Northeast Gulf Science Volume 2 Article 2 Number 2 Number 2

12-1978 A New Species of Poeciliid Fish of the Genus Poecilia from Hispaniola, with Reinstatement and Redescription of P. dominicensis (Evermann and Clark) Luis R. Rivas National Marine Fisheries Service

DOI: 10.18785/negs.0202.02 Follow this and additional works at: https://aquila.usm.edu/goms

Recommended Citation Rivas, L. R. 1978. A New Species of Poeciliid Fish of the Genus Poecilia from Hispaniola, with Reinstatement and Redescription of P. dominicensis (Evermann and Clark). Northeast Gulf Science 2 (2). Retrieved from https://aquila.usm.edu/goms/vol2/iss2/2

This Article is brought to you for free and open access by The Aquila Digital Community. It has been accepted for inclusion in Gulf of Mexico Science by an authorized editor of The Aquila Digital Community. For more information, please contact [email protected]. Rivas: A New Species of Poeciliid Fish of the Genus Poecilia from Hispan

Northeast Gulf Science Vol. 2, No.2, p. 98-112 December 1978

A NEW S.!>ECIES OF POECILIID FISH OF THE GENUS Poecilia FROM HISPANIOLA, WITH REINSTATEMENT AND REDESCRIPTION OF P. dominicensis (EVERMANN AND CLARK)1

Luis R. Rivas National Oceanic and Atmospheric Administration National Marine Fisheries Service Southeast Fisheries Center Miami Laboratory 7 5 Virginia Beach Drive Miami, FL 3 3149

ABSTRACT: Exploration of the streams and lakes of Hispaniola and available collections of poeciliid fishes from that island are discussed, followed by the taxonomic history and generic status of .the two species described. The genera Limia and Mollienesia were synonymized with Poe­ cilia by Rosen and Bailey (1963) and the former Mollienesia dominicensis of Evermann and Clark (1906) became a junior homonym of Limia dominicensis of Valenciennes (1846). Rosen and Bailey, therefore, renamed M. dominicensis of Evermann and Clark as Poecilia montana. Because Mollienesia is het·ein retained as a synonym of Poe cilia but Limia is reinstated as a valid genus, Poe cilia dominicensis of Evermann and Clark is no longer a junior homonym of Limia dominicensis of Valenciennes. The name montana, therefore, becomes a junior ob­ jective synonym of Eyermann and Clark's dominicensis. The new species herein described, although available to previous authors, had been hitherto conf'used with Poe cilia dominicensis of Evermann and Clark. Both species are superficially similar but clearly distinct and more closely related to each other than either is to any of the other species of Poecilia. They are restricted to the island of Hispaniola and show character displacement in the several localities where they occur together.

Before the late forties and early University of Miami Department of fifties, little was known about the Zoology (now Biology) where I was poeciliid fishes of Hispaniola and only a Curator of Fishes at that time. few lots of specimens were available in These expeditions resulted in the museums. Most of the few localities coverage of most of the lakes and stream previously collected were restricted to systems of Hispaniola, including all the vicinities of Port-au-Prince, Haiti, localities previously reported in the or Santo Domingo, Dominican Republic, literature, and in the discovery of and most of the river systems of the several undescribed species of poeciliid island remained unexplored .. fishes. During April, 1949, I conducted an Only two of the new poeciliids, expedition to the Dominican Republic Gambusia pseudopunctata (Rivas, 1969) to obtain specimens, especially poeciliids, and G. hispaniolae (Fink, 1971) have from previously unexplored streams. Sub­ been published and Mollienesia=Poecilia sequently, in April, 1951, I conducted elegans (Trewavas, 1948) was designated another expedition, this time to Haiti, the type species of a new genus, for the same purpose. Both of these Curtipenis=Poecilia by Rivas and Myers expeditions were sponsored by the (1950) on the basis of additional material collected during my 1949 Hispaniola !Contribution number 78-52M, Southeast expedition. Further than that, study and Fisheries Center, Miami Laboratory. publication of the remaining new species 98 Published by The Aquila Digital Community, 1978 1 Gulf of Mexico Science, Vol. 2 [1978], No. 2, Art. 2

New poecilliid fish 99

and the extensive material collected the last two rays, regardless of their could not be continued because of other degree of separation, are always counted commitments. Four of the new species as a single ray. All pectoral fin rays are belong to the genus Linlia and will be counted. described in a preliminary review of the Proportional body measurements in genus now in preparation (see below on relation to standard length are omitted status of Limia). The other, a new for reasons discussed by Hubqs and species of Poecilia, is described herein. Springer (1957) and Rivas (1963; 1969). In poeciliid fishes there is considerable MATERIALS AND METHODS variation in these characters individually, ontogenetically, seasonally, geographical­ The new species of Poecilia described ly, and environmentally and, therefore, in this paper is based on 461 adult they are of little or no value in distin­ males, 860 undeveloped males, 791 guishing species. Fin measurements and adult females, and 2,600 juveniles for a the relative position of fins may be total of 4,712 specimens from 14 better expressed quantitatively by in­ localities. The holotype and a series of dicating their magnitude in relation to paratypes have been deposited in the one another, their relative position to National Museum of Natural History one another, or their position and extent (USNM). Series of paratypes have been in relation to points of reference. This deposited in the Museum of Com­ procedure has been followed in the de­ parative Zoology (MCZ), the Museum scriptions, under the section dealing of Zoology University of Michigan with morphological characters. In the (UMMZ), and the Florida State Museum description of the new species the (FSM). Poecilia dominicensis (montana meristic characters for the holotype are of Rosen and Bailey, 1963; see below) given first followed in parentheses by previously known only from the female those of the male and female paratypes. holotype and two female paratypes is redescribed from 108 adult males, 85 TAXONOMIC HISTORY AND undeveloped males, 233 adult females, GENERIC STATUS and 316 juveniles for a total of 7 42 specimens from eight localities. Series The new species described below of specimens have been deposited in the belongs to the group of poeciliid fishes USNM, the MCZ, the UMMZ, and the formerly placed in the genus Mollienesia FSM. and hitherto represented by only two Gonopodial characters are self-ex­ species in Hispaniola. The first species; planatory or named and counted as pre­ was described as Platypoecilus domini­ viously described (Rivas, 1963). Scales censis by Evermann and Clark (1906) are counted according to methods and the second as Mollienesia elegans, described by Miller (1948). In the dorsal by Trewavas (1948). Platypoecilus and anal fins the last two rays are, in dominicensis was erroneously synony­ some specimens, very close set and mized with Poecilia dominicensis Val­ appear as a single ray, branched, or split, enciennes (1846) by Regan (1913) to the base. In other specimens, the who placed both species in the genus last two rays are distinctly separate but Limia. Subsequently, however, Myers still close to each other. For these reasons, (1931) correctly indicated that https://aquila.usm.edu/goms/vol2/iss2/2 2 DOI: 10.18785/negs.0202.02 Rivas: A New Species of Poeciliid Fish of the Genus Poecilia from Hispan

100 L. R. Rivas

Platypoecilus dominicensis was not a following paragraphs. Limia but a distinct, valid species of Other than introductory philosophical Mollienesia and placed it in that genus remarks on the taxonomic role of genera to stand as M. dominicensis (Evermann as indicators of relationships, Rosen and and Clark). Myers did not explain how Bailey did not give specific reasons to he came to this conclusion since the justify their synonymizing Limia with holotype of dominicensis is a female Poecilia (see critique by Rivas, 1965). and in the absence of adult males, fe­ It is significant, however, that Rosen and male poeciliids are difficult to identify Bailey retained Limia as a subgenus. as to genus and species. Nevertheless, As pointed out by Rosen and Bailey, with males now available for ascertaining the genus can and should serve to express generic relations and by comparing relationships. In keeping with this females of the new material descibed be­ observation, it seems to me that lumping low with the holotype of dominicensis Limia and Poecilia, rather than keeping (USNM 53277), I have confirmed them separate, would mask their affinities Myers' conclusions. Myers (1935) re­ and falsely indicate that they are more described and figured what he thought closely related than they really are. In was M. dominicensis but his description, further agreement with Rosen and figures, and material actually represent Bailey's other remarks, generic groups the species herein described as new. In should be constructed on similar stan­ that paper Myers also designated M. dards of morphological distinctiveness. dominicensis as the type species of This, however, would apply much better Psychropoecilia, a new subgenus of to Limia and Poecilia as separate genera Mollienesia. As to Mollienesia elegans, rather than as a consolidated generic previously known only from four group. I also believe that premature specimens, Riv~as and Myers (1950) shifting of ranks within a family, be­ designated it the type species of a new fore certain groups of species and their genus, Curtipenis, on the basis of relationships are better known, can additional material. lead to misinterpretation of what should In the most recent revision of the constitute a valid genus. Admittedly, family 'Poeciliidae, Rosen and Bailey criteria for generic separation are mostly (1963) synonymized Mollienesia, subjective and, in the end, a genus is Psychropoecilia, Curtipenis, Limia, and accepted or rejected as valid according other genera with Poecilia, but Limia to the personal judgement of the re­ was retained as a subgenus. With their searcher. I do believe, however, that a consolidation of Limia and Mollienesia genus can be more objectively defined with Poecilia the former lvlollienesia= as a group of species more closely Poecilia dominicensis (Evermann and related among themselves than with Clark, 1906) became a junior homonym other species which form similar groups of Limia=Poecilia dominicensis Valen­ within the family (the tribe Poeciliini ciennes, 1846. Rosen and Bailey there­ in this case). This concept is further fore renamed dominicensis of Evermann strengthened if there are, within the and Clark as man tana. I provisionally genus, definable groups of species each accept the consolidation of Mollienesia of which could be interpreted as sub­ and Curtipenis, but not of Limia with genera and/or as species groups. After Poecilia, for the reasons stated in the nearly 40 years of research on the ex-

Published by The Aquila Digital Community, 1978 3 Gulf of Mexico Science, Vol. 2 [1978], No. 2, Art. 2

i'Jew poeciliid fish "101

tensive material of poeciliid fishes (Valenciennes). In accordance with the available to me, I have found that International Code of Zoological Nomen­ according to these crite'ria, Limia should clature, therefore, as provided by Article · stand as a valid genus in the tribe Poe­ 59(c), the name montana of Rosen and ciliini. The characters distinguishing Bailey (1963), proposed in replacement Limia from Poecilia, its closest relative of Evermann and Clark's dominicensis, in the tribe, are given in the key below. becomes a junior objective synonym of Subgenera and species groups within the latter. Limia will be defined in a revision of that genus now in preparation (Rivas, MS). Poecilia hispaniolana, New Species la.- Subdistal segments of ray 3 of Figs. 1 (A, B), 2 (A), and 3 gonopodium with erect or re­ torse spines, or with broadly Mollienisia dominicensis (misidentifica­ T-shaped, somewhat bifurcate pro­ tion, not of Evermann and Clark), cesses; distal arc of ray 5a not bend­ Myers, 1935: 310 (name, original ing abruptly toward ray 4p; ray 5p reference, and synonymy excluded; extending distally to within 2 to diagnosis only), 311 (male and fe­ 6 segments from tip of ray 5a. male figured; Psychropoecilia, new Segments distal to ray 4p serrae subgenus; species erroneously referred 12 or fewer. Three subgenera to dotninicensis of Evennann and (Poecilia, Lebistes, and Pamphor­ Clark; comments excluded), 312 ichthys) as diagnosed by Rosen (USNM material only, except halo­ and Bailey (1963) and about 23 type of dominicensis). species. Mainland of North, Central, Mollienisia (Psychropoecilia) domini­ and South America (20 species) and censis, Trewavas, 1948: 410 (sy­ the island of Hispaniola (3 species) nonymy in part; reference to Myers, Genus Poecilia · 19 35, only as to his diagnosis, figur~, lb.- Subdistal segments of ray 3 of and USNM material, except holotype gonopodium without spines or of dominicensis). processes; distal arc of ray 5a Poecilia montana (in part), Rosen and bending abruptly toward ray 4p; Bailey, 1963: 48 (reference to Myers, ray 5p extending distally to within 1935, only as to his diagnosis, figure, 10 to 15 segments from tip of ray and USNM material, except holotype 5a. Segments distal to ray 4p serrae of dominicensis). 14 or more. Three subgenera to be Synonymy. - As shown below, the diagnosed by Rivas (MS) and 14 diagnosis and figures given by Myers species. Confined to the Greater (1935) clearly refer to this species, not Antilles in Cuba (one species), to P. dominicensis. I have examined Hispaniola (11 species), Jamaica the material (USNM 88884, and 100287) (one species), and Grand Cayman used by Myers for his figures and dia­ Island (one species). gnosis, as well as additional material Genus Limia from the same locality (St. Michel de With the reinstatement of Limia to l' Atalaye). At the time of Myers' full generic rank, Poecilia dominicensis study P. dominicensis was known only (Evermann and Clark) is no longer a from the female holotype (USNM junior homonym of Limia dominicensis . 53277) and two female paratypes https://aquila.usm.edu/goms/vol2/iss2/2 4 DOI: 10.18785/negs.0202.02 Rivas: A New Species of Poeciliid Fish of the Genus Poecilia from Hispan

102 L. R. Rivas

/

c

Figure 1. A. Adult 48mm female paratype of Poecilia hispaniolana, new species from Rio Yaque del Norte at Jarabacoa, Dominican Republic (USNM 218711). B. Adult 27mm male paratype of P. hispaniolana, new species from the same locality (USNM 218711). C. Adult 42mm female of P, dominicensis (Evermann and Clark) from Rio Camu, 9 km west of La Vega, Dominican Republic (FSM 25276). D. Adult 26mm male of P. dominicensis (Evermann and Clark) from the same locality (FSM 25276).

Published by The Aquila Digital Community, 1978 5 Gulf of Mexico Science, Vol. 2 [1978], No. 2, Art. 2

New poecilild fish 103

(USNM 126137). Females of P. his­ 13.4-32.5 mm, 61 adult males 22.5- paniolana and P. dominicensis are 32.4 mm, and 95 adult females 32.5- difficult to distinguish, especially if 51.0 mm, L.R.R. and B.P.H., April 22, only a few specimens of one species 1949 (UMMZ 200218). Rio Yaque del are available. As mentioned above, Norte at Jarabacoa, Prov. La Vega, D.R., with males now available, the identity 44 juveniles and undeveloped males of P. dominicensis and its distinctive­ 12.8-31.0 mm, 22 adult males 24.9-32.0 ness from P. hispaniolana can be esta­ mm, and 44 adult females 31.8-53.3 mm, blished by comparison of females of L.R.R. and B.P.H., April 24, 1949 both species with the holotype of P. (USNM 218711). Rio Camu, 9 km west dominicensis. of La Vega, Prov. La Vega, D.R., 140 Types. The holotype (USNM juveniles and undeveloped males 12.0- 218706) is an adult male 35.5 mm SL 34.5 mm, 23 adult males 27.0-36.0 mm, collected by Luis R. Rivas and Burton and 49 adult females 34.5-58.5 mm, P. Hunt in the Rio Mijo, at road from L.R.R. and B.P.H., April 24, 1949 Azua to San Juan, Province of Bene­ (USNM 218712). Rio Yuna at highway, factor, Dominican Republic, on April north of Monsenor Noel, Prov. La Vega, 21, 1949. Paratypes (USNM 218707), D.R., 687 juveniles and undeveloped collected with the holotype, comprise males 15.6-33.2 mm, 27 adult males 218 juveniles and undeveloped males 25.5-32.0 mm, and 94 adult females 13.6-35 mm, 68 adult males 21.8-36.0 35.0-58.5 mm, L.R.R. and B.P.H., mm, and 38 adult females 35.0-51.4mm. April 24, 1949 (MCZ54057). Riviere The rest of the material collected, from Trois Rivieres at bridge of road from the following 13 localities, are also de­ Gonaives to Port-de-Paix, Dept. de signated as paratypes. Rio Yaque del l'Artibonite, Haiti, 379 juvenile and un­ Sur at bridge of road from Azua to developed males 13.2-34.0 mm, 51 SanJuan, Prov. Azua, D.R., 166juveniles adult males 24.5-35.6 mm, and 66 adult and undeveloped males 14.5-34.3 mm, 28 females 35.0-51.3 mm, L.R.R., April 9, adult males 28.6-34.8 mm, and 46 adult 1951 (USNM 218713). Stream about females 35.5-47.2 mm. L.R.R. and 5 km west of St. Michel de I' Atalaye, B.P.H., April 21, 1949 (USNM 218708). Dept. de I' Arbonite, Haiti, 568 juveniles Rio Vallejuelo at El Cercado, Prov. and undeveloped males 12.0-31.5 mm, 32 Benefactor, D.R., 178 juveniles and adult males 23.7-32.0 mm, and 101 adult undeveloped males 11.0-35.2 mm, 34 females 31.5-43.4 mm, L.R.R., AprillO, adult males 24.0-45.5 mm, and 26 adult 1951 (FSM25275). Riviere Canot at fork females 35.0-47.2 mm, L.R.R. and of road from St. Michel de l'Atalaye to B.P.H., April 21, 1949 (USNM 218709). Hinche, Dept. de l'Artibonite, Haiti, Rio Artibonite at Pedro Santana (Cer­ 19 juveniles and undeveloped males cadillo), Prov. San Rafael, D.R., 136 13.0-26.0 mm, 3 adult males 24.6-25.6 juveniles and undeveloped males 11.0- mm, and 4 adult females 28.0-40.0 mm, 31.0 mm, 9 adult males 22.0-34.5 mm, L.R.R., AprillO, 1951 (USNM 218714). and 12 adult females 31.6-48.0 mm, Riviere Guayamouc at Hinche, Dept. de L.R.R. and B.P.H., April 22, 1949 l'Artibonite, Haiti, 323 juveniles and un­ (USNM218710). Rio Massacre at Loma developed males 12.2-29.5 mm, 53 adult de Cabrera, Prov. Libertador, D.R., males 21.0-32.0 mm, and 58 adult fe­ 413 juveniles and undeveloped males males 30.0-41.0 mm, L.R.R., April 10, https://aquila.usm.edu/goms/vol2/iss2/2 6 DOI: 10.18785/negs.0202.02 Rivas: A New Species of Poeciliid Fish of the Genus Poecilia from Hispan

'!04 L. R. Rivas

1951 (USNM 218715). Riviere Los Pine orbital canal closed, with 3, rarely 2 at road from Hinche to Las Cahobas, 2 pores. Posterior supraorbital canal closed km north of Las Cahobas, Dept. de below, with 2 pores, opening above into l'Ouest, Haiti, 14 7 juveniles and un­ a U-shaped, open pit. Mandibular canal developed males 13.5-29.8 mm, 29 adult closed posteriorly, with 2 pores below males 24.0-32.5 mm, and 118 adult tip of maxillary, opening anteriorly females 29.8-46.0 mm, L.R.R., Aprilll, into a U-shaped groove connecting each 1951 (USNM 218716). Riviere La Tombe side across mandible. Preopercular canal at side road from Mirebalais to Saut closed, with 7 pores. d'Eeau, Dept. de l'Ouest, Haiti, 34 Outer teeth in both jaws uniserial, juveniles and undeveloped males 17.2- forming a weak arc on each side of 28.0 mm, 20 adult males 22.6-26.7 mm, symphysis, close-set, movable, spatulate and 40 adult females 29.4-42.2 mm, or scooplike, slightly curved inward, L.R.R., Aprilll, 1951 (USNM 218717). wider at the crown and tapering toward Name. - This species is named after the base, their cutting edge convex and the Island of Hispaniola to which it is slanted. Inner teeth in both jaws confined and where it occurs in both movable, tricuspid, much smaller than Haiti and the Dominican Republic. teeth of outer row, irregularly arranged, Gonopodial characters. - Terminal forming a band several teeth wide on segments of ray 3, without inner pro­ each side of symphysis. Upper and lower cesses, 14(10-15, usually 12 or 13). jaws weakly united at symphysis. Ray 4p serrae 9(7-11, usually 9). Seg­ Second pelvic ray of males expanded ments distal to 4p serrae 11(9-12, usually medially, forming a bulbous process on 10). Free flap of membranous swelling its outer margin. Gonopodium not of ray 3 much longer than width of reaching to vertical from end of dorsal gonopodium at insertion of flap (Fig. base. 2A), reaching to, or beyond the most distal serra of ray 4p. Processes on anterior margin of subterminal seg­ ments of ray 3 broadly T-shaped, some­ what bifurcate. Subterminal segments of ray 4a smooth. A membranous hook at tip of ray 3 and a bony hook at tip of ray 5p. Anterior margin of ray 5a facing distal spines of ray 4p smooth. Meristic characters. - Dorsal rays 8(8, rarely 7 or 9). Anal rays 9(9). Pectoral rays 16(14-16, usually 15 or l6). Branched caudal rays 15(13-16, usually 14). Lateral scales 28(27-30, Figure 2. Distal end of gonopodium. A. From lJSually 28). Scales around caudal ped­ adult 28.5 mm male paratype of Poecilia uncle 16(16). hispaniolana, new species from Rio Mijo, at road from Azua to San Juan, Dominican Morphological characters. - Preorbital Republic (USNM 218707). B. From adult not free, the skin covering it not folded 26.5 mm male of P. dominicensis (Evermann under its ventral edge. Preorbital canal and Clark) from Rio Camu, 9 km west of La closed, with 4 pores. Anterior supra- Vega, Dominican Republic (FSM 25276).

Published by The Aquila Digital Community, 1978 7 Gulf of Mexico Science, Vol. 2 [1978], No. 2, Art. 2

New poeciliid fish 105

Predorsal contour nearly straight or Relationships. - This species belongs slightly convex longitudinally in males, to the subgenus Poecilia as diagnosed convex, but not strongly, in females; by Rosen and Bailey (1963). In addition transversely convex in both sexes. Body to hispaniolana, this subgenus com­ axis straight, the axis of caudal peduncle prises caucana, elegans, dominicensis, not forming an angle with axis of anterior vivipara, sphenops, butleri, mexicana, portion of body. Origin of dorsal fin formosa, latipinna, petenensis, velifera, above midlength of gonopodium in latipunctata, and sulphuraria. I have males, slightly behind origin of anal in examined material of all these species females, much nearer to upper end of for comparison with hispaniolana. opercle than to middle of caudal base in Poecilia hispaniolana, is readily dis­ males, about midway between upper tinguished from formosa, latipinna, end of opercle and middle of caudal petenensis, velifera, and latipunctata, all base in females. Origin of anal fin mainland species, by its color pattern nearer to upper end of opercle than and fewer dorsal rays: 8, rarely 7 or 9 to middle of caudal base in females. vs. 9 to 19, rarely 9 or 19. It is further Caudal fin truncate or slightly convex, distinguished from these species by about as long as wide, its upper and characters relating to the dentition lower posterior comers angulate. and to the gonopodium. In hispaniolana Coloration. - After original fixation the inner teeth are tricuspid whereas in 10% formalin and 27 to 29 years in in the others they are unicuspid. In 60% ethanol the color pattern is as hispaniolana, the processes on the follows. anterior margin of the subdistal seg­ Ground color yellowish-brown on ments of ray 3 of gonopodium are back and upper sides, becoming lighter, broadly T -shaped and somewhat bi­ more or less abruptly, on lower sides and furcate whereas in the others, these ventral region. Darker margins of scale processes are spinelike and retrorse. pockets producing a reticulate pattern These same gonopodial differenes and on back and upper sides. Nape dark the color pattern distinguish hispaniolana brown, the dark pigment extending as from vivipara. In addition, hispaniolana a middorsal streak to origin of dorsal differs from vivipara in several other fin. Sides of body in males with 8 to 10 gonopodial characters such as the thin, dark crossbars, much narrower subterminal segments of ray 4a which than the interspaces; crossbars also are smooth in hispaniolana and serrate present, and thicker, in juveniles and in VlVlpara. The membranous hook at young females; absent in adult females. tip of ray 3 and the hook at tip of ray Dorsal fin margined with melanophores 5p are present in hispaniolana and absent and with a conspicuous dark blotch on in vivipara. The anterior margin of ray last four rays. Pectoral, pelvic, anal, and 5a facing the distal spines of ray 4p is caudal fins hyaline sometimes a trans­ smooth in hispaniolana and serrate m verse row of dark spots on caudal fin vivipara. of males. The color pattern and the shape of Size. - The largest male measures 36 the processes on anterior margin of ray mm SL and the largest female 59 mm. 3 of gonopodium, as described above, On the average, males are about half separate hispaniolana from caucana, as large as females. butleri, mexicana, sulphuraria, sphenops, https://aquila.usm.edu/goms/vol2/iss2/2 8 DOI: 10.18785/negs.0202.02 Rivas: A New Species of Poeciliid Fish of the Genus Poecilia from Hispan

106 L. R. Rivas

and elegans. In addition, hispaniolana while collecting in the localities where differs from caucana, butleri, mexicana, either species occurs alone, or where sulphuraria, and elegans in having both species are now known to occur tricuspid instead of unicuspid inner together. This may be the reason why teeth, and from sphenops in the pre­ hispaniolana has been confused with orbital which is free in the latter but not dominicensis plus the fact that domini­ in hispaniolana. Also, the anterior mar­ censis was previously known only from gin of ray 5a of gonopodium facing the the three female types. In spite of their distal spines of ray 4p are smooth in very similar appearance and close re­ hispaniolana and serrate in sphenops. lationship however, hispaniolana and Other differences are the posterior dominicensis differ in several characters margin of ray 3 of gonopodium which as shown in the following key. is serrate in hispaniolana and smooth in 1a.- Gonopodium not reaching to ver­ caucana, and the membranous hook tical from end of dorsal base. Free at tip of that ray which is present in flap of subterminal membranous hispaniolana and absent in elegans. In swelling of ray 3 much longer than hispaniolana, the number of dorsal width of gonopodium at insertion rays is 8, rarely 7 or 9 and usually 9 of flap (Fig. 2A) reaching to, or in elegans. Also, in hispaniolana the beyond the most distal serra of membranous swelling of ray 3 forms a ray 4p. Ray 4p serrae 7 to 11, free flap and hood, but not in elegans. usually 9. Segments distal to ray Poecilia hispaniolana is more closely 4p serrae 9 to 12, usually 10. related to P. dominicensis than to any Second pelvic ray of male ex­ other species of the subgenus Poecilia. panded medially, forming a bulbous Both species differ from others of that process on its outer margin. Caudal sub genus in the same characters given fin truncate or slightly convex, its above to separate P. hispaniolana. Of upper and lower posterior corners these 11 characters shared by P. his­ angulate. Pectoral rays 14 to 16, paniolana and P. dominicensis, the usually 15 or 16. Branched caudal most important is the structure of the rays 13 to 16, usually 14. Lateral subdistal segments of ray 3 of the scales 27 to 30, usually 28. In gonopodium. The processes on the fast water with bottom composed anterior margins of these segments are of rock, stones, or gravel. broadly T-shaped and bifurcate in Poecilia hispaniolana, new species P. hispaniolana and P. dominicensis lb.- Gonopodium reaching to, or be­ whereas in all the other species of the yond vertical from end of dorsal subgenus Poecilia these processes are base. Free flap of subterminal spinelike and retrorse. Both species membranous swelling of ray 3 occur together in several localities much shorter than width of gono­ and their ranges are partly sympatric. podium at insertion of flap (Fig. At least in one locality hispaniolana 2B) not reaching to the most distal occurs together with dominicensis and serra of ray 4p. Ray 4p serrae 8 to elegans. Males and females of hispaniolana 11, usually 10. Segments distal to and dominicensis are so similar in ray 4p serrae 7 to 10, usually 8. appearance that the presence of two Second pelvic ray of male not ex­ species was not detected in the field panded medially, not forming a

Published by The Aquila Digital Community, 1978 9 Gulf of Mexico Science, Vol. 2 [1978], No. 2, Art. 2

New poeciliid fish 107

Figure 3. Geographical distribution of Poecilia hispaniolpna (solid circles) and P. dominicensis (open circles) according to 1949 and 1951 locality records by Rivas and recent records by Richard Franz and Fred G. Thompson. Solid circles inside open circles indicate localities where both species occur together.

bulbous process on its outer margin. Haiti eastward to Neiba Bay, Dominican Caudal fin convex, its upper and Republic. Many localities were visited lower posterior comers rounded. throughout the southwestern highlands Pectoral rays 13 or 14, usually 14. and the Cul-de-Sac lowlands and al­ Branched caudal rays 14 to 18, though several species of poeciliids were usually 16. Lateral scales 26 to 28, collected there, no species of Poecilia usually 27. In sluggish water with were obtained. bottom composed of sand or mud. Based on the many localities collected Poecilia dominicensis (Evermann and throughout the central highlands, the Clark, 1906) known range of hispaniolana extends, There is no doubt that hispaniolana on the north slope, from the upper and dominicensis constitute distinct, Trois Rivieres and Artibonite systems reproductively isolated genetic entities, of northwestern Haiti to the upper Rio each representing a valid species. This is Yuna system of northern Dominican well demonstrated by the fact that, Republic. On the south slope, the known where the two species occur together, range extends from a small stream empty­ they show greater divergence and less ing into the northwest comer of Lake individual variation. This phenonmenon, Enriquillo, in southwestern Dominican known as "character displacement," Republic, to the upper Rio Yaque del will be the subject of a separate study. Sur system. The westernmost locality Distribution, poeciliid associations, on the south slope is a recent addition and ecology. - The range of P. hispanio­ to those visited by me in 1949. This new lana is apparently confined to the high­ record was recently obtained by Richard land streams of central Hispaniola Franz and Fred G. Thompson, of the (Fig. 3). This region is sharply separated Florida State Museum, who kindly made from the southwestern highlands by the it available to me (in !itt). Cul-de-Sac Plain (mostly below sea level) As shown on the map, hispaniolana which extends from Port-au-Prince Bay, occurs with dominicensis in four locali- https://aquila.usm.edu/goms/vol2/iss2/2 10 DOI: 10.18785/negs.0202.02 Rivas: A New Species of Poeciliid Fish of the Genus Poecilia from Hispan

108 L. R. Rivas

ties, all of which are on the north slope. perugiile; description of eggs), fig. 2 Only at two localities in the Rio Y aque (female holotype). Regan, 1913: del Norte at Jarabacoa and in the Rio 1015 (erroneously synonymized with­ Yuna was hispaniolana found to occur out comment with Limia domini­ with elegans, in addition to domini­ censis). Myers, 1931:2 (notasynonym censis. In the Rio Yuna, a species of of Limia dominicensis, transferred to Limia was also found to occur with genus Mollienisid; distribution; Rio these three species of Poecilia. In several Camu in part). localities, however, especially in the Limia caudofasciata (misidentification, western section of its range, the only not of Regan), Nichols and Myers, poeciliid found to occur was hispanio­ 1923: 2 (in part?). lana and, in a few others, it occurred Mollienisia domincensis, Myers, 1935: with only one of two species of Limia. 310 (name, original reference, and The known altitudinal distribution synonymy only; diagnosis excluded), of hispaniolana ranges from 91 to 7 32 311 (name only as type species of m (300 to 2,400 ft.) with a mean of the subgenus Psychropoecilia; name 422 m (1,384 ft.). At these elevations, erroneously applied to species de­ shallow riffles and rapids over rock, scribed; male and female figures stone, or gravel bottom, alternate excluded; comments), 312 (holotype with deeper, sluggish pools with sandy only, all. other USNM material ex­ or muddy bottom. According to my cluded). field observations, hispaniolana prefers Mollienisia (Psychropoecilia) domini­ the riffles over hard bottoms. This species censis, Trewavas, 1948: 410 occurs at a much higher elevation than (synonymy in part; reference to dominicensis Myers, 1935, only as to name; his Material examined. - In addition to diagnosis, figures, and USNM material the type material described above, I excluded, except holotype). Rivas and have examined the material reported by Myers, 1950: 289 (gonopodium com­ Myers (1935) as dominicensis, from pared with Curtipenis). St. Michel de 1' Atalaye, Haiti (USNM Limia noblei (nomen nudum, authorship 88884, 100286, and 100287). I have not attributed to G. S. Myers), Rosen and examined the material in the Florida Gordon, 1953: 30, fig. 37 (drawing of State Museum recently collected by tip of gonopodium), 35 (system of Richard Franz and Fred G. Thompson. nerve f~bers of gonopodium). I have verified (in !itt.), however, their Poecilia montana Rosen and Bailey, identification of this species in their 1963: 48 (new name; synonymy and collections, from a drawing of the references in part; renaming of domini­ gonopodium. censis Evermann and Clark, 1906 a Poecilia dominicensis junior homonym of dominicensis (Evermann and Clark) Valenciennes, 1846 because of con­ Figs. 1 (C, D), 2 (B), and 3 solidation of Limia with Poecilia; Platypoecilus dominicensis Evermann holotype of dominicensis E. and C. and Clark, 1906: 852 (original de­ becomes holotype of montana; mat­ scription; stream in San Francisco erial; range). Mountains, Dominican Republic), 853 Synonymy. - Most of the above (compared to Platypoecilus=Limia synonymy is self-explanatory and the

Published by The Aquila Digital Community, 1978 11 Gulf of Mexico Science, Vol. 2 [1978], No. 2, Art. 2

New poeciliid fish 109

majority of items have been discussed segments distal to ray 4p serrae (8). above in the section on taxonomic Types. -The holotype (USNM 53277) history and under the synonymy of is an adult female 38.7 mm SL collected hispaniolana. The name noblei, however, by August Busck in a small stream in requires clarification. This name appears the San Francisco Mountains, "some 40 twice in a paper on the functional miles from Santo Domingo City," anatomy and evolution of male genitalia Dominican Republic, September, 1905. in poeciliid fishes by Rosen and Gordon The paratypes, collected with the halo­ (1953). There is no reference to that type, comprise two adult females 34.5 study to the original publicaton or date mm (USNM 126137) and 29.0 MM SL of the name noblei and no description, (CAS-SU 9350). The type locality diagnosis, or mention of where that described above is a small creek tributary nominal species occurs,museum material, to Rio Raina near Villa Altragracia, etc. On page 30, "Limia noblei Myers" just off the road from Santo Domingo. is simply given in the caption of fig. 3 7 A series of topotypes (USNM 218722) (C) as referring to the drawing of the tip was collected by me on that locality of a gonopodium, with no other com­ on April 24, 1949. This material com­ ment. On page 35, the name Limia prises 26 juveniles and undeveloped noblei is listed, without Myers appended males 19.5-26.2 mm, 13 adult males to it, with several other species sharing 21.4-25.4 mm, and 59 adult females a system of gonopodial nerve fibers. 24.8-43.0 mm. Shortly after the publication of Rosen Name. - This species was named and Gordon's study, I contacted Dr. after Santo Domingo, better known Myers (in !itt.) seeking an explanation today as the Dominican Republic. for the name noblei. He promptly Gonopodial characters. - Terminal replied (in !itt.) that Limia noblei was segments of ray 3, without inner pro­ a manuscript, label name that he had cesses, 10-14, usually 12. Ray 4p serrae put in some museum jars containing 8-11, usually 10. Segments distal to "Noble's Mollienesia dominicensis" from 4p serrae 7-10, usually 8. Free flap Central Dominican Republic and that he of membranous swelling of ray 3 much had never published it. He also implied shorter than width of gonopodium at that he did not intend to publish the insertion of flap (Fig. 2B), not reaching name. As of this writing (August 7, to the most distal serra of ray 4p. Pro­ 1978), the name has not been formally cesses on anterior margin of subterminal published by Myers or by anyone else. segments of ray 3 broadly T-shaped, According to the International Code of somewhat bifurcate. Subterminal seg­ Zoological Nomenclature the name ments of ray 4a smooth. A membranous noblei is, therefore, a nomen nudum hook at tip of ray 3 and bony hook at because it fails to satisfy the conditions tip of ray 5p. Anterior margin of ray 5a of Article 13a. The gonopodium figured facing distal spines of ray 4p smooth. as Limia noblei by Rosen and Gordon, Meristic characters. - Dorsal rays however, can be identified as that of 8(8, rarely 7 or 9). Anal rays 9(9). Poecilia dominicensis by the small size­ rays 13 or 14, usually 14. Branched of the free flap of the membranous caudal rays 14-18, usually 16. Lateral swelling, by the number of ray 4p scales 26-28, usually 27. Scales around serrae (10), and by the number of caudal peduncle 16. The holotype has https://aquila.usm.edu/goms/vol2/iss2/2 12 DOI: 10.18785/negs.0202.02 Rivas: A New Species of Poeciliid Fish of the Genus Poecilia from Hispan

110 L. R. Rivas

8 dorsal rays, 9 anal rays, 14 pectoral between upper end of opercle and middle rays, 27 lateral scales, and 16 scales of caudal base in females. Origin of around the caudal peduncle; the caudal anal fin about midway between upper fin and the branched rays could not be end of opercle and middle of caudal counted. base, or nearer to middle of caudal base Morphological characters. - Preorbital in females. Caudal fin convex, about as not free, the skin covering it not folded long as wide in males, wider than long in under its ventral edge. Preorbital canal females, its upper and lower posterior closed, with 4 pores. Anterior supra­ comers rounded. orbital canal closed, with 3, rarely 2 Coloration. - After original fixation in pores. Posterior supraorbital canal closed 10% formalin and 27 to 29 years in 60% below, with 2 pores, opening above ethanol the color pattern is as follows. into a U-shaped, open pit. Mandibular Ground color yellowish-brown on canal an open U-shaped groove across back and upper sides becoming lighter, mandible. Preopercular canal closed, more or less abruptly, on lower sides with 7 pores. and ventral region. Darker margins of Outer teeth in both jaws uniserial, scale pockets producing a reticulate forming a weak arc on each side of pattern on back and upper sides. Nape symphysis, close-set, movable, curved dark brown, the dark pigment extending inward and away from symphysis, as a middorsal streak to origin of dorsal narrowly spatulate, slightly tapering fin. Sides of body in males with 6 to 9 toward the base, bluntly pointed. Inner thick, dark crossbars, narrower than the teeth in both jaws movable, tricuspid, interspaces; crossbars also present, but much smaller than teeth of outer row, thicker, in juveniles and young females; uniserial and regularly arranged or present or absent in adult females. irregularly arranged and forming a band Dorsal fin margined with melanophores, several teeth wide on each side of the dark margin more conspicuous in symphysis. Upper and lower jaws weakly males, faint or absent in females; a united at symphysis. conspicuous, dark blotch on last four Second pelvic ray of males not ex­ rays. Pectoral, pelvic, anal, and caudal panded medially, not forming a bulbous fins colorless, sometimes a faint, tr~ns­ process on its outer margin. Gono­ verse row of dark spots on caudal fin podium reaching to, or beyond vertical of males. from end of dorsal base. Size. - The largest male measures 27 Predorsal contour nearly straight mm SL and the largest female 52 mm. or slightly convex longitudinally in On the average, males are about half males, convex, but not strongly, in as large as females. females; transversely convex in both Relationship. - This species differs sexes. Body axis straight, the axis of from all the others of the subgenus caudal peduncle not forming an angle Poecilia, except hispaniolana, in the with axis of anterior portion of body. same characters already given above Origin of dorsal fin slightly behind mid­ for the latter. The relationships and length of gonopodium in males, above differences between dominicensis and origin of anal in females, much nearer hispaniolana have also been discussed to upper end of opercle than to middle under the latter. of caudal base in males, about midway Distribution, poeciliid associations,

Published by The Aquila Digital Community, 1978 13 Gulf of Mexico Science, Vol. 2 [1978], No. 2, Art. 2

New poeciliid fish 111

and ecology. - As is the case with or muddy bottom. This species occurs hispaniolana, the range of dominicensis at much lower elevations than his­ is apparently confined to the highland paniolana. streams of Central Hispaniola (Fig. 3). Material examined. - In addition to As already explained in more detail the holotype, paratypes, and topotypes under hispaniolana, this region is se­ described above I have examined the parated from the southwestern highlands following lots from seven localities. by the Cul-de-Sac Plain. Rio Massacre at Lorna de Cabrera, Prov. Based on the many localities collected Libertador, Dominican Republic, 9 un­ throughout the central highlands, the developed males 20.8-25.0 mm, 9 adult known range of dominicensis extends, males 22.0-25.0 mm, and 30 adult on the north slope, from the Rio females 25.5-42.0 mm, collected by Massacre, which forms part of the Luis R. Rivas and Burton P. Hunt, boundary between Haiti and the Domin­ April 22, 1949 (MCZ 54058). Rio ican Republic, to the lower Rio Yuna Tomines at Santiago Rodriguez, Prov. system of northern Dominican Re­ of Monte Cristi, D. R., 268 juveniles public. On the south slope, the known and undeveloped males 11.3-25.5 mm, range appears to be confine~ to the 42 adult males 22.0-25.5 mm, and 51 closely adjacent systems of the Rio adult females 24.2-39.0 mm, L. R. R. Haina and the Rio Ozama. The six and B. P. H., April 23, 1949 (USNM easternmost localities on the north 218718). Rio Yaque del Norte at bridge slope (Rio Yuna) and the two eastern­ of road between Valverde and Guay­ most (Rio Ozama) localities on the acanes, Prov. Santiago, D. R., one south slope are recent additions to those adult female 27.5 mm, L. R. R. and B. visited by me in 1949. These new re­ P. H., April 23, 1949 (USNM 218719). cords were recently obtained by Richard Rio Yaque del Norte at Jarabacoa, Prov. Franz and Fred. G. Thompson, of the La Vega, D. R., 20 juveniles and un­ Florida State Musuem. developed males 15.6-28.5 mm, 11 adult As already indicated, Poecilia domini­ males 21.0-26.0 mm, and 22 adult censis occurs with P. hispaniolana in females 26.3-46.5 mm, L. R. R. and four localities, all of which are on the B. P. H.,April24, 1949 (UMMZ 200219). north slope (Fig. 3). Only at two locali­ Rio Jimenoa at road between Jarabacoa ties was P. dominicensis found to occur and L

112 L. R. Rivas

22.0-26.0 mm, and 37 adult females Nov. No. 503,2 p. 28.3-47.0 mm, L. R. R. and B. P. H., . 1935. An annotated April24, 1949 (USNM 218721). list of the cyprinodont fishes of Hispaniola, with descriptions of two I have not examined the material in new species. Zoologica, N. Y. 10 (3): the Florida State Museum recently 301-316. collected by Richard Franz and Fred Nichols, J. T. and G. S. Myers. 1923. G. Thompson. I have verified (in litt. ), A new Limia from San Domingo. however, their identification of this Amer. Mus. Nov. No. 79,2 pp. species in their collections from a Regan, C. T. 1913. A revision of the drawing of the gonopodium. cyprinodont fishes of the family Poeciliinae. Proc. Zool. Soc. Lond. 11:977-1018. ACICNOWLEDGEMENTS Rivas, L. R. 1963. Subgenera and species groups in the poeciliid fish I am grateful to Michael Endicott genus Gambusia Poey. Copeia 1963 for the photomicrographs of the gono­ (2):331-347. podia and to Grady Reinert for drawing _____ . 1965. [Review of] The the map. As already mentioned, Richard poeciliid fishes (Cyprinodontiformes), Franz and Fred G. Thompson provided their structure, zoogeography and systematics, by D. E. Rosen and R. additional locality records. Special thanks M. Bailey. Copeia 1965 (1):117-118. are due to Dr. Burton P. Hunt, Professor ~----. 1969. A revision of the of Biology, University of Miami, for his poeciliid fishes of the Gambusia help in the field during the 1949 ex­ punctata species group, with de­ pedition to the Dominican Republic. scription of two new species. Copeia 1969 (4):778-795. LITERATURE CITED ~----· MS. A revision of the poeciliid fishes of the genus Linda, Evermann, B. W. and H. W. Clark. 1906. with descriptions of four new species. New fishes from Santo Domingo. _____ . and G. S. Myers. 1950. Proc. U. S. Nat. Mus. 30(1478): A new genus of poeciliid fishes from 851-855, figs. 1-3. Hispaniola, with notes on genera Fink, Vj. L. 1971. A revision of the allied to Poecilia and Mollienesia. Gambusia nicaraguensis species group Copeia 1950 (4):288-294. (Pisces: Poeciliidae). Pub. Gulf. Coast Rosen, D. E. and M. Gordon. 1953. Res. Lab. Mus., 2:47-77, figs. 1-9. Functional anatomy and evolution Hubbs, C. and V. G. Springer. 1957. of male genitalia in poeciliid fishes. A revision of the Gambuis nobilis Zoologica, N.Y. 38(1):1-47. species group, with description of _____ . and R. M. Bailey. 1963. three new species, and notes on their The poeciliid fishes (Cyprinodonti­ variation, ecology, and evolution. formes), their structure, zoogeography, Texas]. Sci. 9(3):279-327. and systematics. Bull. Am. Mus. Nat. Miller, R. R. 1948. The cyprinodont Hist. 126 (1):1-176. fishes of the Death Valley system of Trewavas, E. 1948. Cyp1inodont fishes eastern California and southwestern of San Domingo, Island of Haiti. Nevada. Misc. Pub. Mus. Zool., Univ. Proc. Zool. Soc. 118 (2):408-415. Mich. No. 68, 155 pp. Valenciennes, A. 1846./n Cuvier, G. and Myers, G. S. 1931. Poeciliid fishes o( A. Valenciennes. Histoire naturelle the genus Mol/ienisia in Hispaniola, des poissons. Paris, vol. 18:1-505. with notice of a new Limia from the Samana Peninsula. Amer. Mus.

Published by The Aquila Digital Community, 1978 15