146084) is an adult male, 38.3 mm. in standard length, seined in the Rio Polochic at the “playa,” about 0.5 km. east of Panzos, Alta Vera Paz, Guate- mala, on March 6, 1940, by Laurence C. Stuart. The allotype (UMMZ 172455), an adult fe- male 50.5 mm. long, and 290 other specimens (UMMZ 146078) including half-grown and adult males and females from 29 to 55 mm. long, were taken with the holotype. UMMZ 146093, 5 subadult to adult specimens from 29 to 39.5 mm. long, seined in “El Canal,” a diversion of Rio Polochic, about 10 km. by river below Pan- zos, Alta Vera Paz, Guatemala, on March 8, 1940, by Stuart. Diagnosis.— A large, deep-bodied, large-headed species of Carlhubbsia with 8 to 13 well to moderately developed narrow dusky bars on the side. Dorsal fin rather bluntly rounded anterior- ly, projecting beyond succeeding rays to give the fin a falcate outline, with dusky distal band. Gonopodium with series of 15 to 20 terminal serrae on lateral margin of left half of ray 5a, the proximal members usually bidentate, the last member continuous with segments of ray 5p; tip of ray 4p with a cluster of approximately 8 serrae, the subdistal elements of this ray ex- tremely weak but not obsolescent. Gonopodial suspensorium with uncini developed on gonapo- physes I and II. Vertebrae 28. Preorbital canal with 4 pores in adult. Scales in lateral series 25 or 26. Body-circumference scales 23 to 26. Pec- toral fin rays 13, rarely 14. Head angle 47° to 51°. For the distinctive features of body and fin form, pigmentation and skeletal morphology, see also Tables 1-3, 13-18. General Description.— A deep-bodied, robust species with moderately high and angular con- tours. In adult males the predorsal profile is •6 Zoologica: New York Zoological Society [44: 1 Text-fig. 1. Skeleton of adult male Carlhubbsia stuarti, n. sp. Compare the following positional and struc- tural features with these items in Text-fig. 2: (a) development of neural plates on anterior trunk vertebrae, (b) position of pelvic girdle, (c) curvature of distal tips of posterior pleural ribs, (d) form of first three sexually modified, attached hemal spines (gonapophyses), (e) relative positions of dorsal and anal fin origins, and (f) length of sexually modified anal fin (gonopodium). Tracing from an X-ray. Text-fig. 2. Skeleton of adult male Phallichthys fairweatheri, n. sp. See caption of Text-fig. 1. Tracing from an X-ray. flat or slightly arched from nape backward and rises sharply to the dorsal origin; the dorsal and ventral margins of the caudal peduncle also are quite straight but taper rather gradually to the caudal base. In adult females the dorsal profile usually is slightly arched, and rises less abruptly to the dorsal origin; the anterior part of the caudal peduncle is relatively heavier in adult females and its slightly concave dorsal and ven- tral margins taper more abruptly to the caudal base. In both sexes the head angle measures 47° to 51°. In both sexes the snout is distinctly blunt and the lower jaw rises sharply and obliquely upward instead of outward from its articulation. The male is relatively deeper bodied than the female, and has more angular contours. The dorsal fin is distinctly falcate in both sexes. The anal fin is broad and fan-shaped. In adult females the dorsal originates closer to caudal base than to snout, and the anal originates slightly behind the vertical from the dorsal origin. In adult males the two fins originate in the same plane as a result of the forward migration of the anal fin during sexual differentiation. The tip of the gonopodium extends posteriorly ap- proximately two-thirds the distance from anal origin to caudal base. In both sexes the caudal fin is broad, subtruncate and only slightly round- ed at its dorsal and ventral trailing edges, and is yellow toward the base. There are from 17 to 19 principal caudal rays, most often 17. The pelvic fins are well developed in both sexes 1959] Rosen & Bailey: Middle-American Poeciliid Fishes: Carlhubbsia & Phallichthys 1 and are bright yellow. In adult females the pel- vics are almost as long as the anal, originat- ing on the belly one-half the distance from the anal origin to the edge of the opercle. When folded back the tip of the longest pelvic ray extends to the anal opening or barely overlaps the anal fin origin. In adult males the pelvics originate just anterior to the base of the gono- podium and the longest ray overlaps and extends beyond the bases of modified anal rays 1 and 2. In both sexes the pectoral fins are broadly spatu- late and originate well below the midlateral line just behind the opercular margin; they extend obliquely upward and backward to a vertical from the dorsal origin. The gillrakers on the outer face of the first arch, though well developed, are short and slen- der; they number 18 to 21. Two principal pigment patterns on the trunk and caudal peduncle consist of a cross-hatched reticulum, most evident on the dorsum but clear- ly defined on the sides and venter as well, and a series of as many as 13 to 15 vertical bars su- perimposed on the reticular network. These bars are most pronounced in adult males. The bars are best developed on the caudal peduncle but extend anteriorly almost to the pectoral base. Each bar is long and slender, extending usually to within one scale row from the middorsal and midventral lines. There is no dark subocular bar or tear drop. In adults of both sexes the distal third of the dorsal fin is dusky. There is a basal band consisting of darkened interradial mem- branes in the proximal third of the dorsal fin. Between the darkened bands, the middle third of the fin is lighter; anteriorly there is a light sprinkling of melanophores but posteriorly the membranes are clear. Probably this area was colored in life. Other fins are clear, with only a scattering of fine melanophores at their bases. Skeletal Morphology .—The vertebral axis con- sists of 28 elements in each of 23 specimens X-rayed, including the holotype. The column is divided approximately in half into a precaudal or trunk series and a caudal series, the division being determined by the first hemal spine. In adult females, the first hemal spine usually emerges from the 15th vertebra. In adult males, the first hemal spine or gonapophysis is on the 14th vertebra. The vertebral axis takes the form of a gentle sigmoid curve, the precaudal section arching upward, especially in large adults. The first cervical vertebra bears no rib, but a pleural rib is present on most of the remaining precaudal vertebrae. The first rib is long and slender and is loosely articulated with the posterodistal margin of the transverse process of the second vertebra; the rib lies against the medial surface of the pectoral girdle. Near the distal end of the rib there lies a free stylet-shaped bone which is expanded proximally and slender distally. This was termed the postcleithrum by Woods & Inger (1957: 247). This bone is well separated from the cleithrum and there is no other bone in series with it. The first 9 or 10 pairs of ribs are long, after which they diminish in size. The last de- finitive pleural rib usually occurs on the 13 th vertebra but a delicate rudimentary rib may occur on the 14th. Minute epipleurals are pres- ent on all but the last two or three pleural ribs. In adult males the last five or six ribs are sexually modified. The last three or four are relatively small, extremely slender, and are widely sepa- rated at their tips; the tips of the preceding two or three ribs are bent sharply forward toward the pelvic girdle but do not converge. In adult males parapophyses occur on the first and second caudal vertebrae though they are usually quite small and closely applied to the base of the modi- fied hemal spines or gonapophyses. In adult females parapophyses occur on the first two or three caudals. The so-called parapophyses of the more posterior trunk vertebrae, and of the an- terior caudals in other species, actually are serial- ly homologous with the transverse processes of the cervical and postcervical elements. The gonopodial suspensorium of adult males receives a contribution from the vertebral axis of four specialized hemal spines. The first, or ligastyle, is a long, slender rod that originates on the 13 th vertebra and migrates forward so that its dorsal margin comes to lie directly be- neath the centrum of the 10th vertebra. The remaining three hemal spines become special- ized chiefly by the addition of bony substance to their distal and posterior surfaces and are referred to as gonapophyses. The first two incline forward. Gonapophysis III is either slightly arched forward or essentially vertical. Only gonapophyses I and II bear uncini, which emerge proximally on the spines close to the vertebral centrum; the uncini of the second spine, how- ever, are slightly farther down the shaft and more robust. The actinosts of the anal fin in the adult male also are specialized and are incorporated into the suspensorial system as the gonactinosts (for general orientation see Rosen & Gordon, 1953: 11-13, Text-figs. 14-16). The first is a short, heavy, blunt rod that inclines sharply forward. It articulates with fin rays 1 and 2. Gonactinosts 2, 3 and 4 are fused to form a single highly complex plate of bone, the primary gonactinostal complex, which supports fin rays 3, 4 and 5. The complex is greatly dilated antero-posteriorly, having a distinct rostral bulge just above the tip 8 Zoologica: New York Zoological Society [44: 1 of the first gonactinost. The posterior lateral wings that are produced symmetrically from eccentric positions along either side of the core of gonactinost 4 are rather well developed, par- ticularly at the dorsal third of the complex where they flare broadly. Gonactinost 5 lies embedded in the trough formed by the lateral wings of the primary complex. Gonactinosts 6, 7, 8 and 9 are slender and rod-like; they are closely joined and their tips converge slightly. Actinost 10 ot the anal fin becomes much reduced or obsoles- cent and is incorporated into the base of gonac- tinost 9 as a tiny sliver of bone. The gonactinosts are firmly anchored to the vertebral axis by means of the ligamentous tissues associated with the ligastyle and the gonapophyses. The primary gonactinostal complex is associated with the ventral margins of the ligastyle and gonapophysis I. The tip of gonapophysis II interdigitates with gonactinosts 7 and 8. The tip of gonapophysis III curves forward toward the base of the final gonactinost. The gonopodium in the adult male consists of the produced and modified rays 3, 4 and 5 of the anal fin. Together these three rays are folded to form a broad groove along the right side of the gonopodium. When held at rest, i.e., pointed caudally, ray 3 forms the ventral border, ray 5 the dorsal border and ray 4 the lateral wall of the dextral groove. The paired halves of ray 3 are segmented to the tip of the fin and are never consolidated. On ray 3 there is a series of from 15 to 20 long spinous processes which are fol- lowed distally by 5 to 8 simple terminal seg- ments. The spines are present on the right half of the ray only; they form the ventral margin of the dextral groove. The terminal elements of ray 4a are expanded dorso-ventrally, and their segmentation axes are at right angles to the long axis of the ray. The elements of ray 4p are ex- tremely delicate terminally but are never obso- lescent. Its sinistral branch terminates in a cluster of 7 or 8 subtriangular serrae which face out- ward, away from the gonopodial groove on the convex surface of the permanently folded fin. On ray 5p there is a single series of unpaired retrorse distal serrae; they originate on the sinis- tral portion of this ray and are fused with the underlying segments of 5a. The unpaired serrae are subtriangular, approximately 15 in number. Several of the proximal members of this series are bidentate. Proximally on 5p, just below the tip of ray 6, there is a series of 5 to 10 minute retrorse serrae. The elements of ray 5a are flattened dorso-ventrally but are otherwise sim- ple; they extend to the tip of the fin. The tips of all the rays became slender distally but show no significant displacement from their original axes; they are grouped rather tightly together and are never separated by distinct tissue spaces. The gonopodium as a whole has a distinctly acu- minate profile and is terminated by a rigid, hook- like cap of tough membranous tissue. Relationship— This species is clearly referable to the genus Carlhubbsia on the basis of details in the gonopodium and gonopodial suspenso- rium (see Text -figs. 3 and 5 and discussion on pp. 29-32). It may be separated readily from the only other known species, C. kidderi, by the characters listed in the diagnoses and in Table 1. The two species are allopatric, being sepa- rated geographically by a linear distance of ap- proximately 45 miles in drainages which empty on opposite sides of the Yucatan Peninsula (Map 1). The origin and relationship of the members of the genus Carlhubbsia are discussed below (pp. 35-39). Habitat— At the two known localities Dr. Stuart recorded the water as muddy, but not excessively so, the bottom as sandy mud, the current as moderate and as slow, the tempera- ture as warm, and the depth of capture as up to 5 feet. Range.— Carlhubbsia stuarti is recorded only from the drainage of Rio Polochic, Guatemala (Map 1). Robert R. Miller, in 1946, found this species to be common also in Lake Izabal, the terminus of Rio Polochic. Etymology.— This species is named in honor of Dr. Laurence C. Stuart in recognition of his efforts in collecting this and many other species of freshwater fishes in Guatemala during the past quarter-century. Carlhubbsia kidderi (Hubbs) PI, II; Text-figs. 3, 5 Allophallus kidderi. — Hubbs, 1936: 232-238 (original description; Rio Champoton, Cam- peche, Mexico). Scrimshaw, 1944: 182 (su- perfetation). Scrimshaw, 1945: 234-237 (viviparity). Scrimshaw, 1946: 21-22 (egg size). Aulophallus kidderi.— Scrimshaw, 1945: 239- 241 (lapsus for Allophallus). Carlhubbsia kidderi.— Whitley, 1951: 61 (re- placement synonym tor Allophallus). Rosen & Gordon, 1953 : 2, 28 (mechanics of gono- podium; reference by generic name only). Material.— Rio Champoton drainage (Cam- peche, Mexico) : UMMZ 102206 (71 hf.-gr., ad. males and females, 16 to 40 mm. standard length), Rio Champoton, at Janateya, 7 leagues east of Champoton, July 9, 1932, E. P. Creaser and A. S. Pearse; UMMZ 102199 (holotype of 1959] Rosen & Bailey: Middle-American Poeciliid Fishes: Carlhubbsia & Phallichthys 9 Map i. Lower Middle America giving distribution by record stations for the forms of Carlhubbsia and Phallichthys. Allophallus kidderi Hubbs, 16 mm.) and UMMZ 102200 (3 subad., 11 to 16 mm.), Rio Cham- poton, about 11 miles from mouth, July 11, 1932, E. P. Creaser. Rio San Pedro de Mdrtir drainage (El Peten, Guatemala) : UMMZ 144206 (4 subad., 9 to 32 mm.), Laguna de Yalac, in course of Rio San Pedro, about 6 leagues by river (easterly of El Paso de los Caballos), March 17, 1935, Carl L. Hubbs and Henry van der Schalie; UMMZ 144204 (185 hf.-gr., ad. males and females, 16 to 51 mm.). Laguna de Yalac, in course of Rio San Pedro, about 6 leagues by river above El Paso de los Caballos, March 16, 17, 1935, Hubbs and van der Schalie; UMMZ 144205 (one ad. female, 39 mm.), Rio San Pedro, about !4 mile below Laguna de Yalac, about 6 leagues by river east of El Paso de los Caballos, March 18, 1935, Hubbs and van der Schalie; UMMZ 144201 (28 hf.-gr. to ad., 17 to 42 mm.), north end of Laguna Perdida, 6 leagues south of El Paso de los Caballos, March 6, 7, 1935, Hubbs and van der Schalie; UMMZ 144202 (138 yg. to ad., 13 to 34 mm.), lagoon- like arm of Rio San Pedro at El Paso de los Caballos, March 10-14, 1935, Hubbs and van der Schalie; UMMZ 144203 (94 hf.-gr. to ad. males and females, 16 to 35 mm.), lateral waters of Rio San Pedro at Desempeno, just below El Paso de los Caballos, March 1 2, 1 935, Hubbs; UMMZ 144207 (two ad. females 26 and 32 mm.), Rio San Pedro, at Mactun Rapids, about 10 Zoologica: New York Zoological Society [44: 1 o o CL, o C V. . §>.8 U * >,5 ^ 2! £ 0 ~ - « * »v Q d w e « u o ^ ® -a - d 'd ,S „ OX) .5 ofl ftc « S3 — K j « '-3 eg d S d 2 u u c » J J ■ r „ d 5 pQ •“ 8 § ~ - <+-i Xl « T3 cfl ’ a o • — M § s £p a • L (d (Q d « qj T .,= rgo»? S§3hS - _ aii^oJa « — .22 « £ 8 -s 5 | .S £ r g •a u ‘C j> ^ ' — * — «» II 5 ° ^ £ w H >,_ o _ f= d 5 — « aJ I S 2 P< -a g § ;§> *■» -*-1 s — ' d> Q OX) ^ < ■g > -C !§>.| |T| 55 -•«§ Is 2 - <4H <4-« C ■ C a -e &< - S*s - .Sf C.2 C S J3 >- . si . - .is ~ j§ 's H i - « — -C J3 -5 „• Ji C Q y a oc S ^ 5 ^ ^ g > 2 s o s -b e ~ 3 u, a b * 2 :%i J CL, . S)"S o ft* £ *+-» C3 Q d, S ° 1 § o c O o © .2 2 c C <-» ^ ~ a. a “■s*-- XS ■g 2^-Sj-2 W «3 ^ 2 *~2 *D u-t O DX) JO X3 O- (U ‘ui ^ 2 u; O ■~ ” M ^ 3 -o a c a >> <*> O X .s ~ OJJ Q. o > *0 -*-» *-> O W > > TOTAL A V / 3 ' V V. \ O-J - \ 2 x /!3 7 3 / /so 124/ y/6 Text-fig. 6. Frequency distribution of dorsal and pectoral fin ray counts in the subspecies of Phallich- thys amates. In each box the number of specimens appears in the upper left for P. a. amates, in the lower right for P. a. pittieri. [?=Nigra], Conquantu, Western Panama). Behre, 1928: 316 (synonymy; Skoon Creek, tributary to Rio Tiliri, tributaries to Almirante Bay and Chiriqui Lagoon, west- tern Panama). Hildebrand, 1930: 6 (Si- quirres, Costa Rica; characters; taxonomy) . Jordan, Evermann & Clark, 1930: 190 (sy- nonymy, in part; type locality). Poeciliopsis isthmensis.— Regan, 1913: 997, pi. 100 (original description; Colon, Panama). Meek & Hildebrand, 1916: 325 (Colon, Panama) . ?Breder, 1925: 141 (Gatun Spill- way, Panama; a young female, questionable reference). Phallichthys isthmensis.— Myers, 1925: 370 (distinct from pittieri and amates). Material.— CNHM 7841 and UMMZ 177277 (8 yg. to ad. males and females, 19 to 50 mm. standard length), Guapilis, Limon, Costa Rica, 1912, S. E. Meek; CNHM 7839 (paratypes of Poecilia pittieri Meek) (14 ad. males and fe- males, 26 to 52 mm.), Rio Reventazon, La Junta, Limon, Costa Rica, April 7, 1912, Meek; UMMZ 177276 and USNM 92157 and 92158 (29 ad. males and females, 20 to 41 mm.), Si- quirres, Limon, Costa Rica, Oct. 12, 1928, A Alfaro; UMMZ 72585 (7 hf.-gr. to ad. males 20 Zoological New York Zoological Society [44: 1 Table 7. Proportional Measurements of Ten Adult Males of Phallichthys amates pittieri Proportions are expressed in thousandths of the standard length. See text for source of specimens. Measurement CNHM 7839 UMMZ 177277 USNM 94108 UMMZ 177277 (CNHM 7841) CNHM 7839 UMMZ 177276 USNM 92158 Standard length (mm.) 34.5 33.5 27.2 27.1 26.5 26.0 24.5 23.6 22.7 21.2 Body, greatest depth 383 388 331 328 377 346 327 318 352 330 Caudal peduncle, least depth. . 232 230 206 221 226 231 225 212 211 203 Dorsal origin to snout tip 522 543 529 517 509 538 498 508 533 566 Anal origin to mandibular symphysis 528 555 515 517 547 515 527 530 524 557 Dorsal origin to caudal base . . 554 567 537 517 566 538 555 517 520 557 Anal origin to caudal base . . . . 560 585 559 568 566 558 567 542 551 495 Head length 275 278 279 284 279 288 318 284 286 311 Head width 171 161 165 170 159 165 176 191 198 203 Snout length . 78 84 85 77 75 77 85 75 80 Orbit length 90 84 103 96 94 96 98 97 97 94 Postorbital length of head . . . . 116 107 118 125 113 115 143 140 141 127 Orbit to angle of preopercle. . 61 75 70 66 68 69 69 68 66 71 Interorbital, bony width 133 119 118 125 125 123 131 140 141 132 Mouth, over-all width 99 104 92 89 98 108 102 104 Dorsal fin, depressed length . . . 464 400 393 406 442 415 396 390 396 425 Anal fin, depressed length . . . . 545 513 507 531 559 558 486 525 551 552 Caudal fin length 371 331 360 343 385 369 383 373 Pectoral fin length 252 227 243 240 272 265 267 247 255 Pelvic fin length 223 239 202 203 234 238 245 212 238 212 Table 8. Proportional Measurements of Ten Adult Females of Phallichthys amates pittieri Proportions are expressed in thousandths of the standard length. See text for source of specimens. Measurement CNHM 7839 Standard length (mm.) 52.0 49.0 46.8 45.5 45.4 44.0 38.6 35.0 34.6 33.0 Body, greatest depth 404 422 395 398 414 409 402 391 384 379 Caudal peduncle, least depth 223 227 214 215 220 216 207 214 202 212 Dorsal origin to snout tip 560 571 551 569 564 557 583 557 552 545 Anal origin to mandibular symphysis. 650 665 645 659 641 639 661 663 647 639 Dorsal origin to caudal base 498 531 515 495 504 516 490 506 512 506 Anal origin to caudal base 435 416 415 424 412 416 391 403 410 415 Head length 267 271 271 266 271 280 277 297 280 285 Head width 194 206 192 191 200 200 187 186 182 158 Snout length 85 88 88 79 84 86 86 89 87 85 Orbit length 88 90 96 97 93 91 104 103 101 106 Postorbital length of head 117 114 111 112 110 111 109 114 116 121 Orbit to angle of preopercle 71 71 68 70 68 70 67 66 66 64 Interorbital, bony width 140 149 139 143 148 145 153 149 145 142 Mouth, over-all width 102 116 107 108 107 117 109 116 109 Dorsal fin, depressed length 352 371 368 356 357 364 352 351 364 364 Anal fin, depressed length 252 259 252 248 256 259 259 251 260 255 Caudal fin length 346 358 343 365 337 367 Pectoral fin length 240 249 239 235 249 245 259 249 237 255 Pelvic fin length 215 229 212 215 207 223 220 217 220 233 1959] Rosen & Bailey: Middle- American Poeciliid Fishes: Carlhubbsia & Phallichthys 21 Table 9. Proportional Measurements of Ten Adult Males of Phallichthys amates amates Proportions are expressed in thousandths of the standard length. See text for source of specimens. Measurement | UMMZ 173364 UMMZ 65220 UMMZ 173280 CNHM 56168 USNM 101780 UMMZ 113403 Standard length (mm.). . 29.0 28.0 28.0 25.5 24.8 22.5 20.0 19.7 16.0 14.0 Body, greatest depth .... 407 411 411 373 383 382 365 360 313 322 Caudal peduncle, least depth 238 236 239 235 226 227 230 218 206 215 Dorsal origin to snout tip 562 578 571 565 544 578 560 584 506 572 Anal origin to mandibular symphysis 510 525 518 506 504 507 500 497 488 479 Dorsal origin to caudal base 472 486 489 549 504 560 525 492 481 479 Anal origin to caudal base 569 582 578 596 564 578 595 584 563 565 Head length 293 311 311 302 290 302 250 315 250 286 Head width 207 214 211 196 194 191 200 193 181 193 Snout length 107 103 107 78 81 76 80 96 113 93 Orbit length 93 96 96 94 97 102 105 102 100 107 Postorbital length of head 124 132 132 137 121 129 130 122 125 107 Orbit to angle of preopercle 76 75 75 63 60 67 70 66 44 43 Interorbital, bony width. . 159 153 153 141 125 142 135 142 106 122 Mouth, over-all width. . . 117 121 114 114 101 111 110 96 100 107 Dorsal fin, depressed length 348 343 386 353 363 338 375 345 300 300 Anal fin, depressed length 466 475 479 478 480 490 533 475 544 Caudal fin length 345 378 411 375 378 380 386 300 329 Pectoral fin length 272 268 289 275 278 249 250 274 244 215 Pelvic fin length 234 250 239 235 234 200 215 233 200 193 and females, 23 to 37 mm.), Skoon Creek, trib- utary to Rio Tiliri, tributary to Rio Sixaola, Talamanca, Costa Rica, Jan. 25, 1923, E. Behre and Chambers; USNM 94201 (8 ad. males and females, 28 to 39 mm.), Descampos, 1200 me- ters, Costa Rica, 1928, A. Alfaro; USNM 94108 (4 ad. males and females, 25 to 41 mm.), Tiribi at 1,200 meters elevation, Costa Rica, May 7, 1928, A. Alfaro; UMMZ 72587 (3 ad. females, 34 to 52 mm.), San San Creek, tributary to San San River at old San San Farm, Bocas del Toro, Panama, Feb. 5, 1923, Behre and Chambers; UMMZ 72586 (4 hf.-gr. to ad. males and fe- males, 29 to 51 mm.), Fruitdale Creek, along railroad spur back of Almirante, July and Aug., 1921, Behre; UMMZ 72588 (one yg., 14 mm.), Quebrada Nigra, flowing into Almirante Bay, Panama, July 8, 1921, Behre; UMMZ 72584 (one ad. female, 35 mm.), Guibari Creek, trib- utary to Rio Cricamola, below Conquantu, Pan- ama, Feb. 23, 1923, Behre and Chambers; UMMZ 72590 (4 ad. males, 24 to 33 mm.), Nomonuen Creek, tributary to Rio Cricamola above Conquantu, Feb. 22, 1923, Behre and Chambers; UMMZ 72589 (one ad. male, 24 mm.), small creek tributary to right bank of Rio Cricamola below Conquantu, Feb. 26, 1923, Behre and Chambers. Regan’s (1913) record of this form from Colon was questioned by Hildebrand (1938), who stated that “Having collected rather exten- sively in 1911 and 1912, and again in 1935 and 1937, in the vicinity of Colon, from whence the types of P. istlimensis were reported, I am obliged to conclude that the species either is very rare there, or that a mistake was made in the earlier record.” Diagnosis. — A subspecies of Phallichthys a mates with dorsal rays about equally 10 or 11, the first 2 rays (one only in 9% of specimens) simple in adult (Table 13 and Text-fig. 6). Pec- toral rays 12 or 13, most often 13 (82% have 13 rays on at least one side). Dorsal fin larger than in amates: the depressed length 39 to 47% of standard length in adult males and 35 to 37% in adult females. Vertical bars usually present, though poorly defined, in adult females. Al- though our material of this subspecies averages larger than that of P. a. amates, the maximum sizes do not differ appreciably. The longest speci- mens examined are 40 mm. standard length 22 Zoologica: New York Zoological Society [44: 1 Table 10. Proportional Measurements of Ten Adult Females of Phallichthys amates amates Proportions are expressed in thousandths of the standard length. See text for source of specimens. Measurement UMMZ 173280 CNHM 56168 UMMZ 173280 CNHM 56168 UMMZ 173280 Standard length (mm.). . 37.0 32.6 31.1 29.1 27.8 27.5 26.3 25.9 25.1 25.0 Body, greatest depth. . . . Caudal peduncle, 389 377 395 385 370 364 361 347 371 400 least depth 214 209 222 210 205 207 202 201 211 228 Dorsal origin to snout tip Anal origin to mandibular 589 583 563 581 594 585 586 591 578 572 symphysis Dorsal origin to caudal 646 641 653 632 633 647 635 656 629 644 base 473 488 508 484 482 476 460 452 482 476 Anal origin to caudal base 403 411 408 409 414 393 399 386 422 424 Head length 303 307 302 299 309 309 304 313 307 296 Head width 203 205 215 206 209 211 209 216 219 208 Snout length 92 86 96 86 90 95 91 85 84 80 Orbit length 97 95 100 103 112 102 103 104 108 104 Postorbital length of head Orbit to angle of 124 132 135 127 137 135 137 131 135 140 preopercle 62 67 64 76 72 69 65 69 76 76 Interorbital, bony width . . 157 166 161 165 165 164 156 166 167 164 Mouth, over-all width . . . Dorsal fin, depressed 124 117 122 120 119 120 106 120 124 124 length 297 304 293 278 281 284 278 274 288 Anal fin, depressed length 270 276 283 261 255 258 270 259 271 264 Caudal fin length 346 350 354 333 353 367 346 351 374 356 Pectoral fin length 257 245 264 251 259 258 247 251 239 264 Pelvic fin length 224 227 231 227 205 200 198 201 203 224 (male) and 52 mm. (female). Most adult males are between 22 and 34 mm. long. Habitat— Meek (1914) reported this form from swift rocky streams in Costa Rica, and specimens were collected by Alfaro at Tiribi and Descampos, Costa Rica, at altitudes of 1,200 meters; other stations are lower. Range.— Known from the Caribbean slope of Costa Rica and western Panama (Map 1). Phallichthys amates amates (Miller) PI. IV; Text-figs. 4, 5 Poecilia amates.— Miller, 1907: 108 (original description; Los Amates, Guatemala). Poeciliopsis amates— Henn, 1916: 120 (tax- onomy; gonopodium; coloration). Phallichthys amates— Hubbs, 1924: 10 (tax- onomy; Tela, Honduras) . Myers, 1925 : 370 (distinct from pittieri and isthmensis). Hubbs, 1926: 70 (synonymy; records). Jor- dan, Evermann & Clark, 1930: 190 (syn- onymy, in part; type locality). Rosen & Gordon, 1953: 24, 29, 33, 38 (mechanics of gonopodium; sexual behavior) . Material. — Guatemala: UMMZ 65220, CNHM 56168 (12 yg. to ad. males and females, 18 to 26 mm. standard length), Los Amates, Izabal, Jan. 17, 1905, N. Miller. Honduras: UMMZ 173280 (25 yg. to ad. males and females, 15 to 37 mm.) , Rio Mapache at Masca, Cortes, April 5, 1951, Gordon and Wheeler; UMMZ 173297 (29 yg. to ad. males and females, 18 to 40 mm.), Rio Tulian, Tulian, west of Puerto Cortes, Cortes, April 6, 1951, Gordon and Wheeler; USNM 101780 (one ad. male, 16 mm.), Rio Chamelecon, 8 miles above San Pedro, Cortes, Jan. 19, 1936, Blanchard; UMMZ 173147 (8 yg. to ad. males and females, 12 to 24 mm.), Rio Benejo, tributary to Rio Chamelecon, just north of San Pedro Sula, Cor- tes, Mar. 18, 1951, Gordon and Chable; UMMZ 173156 (27 yg. to ad. males and females, 15 to 36 mm.), tributary to Rio Ulua, Agua Priete, north of San Pedro Sula, Choloma Road, Cortes, Mar. 18, 1951, Gordon and Chable; UMMZ 56875 (one subad. male, 18 mm.). Tela, At- lantida, Mar. 14, 1923, T. H. Hubbell; UMMZ 113403 (one subad., 15 mm. and one ad. male, 16 mm.), river just outside of Tela, Atlantida, Spring, 1936, A. Greenberg; UMMZ 173177 ( 13 yg. to ad., 18 to 35 mm.), tributary to Rio 1959] Rosen & Bailey: Middle- American Poeciliid Fishes: Carlhubbsia & Phallichthys 23 Table 11. Proportional Measurements of Ten Adult Males of Phallichthys fairweatheri Proportions are expressed in thousandths of the standard length. See text for source of specimens. Measurement UMMZ 144186 Holo- type UMMZ 172456 UMMZ 144186 Standard length (mm.). 30.1 29.7 28.7 27.4 25.5 25.5 25.5 25.5 24.0 23.0 Body, greatest depth. . . Caudal peduncle, 415 438 411 394 392 412 361 404 354 348 least depth 249 253 230 237 235 235 231 235 233 217 Dorsal origin to snout tip 598 Anal origin to mandibular 606 610 584 588 588 588 596 567 565 symphysis Dorsal origin to caudal 498 525 537 511 510 518 510 537 504 522 base 498 495 495 493 482 506 471 471 496 461 Anal origin to caudal base 588 593 606 599 569 588 580 584 558 565 Head length 299 303 303 310 314 314 302 302 313 313 Head width 179 162 174 153 173 176 173 173 167 174 Snout length 86 88 91 88 86 90 90 90 83 87 Orbit length 106 101 105 99 110 106 110 114 104 109 Postorbital length of head 123 Orbit to angle of 114 122 117 122 122 118 118 125 126 preopercle 63 61 63 66 67 71 71 63 63 65 Interorbital, bony width . 116 121 122 110 110 122 114 118 125 130 Mouth, over-all width . . Dorsal fin, depressed 100 98 91 91 94 94 94 98 88 87 length 342 374 341 350 353 349 357 333 354 330 Anal fin, depressed length 585 606 610 602 580 596 608 624 617 622 Caudal fin length 365 380 366 365 392 373 373 377 371 369 Pectoral fin length 239 253 230 230 239 247 254 261 Pelvic fin length 219 202 195 201 212 196 196 212 204 196 Lancetilla, 1 mile south of Tela, Atlantida, Mar. 20, 1951, Gordon and Chable; UMMZ 173193 (1 ad. female, 29 mm.), Rio Lancetilla at Lancetilla near Labor Camp swimming pool, near Tela, Atlantida, Mar. 22, 1951, Gordon and K. J. Davidson; UMMZ 173206 (20 yg. to ad., 15 to 39 mm.) and UMMZ 173214 (24 hf.- gr. to ad. males and females, 14 to 36 mm.), Lily Pond, Lancetilla Experimental Station, Lancetilla, near Tela, Atlantida, Mar. 22, 1951, Gordon and Davidson; UMMZ 173231 (246 yg. to ad., 15 to 37 mm.) , drainage ditch on Sec. 8, African Oil Palm Plantation, San Alejo, At- lantida, Mar. 24, 1951, Gordon and Davidson; UMMZ 173221 (26 yg. to ad., 14 to 47 mm.), tributary to Rio San Alejo, San Alejo, Atlantida, Mar. 24, 1951, Gordon and Davidson; UMMZ 173344 (35 yg. to ad., 14 to 30 mm.), near San Juan Benque, 48.5 km. west of La Ceiba, Atlan- tida, April 11, 1951, Gordon, Chable and George; UMMZ 173356 (6 hf.-gr. to ad., 20 to 27 mm.), near San Juan Benque, 47.3 km. west of La Ceiba, Atlantida, April 11, 1951, Gordon, Chable and George; UMMZ 173364 (63 yg. to ad., 19 to 39 mm.), Rio Cuero, near La Masica, Atlantida, April 11, 1951, Gordon, Chable and George; UMMZ 173372 (29 yg. to ad., 18 to 37 mm.), Rio Salado canal, Atlan- tida, April 11, 1951, Gordon, Chable and George; UMMZ 173318 (20 yg. to ad., 17 to 56 mm.), stream 6.6 km. east of La Ceiba, at Standard Fruit Company, Atlantida, April 10, 1951, Gordon and Chable; UMMZ 173329 (149 yg. to ad., 11 to 43 mm.), 18.3 km. E. of La Ceiba, Atlantida, April 10, 1951, Gordon and Chable. Diagnosis— A subspecies of Phallichthys amates with dorsal rays 8 to 10, usually 9 (10 in 4% of specimens), the first ray (2 rays in 3% of specimens) simple in adult (Table 13 and Text-fig. 6). Pectoral rays 11 to 13, usually 12 (only 4% have 13 rays on one or both sides). Dorsal fin smaller than in pittieri; the depressed length 30 to 39% of standard length in adult males and 27 to 30% in adult females. Vertical bars not evident in adult females. The longest specimens examined are 30.6 mm. standard length (male) and 56 mm. (female). Most males are between 20 and 30 mm. long. Habitat.— As judged from the field records of Myron Gordon during 1951 in Honduras, this 24 Zoologica: New York Zoological Society [44: 1 Table 12. Proportional Measurements of Ten Adult Females of Phallichthys fainveatheri Proportions are expressed in thousandths of the standard length. See text for source of specimens. Measurement UMMZ 144186 Allo- type UMMZ 172457 UMMZ 144186 Standard length (mm.) . . 38.3 33.3 33.0 31.8 31.2 31.0 29.5 28.6 28.5 27.1 Body, greatest depth. . . . 381 408 382 377 394 406 390 388 379 387 Caudal peduncle, least depth 225 216 221 204 218 210 220 220 218 214 Dorsal origin to snout tip 632 616 639 613 615 629 631 626 628 635 Anal origin to mandibular symphysis 674 666 651 670 683 690 671 682 667 668 Dorsal origin to caudal base 423 453 445 425 445 442 447 437 432 432 Anal origin to caudal base 418 423 406 396 417 400 417 395 421 410 Head length 313 324 324 321 330 332 325 332 337 336 Head width 183 189 191 186 192 203 193 192 193 199 Snout length 94 99 91 94 96 100 95 94 102 100 Orbit length 110 111 109 113 109 116 112 112 116 122 Postorbital length of head 133 132 133 126 128 139 136 140 133 140 Orbit to angle of preopercle 57 66 61 63 64 65 68 66 63 63 Interorbital, bony width . . 151 159 152 154 154 161 159 161 158 162 Mouth, over-all width. . . 99 105 106 101 103 106 102 112 116 103 Dorsal fin, depressed length 300 309 303 305 304 294 322 294 309 295 Anal fin, depressed length 264 279 267 252 281 269 274 273 Caudal fin length 352 360 361 368 361 366 367 375 365 Pectoral fin length ...... 243 239 250 239 254 252 246 244 Pelvic fin length 172 171 173 179 173 174 170 168 175 166 Table 13. Frequency Distribution of Dorsal Fin Rays in Carlhubbsia and Phallichthy Total Dorsal Rays Simple Dorsal Rays OUCL1C5 All OU USUCW1CS 8 9 10 11 1 2 Carlhubbsia stuarti 5 116 2 1 19 Carlhubbsia kidderi 1 19 20 Phallichthys amates pittieri 45 37 7 72 Phallichthys amates amates 16 108 5 124 4 Phallichthys fairweatheri 31 11 39 2 form has rather broad tolerance to amount of current and type of bottom. At stations where moderate to large samples were taken, the cur- rent was recorded variously as none, slight, swift and rapid; the bottom as mud, mud and sand, hardpan and rubble; the water as clear, brown, stagnant and brown, and cloudy; vegetation (representing several species) as wanting or present; temperatures varied from 22° to 28° C. All specimens were seined from shallow water, usually less than three feet deep. Range. — Known from the Atlantic coastal lowland of the Motagua River system, eastern Guatemala, east to near La Ceiba, north-central Honduras (Map 1). Phallichthys fairweatheri, n. sp. Pis. V, VI; Text-figs. 2, 4, 5 Dextripenis evides ( nomen nudum) .—Turner, 1940: 89 (superfetation). Scrimshaw, 1944: 182 (superfetation). Scrimshaw, 1945: 234- 241 (embryonic development). Scrimshaw, 1946: 21-22 (unnamed form from Guate- mala; egg size). Material. — Holotype (UMMZ 172456), an adult male, 29.7 mm. in standard length, col- lected in Rio San Pedro de Martir, or a branch, about Va, mile below Laguna de Yalac, some 6 leagues (by river) upstream (east) from El Paso de los Caballos, in the Usumacinta River 1959J Rosen & Bailey: Middle- American Poeciliid Fishes: Carlhubbsia & Phallichthys 25 basin, El Peten, Guatemala, on March 18, 1935, by Carl L. Hubbs and Henry van der Schalie. The allotype (UMMZ 172457) , an adult female, 33.3 mm. long, and 6 additional half-grown to adult specimens (UMMZ 144191, 21 to 42 mm. long, were taken with the holotype (see Map 2). Additional specimens are as follows: Rio San Pedro de Martir drainage (El Peten, Guatemala): UMMZ 144190 (19 hf.-gr. to ad. males and females, 13 to 44 mm.), Laguna de Yalac, in course of Rio San Pedro de Martir about 6 leagues by river (east) above El Paso de los Caballos, in front of old chicle station, Mar. 16, 1935, Hubbs and van der Schalie; UMMZ 144189 (76 hf.-gr. to ad. males and females, 19 to 42 mm.), Laguna de Yalac, both sides of old chicle station, Mar. 16-17, 1935, Hubbs and van der Schalie; UMMZ 144188 (one hf.-gr. and one female, 15 and 35 mm.), lagoon- like arm of Rio San Pedro de Martir at El Paso de los Caballos, March 10-14, 1935, Hubbs and van der Schalie; UMMZ 144186 (792 yg. to ad. males and females, 12 to 40 mm.), Rio San Pedro, at or opposite Desempeno, just below El Paso de los Caballos, Mar. 12, 1935, Hubbs. Rio de la Pasion drainage (El Peten, Guate- mala): UMMZ 144185 (9 hf.-gr. to subad., 16 to 19 mm.), Laguna de Eckibix, in savanna region southeast of Santa Ana on south shore about Va mile from west end, Feb. 26, 1935, Hubbs and van der Schalie; UMMZ 144187, (3 hf.-gr. to ad. females, 11 to 19 mm.), Laguna de Eckibix, extreme west end, Feb. 26, 1935, Hubbs; UMMZ 144193 (44 hf.-gr. to ad. males and females, 15 to 39 mm.), Arroyo Subin, at Trinidad, about 2 miles east of Santa Teresa, April 2, 1935, Hubbs and van der Schalie; UMMZ 144197, 144194, and 144192 (77 yg. to ad. males and females, 1 1 to 37 mm.) , Arroyo Subin at Santa Teresa, 13 miles south of La Libertad, April 2-3, 1935, Hubbs, van der Schalie and Taintor; UMMZ 144195 (70 yg. to ad. males and females, 12 to 32 mm.), Arroyo Subin, at second rapids (about 2 miles) above mouth into Rio de la Pasion, April 25, 1935, Hubbs; UMMZ 144196 (7 hf.-gr. to ad. males and females, 19 to 36 mm.), Arroyo Subin, in small bay connected with stream beside third rapids (about 2V2 miles) from mouth into Rio de la Pasion, April 25, 1935, Hubbs. New River drainage (British Honduras): NYZS-GAF 5 (49 yg. to ad. males and females, 11 to 30 mm.), Hill Bank opposite campsite, inlet to lagoon of New River, Mar. 16, 1954, Gordon, Williams and Hamilton. Rio Hondo drainage (British Honduras): NYZS-GAF 6 (71 yg. to ad. males and females, Table 14. Frequency Distribution of Anal Fin Rays in the Forms of Carlhubbsia and Phallichthys Species or Subspecies Anal Rays 9 10 11 Carlhubbsia stuarti 19 1 Carlhubbsia kidderi 1 18 1 Phallichthys amates pittieri 79 Phallichthys amates amates 25 2 Phallichthys fairweatheri 22 12 to 32 mm.), lagoon and creek on east bank of east branch of Rio Hondo, opposite San An- tonio, Orange Walk, Mar. 20, 1954, Gordon, Fairweather and Chaveria; NYZS-GAF 7 (123 yg. to ad. males and females, 11 to 31 mm.), lagoon opposite San Antonio, connected with creek to east branch of Rio Hondo, Orange Walk, Mar. 21, 1954, Gordon and Chaveria. Diagnosis— A deep-bodied, large-headed spe- cies of Phallichthys with severely angular con- tours. There is a series of 2 to 4 broad dusky bars on the caudal peduncle that may fuse ven- trally to form a conspicuous postanal blotch, especially pronounced in adult males. In both sexes, from 6 to 7 rows of bright orange dots extend along scale rows from opercle to caudal base.3 Adult females with several small and irregu- lar black spots above genital opening. Usually with an indistinct subocular dark bar that ex- tends more backward than downward across cheek. Tissue of lower lip developed into a fleshy pad. Dorsal fin rounded, with a black margin, the middle rays longest. Anal fin of female angulate, with conspicuous dark border. Gono- podium dextrally folded, long and tapering, ex- ceeding vertical from caudal base, with a minute terminal swelling of tough membranous tissue; ray 4p with a single row of large, retrorse proxi- mal serrae and smaller erect terminal serrae on left half and 5 or 6 minute serrae on right half; dorsal margin of ray 3 without denticles; right half of ray 3 with series of broad, flat spines, not much incurved and tapering abruptly to the 8 or 9 slender terminal segments; left half of ray 3 without definite spinous processes. Adult males with tips of posterior anal rays black. Gonopodial suspensorium with recurved uncini, when present, developed near base of gonapo- physis I and heavy subtriangular uncini near 3 Red and yellow pigments are water soluble and for this reason are not observed in preserved specimens. The color description is taken from living and recently fixed animals from British Honduras. 26 Zoologica: New York Zoological Society [44: 1 Table 15. Frequency Distribution of Pectoral Fin Rays in the Forms of Carlhubbsia and Phallichthys The counts from left and right fins for each specimen are summed Species or Subspecies 18 Pectoral Rays 19 20 21 22 23 24 25 26 27 Carlhubbsia stuarti Carlhubbsia kidderi 3 Phallichthys amates pittieri Phallichthys amates amates Phallichthys fairweatheri 13 2 3 5 4 10 19 14 13 51 3 115 2 3 1 1 Table 16. Frequency Distribution of Scales in Lateral Series in the Forms of Carlhubbsia and Phallichthys Species or Subspecies Lateral Scales 22 23 24 25 26 27 Carlhubbsia stuarti 11 9 Carlhubbsia kidderi 6 14 Phallichthys amates pittieri 13 62 Phallichthys amates amates 28 6 Phallichthys fairweatheri 5 13 4 Table 17. Frequency Distribution of Circumferential Scales in the Forms of Carlhubbsia and Phallichthys Species or Subspecies Scale Rows Around Body 21 22 23 24 25 26 Carlhubbsia stuarti 2 4 10 4 Carlhubbsia kidderi 9 8 3 Phallichthys amates pittieri 13 7 Phallichthys amates amates 1 9 9 1 Phallichthys fairweatheri 1 10 9 2 tip of gonapophysis II; uncini near tip of gonapophysis III, when present, short, heavy and sharply pointed. Teeth of inner series in broad, strongly curved bands that have well- developed, lateral, posterior projections. Verte- brae usually 28, infrequently 27 or 29. Scales in lateral series 22 to 24. Dorsal fin rays 9 or 10, usually 9, typically with only the first ray simple, rarely with 2 simple rays. Anal fin rays 10. Pectoral fin rays 10 or 11, rarely 12. Head angle 41° to 49°. For the distinctive features of body and fin form, pigmentation and skeletal morphology, see also Tables 6, 11-18. General Description— A deep-bodied, robust species with high, severely angular contours. In adult males the predorsal profile is flat and rises sharply to the dorsal origin; the dorsal and ventral margins of the caudal peduncle are straight and taper rather abruptly to the caudal base. The head angle is 41° to 49°. In adult females the predorsal profile rises less sharply to the dorsal origin; the dorsal and ventral mar- gins of the caudal peduncle slope gently toward the plane of the body axis and taper only gradu- ally toward the caudal base. The head angle is 41° to 46°. In both sexes the snout is distinctly pointed and long, and the head deep and tri- angular in profile. The sharply-pointed appear- ance of the snout in profile view is due, in part, to the development along the outer margin of the lower jaw of a variously developed fleshy pad. In general the male is relatively deeper than the female and has conspicuously rhombic con- tours; those of the female are more curvilinear. The median elevated fins in adult females originate slightly closer to the base of the caudal peduncle than to the tip of the snout and are 1959] Rosen & Bailey: Middle- American Poeciliid Fishes: Carlhubbsia & Phallichthys 27 nearly opposite. In adult males the anal fin or gonopodium originates slightly in advance of the more posteriorly inserted dorsal fin and this is essentially the median position with reference to the body axis. The dorsal fin in both sexes is broadly rounded; the middle and posterior rays exceed the anterior ones in the depressed fin. In males the fin is more elevated and may extend almost to the caudal base. In females the anal fin has an acute anterior angle, the free margin somewhat falcate, and the third and fourth rays are much the longest. The tip of the fully de- veloped gonopodium of the adult male extends backward to slightly beyond the caudal base. The caudal fin in both sexes is broad, subtrun- cate, only slightly rounded at its upper and lower posterior margin. There are 15 or 16 principal caudal rays. The rounded pelvic fins in adult females are small, about one-half the length of the anal fin; they originate approximately one- third the distance from the anal origin to the lower edge of the subopercle. In adult males pelvic rays 2 and 3 are produced; they originate just anterior to the origin of the gonopodium. The pectoral fins are broadly spatulate in both sexes and originate well below the midlateral line just behind the opercular margin, extending backward to a vertical from the dorsal origin. The gillrakers on the outer face of the first arch, though well developed, are short and slen- der; they number 20 to 25. There are three principal pigment patterns on the trunk and caudal peduncle. Two are pro- duced by melanophores : a reticular pattern that is usually most evident above the midlateral line, and a series of vertical bars that is more pronounced in the adult male. The bars are more or less uniform and are restricted largely to the caudal region, only one or two bars of a series of 6 or more being situated anterior to the vertical from the anal origin. Two or three of the posterior bars are darker at their upper and lower extremities than midlaterally, and may ring completely the caudal peduncle. Ventrally the bars, particularly in adult males, may be so intensely black and broad that they fuse to form a large, postanal blotch. Lateral striping con- sists of 6 to 7 rows of bright orange dots that extend along the scale rows from the opercle to the caudal base. Adult females have three or more irregular but intensely black spots just above the anal base. In adult females both ele- vated median fins are evenly edged with black, except for the tips of the first three or four anal rays. In adult males the dorsal fin is black-edged; the gonopodium is unmarked, except for a stip- pling of melanophores and erythrophores at its base, but the tiny posterior anal rays have black Table 18. Frequency Distribution of Number of Vertebrae in the Forms of Carlhubbsia and Phallichthys Species or Subspecies Vertebrae 27 28 29 30 Carlhubbsia stuarti 23 Carlhubbsia kidderi 2 36 3 Phallichthys amates pittieri 1 53 1 Phallichthys amates amates 1 19 1 Phallichthys fairweatheri 1 36 2 tips. The other fins are without distinctive color pattern, but may show a fine dusting of melano- phores, xanthophores and erythrophores, espe- cially near their bases. The lips and the interorbital region are suffused evenly by melan- ophores. A short and faint suborbital bar is occasionally present. Skeletal Morphology— The vertebral axis, consisting usually of 28 elements (Table 18), is divided approximately in half into a precaudal or trunk series and a caudal series, the division being determined by the first hemal spine. In adult females, the first hemal spine usually emerges from the 15th vertebra. In adult males, the first attached hemal spine, which becomes sexually modified to form the first gonopophysis of the gonopodial suspensorium, is on the 14th vertebra. The precaudal portion of the axis rises gently from the vertebral centrum carrying the first hemal and then flattens out again as it approaches the base of the skull. The vertebral axis takes the form of a gentle sigmoid curve, the spinal curvature being especially pronounced in large adults. A pleural rib is present on each precaudal vertebra except the first. The first rib is long and slender and is loosely articulated with the postero-distal margin of the transverse pro- cess of the second vertebra; the rib lies against the medial surface of the pectoral girdle. Near the distal end of the rib there lies an ex- panded, stylet-shaped bone (see p. 7), the “post- cleithrum.” Each successive rib is joined to a vertebra in the same manner as the first rib; they gradually diminish in size posteriorly. In adult males the last 6 or 7 ribs become sexually modified. The last three are quite small, slender and widely separated; the preceding three or four ribs arch gently forward at their tips and converge, just above and behind the pelvic girdle. Minute epipleurals are joined loosely with the postero-proximal surfaces of the pleural ribs just below or adjacent to the level of the trans- verse processes; they are present on all but the last two pleural ribs. Parapophyses (see p. 7) 28 Zoologica: New York Zoological Society [44: 1 do not occur on any of the anterior caudal vertebrae in Phallichthys fairweatheri. The contribution of the vertebral axis to the gonopodial suspensorium of the adult male con- sists of four specialized hemal spines. The first, or ligastyle, is a long, slender rod that lies in the primary suspensory ligament. Its dorsal detached margin migrates forward during sexual differ- entiation and comes to lie beneath the centrum of the 10th precaudal vertebra. The next three hemal spines or gonapophyses are specialized chiefly by the addition of bony substance to their distal and posterior surfaces. The first inclines slightly forward. It is straight except for the distal fourth which is bent sharply forward to meet the dorsal edge of the primary gonactinostal complex. A single pair of well-developed curvi- linear uncini is usually present; they extend backward from the base of gonapophysis I and overlap the base of gonapophysis II. The second gonapophysis is more or less vertical. The base of the spine’s shaft is slender, but it widens gradually toward the tip where it merges with a pair of extremely heavy, subtriangular uncina- toid processes. The third gonapophysis is either vertical or inclines slightly backward. It is the longest of the three gonapophyses, and its uncini, when present, are quite small, subtriangular, and are always situated near the distal tip of the spine’s shaft. The actinosts of the anal fin in the adult male also are specialized and are in- corporated into the suspensorial system as the gonactinosts. The first is a short, heavy, blunt rod that inclines sharply forward. It articulates with fin rays 1 and 2. Gonactinosts 2, 3 and 4 are fused to form a single highly complex plate of bone, the primary gonactinostal complex, which supports fin rays 3, 4 and 5. The complex in this species is compressed antero-posteriorly. The posterior lateral wings that are produced symmetrically outward from eccentric positions along either side of the core of gonactinost 4 are scarcely developed except at the dorsal third where they flare broadly. Gonactinost 5 lies embedded in the depression formed by the lat- eral wings of the primary complex. Gonactinosts 6, 7, 8 and 9 are slender and rod-like; their tips flare apart. Actinost 10 of the anal fin becomes much reduced and is incorporated into gonacti- nost 9 as a tiny sliver of bone. The gonactinosts are firmly anchored to the vertebral axis by means of the ligastyle and the gonapophyses. The primary gonactinostal complex is attached by ligaments to the ventral margins of the liga- style and gonapophysis I. The tip of gonapo- physis II interdigitates with gonactinosts 7 and 8, where it is held in place by means of tendons and ligaments. The tip of gonapophysis III curves forward toward the posterior surface of the final gonactinost, where it, too, is attached by means of ligamentous connective tissues. The gonopodium in the adult male consists of the produced and modified rays 3, 4 and 5 of the anal fin. Together these rays are folded to form a hemicylinder with a broad groove along the right side of the gonopodium. When held at rest, i.e., pointed caudally, ray 3 forms the ventral border, ray 5 the dorsal border and ray 4 the lateral wall of the dextral trough. The paired halves of ray 3 are segmented to the tip of the fin and are never consolidated. The broad and flat spinous processes of the subterminal 20 to 30 segments of the right half of ray 3 are only slightly dextrally incurved to form the ventral border of the open groove; they taper abruptly to 8 or 9 paired terminal segments that form a hooked ramus. Definitive spines are wanting on the left half of ray 3 and the individual, rela- tively unspecialized, segments are shallow and rather poorly developed. The subterminal seg- ments of ray 4a are flattened dorso-ventrally and their long axes are greatly extended. The ter- minal 10 or 11 shorter segments are closely applied to the dorsal margin of the hooked distal ramus of ray 3. A loaf-shaped membranous swelling arises from the right side of the com- posite terminal bony ramus. The proximal seg- ments of the right half of ray 4p form a high, knife-like ridge. Penultimately at the distal fourth of the left half of ray 4p there is a series of approximately 15 retrorse serrae that face upward and outward away from the fin’s long axis; they are preceded by 4 or 5 smaller, slender and erect, terminal serrae that extend laterally at right angles to ray 5. The elements of the distal fourth of the right half of ray 4p are delicate and thread-like except where they are modified as a series of 5 or 6 minute retrorse serrae adjacent to the bases of the larger serrae on the left half of this ray. Ray 5a is bilaterally asymmetrical; the segments of the right half of the ray are wider and more flattened than those of the left half. The segments of both, however, are greatly dilated longitudinally and extend to the tip of the fin beneath the membranous swell- ing as slender hair-like rods. The segments of ray 5p are extremely delicate. They are joined to the outer edges of the paired members of 5a below; those of the left half are poorly devel- oped, becoming obsolescent at the level of the minute serrae on the right half of ray 4p; the right half of ray 5p is developed as a high, laterally compressed ridge that extends distally to the level of the smaller, erect serrae at the tip of the sinistral component of 4p. The tips of all the rays become rather slender distally 1959] Rosen & Bailey: Middle- American Poeciliid Fishes: Carlhubbsia & Phallichthys 29 and at the extreme tip of the fin they form the beginnings of a tight spiral in which the upper- most or 5th ray shows the greatest displacement from its original axis. The gradual decrease in height of all bony elements toward the tip of each ray gives to the gonopodium as a whole a distinctly acuminate profile. Other than the hooked distal rami of rays 3 and 4a there is no distinctive terminal segment. Relationships.— This new species is referable to the genus Phallichthys on the basis of the structure of the gonopodium and gonopodial suspensorium (see Text-figs. 4 and 5, and dis- cussion on pp. 29-32). It may be separated readily from the only other known species, P. amates, by the characters listed in the diag- noses and in Table 6. The two species are allo- patric (Map 1). Nomenclature— Phallichthys fairweatheri was first collected by C. L. Hubbs in El Peten, Guatemala. Hubbs thought that this fish should be the type of a new genus and assigned it the unpublished manuscript name “Dextripenis evides,” the generic name referring, of course, to the dextrally folded gonopodium. Specimens bearing this name were made available to C. L. Turner and N. S. Scrimshaw for their investiga- tions on the reproductive behavior of poeciliid fishes. Both men used the name or otherwise referred to this species in their studies (see synonymy, p. 24), but did not accompany it with an adequate description. Thus, Dextripenis evides is a nomen nudum and has no validity under the International Rules of Zoological Nomenclature. This species is named in honor of the Rev. Gerald Fairweather in acknowledgment of his participation in obtaining extensive scientific collections of fishes in British Honduras. Range.— Phallichthys fairweatheri is known to occur in three apparently separated areas in British Honduras and northern Guatemala (Map 1 ) . The fish from the Rio Hondo and New River systems, tributary to Chetumal Bay, British Honduras, probably represent a single stock since these drainages are connected by flood plains at high water, according to L. C. Stuart (personal communication). The stocks in the Rio de la Pasion and Rio San Pedro de Martir systems of Guatemala are now well sepa- rated, but future collecting in the connecting waters of the Rio Usumacinta system (Map 2) may close this gap. Habitat— In 1935, Drs. Carl L. Hubbs and Henry van der Schalie, on the Fifth Carnegie Institution-University of Michigan Expedition, took this species 13 times in El Pet6n. At all stations the water was quiet or stagnant, or at least some quiet water was present. The water was clear, often blackish, at most stations though dirty at a few; vegetation was commonly present and often dense. The bottom consisted of or included soft mud at all stations. The habitat is perhaps best indicated by the conditions in Rio San Pedro de Martir (UMMZ 144186), where 792 specimens were taken: water rather dirty; vegetation slight to thick; bottom very soft mud with much hydrogen sulphide; virtually no current; shore a recently exposed mudflat. Assessment of Taxonomic Criteria The species of Carlhubbsia and Phallichthys are so similar in appearance that early in this study we thought they constituted a single genus. Largely on the basis of the detailed anatomical differences in the gonopodium and gonopodial suspensorium discussed below, we conclude not only that generic separation is called for, but that two distinctive phyletic lines are involved. The common feature of an asymmetrically folded gonopodium is believed to be independ- ently evolved in these lines. Gonopodium The gonopodia of Carlhubbsia and Phallich- thys are permanently folded to one side (sinis- trally in P. amates, dextrally in other species), and show little internal symmetry. Although the gonopodia are superficially similar because of asymmetry, for a phylogenetic study it is essen- tial to obtain more detailed information on indi- vidual gonopodial structures. To this end, special attention is directed to form and frequency of specialized terminal features. Some of the salient distinctions between the gonopodia of these genera are set forth in Table 19 (see also Text- figs. 3 and 4). More extended descriptions and discussions of gonopodial characteristics appear in the systematic accounts of genera and species. In both genera the elements in ray 3 reflect the over-all symmetry of the fin, as do all the bony elements of the gonopodium. Segments which abut directly onto the permanently developed groove along one side of the fin, and which are functional components of this groove, are gen- erally better developed than segments which arise on the side away from the direction of folding. Thus, in the sinistral Phallichthys amates, the dextral spinous segments near the tip of ray 3 are short with miniscule ventral processes, whereas the sinistral elements are long and curved, the ventral processes folding inward to form the ventral margin of the gonopodial groove. In P. fairweatheri and the species of 30 Zoologica: New York Zoological Society [44: 1 Carlhubbsia the spines are present only on the right half of the ray and face into the gonopodial groove, again to form the ventral margin of the partially closed channel. An exception to this generalization involves those specialized hold- fast structures that are believed to assist in copulation; for example, the distal serrae on ray 4p in all four species face outward, that is, away from the gonopodial groove on the convex sur- face of the permanently folded fin. In Phallichthys the gonopodium is compara- tively simple, having few specialized terminal features; there are unilateral distal spines on ray 3 and serrae on ray 4p, the subterminal elements of all rays are uniformly simple and the fifth ray is unspecialized (Table 19). In Carl- hubbsia, by comparison, the gonopodium is complex. Each ray with the exception of the anterior branch of 5 is terminated by a series of specialized elements: in addition to spines on ray 3 and distal serrae on ray 4p there is a small terminal hook on ray 3, a series of laterally dilated terminal segments on ray 4a, a disrupted ray 4p and posterior serrae on ray 5p. Gonopodial Suspensorium In poeciliid fishes the development of such suspensorial structures as the ligastyle and the gonapophyses is influenced by over-all growth patterns and time and rate of sexual differentia- tion (Rosen, ms.). Growth and maturation, in turn, are affected by a variety of environmental factors, such as nutrition, light and temperature. But underlying genetic patterns appear largely to control the expression of some structures ir- respective of body size or form; if these can be properly identified they provide indications of natural relationship. The position of uncini on the gonapophyses and their form, as well as the shape of the primary gonactinostal complex, have always proved to be relatively constant within a related group of species. The form of the ligastyle and the orientation of the gonapophyses are apparently controlled in large part by the form of the body. In Carl- hubbsia hidden, the slenderest of the species under consideration, the ligastyle is reduced to a rudiment of bone embedded in the primary suspensory ligament and all three gonapophyses incline forward at a sharp angle. In Phallichthys antates, of intermediate body depth, the ligastyle is quite small but in some individuals is pro- longed into a slender rod of bone equal in length to the diameter of a vertebral centrum; only the first gonapophysis is bent sharply forward. In C. stuarti and P. fairweatheri, the two deepest- bodied species, the ligastyle is a well-developed, long bony rod and the anal fin supports have a more nearly vertical orientation. In each, the angle between the gonapophyses and the verte- bral column is roughly proportional to the dis- tance between the gonactinosts and the vertebral axis (Text-fig. 5). Positional relationships and form of the uncini on the three gonapophyses have proved ex- tremely effective in interpreting the relationships in these fishes. As may be seen from Plate VI, Text-fig. 5, and Table 19, there is measurable individual variation within species and between species of Carlhubbsia and Phallichthys in the orientation and extent of development of uncini. But what matters from the viewpoint of me- chanical control is not the precise point at which an uncinus arises per se, but the total adaptation for the job of suspension, which if the point is constant can be achieved by fine adjustments in size, orientation and rigidity of the uncini. In the species of Carlhubbsia, for example, C. stuarti lacks uncini on gonapophysis III, but the uncini on gonapophysis II are so long that they overlap the shaft of number III at the same point at which two tiny uncini arise on element III in C. kidderi. Two specimens of C. kidderi were found with well developed uncini on gonapo- physis II, but none on III. The same broad functional problem of suspension has been solved repeatedly in different ways within the Poeciliidae as a whole. For this reason it is sig- nificant that, despite individual and specific differences, each genus presents a relatively dis- tinctive pattern with respect to the topographic relations and basic morphology of these gonapo- physeal processes. In both Carlhubbsia and Phallichthys the rela- tive size of the primary gonactinostal complex remains fairly constant, without reference to variations in body proportions or ultimate size attained. As a result, in the more slender species, C. kidderi and P. amates, this bony complex closely approaches the vertebral axis, whereas in the deep-bodied forms, C. stuarti and P. fair- weatheri, it is well separated. As previously noted, the length of the ligastyle and the angle of inclination of the first gonapophysis compen- sate for these differences. Thus, the relative con- stancy within each genus suggests conservative- ness in the gonactinostal complex, and the find- ing of differences between the groups of species (genera) emphasizes its reliability. In the species of Carlhubbsia the primary complex is greatly dilated along the antero-posterior dimension; in the species of Phallichthys it is narrow (Text-fig. 5). This difference is brought about in three ways: when incorporated into the complex acti- 1959] Rosen & Bailey: Middle- American Poeciliid Fishes: Carlhubbsia & Phallichthys 31 Table 19. Significant Contrasting Characters of Carlhubbsia and Phallichthys Character Carlhubbsia Phallichthys Gonopodium (adult male) : Ray 3 15 to 20 subterminal segments with flat- tened incurved spinous processes; 5 to 8 distal elements simple; with small curved terminal hook 20 to 30 subterminal segments with flattened incurved spinous processes; 8 or 9 distal ele- ments simple, forming a ven- trally hooked ramus; no terminal hook Ray 4a Approximately 15 distal segments much elevated; not forming a hooked ramus 5 terminal segments compose a minute hooked ramus that arches downward over the hooked tip of ray 3 Ray 4p Very thin distally, closely joined to ray 4a; obsolete from segments 20 to 25 of ray 4a (counting apico-basally) to about segment 15 (ray 4a) where it re- appears as a clustered series of 4 to 8 unpaired retrorse serrae Terminal series of up to 17 or 18 well-developed, retrorse serrae Ray 5 Ray 5p with sinistral, retrorse serrae on 6 or 7 ( kidderi ) or about 15 ( stuarti ) terminal segments; serrae fused with dis- tal elements of 5a and separated from unspecialized segments of 5p ( kidderi ) or distinct and continuous with unspe- cialized segments ( stuarti ) Little specialized except in sym- metry; without serrae Gonapophyses (adult male) : Uncini on I Slender and straight, tilted downward, of moderate length ( kidderi ) or long (stu- arti) ; arising near base of gonapophysis Slender and short, curved up- ward, arising near base of gon- apophysis ( fairweatheri ) or absent (amates) Uncini on II As on gonapophysis I but longer and stronger, situated slightly farther down gonapophysis but in same plane as un- cini on I Stout, broad based, short, curved downward, arising on distal half of gonapophysis; not in same plane as uncini on I Uncini on III Absent (stuarti, rarely in kidderi) or rather short, moderately broad and straight, arising near middle of gonapophysis in same plane as uncini on I and II (kid- deri) Short and very broad with tips obtuse, frequently curved downward, arising in distal third of gonapophysis, not in same plane as uncini on I and II (usually absent in fair- weatheri) Primary gonactinostal complex (adult male): Broad; antero-posterior breadth about 1/3 to 2/5 length Narrow; antero-posterior breadth about 1/4 length Preorbital bone More sculptured; with well - developed process projecting backward Simple; roughly subtriangular in outline Dorsal fin More or less angulate; the free edge falcate Rounded nosts 2, 3 and 4 are distinctly separated in Carl- hubbsia, in close apposition in Phallichthys; the anterior plate of bone on gonactinost 2 is broadly dilated or distended anteriorly in Carlhubbsia, only narrowly so in Phallichthys; and the pos- terior lateral wings on gonactinost 4 are accentu- ated in Carlhubbsia, only moderately developed in Phallichthys. In summary, there are three principal differ- ences between the suspensoria of Carlhubbsia and Phallichthys. In Carlhubbsia the uncini al- ways originate along the proximal half of the shaft of a gonapophysis; in Phallichthys the uncini of gonapophysis I originate on the proxi- mal half, but on gonapophyses II and III they originate on the distal half of the spine’s shaft. 32 Zoologica: New York Zoological Society [44: 1 In Carlhubbsia the uncini are long, straight, and they merge with the gonapophyses rather ab- ruptly; in Phallichthys the uncini are short, they are curvilinear, and they merge with the gona- pophyses gradually, causing the spines to appear swollen at their tips. In Carlhubbsia the primary gonactinostal complex is greatly dilated antero- posteriorly; in Phallichthys it is relatively nar- row. Head Skeleton and Dentition In both Phallichthys and Carlhubbsia the head skeleton is typical of such poeciliid species as Poecilia vivipara and Xiphophorus maculatus whose principal diet consists of organic debris, minute aquatic organisms and plant material. In both genera, the base of the cranium is high and firmly fixed in position by means of well- developed supraoccipital and epiotic processes, which join the high, expanded neural crests of the cervical vertebrae by means of strong ligaments. This type of deep, immobile skull is character- istic of other sluggish, forage-feeding fishes as well. Similarities in basic skull form in the two genera may as reasonably be interpreted to re- flect similar feeding behavior as intimacy of re- lationship. The only superficially obvious skull differences occur in the orbital bones. In Carl- hubbsia the preorbital (lacrymal) is more sculp- tured and has a well developed process produced backward toward the lateral ethmoid; in Phal- lichthys it is simple, roughly subtriangular in outline. The dentigerous features of premaxillae and dentaries are similar in Carlhubbsia and in Phal- lichthys, although the shape of the inner bands is distinctive in P. fairweatheri. Such nutritional adaptations as tooth structure and size and ori- entation of dentigerous borders are notably vari- able in some poeciliid genera (e.g., Poeciliopsis) , and the taxonomic usefulness of these characters is limited accordingly. To some degree, tooth structure varies independently of basic architec- ture of mouth parts. In Carlhubbsia and Phal- lichthys, for example, the jaws are weak, blunt and loosely joined at the midline. Yet Phallich- thys fairweatheri has a broad, curved inner band of minute teeth and P. amates and the species of Carlhubbsia have a nearly straight, narrow inner band of longer teeth. Jaw structure is fun- damentally similar in the poeciliid genera Aulo- phallus, Quintana and Girardinus. In Aulophal- lus the teeth of the outer series are setiform, in Quintana they are more or less conical and sharply pointed, and in Girardinus they are broadly oblanceolate. In general, it appears that in this family dental characters offer little hope of aiding phylogenetic analysis. Systematic ar- rangements of the Poeciliidae and other cyprino- dontiform groups based on these and other nu- tritional features have been subjected to just criticism (e.g., Hubbs & Turner, 1939; Miller, 1956). Sensory Canals Both Carlhubbsia and Phallichthys exhibit moderately well developed sensory canal sys- tems, although problems in their direct compari- son arise due to changes in the canals associated with size and age. Closed canals of large females may be represented only as open grooves in juveniles, in small adults, or even in mature males. Thus, of ten adult females of Carlhubbsia hidden (UMMZ 144203) 28 to 34.5 mm. in standard length, the supraorbital canal is closed between pores 2 to 4 and 6 and 7 in nine, and from 2 to 3 and 6 to 7 in one, the smallest; the preopercular canal is closed and has 7 pores in eight, has 8 pores in one, and has 6 pores and a short groove in the smallest; and the preorbital canal has 3 pores in all. But of ten adult males from 16 to 20.5 mm. long from the same series, the supraorbital canal is developed from 2 to 4 and from 6 to 7 in six, from 2 to 3 and 6 to 7 in one, from 6 to 7 in one, and is represented only by grooves in two; the preopercular canal consists of 6 pores in seven, of 4 in two, and of 3 in one, the remaining portions being evidenced by open grooves; and the preorbital canal has 2 pores and a groove in three and consists of a groove only in seven. In Carlhubbsia and Phal- lichthys the mandibular canal is never devel- oped, in the definitive condition the preopercu- lar canal has 7 pores in all species, and the pre- orbital canal has 3 pores except for C. stuarti which has 4. The apparently definitive condition for the supraorbital canal is to have a tube with three pores above the eye (pores 2, 3 and 4a) and one with 2 behind it (pores 6 and 7) in all species. This arrangement is usually found in both species of Carlhubbsia, at least in adult females, and sometimes, especially in very large females, in the forms of Phallichthys. In most adults of Phallichthys, however, only the short postorbital section of the canal is covered. In three of ten adult females of C. stuarti examined, there is a third remnant of the supraorbital canal connecting pores 4b and 5. In general, Carlhubbsia and Phallichthys are similar in the pattern of their sensory canals. But this pattern is essentially the same as that reported by Gosline (1949) for Mollienesia latipinna and Girardinus metallicus and is close to that found in Platypoecilus (= Xiphophorus ) maculatus and Poeciliopsis sp. We therefore find it impossible at this time to treat the simi- 1959] Rosen & Bailey: Middle- American Poeciliid Fishes: Carlhubbsia & Phallichthys 33 larities in sensory canal pattern in Carlhubbsia and Phallichthys as indicative of a phylogenetic relationship. Relationships of Phallichthys and Status OF THE POECILIOPSINAE Based chiefly on similar asymmetric modifica- tions of the gonopodium of adult males, the genera Phallichthys, Poeciliopsis, Poecilistes, Aulophallus, Xenophallus, Phalloptychus and Carlhubbsia (as Alio phallus ) have been asso- ciated as a subfamily, the Poeciliopsinae, by Hubbs (1924, 1926, 1936). It has been inferen- tially suggested (Miller, 1955:50) that the mon- otypic genus Poecilistes is a generic synonym of Poeciliopsis. We find that the inner teeth in Poecilistes pleurospilus are much reduced in number, uniserial, and are restricted to the lat- eral ends of a row in each jaw. The supposed absence of the inner teeth has been employed as the primary basis for removal of Poecilistes from Poeciliopsis (Hubbs, 1936:233), a sepa- ration that can no longer be accepted. As a result of comparative morphological study we believe that the above association consists of four divisions, one comprising Phallichthys, Poe- ciliopsis and Aulophallus, the others Phallopty- chus, Xenophallus and Carlhubbsia respectively. Each division is representative of what appears to constitute a distinctive phyletic line. The diagnostic features of Poeciliopsis (see also Hubbs, 1936, and Hubbs & Miller, 1954) include: (1) body slender with posteriorly- placed dorsal fin; (2) gonopodium tightly rolled into a partially closed tube; ray 4p with paired, asymmetrical terminal retrorse serrae; terminal segments of ray 5 simple, without ornaments; rays 7 and 8 converging or in contact along mid- dle of their lengths; (3) gonapophyses with un- cini nearer distal than proximal end of shaft; uncini with broad bases and ventrally-arched, blunt tips; and (4) primary gonactinostal com- plex greatly dilated antero-posteriorly and deep- ly notched at dorsal margin, the incorporated shafts of actinosts 2, 3 and 4 flaring outward and upward away from the consolidated basal components. With two exceptions, Aulophallus shares the above features with Poeciliopsis. It differs prin- cipally in the development of unpaired serrae on ray 4p of the gonopodium and in the trans- verse widening of terminal elements of ray 5a which in Poeciliopsis are greatly compressed. Dentition, used formerly to separate these groups, should be carefully reviewed, since within Poeciliopsis tooth form and orientation and size of the dental ridges are highly variable, according to Robert R. Miller (personal com- munication—see also Hubbs, 1936: 235). The demonstrable intimacy of relationship between Poeciliopsis and Aulophallus may eventually necessitate their merger. That the species of Phallichthys may have evolved directly from a form having symmetri- cal or nearly symmetrical gonopodial elements rather than from a Poeciliopsis- like fish, is sug- gested by the presence of only moderately in- curved spinous processes on ray 3, the presence of a high, laterally compressed ridge on 4p and the lack of definitive closure of the groove at any point along the gonopodium. In the evolu- tion of asymmetrical genitalia in Poeciliidae, formation of a fixed unilateral groove is made possible by folding of the rays, incurving of ele- ments of ray 3 to form the ventral wall of the groove and compaction and superimposition of rays following loss of basic rotatory mechanisms (Rosen & Gordon, 1953). The segment ridge on ray 4p of the gonopodium in Phallichthys almost certainly is a remnant of the dorsal center of rotation (between rays 4 and 5) that is a constant feature of almost all symmetrical gonopodia. This, in turn, suggests that in Phal- lichthys some rotatory movements may occur during fin erection; additional mechanical adjust- ments would seem almost obligatory in order for the ray 5 complex to be apposed with the only moderately incurved spines of ray 3. The low degree of closure of the unilateral groove of the resting fin in both P. fairweatheri and P. amates contrasts sharply with the compact, twisted or even helical gonopodia in species of Poeciliopsis. In the latter forms, maximal fold- ing of rays 3, 4 and 5 occurs during develop- ment, and no further positional adjustments accompany fin erection. The morphological dif- ferences separating these genera are summarized in Table 20. Despite many differences in the details of their gonopodia and gonopodial suspensoria, Phallich- thys and Poeciliopsis display a number of broad similarities. In their gonopodia we find: (1) segments on one-half of ray 3 rolled inward to form the ventral wall of the unilateral groove; (2) segments of ray 4a simple and without orna- ments; those of 4p with a series of terminal retrorse serrae (paired though asymmetrical in Poeciliopsis and Phallichthys fairweatheri, un- paired in Phallichthys amates and Aulophallus) ; and (3) segments of ray 5 simple and much reduced. In their gonopodial suspensoria, we find: (4) uncini usually arising well out on gonapophyseal shaft; and (5) uncini with heavy bases, their tips blunt and developed in an arch downward. In view of over-all similarities in form and orientation of their male secondary 34 Zoologica : New York Zoological Society [44: 1 Table 20. Comparison of Phallichthys and Poeciliopsis * Character Phallichthys Poeciliopsis Form Body depth (percent, of Body deep and compressed Body slender and more terete standard length) 32 to 44 About 22 to 31 Body circumference scales 21 to 24 19 to 21 Dorsal origin (females) Slightly in advance of anal origin Decidedly behind anal origin Dorsal rays 8 to 11; infrequently 8 Usually 7 or 8 Vertebrae Suborbital dark bar 27 to 30, modally 28 or 29 Present, more or less oblique; sometimes faint 29 to 33, modally 30 to 32 Absent Mandibular canal Gonopodium: Never developed Often present; sometimes absent Form Spinous processes Asymmetry less extreme; no definitive closure of groove Twisted into a tightly rolled and partially closed tube on ray 3 Subterminal segments Moderately incurved Strongly incurved of ray 3 Not consolidated Usually several consolidated to form a slender rod Ray 4p With high laterally compressed ridge; ter- minal retrorse serrae paired or unpaired Without high ridge; terminal re- trorse serrae paired Rays 7 and 8 Suspensorium: Symmetrical; well separated at middle of their lengths Distorted; converging or in con- tact at middle of their lengths Gonapophyseal uncini Stout, broad-based, short, curved down- ward, usually emerging on distal half of spine; not lying in same plane Stout, broad-based, usually short, curved downward, emerging on distal half of spine; not ly- ing in same plane Gonactinostal complex Narrow antero-posteriorly; neither dilated nor notched at dorsal end Dorsal end dilated antero-pos- teriorly and deeply notched * Dr. Robert R. Miller is engaged in a revisionary study of Poeciliopsis which, when completed, will permit a much more adequate contrast of these genera. sexual specializations, it is possible that Phallich- thys and Poeciliopsis, with Aulophallus, may have radiated from a common prototype in which developmental patterns for gonopodial asymmetry were first becoming established. The difference in direction of gonopodial asymmetry of Phallichthys fairweatheri (dextral) and P. amates (sinistral) could have come into exis- tence by means of a genetic “switch” mechanism that controlled direction of asymmetry at a criti- cal point during epigenesis; it does not neces- sarily represent a fundamental divergence be- tween these two species. The Uruguayan genus Phalloptychus is char- acterized by having extremely well developed, projecting and more or less vertically oriented, unpaired serrae at the tip of gonopodial ray 4p, and long, straight and slender suspensorial un- cini that emerge from the bases of and extend horizontally backward from gonapophyses II and III. With reference to their zoogeography and comparative morphology, the two tiny spe- cies of this South American genus are highly distinctive. We find no basis for an alliance be- tween this genus and the group including Phal- lichthys, Poeciliopsis and A ulo phallus. Similarly, the Central American Xenophallus has no special combination of gonopodial or suspensorial traits which, in our opinion, asso- ciate it with Poeciliopsis and its allies. The gono- podium (see Rosen & Gordon, 1953:29) is re- markably simple, without serrae or spines; it is specialized only in the sinistral or dextral fold- ing and in the prolongation of the tip of ray 4a as a single, consolidated curved bony rod. In its suspensorium, uncini on gonapophyses I, II and III arise on the basal half of the gonapophy- seal shafts along a single axis, and they are linear and rather slender. The relationships of both Phalloptychus and Xenophallus will be dis- cussed more fully in a forthcoming publication (Rosen, ms.). 1959] Rosen & Bailey: Middle-American Poeciliid Fishes: Carlhubbsia & PhaUichthys 35 Evidence has been presented above suggest- ing that Carlhubbsia is not intimately related to PhaUichthys or to any other member of the heterogeneous assemblage heretofore lumped as the subfamily Poeciliopsinae. Nowhere in the group are there other forms having suspensorial and gonopodial details like those of Carlhubbsia. Among the remaining major groups of poeciliids we find a constellation of features which most closely resembles those of Carlhubbsia in the endemic Cuban genera Quintana and Girardinus (see below, pp. 35-39). From a functional standpoint it may seem elementary to suggest that asymmetric folding of the primary anal rays to form a permanent closed or partially closed tube would produce a highly adapted vehicle for sperm transfer. All poeciliid fishes not so equipped create a transi- tory tube by folding the anal rays at each copu- latory attempt. Thus all members of this group, and indeed others that similarly employ the anal fin to effect internal fertilization, are in a sense preadapted to the evolution of a permanently asymmetric genitalium. Permanent folding per- mits structural modification to enhance the ef- fectiveness of the mechanism. That a develop- ment so useful in the maintenance of species should evolve but once in this family in which gonopodial plasticity is abundantly demon- strated is conceivable, but a polyphyletic origin of asymmetry is certainly to be anticipated. Justification for dismemberment of the Poe- ciliopsinae rests solidly with the weight of evi- dence from study of the fine details of the gonopodium (apart from its asymmetry) and gonopodial suspensorium. Gonopodial asym- metry, far from being a uniting character, almost certainly has appeared independently at least five times within the Poeciliidae [in Phallich- thys - Poeciliopsis - A ulophallus, Phalloptychus, Xenophallus, Carlhubbsia and in Xenodexia ctenolepis, regarded by Hubbs (1950) as consti- tuting a distinct subfamily], and in two other cyprinodontiform families, the Jenynsiidae and Anablepidae. The beginnings of such a pattern can be seen in still another poeciliid genus, Quin- tana, in which serrae on ray 4p of the gono- podium are always twisted sinistrally.4 4 C. L. Hubbs (Occ. Pap. Mus. Zool., Univ. Mich., 302: 1-3, 1935), in an article entitled “Studies of the fishes of the Order Cyprinodontes. XIV. Plectrophallus regarded as a distinct genus,” figured the gonopodium of the poeciliid Plectrophallus tristani (Fowler) to illus- trate the asymmetric folding of the rays. This would, of course, represent yet another example of the independ- ent origin of asymmetry in the family, since P. tristani is probably allied to the species of Brachyrhaphis. P. tristani is known from but one specimen, however, and we have not seen this. Carlhubbsia and the Cuban Endemic Poeciliids Our attempt to decipher the relationships of Carlhubbsia has led us to investigate the Cuban poeciliids which in current classification consti- tute the tribes Girardinini and Quintanini. Howell Rivero & Rivas (1944) called attention to the integrity of the girardinins, and pointed to the distinctiveness and uniformity of their gonopodia and suspensoria. Suspensorial struc- tures were studied further by Howell Rivero (1946) and the compactness of the Girardinini was again emphasized. This group, now num- bering about 10 recognized species, was arranged in five genera that were separated chiefly on the basis of dentition and mouth structure. There are at most minor differences, mostly average features, in the gonopodia of all groups except Toxus, which, though sharing the prominent horn-like terminal appendages and all specialized bony structures, has the most distinctive gono- podium of the five genera. Howell Rivero & Rivas (1944: 14) summarized their study of these fishes as follows: “In their fundamental features, therefore, the genera of the Girardinini are in almost complete agreement. The radiative adaptation of these genera seems to have been related chiefly to food habits, for most of the generic characters in- volve the structure of the jaws, mouth and teeth (see key to genera). There is every reason to believe that the genera of the tribe have evolved in Cuba, after a single ancestral species migrated into what is now that island.” As mentioned above, dentitional characters have been seriously over-emphasized in the sys- tematics of the cyprinodontiform fishes. The gonopodia and suspensoria of the girardinin fishes (Howell Rivero & Rivas, 1944; Howell Rivero, 1946; Rosen & Gordon, 1953: 27) show numerous and striking similarities. This is so manifestly a compact group phylogenetically and zoogeographically that we prefer to classify the species in a single genus, Girardinus Poey, of which Glaridichthys Garman (including Glaridodon Garman), Toxus Eigenmann, Dac- tylophallus Howell Rivero & Rivas and Allodon- tium Howell Rivero & Rivas are generic syno- nyms. Thus, Girardinus as amended is equivalent to the Girardinini of recent authors. Since Quintana is the only genus in the tribe Quin- tanini, these terms also are equivalent in scope. Girardinus and Quintana That Quintana atrizona is highly distinctive was clearly indicated by Hubbs (1934) in the original account. He placed the genus in the 36 Zoologica: New York Zoological Society [44: 1 Text-fig. 7. Distal tips of the gonopodia of: A. Girardinus denticulatus (Garman), and B. Quintana atrizona Hubbs, as seen from the left side. Bony ele- ments being compared are shown in solid black. subfamily Gambusiinae and, hesitantly, in the tribe Heterandriini. Certain similarities to Gam- busia, Allogambusia and Girardinus ( sensu lato) he regarded as more plausibly due to parallelism and convergence than to common origin. Howell Rivero & Rivas (1944: 13) granted separate tribal status to Quintana, but commented that “Our continued studies have emphasized the integrity of the Cuban group Girardinini, al- though the gonopodial characters of Quintana (see key) somewhat confuse our views as to the isolated position of this group.” Howell Rivero’s description and figure of the suspensorial ap- paratus of Quintana (1946) clearly indicate the unique features. He reported that only the sec- ond of the three gonapophyses is appreciably specialized and complex, an observation not fully substantiated by our material. Recent evidence has been found to support the hypothesis that Quintana and Girardinus arose on Cuba from invasions of a single or of two closely related forms (Rosen, ms.). If this is true the affinities of these genera should be emphasized by their close association in the sys- tematic structure of the family. For the present purposes it is sufficient to note certain structural similarities. In the gonopodia and gonopodial suspensoria of Quintana and Girardinus (Text-figs. 7, 8 and 9), we find the following diagnostic structures that we interpret as homologous in these genera: (1) a minute, recurved, terminal hook on ray 3; (2) weak retrorse serrae on ray 4p; (3) moder- ately to well-developed serrae on ray 5p; (4) three highly specialized gonapophyses; and (5) uncini developed on all three gonapophyses. To these may be added similarities of body form and fin shape. All species are streamlined, all have the dorsal and ventral trunk profiles sym- metrically arched, and in all the caudal peduncle is long and slender. The median fins, particularly the dorsal, tend to be sharply pointed and even falcate. Characters Indicating Relationship of Carlhubbsia with Quintana and Girardinus If we deal first with the primary terminal specializations on rays 3, 4 and 5 of the gono- podium (Text-fig. 10) and allow for the minor modifications which are probably to be attrib- uted to asymmetric growth, we find that with but one important exception the gonopodia of Quintana and Carlhubbsia are much alike. They share the following similarities: in ray 3 both have a series of stout proximal elements that become considerably compressed apico-basally 1959J Rosen & Bailey: Middle-American Poeciliid Fishes: Carlhubbsia & Phallichthys 37 Text-fig. 8. Axial division of the gonopodial sus- pensorium of Girardinus creolus (Eigenmann), showing the three gonapophyses and a slightly modi- fied hemal spine. Anterior to the left. Text-fig. 9. Axial division of the gonopodial sus- pensorium of Quintana atrizona Hubbs. showing the three gonapophyses and a slightly modified hemal spine. Anterior to the left. towards the tip of the ray. At the exposed margin of the ray about ten terminal segments that are decidedly higher than long are produced into broad, flat spines; the most distal extent of the posterior spinous processes contribute to the formation of the eccentric groove that arises obliquely from the ventral margin of this ray. In Carlhubbsia the single, dextral eccentric groove forms the permanent anterior edge of the partially folded gonopodium. In Quintana the eccentric grooves on both sides of the ray serve as the anterior edge of the transitory channel when the gonopodium is swung forward and to one side during fin erection (Rosen & Gordon, 1953: 18-23). In both, ray 3 is terminated by an abruptly widened, segmented or consolidated, bony complex and by a minute recurved bony hook (in C. stuarti, the hook is replaced by an uncalcified though rigid membranous structure) . In both Carlhubbsia and Quintana ray 4a is slender proximally and abruptly shortened and widened distally. Ray 4p in both genera is slen- der distally (usually obsolescent in Carlhubbsia) and bears a cluster of retrorse serrae. These serrae are rotated sinistrally in both genera. Ray 5p in both Quintana and Carlhubbsia is termi- nated by a series of erect though short serrae. The subdistal segments of ray 5a, which in Quintana develop as an “elbow-like” structure, and the profound asymmetry in Carlhubbsia are the only gonopodial features in which the two groups differ significantly. In the gonopodium of Girardinus, ray 4a is not expanded distally, the spines of ray 3 are smaller, less numerous, and are delicately pointed, and a conspicuous pair of fleshy sub- radial processes originates below the spines of ray 3 and extends in an arc forward and laterally. Of all the taxonomically significant morpho- logical details, those of the gonopodial suspen- sorium show the most striking similarities among the species of Carlhubbsia, Quintana and Girar- dinus (Text-figs. 5, 8 and 9). In each, the three gonapophyses all bear uncinatoid processes that lie more or less in the same plane at an angle of approximately 30° with the vertebral axis. The uncini in all are limited to the proximal portion of the spine’s shaft, and are always linear, pointed and rather slender. The gonapophyses of the suspensorium of Quintana were described and figured by Howell Rivero (1946) as being simple and having but a single pair of uncini. Actually the suspensorium of a paratype of Q. atrizona (Text-fig. 9) shows small uncini on all three gonapophyses. Quintana and Girardinus are considerably more alike than had been thought from earlier studies. In sus- pensorial structure they differ only in the extent of development of the uncini; in Girardinus (as well as in Carlhubbsia ) the uncini are relatively large, strongly produced, and overlapping. Howell Rivero’s Figure 10, which shows almost complete lack of specialization of gonapophyses I and II of Quintana, indicates that he may have worked with subadult males in which the fine details of suspensorial structure had not yet fully developed. Resume of Morphological Analysis and Conclusions Since the similarities in certain skeletal fea- tures in Carlhubbsia, Quintana and Girardinus are not always clearly defined and the alleged homologies may seem questionable, it may be contended that many or all of the resemblances are examples of evolutionary parallelism. It is 38 Zoologica: New York Zoological Society [44: 1 to >z <£ tr < or ro ” £ 2 (U.SH *S u. ^ • 03 2 .2 « - u 2 -2 i2 j~ h ^ o JS ■Or'* 4> O 00 gtJTJ°'-C U Tl — ° S3 2 Ja j§ cug ~ X) T3 S O O 3 fl O <- ~ 53 a ° a ° ^ ^3 W • C3 w ?3 -3 03 K 13 c3 ~ .-3 O '!? >, " 3 ■Ci 3 ^ cira O « a 3 l. ^ -a c o ■? 35 -5 .«> S -3 « Z 6 > ^ g ^1 g-3« S- , J3 I & °.2Eo2'3 «cEc“ii « -a — > « « £>.S> ~ c g a c „ a 2 G d> .22 es C g-.S I ^ i3 E c "o a o n O ^ ‘W W C S)ftS0:>u a g .2-.S £ ivO Vh ^ oo _ « >> o 5 g « « .2 15 Oh^£2 TD > X- o3 _. 3 u. o ^ ^ ^ - o ¥> u-t ^ .i5 « t2 .g ^ _ u 3“^ -frl _o u 42 -a £ X3 01) & ^ ~ ' 3 s C .s.9^S Q 5> EO° u 13 53 o 2, -a n-i t- «*-i 0 ^ d a co e o o 2 C/5 Si n p -5 2 01 01 Oh E o O 03 g t> .2 co o .22 ;u>s 03 <3 1 o cs *o 'S5 § i .£ - ~ '"D -fl T3 ^ 1 C W>T3 O 1 •-• T* cS X w -a ►h -j-i r- ^h H H -S < 52 c