Amerncan Museum

ovitates

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N. Y. 10024

NUMBER 2500 AUGUST 21, 1972

Origin of the Characid Fish Genus Bramocharax and a Description of a Second, More Primitive, Species in Guatemala

By Donn ErRIc Rosen!

ABSTRACT

A new species of the characid fish genus Bramocharax is described from a karst region along the northern foothills of the Sierra de Chama in Guatemala. The region is characterized by sinks and caverns into which streams of various sizes disappear. The points of emergence or hydrographic relationships of these streams are not always evident, but the ichthyofaunal evidence presented helps to tie some of them to two of the major tributaries of the Rio Usumacinta system. The phyletic relationships of the three presently known forms of Bramocharax (the species being described here and the two subspecies of bransfordi) are analyzed. The zoogeo- graphic inferences drawn from the proposed phylogeny are that the genus Bramo- charax arose in the Rio Usumacinta system and then spread southward to the Rio San Juan system of Nicaragua and Costa Rica. The populations of Bramocharax inhabiting lakes Nicaragua and Managua are identified as historically the most recently established.

INTRODUCTION

Continued exploration in the mountainous karst region of Alta Verapaz, Guatemala, has revealed the existence of additional endemic fishes in isolated river basins of the Rio Usumacinta system. This region, along the northern foothills of the Sierra de Chama, is typical karst topography of

1 Chairman and Curator, Department of Ichthyology, the American Museum of Natural History.

2 AMERICAN MUSEUM NOVITATES NO. 2500

. Rio SAN A, Ede,

Ao

> = ms. a = a

Fig. 1, Four main tributaries of the Rio Usumacinta in Guatemala and Mexico. This report is concerned with the karst region of the Sierra de Chama, shown in box at bottom of map and enlarged in same section of figure 2. The inset (upper right) of southern Mexico and northern Central America shows the entire region covered in this map section.

sinks, caverns, and valleys with unsystematic drainage patterns and disap- pearing streams. At present five such streams or rivers with subterranean drainage have been explored by Reeve M. Bailey and me, although there are many more that have been recorded on recent topographic maps and viewed during our aerial surveys. Collecting has thus far been concen- trated in isolated basins associated with the headwaters of two of the main branches of the Usumacinta, the Rfo de la Pasién and the more westerly

1972 ROSEN: BRAMOCHARAX 3

’ opts ASIN ‘J SOE LArg fg f aS

1A YALICAR J “ alae

— - R ey 0 CHIAIN ¥\ '

Y

ee | Statute Miles a Rio CAHABON

Bi) ‘1 Boone” “fF

[-} TUCURU RIO P Rig PANIMS

Fic. 2. Karst region of Alta Verapaz, Guatemala, along foothills of Sierra de Chama. See figure 1. Rivers and streams belonging to different hydrographic basins are set off by boundary lines of long and short dashes. To the west /eft is the Rio Salinas Basin. Northeastern basin top right is the Rio de la Pasién. At the lower right are main branches of Rio Polochic Basin. Heavy arrows point to subterranean channels, sinks, or caverns associated with major disappearing streams of this region. Slender arrows near watercourses show direction of current. Closed and open circles are sites from which fish were collected during 1963, 1968, and 1971. Closed circles, numbered 1, 2, and 3, are localities from which Bramocharax baileyi was collected.

Rio Salinas (fig. 1). Until 1971, Bramocharax was twice collected in the Rio Chajmaic (with an underground connection to the Rfo de la Pasién) and once in the Rio Dolores (with an underground connection to the Rfo Salinas). The only two, small, apparently juvenile specimens taken in the Rio Dolores were regarded as representatives of a disjunct population of the recently described Bramocharax bransfordi doriont Rosen (1970), which was otherwise based on the several hundred specimens from the Rio Chajmaic Basin. Additional collecting during March, 1971, in the general area of the Rio Dolores, to the west of Chajmaic, yielded two more samples

4 AMERICAN MUSEUM NOVITATES NO. 2500

of Bramocharax, totaling 11 specimens. All 11 appear to represent a single form and are inseparable from the original two specimens caught earlier in the Rfo Dolores. Study of these 13 individuals has shown them to be distinct in body proportion, pigmentation, and dentition from members of the Rio Chajmafc population, and they are described herewith.

The field work for this and related studies of Guatemalan fishes is sup- ported by funds from Mr. James C. Greenway, Jr.

Bramocharax baileyi, New sPEciEs Figures 1-6; tables 1, 2 Bramocharax bransfordi dorionti Rosen, 1970 p. 3 (specimens 33.5 and 66.5 mm.

standard length [erroneously given as 24, rather than 34, and 66 mm.] from the Rio Dolores [AMNH! 29415]).

MatTERIAL: The holotype (AMNH 30197) is a subadult 68.5 mm. in standard length, obtained with rotenone in an isolated basin, the Rfo San Simon, 6 km. due west of Chiséc, Alta Verapaz, Guatemala, March 6, 1971, by R. M. Bailey, D. E. Rosen, and party. Taken with the holotype were a subadult 63.5 mm. and a juvenile 35.5 mm. in standard length (AMNH 30198). Additional specimens were secured with rotenone in 1968 and 1971 in Alta Verapaz by R. M. Bailey and D. E. Rosen, as follows: 1968, 2 specimens (AMNH 29415) 33.5 and 66.5 mm. in standard length in the Rio Dolores, along the shore, at Yaxcabnal, March 21; 1971, 8 speci- mens (AMNH 30199) 31.1 to 64.8 mm. in standard length in an outlet tributary of a lagunetta between the Rio Rocié Pemech and the Rfo Canilla due west of Finca Taquec Canguinic, March 23. The 1968 locality was reached by light plane and the two 1971 sites by helicopter.

Diacnosis: A slender bodied, long-snouted form of Bramocharax (depth 3.2-3.8 in standard length; snout 26 to 29 percent of head length) with the humeral spot vertical, extending below lateral line; outer premaxillary teeth well-developed, perforating upper lip, and separated from the inner teeth in a distinct row of four evenly spaced teeth or the anterior three teeth well separated from the fourth; the premaxillary canines (of the inner tooth row) only slightly larger than succeeding teeth and fitting within sockets in the lower jaw when the mouth is closed; second tooth of mandible only slightly smaller than first and third; diastemas between second and third, and third and fourth mandibular teeth slight or absent; all enlarged mandibular and maxillary teeth and all inner premaxillary teeth notably cuspidate, with as many as six lateral cusps.

Description: Dorsal fin rays, 11 (13); anal fin rays, 25 (1), 26 (3), 27 (6),

1 AMNH, the American Museum of Natural History.

1972 ROSEN: BRAMOCHARAX 5

28 (3); pelvic fin rays (sum of right and left counts), 16 (13); pectoral fin rays (sum of right and left counts), 25 (1), 28 (4), 29 (3), 30 (4), 32 (1); lateral line scales, 36 (1), 37 (9); scale rows on left side (counted obliquely backward from pelvic fin origin), 144% (7), 1514 (5). Proportional meas- urements that distinguish bazleyi are given in figure 5 and table 1. As with Bramocharax bransford: bransford: Gill and B. b. dorioni a ratio of snout to head length plotted against standard length provides the simplest morpho- metric separation of baileyi. Measurements of greatest depth of body in relation to standard length show that bazley: is consistently slender bodied, comparable with relative depth measurements of adult male bransford: bransfordi (standard length divided by greatest depth in baileyi: 3.2 to 3.8; in male bransfordi bransfordi: 3.4 to 3.5). Young and adult females of bransford: bransfordi have values comparable with those of young and adult males and females of bransfordi doriont: 2.7 to 3.3. In life, the body color of bailey: is silvery with a pale yellow or yellow-green cast. The pelvic, anal, and caudal fins are a pale rosy color, and the dorsal is pale yellow-orange. The distinguishing humeral blotch has a ventral vertical component that extends below the lateral line. Melanophore patterns are shown in figures 3 and 4.

Dentitional characters are discussed at length below.

GrocraPpHy: The first known sample of bailey: was taken in the Rio Dolores, along the shore at the village of Yaxcabnal (fig. 2, loc. 2). Speci- mens collected subsequently are from a surface tributary to the Rio Dolores (a small stream between the Rfo Rocid Pemech and the Rio Canilla; fig. 2, loc. 1) and from a disappearing stream section of the Rfo San Simén (fig. 2, loc. 3). The latter collection is especially interesting because the Rio San Simén is shown on recent topographic maps of the region (part of a series produced by the Direccién General de Cartagraffa, Republica Guatemala) to drain into the upper Rfo de la Pasién, whereas the other two localities are hydrographically part of the Rfo Salinas (Rfo Chixoy- Rio Negro) Basin. The topographic sheets show two separate disappearing stream sections of the Rio San Simén. The first of these, from which batleyt was collected, is about 4 miles long and passes under a low ridge to the north to join the second and larger disappearing stream section. The latter is about 30 miles long and flows to the east where it divides into two subterranean passages at the foot of a NW-SE ridge. It emerges to the east of this ridge as two streams that join and flow directly into the upper Rio de la Pasién (Rfo Sebol). That the first two and the third col- lecting stations are in different hydrographic basins is suggested by some of the faunal associates of bazley: in the two regions. In the two collections in the Rio Dolores Basin daileyi is associated with the same as yet unde-

‘yyZua] peoy jo uso stad G°/Z St yey} YIU] Inous & se passoidxa aq osje Aevul G/Z JO xopUI Y}3ua] pesy/Y)3us] nous y *(UIZLIO UY [esIOp Jo JUOY UT SMOI aTeIS OM} WO preMyoEQ Afonbi[qo pajunos) sMod ayeos “Y‘og ‘sayeos oul] yesaye] “og’ TT] ‘(pequnos jou jurds ‘yay snyjd yyst1) shes uy oaged “T+y Swg f(er snd 34311) shea uy [esojood “T+ y ‘yg ‘(seseq oyeredas YUM sAvs lr) shea uy yeue ‘yay ‘(saseq oyesedas yqM ses [[e) sher uy [esiop “YC ‘(wT yewqso 03 dy ynous) y38u9] InNous “y-ug {(auesquIsUI Ze[NO -1ado jo 23pa 0) dy ynous) yyBue] pesy “TH ‘Apoq jo yidop ysoyead “q's ¢(aseq [epnes o} dy ynous) ysus] prepueys “TG :suouviesiqqy

LOO ¢00F OOF 9L0F s00F OOF FLOF 10°OF Jo1q piepurys 6 FI 69 91 6°82 8°9G Il ELE ¥'¢ ue {Cl = 91 63 92 II $92 6% OT #E 0°6 Vg — — 91 66 4 Il 19% ¥¢ O's! cs 801 +'8¢ “Fl LE 91 Og 8% Il 812 cs 91 $'¢ FI S66 “SI LE 91 cS 9% Il 912 Ls FEI os Lal 9'OF eal LE 91 8z 82 Il ZLG rags l'6I #'§ O'LI Ges BGI LE 91 OS KG Il 612 0°9 CS 9°§ CLI 6'§9 “Fl LE 91 oS 9G Il oLG GG Z'0% $'¢ 9°61 Z49 “FI LE 91 Og LZ Il L96 9S O12 ¥E o6I 8°+9 66106 HNWV FyaepD ONT pue Yootsag ED0Yy OVW us9Mjaq wesyas {GI — 91 82 LG Il PLS I Cl oan G6 GES {GI 9¢ 9] Og LG Il 882 U9 1S 8 “LI ¢"99 GIt6Z HNWV yeuqeoxe x ‘solojoq ofy Ahal LE 91 82 L Il 69% Vs Lil os Vil CSE “Pl LE 91 8% LG I E86 09 ZS oe 9°61 c'¢9 86106 HNWV (adAjo[0}]) “rl LE 9I 6% 8% Il GLZ i) OCS HE 0°02 G89 £6106 HNWV UugUIIg URS OY Itu 14+4 OOOT x

‘WIS OSTIT Ud wd Wiv Wid “TH/T4s “Tus ‘TH ‘GWO/TIS ‘AO "TS

(SAdJIUITT]IPA, Ul aie syUSUTZINSEIPY ) wajing xDapys0MDIgG AO SAMALVAY OLLSIMAP] GNV OlYLANOHAAOP[

1 GIaVviL

1972 ROSEN: BRAMOCHARAX 7

scribed forms of the poeciliid genera Xiphophorus and Heterandria. It also occurs in the small stream between the Rfo Rocia Pemech and Rio Canilla with the recently described Scolzchthys greenwayi Rosen; S. greenwayi is otherwise known both from the Rio Dolores Basin at Cubilguitz and in an upper tributary to the Rio Senizo that flows directly into the upper Rfo Salinas (Rio Negro). In the Rfo San Sim6n, baileyi was taken together with Aiphophorus helleri guentheri Jordan and Evermann and Heterandria bimaculata (Heckel), both widespread forms that are in all the main branches of the Rfo Usumacinta system exclusive of the Rio Dolores. Although Xiphophorus hellert guentheri is not known from other isolated karst valleys of this region of Alta Verapaz, Heterandria bimaculata is abundant in the isolated Rfo Chajmafc that drains into the upper Rfo de la Pasién. Between the Rio San Simon and the Rfo Dolores Basin there is a small oval valley with one major disappearing stream, the Rio Candelaria Yalicar, and two smaller, possibly intermittent ones. The Rio Candelaria Yalicar contains, among other fishes, forms of Xiphophorus and Scolichthys that are characteristic of the Rio Dolores, although the Heterandria there is distinct from all others. Bramocharax bailey: does not occur in that area, however. The Rfo Can- delaria Yalicar Valley is equidistant between the Rfo Dolores and the Rio San Siméon, about 4 miles from each, and shares some of the unique karst region faunal components only with the Rfo Dolores. We may there- fore tentatively accept the hydrographic interpretation of the Rfo San Simon as a stream historically distinct from the Rfo Salinas and probably related to the Rio de la Pasién Basin. If this interpretation is sustained by future work then it is evident that Bramocharax baileyi, at one time, may have been widespread in the two main tributaries of the Rio Usumacinta, or that latterly bazley: has transferred from one basin to the other, and that it owes its survival in the upper reaches of the Rfo de la Pasiédn and Rio Salinas systems to the development of the topographically isolated refugia that characterize this karst region. The evident possibility also exists that Bramocharax baileyi was at one time sympatric with Bramocharax bransfordi dortont in the upper Rio de la Pasién. Bramocharax bransfordi dorioni is presently known only from the isolated, intermontane Rio Chajmafc, but the Rio Chajmafc is now separated from the lower Rfo San Simén by about 10 airline miles and from the upper Rfo San Simén by half that distance.

A more comprehensive report on the karst fauna of Guatemala is pre- sented in a forthcoming article.

Etymo.ocy: This new form is named for Dr. Reeve M. Bailey: teacher, friend, and colleague of 26 years, and field companion in Guatemala during 1966, 1968, and 1971.

8 AMERICAN MUSEUM NOVITATES NO. 2500

Fic. 3. Guatemalan forms of Bramocharax. Top: Bramocharax bailey, paratype, AMNH 30199, 63.9 mm. standard length. Center: Bramocharax baileyi, holotype, AMNH 30197, 68.5 mm. standard length. Bottom: Bramocharax bransford: doriont, AMNH 29412, 73 mm. standard length.

TAXONOMY AND ComPARISONs: Preliminary comparisons of Bramocharax baileyi with Bramocharax bransfordi doriont Rosen suggested a close relation- ship between them, and indeed the first two specimens of bazleyz collected in 1968 were not separated as a distinct taxon. So striking are the super- ficial resemblances between bazleyi and 6. dorioni, that early in the present investigation it was thought best to treat the two as subspecies of a single form distinct from bransfordi bransfordi in which, alone among the three

Fic. 4. Guatemalan forms of Bramocharax, anterior third of body, to show details of head and humeral spot. Specimens as in figure 3. Top: B. baileyi, paratype. Center: holotype. Bottom: B. bransford: doriont.

10 AMERICAN MUSEUM NOVITATES NO. 2500

KEY ©R/O DOLORES, RIO SAN SIMON @ RIOCHAJMAIC, GUATEMALA 4 LAKE NICARAGUA @ LAKE NICARAGUA, ARROYO TULI = LAKE MANAGUA * RIO CHIRIPIQUI, COSTA RICA

b. bransfordi

STANDARD LENGTH IN MILLIMETERS

220 230 240 250 260 270 280 290 300 310 320 330 340

SNOUT LENGTH IN MILLIMETERS

X 1000 HEAD LENGTH IN MILLIMETERS

Fic. 5. Snout length/head length indexes in populations of Bramocharax. Note that baileyi tends to approach 5b. doriont at smaller sizes and 6. bransford: at larger sizes, although smallest individuals of each taxon are readily separated.

members of this group, the upper canines lie exposed along the outside of the lower lip when the mouth is closed. Such an interpretation of relation- ships seemed also to accord well with their geographic relationships: baileyi and b. doriont in nearby rivers of the same drainage basin (Rio Usumacinta) and 8. bransfordi isolated far to the south in the Rio San Juan system and the great lakes of Nicaragua. Subsequent detailed comparisons of the morphometric, pigmentary, and dentitional characteristics in the three populations of Bramocharax and in similar and related characids such as Astyanax showed that these early interpretations were without foun- dation. In relative snout length baileyi is intermediate between 6. dorioni and b. bransfordi (fig. 5), and in five of six diagnostic dentitional features b. doriont and b. bransfordi are similar, but distinct from baileyi (table 2). Likewise in the single diagnostic pigmentary feature b. dorioni resembles large 6. bransfordi which also has the humeral spot well developed only above the lateral line. Smaller b. bransfordi, including some adult males, have a strong ventral extension of the humeral spot below the lateral line, as in bailey. All the known bailey: are small, however, and a test of the taxonomic utility of this pigmentary trait must await the collection of

1972 ROSEN: BRAMOCHARAX I]

TABLE 2 PRIMITIVE CHARACTERS AND DERIVED CHARACTERS (IN PARENTHESES) IN Bramocharax bransfordt bransfordi, B. bransfordi doriont, AND B. baileyi, WITH A COMPARISON OF OTHER NEw WorLD CHARACID GENERA

bransfordi bransfordt bransfordi dorioni

(Snout much longer than eye) Snout and eye subequal

(Upper canines overlying lower lip) Upper canines concealed

(Inner premaxillary and mandibular (Inner premaxillary and mandibular teeth separated by large diastemas) teeth separated by large diastemas)

(Premaxillary and mandibular teeth (Premaxillary and mandibular teeth slightly or not at all cuspidate) slightly or not at all cuspidate)

(Second mandibular tooth notably (Second mandibular tooth notably smaller than first and third) smaller than first and third)

(Outer premaxillary teeth reduced and (Outer premaxillary teeth reduced and closely aligned with inner teeth; closely aligned with inner teeth; covered by upper lip) covered by upper lip)

(Humeral spot diffuse, well developed (Humeral spot diffuse, well developed above lateral line in most individuals) above lateral line only)

baileyi OTHER TETRAGONOPTERINE GENERA

(Snout slightly longer than eye) Snout subequal to, or smaller than, eye

in most

Upper canines concealed Upper canines, when present, concealed

Inner premaxillary and mandibular Teeth in principal tooth rows not teeth with small or no diastemas separated by diastemas

Premaxillary and mandibular teeth Premaxillary and mandibular teeth notably cuspidate notably cuspidate in most

Second mandibular tooth only slightly Second mandibular tooth equal in size to, smaller than first and third or only slightly smaller than, the first

and third

Outer premaxillary teeth well developed Outer premaxillary teeth well developed, and well separated from inner teeth; separated from inner teeth, and perforating upper lip perforating the upper lip in most

Humeral spot with a strong vertical Humeral spot with vertical component component below lateral line below lateral line in many species of

Astyanax (incl. A. fasciatus), and in Moenkhausia, Gymnocorymbus, Hemigrammus, Ayphessobrycon, Deuterodon, Creagrutus, Bryconamericus, Glandulocauda, Hemibrycon

larger individuals. In fact, 6. dortont and baileyi exactly resemble each other only in having the upper canines concealed, rather than overhanging the lower lip, when the mouth is closed. A comparison of the eight traits analyzed in table 2 with these same attributes in Astyanax shows, further, that bazleyz is more similar in some dentitional and pigmentary characters to Astyanax than it is to the other two forms of Bramocharax (figs. 6-9).

12 AMERICAN MUSEUM NOVITATES NO. 2500

Specifically in Bramocharax baileyi and in various Guatemalan examples of Astyanax the upper canines are concealed, the inner premaxillary and mandibular teeth have small or no diastemas between them, the pre- maxillary and mandibular teeth are notably cuspidate, the outer premax- illary teeth are well developed, perforate the upper lip, and are well separated from the teeth of the inner series, and the second mandibular tooth is only slightly smaller than the first and third. Apart from the concealment of the upper canines—an arrangement common to most characids—the dentitional similarities of bazley: and Astyanax occur also in various other genera of New World characids. For example, having the second mandibular tooth smaller—but not much—than the first and third is characteristic also of Bryconamericus, Argopleura, and Hemibrycon. Having upper canines—but not notably enlarged ones—is common also to Hemi- grammus, Hyphessobrycon, Thayerta, Bryconamericus, and Hemibrycon. Bramo- charax baileyi, the species of Astyanax, and those of numerous other genera, have the outer series of premaxillary teeth well separated from the inner series. The single distinguishing pigmentary feature of baileyi, a humeral spot with a ventral vertical component that extends below the lateral line, also is widespread among New World characids, including many forms of Astyanax, and is possibly also primitive with respect to the condition of the spot in b. doriont and 6. bransfordi. It appears, then, that many of the features in which baileyi is distinguished from b. doriont and b. bransfordi, but not from various other groups of characids, may be primitive features of the group called the Tetragonopterinae by Eigenmann (1917-1927) aud Eigenmann and Myers (1929). Conversely, the dentitional and pig- mentary characters that align b. doriont and 6. bransfordi seem to represent derived character states with respect to New World characids.

Two questions are raised: On what evidence is bailey: included in the nominal genus Bramocharax, and on what grounds, if any, can the genus Bramocharax be retained as a useful and phyletically meaningful taxon?

Initial comparison of the forms of Bramocharax and Astyanax suggests that Bramocharax is immediately distinguished by its longer snout, longer and more consistently concave anterodorsal profile, and longer head. In apparent snout length Astyanax is enormously variable for a number of reasons, which include actual differences in snout length in relation to standard length, snout length in relation to head length, and in relation to orbital diameter. Head length and eye size also vary greatly, even within a single population or between contiguous populations (fig. 11). Comparisons of snout length and orbit size will not, therefore, separate all individuals of Bramocharax and Astyanax, for the relative proportions of snout and eye in relation to head length are about the same in both.

1972 ROSEN: BRAMOCHARAX 13

Fic. 6. Upper and lower jaw teeth in Bramocharax baileyi of 64.8 mm. standard length, AMNH 30199. Outer premaxillary teeth in solid black.

The two groups are generally separable, however, in regard to relative head length, which in Bramocharax is 30 to 33 percent of standard length and in Astyanax, only about 25 percent of standard length. In dentition Bramocharax bransfordi bransfordi and B. b. doriont are clearly differentiated from Astyanax in the reduction and grouping of the outer premaxillary teeth, in the close alignment of the outer and inner series of premaxillary teeth, in the height of the lower canines, the relative smallness of the second tooth in the mandibular series, the diastemas between the mandi- bular teeth, in the reduction in number and size of lateral tooth cusps, and in the large number of maxillary teeth (as many as 19) that can develop in large individuals. These differences, except for number of lateral tooth cusps and number of maxillary teeth, are consistent even in comparisons of the smallest individuals. In contrast, Bramocharax baileyi is more difficult to separate from Astyanax in dentitional characteristics and part of the difficulty may be related to the relatively small size of the presently known examples of baileyi. The largest number of well-developed maxillary teeth

14 AMERICAN MUSEUM NOVITATES NO. 2500

Fic. 7. Upper and lower jaw teeth in Bramocharax bransfordi doriont of 66 mm. standard length, AMNH 29412. Outer premaxillary teeth in solid black.

known in baileyi is seven, although one specimen (fig. 6) shows evidence of incipient development of three additional teeth. Eigenmann (1921) recorded as many as nine maxillary teeth in Astyanax nicaraguensis Eigen- mann and Ogle, and not uncommonly five to seven in various species of the subgenus Astyanax. When and if larger examples of baileyi are collected, lateral cusp reduction and an increased maxillary dentition may be found to typify this form also. Nevertheless, baileyi is at present distinguished in dentition from the forms of Astyanax only by the slightly enlarged mandi- bular canines and by the presence in five of the seven largest of the 13 known individuals of small diastemas between the second and third and between the third and fourth mandibular teeth (compare figs. 6 and 9). These two rather slight differences in dentition between baileyi and Astyanax are the same dentitional traits that support the inclusion of bailey: in Bramocharax, but, at the same time, demonstrate that it is the

1972 ROSEN: BRAMOCHARAX 15

Fic. 8. Upper and lower jaw teeth in Bramocharax bransfordi bransford: of 98.5 mm, standard length. Outer premaxillary teeth in solid black.

proportional elongation of the head and its parts (see, for example, mandibular shape in figs. 6-10) that primarily distinguishes Bramocharax from Astyanax. It is tempting to hypothesize that existing dentitional differences between the two are directly the result of differences in jaw length, i.e., that as the jaw elongates the outer premaxillary teeth are reduced and are aligned with the inner series, the number of maxillary teeth increases (corresponding to an increase in maxillary length?), and that diastemas develop between the teeth of the main tooth rows. No such simple relationships are suggested, however, by comparison of the long- headed, long-snouted Astyanax nasutus Meek with the shorter headed and shorter snouted Astyanax fasciatus (Cuvier) (figs. 9, 10, 12). Moreover, Bramocharax baileyi, which has a relatively longer snout in relation to head length than 3b. dorioni (fig. 5), should, according to this hypothesis, also have smaller outer premaxillary teeth aligned more like those of 5.

Fic. 9. Upper and lower jaw teeth in Astyanax fasciatus of 65 mm. standard length, AMNH 30163, taken together with Bramocharax baileyi in the Rio San Simon. Outer premaxillary teeth in solid black.

bransfordi, and should have larger diastemas among the teeth of the princi- pal tooth rows than does 6. dorioni. The reverse is true, and the distribution of outer premaxillary teeth follows as Astyanax pattern in bailey: and a b. bransfordi pattern in b. doriont. It is therefore evident that tooth size may be independent of tooth pattern and that both may be independent of jaw length. This conclusion tends to reinforce the significance of denti- tional traits in the species of Bramocharax as evidence of the phyletic unity of the three forms. A corollary conclusion is that the genus Bramocharax should be maintained, at least until there is a thorough review of all the rather similar and generalized New World tetragonopterines. !

1 Another New World characid, Scissor macrocephalus Giinther (see Eigenmann and Myers, 1929, pp. 453-454, pl. 64, fig. 7), is very similar in body proportions to the

1972 ROSEN: BRAMOCHARAX 17

Fic. 10. Upper and lower jaw teeth in Astyanax nasutus. Composite made from paratypes (Field Museum of Natural History 5908) 69 mm. to 78 mm. standard length. Outer premaxillary teeth in solid black.

species of Bramocharax. Scissor macrocephalus is known only from the holotype, and its place of origin, although not certainly known, was believed by Giinther to be Surinam. It is like Bramocharax also in meristic details; counts made from the holotype (British Museum [Natural History] 1858.6.14.1) are: dorsal rays, 11; anal rays, iii-26; pectoral rays (right + left), 13 + 15; pelvic rays (right + left), 8 + 9; lateral line scales, 37 (Eigenmann and Myers, loc. cit., gave 38). In the proportions of snout length (10 mm.) and head length (37.3 mm.) this specimen, of 117.5 mm. standard length, falls within the range of Bramocharax bransfordi dorioni. In dentition, however, S. macrocephalus differs from all species of Bramocharax in having no diastemas among the inner premaxillary and mandibular teeth and in having the second and third mandibular teeth of equal size. The holotype was kindly sent for study by Dr. P. Humphry Greenwood.

18 AMERICAN MUSEUM NOVITATES NO. 2500

Fic. 11. Variations in body form, eye size, and snout length in Astyanax fasciatus from karst region of Alta Verapaz, Guatemala. Top: Rio Candelaria Yalicar, 81 mm. standard length, AMNH 30162. Second from top: Rio San Simén, 70 mm. standard length, AMNH 30163. Bottom two: Rio Senizo, 71 mm. (second from bottom) and 75 mm. standard length (bottom), AMNH 30161.

1972 ROSEN: BRAMOCHARAX 19

Fic. 12. Paratypes of Astyanax nasutus from Lake Nicaragua, Field Museum of Natural History 5908. Top: adult male, 66 mm. standard length. Bottom: adult female, 82 mm. standard length.

The analysis above, summarized in table 2, indicates that Bramocharax baileyi is the most primitive member of its group, sharing few of the special- ized features of dentition, and no advanced pigmentary traits with b. doriont or b. bransfordi. In the matter of relative snout length it is less specialized than 6. bransfordi) and somewhat more so than b. dorioni. Bramocharax baileyz is therefore inferred to be the sister group of the more closely related 5. dortont and b. bransford:. That phylogenetic inference has three noteworthy zoogeographic implications, namely, that the genus

20 AMERICAN MUSEUM NOVITATES NO. 2500

bransfordi dorioni

baileyi

bransfordi bransfordi

YD

USUMACINTA

Fic. 13. Zoogeographic hypothesis of origin and dispersal of Bramocharax in Central America based on fewest assumptions regarding actual distributional evidence. Hypothetical common ancestor of group (A) is inferred to have been in the Rio Usumacinta where baileyi (primitive sister group of the other forms) is now. Hypothetical common ancestor of b. dorioni-b. bransfordi complex (B) is also inferred to have been in the Rio Usumacinta where 8. dorioni is now confined. Hypothetical ancestor (B) would therefore also have given rise to a population that dispersed southeastward and would most likely have entered the Rio San Juan drainage of Nicaragua and Costa Rica via its tributary waters. Once having entered the Rio San Juan Basin, and having differentiated as b. bransfordi, it would have moved upstream to the great lakes of Nicaragua. This hypothesis implies only a single major dispersal. All other hypotheses would require the dispersal of two or more ancestral taxa and the subsequent disappearance of descendent taxa from their ancestral point of origin.

originated in the Rfo Usumacinta Basin, that the dorioni-bransfordi complex also originated there, and that a member of the dorioni-bransford: complex might be expected to be found somewhere in the largely unexplored waters of eastern Guatemala, Honduras, and Nicaragua between the Rio

1972 ROSEN: BRAMOCHARAX 21

Usumacinta and the Rio San Juan (fig. 13). Bussing (1967) proposed that the presence of bransfordi in the lower San Juan and its tributaries may indicate that this form is spreading southward from a population center in the great lakes of Nicaragua. Our present knowledge of the forms of Bramocharax and the overall zoogeographic picture hypothesized here is more consistent with an interpretation of the Rfo San Juan population as stemming from an older invasion of Nicaragua that ultimately found its way upstream to the lake basins. Hence, the populations in Lake Nicaragua and Lake Managua might be of more recent origin than all other populations of Bramocharax.

LITERATURE CITED

Bussinc, W. A. 1967. (April) New species and new records of Costa Rican freshwater fishes with a tentative list of species. Rev. Biol. Trop., for 1966, vol. 14, no. 2, pp. 205-249. EIGENMANN, C. H. 1917-1927. The American Characidae. Mem. Mus. Comp. Zool., Cambridge, vol. 43, pts. 1-4, pp. 1-428. EIGENMANN, C. H., AND G. S. Myers 1929. The American Characidae. Mem. Mus. Comp. Zool., Cambridge, vol. 43, pt. 5, pp. 429-558. Rosen, D. E. 1970. A new tetragonopterine characid fish from Guatemala. Amer. Mus. Novitates, no. 2435, pp. 1-17.