Rev. Biol. Trop., 45(3): fl 85-1208, 1997

Phylogenetic and taxonomic implications of protein variation in the Mesoamerican genus (Caudata: )

David A. Goodl andDavid B. Wake2 1 Museum oC Natural Science, 119 FosterHall, Louisiana State University, Baton Rouge, LA 70803, U.S.A. Museum oC Vertebrate Z00logy and Department oC lntegrative Biology, University of California, Berkeley, CA 94720- 3160, U.S.A.

(Recibido: 10-lV-96 Revisado: 14'-I1I-97 Aceptado: 14'-I1I-97)

Abstraet: An analysis of protein variation was conducted among 28 population samples of of .the genus Oedipina. collected in Guatemala, Honduras, Costa Rica, and Panama, in order to assess the validity oC currently recognized specles and to evaluate their phylogenetic relationships. A set oC eight species of the related genus Nototriton served as an outgroup. There are four major clades, with Nei geneuc distances between elades ranging frorn 0.46 to 2.67. Patteins of geographic variation within O.u.niformis as currently recognized indicate thatpopulations withinit should be placed ioto fourspecies: O. alfatoiis a valid species, and O. gracilis and O. pacificensis are �surrected from the synonymy of O. uniformis, while O, bonitaensis, O, inusitata, O.longissima, and O. syndactyIqremain in synonymy. Geographic variation also is apparent io O.cyclocauda, O. poelzi, and O. pseudouniformis, but.no taxonomic changes arewarranted at thistime. The speciesstatus of O. altura andO. grandis issupported. The lour samplesrepresenting the parvipes group are the rnostdistantly related to the other samples, andgenetic variation within thegroup is sufficientto warrant resurrectingthe Costa Rican O.alleni from the synonyrny of O.parvipes of Panama and Colombia All motphologically differentiated samples are also genetically differentiated. Our results demonstrate that 12 specieswere sampled (O.alfaroi, O. allelli, O.a.ltura, O. cyclocauda, O.gr acilis, O.grandis , O.ignea, O.pac ificensis, O.parvipes, O.poelzi, O. pseudouniformis, and O. uniformis).no samples were available for O.éarablanca, O� collarís, O.complex, O. elongata, O. gephyra, O. paucidentata. O.stenapodia, O.stuarti, or O.taylori, but all are recognizable as distinct species on motphological

grounds. ,

Key words: Caudafa, Oedipina, Allozymes, Proteins, Variation, Phylogeny,

The great radiation of plethodontid and intrapopulational variation in this character. salamanders in the New World tropics has Amongthe tropical species in the supergenus, produced a large number of species that· are onlythe elongated, attenuate salamanpers in the diverse in botb morphology and ecology. These genus Oedipina have increased vertebral species, aH members of the monophyletic numbers, and hence Oedipina constitutes one of supergenus Bolitoglossa, comprise about 45% the best supported monophyletic groups of of the living species of caudate tropical salamanders. Although a few species of (Wake.19 87). Members of this supergenus are Bolitoglossa occur farther south, Oedipina has unique in that the vast majority have a fixed the most southerly overaH distribution of any number (14) of trunk vertebrae, regardless of salamandergenus. The species are ecological whether the body is elongated or noto Most specialists for the IDOSt part, having fossorial other élades ofsalamanders display both inter- habits even in the lowlands where other 1186 REVISTA DE BIOLOGIA TROPICAL salamanders typically are arboreal (Wake and views are generally concordant with those of Lynch 1976). Frost and Hillis (1990), and we accept their Many of the species of Oedipina are main recommendations. Briefly, we consider confusingly similarin structure and coloration. species to be genetically cohesive units that are Most of .the 18 species currently recognized are evolutionarily independent of other such units. dark gray to brown, a1though sorne species In the case of allopatric populations, we are have lightly pigmented dorsal patterns, concerned with historical processes that have especially on the head. They all have very long led to the allopatry, and with estimates of the tails (sometimes 3 times the snout-vent likelihood of future genetic interchange. These length), short legs, and reduced hands and feet. estimates are based on ecology and geography, This high degree of similarity has led to much as well as on measured differences in confusion with regardto the identification and morphological and bíochemical traits. We systematics of species in the genus. The most attemptto estimate relative length of separation difficultcomplex is the taxon currently known on the basis of these characters (primarily as O. uniformis,which ineludes seven described morphology and biochemistry). We assume entities (Brame 1968. O. bonitaensis, O. that populations that have been in isolation gracilis, O. inusitata, O. longissima. O. sufficiently long to permit substantial pacificensis, O. syndactyla, and O. uniformis). morphological and biochemical differentiation It ranges from sea level to over 2000 m are more likely to be reproductively elevation, resulting in its occurrence in a incompatible than are more recently isolated broader range of hábitats than any other populations in which little differentiation has salamander. occurred. This approach is necessarily Two workers, TayJor (1948, 1949, 1952, subjective, but it avoids both the acceptance of 1954) and Brame(1960, 1963, 1968, Brame extremely polytypic species on the one hand andDuellman 1970), have been responsible for and the acceptance of species status for all most of the species descriptions and current undifferentiated (or poorly differentiated) . taxonomy in Oedipina. In the most recent and geographic isolates on the other. in cases in comprehensive reVISlon, Brame (1968) which tests of sympatry are impossible, our recognized 15 species in two species groups. species conceptinvolves an inference regarding He described five new speciesand reduced nine the probability of reproductive isolation. names to synonymy. Brame and Duellmart Sympatry tests probably are rare among (1970), Brodie and Campbell (1993), and relatively young species, but when they are McCranie et al. (1993) each added an additional available the degree of differentiation .among species. closely related sympatric species can guide In the course of field work in Central decisions concerning allopatric populations. In Americaover thelast 15 years(see Wake 1987, general, we believe thatit is better to err in the Wake el al. 1992), an attempthas been made to direction of too many rather than too few obtain samples of these elusive for species, and when evidence for differentiation is morphological and biochemical studies. Recent present andwe have sufficientmaterial to make work has concentratedon Costa Rica, where the a reasonable judgement, we choose to recognize majority of species occurs. This report taxa as full species. summarizes a re-examination of the species status of several populations of Oedipina using biochemical information unavailable to MATERIALS AND METHODS previousworkers. In recent papers we have discussed our A total of 26 populations (Table 1) species concept and our operational taxonomic representing 11 specíes (see discussion below) procedures (Good and Wake 1992, 1993). Our were analyzed for variation at all of the 22 Good &Wake: Mesoamerican salamander genus Oedipina 1187 protein loci listed in Table 2. The distribution unavoidable in this analysis because many of of alle1es al these loei is listed in Table 3, as is the species are very rare (e.g., our specimen of the outgroup state, where appropriate, based on O. altura was only the second specimen found the alleles present in eight species of since the species was firstcollected in 1964). Nototriton, a close relative of Oedipina (Wake AH genetic distances reported in this paper and Elias 1983). In addition, one individual of are calculated according to Nei (1978). O. cyclocauda from Honduras al1d five O. FuU details of synonym for the taxa grandis from Costa Rica were analyzed (Tabie discussed in this paper are provided by Brame 1), but only a subset of the loci were scorable. (1968). data for these specimens are presented in the text onIy. Liver and intestine samples were collected RESULTS from freshly killed specimens and stored at - 76°C until used. Tissues were combined and We found 129 alleles distributed among fue homogenized in approximately equal parts 22 proteins examined in the 26 samples of tissueand deionized water, and fuen subjected to Oedipina (Table 3). No proteins are horizontal starch-gel electrophoresis using monomorphic among all samples, although standardtechniques (Selander et al. 1971, Harris Icdhl and Icdh2 vary only in fue presence of andHopkin son 1976, Richardson et al. 1986, polymorphisms involving unique alleles in Hillis and Moritz 1990). The allele patterns single populations. Alleles per protein range resultingfrom this analysis were subjected to a fromtwo (Icdh2 and Pk) to ten (Ldh and Pepl). variety of phenetic and phylogenetic analyses The maximum number of alleles per protein in both by hand and using the BIOSYS· (Swofford a single sample is three (Pepl in sample 20. and Selander 1981) and PAUP (Swofford 1985) see Table 1). computer packages. Voucher speeimens were There areeight distinct morphotypes (Figs. deposited in the Museum of Vertebrate 1 and 2) among the populations of Oedipina Zoology, University of California (MVZ), the surveyedfor aH loei in this analysis, and these University of Miami (CRE), and the University were considered by Brame (1968) to represent of Costa Rica (UCR). eight species (O. alfaroi, O. altura, O. In the phylogenetic analyses that follow, a cyclocauda, O. ígnea, O. pa rvipes, O. poelzi, conservative approach to character state O. pseudouniformis, and O. un iformis). There transformation designation is taken in that all arecomplex patterns of sympatry among these alleles present together in a polymorphic state morphotypes, which taken together represent a in any taxon are considered to represent a single IDlnImUm estímate of species diversity. character state (Buth 1984). In this way, if, in a Oedipina alfaroi, O. altura, and O. ígnea were hypothetical case, the allele combinations A, represented in this analysis by single samples AB,B, andBC occur in four taxa, all four are and genetic distances within the O. considered to show the same character state. an pseudouniformismorphotype are slight (0.01- alle1e D in a fifth taxon is considered to be a 0.02. Table 4). Within the other morphotypes, separate state if there areno taxa with the allele however, fuere is considerable protein diversity. combinations AD, BD, or CD. This approach Genetic distances within the O. cyclocauda misses potentially useful information in morphotype range up to 0.31, in the O. poe lzi transformalÍon like A-AB-B-BC in the case morphotype to 0.39, in the O. uniformis aboye (see Mabee and Humphries 1993, morphotype to 0.70, and in the O. parvipes Murphy 1993), but the disadvantage is morphotype to 0.83. outweighed by the advantage of also missing In order to determine the degree of genetic potentially obscuring patterns resulting from cohesion within morphotypes, the relation of small sample sizes. Small samples were genetic distance to geographic distance was 1188 REVISTA DE BIOLOGIA TROPICAL examined. Under an isolation-by-distance synonymized them with O. unifo rmis. The model, if populations are in genetic contact relatively high elevation populations (the El (i.e., gene flow occurs) and population genetic Angel, etc., cluster) retaín the name O. characters (mutation, migration, etc.) are in uniformis. The Atlantic lowland form (La equilibrium and more-or-less uniform over Selva) is O. gracilis, and the Pacific lowland geography, points representing pairs of form (Corcovado and Quepos) is O. populations on a pIot of genetic distance pacificensis. Oedipina alfaroiis as similar to O. against geographic distance are expected to faH uniformis as is O. pacificensis, and more on a straight Hne intersecting the origin, similar to O. uniformis than is O. gracilis (but because genetic divergence between any two see discussion of evolutionary rates below). populations should be directly proportional to The pIot of comparisons within the O. their geographic separation (Nei 1972). If there poelzi morphotype (Eg. 4) shows similarities is no genetic interchange among populations, to the analysis of the O. uniformis there is no reason to expect this intersection morphotype. The Volean Barba, Monteverde, with the origin (unless genetic contact only and Las Nubes populatíons (samples 10, 11, recently has been 10st), because genetic and 13) have been contiguous in terms of gene divergence should be independent of geographic flow, at least until recently, but the Moravia re proximity. Fig. 3 is a pIot of Nei genetic Chirripo population (sample 12. Table 1) is distance against geographic distance for the genetically distínct and may be genetically populations of the O. uniformis morphotype. independent. While the degree of differentiation The Corcovado and Quepos populations is such that species recognition of the Moravia (samples 18 and 19. Table 1) are almost population may be warranted, our genetic identical to each other (Table 4). The El Angel, analysis is inconclusive due to límited samples Capelladas, Monteverde, Las N ubes, Tapanti, and we have inadequate preserved material to Turrialba, and Juan Viñas populations (samples properly categorize and diagnose the putative 20-26. Table 1) form a regression line undescribed species. Therefore we defer further suggesting genetic interchange. These discussion of this topic to a later time. populations, however, do not show a similar Genetíc distances between the kind of relationship with the La Selva (sample morphologically distinct Costa Rican and 17) or Corcovado/Quepos populations, nor 00 Panamanian populations of the O. parvipes the La Selva and Corcovado/Quepos morphotype range from 0.61 to 0.83, and we populations show a pattern suggestive of consider the Costa Rican populations (samples genetic contact with each other. There are four 6-7) to represent a distínct species, for which fixed allele differences between the El Angel, the name O. alleni is avaílable (Taylor 1954). etc., cluster and Corcovado/Quepos, seven The populations from Panama (samples 8-9) between the El Angel, etc., cluster and La are retaíned as O. parvipes for the present, but Selva, and 11 between Corcovado/Quepos and they, too, are probably specifically distínct La Selva. The corresponding genetic distances (based on morphology. see Brame 1968) from are 0.32-0.48, 0.51-0.63, and 0.53. The genetic the South American populations (the source of distances from O. alfaroi (representing the O. the holotype). agaín we defer further discussion alfaroi morphotype) to the El Angel, etc., to another time when more material is cluster (0.36-0.48) arelower than many of the avaílable. distances within the O. uniformis morphotype. The aboye discussion suggests that there are Ourinterpretation of the aboye result is that at least 11 genetically independent units the O. uniformis morphotype consists of at (species) amongthe samples analyzedfor all 22 least three genetically independent units. These proteins. These will be discussed hereafter as units have been described in the past (Taylor the distinct OTU's O. alfaroi, O. alleni, O. 1952) as separate species, but Brame (1968) altura, O. cyclocauda, O. gracilis, O. ígnea, O.

1190 REVISTA DE BIOLOGIA TROPICAL genetic distances from O. alfaroi, O. gracilis, increased protein eyolution in the ancestor of O. pacificensis, and O. uniformis to any other the uniformis dade is the more likely Oedipina are expected to be roughly equal if explanation (but recall the comments aboye evolutionary rates are approximately equivalent conceming the unreliability of large Nei along all lineages within the c1ade containing distances). Distances from O. altura are those taxa. This is not the case: O. gracilis is consistently higher than di,stances from O. consistentlymore distantfrom each of the other cyclocauda or O. pseudouniformis to each of Oedipina(except the parvipes clade) than are the other Oedipina species, suggesting an any of the other three species. Similarly, O. increase in rate of protein eyolution in O. uniformis is consistently doser to each of the altura. other. Oedipina (again exduding the parvipes In summary, examination of genetic clade) than are O. pacificensís or O. alfaroi. distances suggests that there haye been rate This suggests thatthere has been arate increase increases in the common ancestor of the of protein evolution in O. gracilis and a uniformis dade, in O. altura, and in O. gracilis, decrease in O. uniformis. On the other hand, if and a rate decrease in O. unifo rmis. If this such rate changes occurred, a similar pattem of argument is accepted, the genetic distances relative rate comparisons should be expected inyolving O. ígnea can be used to estimate its when comparing these populations to O. alleni phylogenetic position. These distances are: and O. parvipes. This was not seen. However, 1.66-1.96 to the parvipes dade, 1.29-1.47 lo the magnitude of the distances inyolyed the cyclocauda dade, 1.04-1.13 to the poelzi (DN=1.84-2.67) casts doubt on their utility. dade, and 1.13-1.18 to the uniformís clade differences between Nei distances less than 1.0 species that haye not undergone rate changes. are much more meaningful than differences The distance between O. ígnea and the parvípes betweendistances aboye 1.5 because the greater dade are within the range of distances from the the difference, the greater the probability that parvipes dade to either the cyclocauda clade or observed similarities are due to chanceo Because the lineage induding the poelzi and uniformis the rate differences discussed aboye are clades, and accordingly it is not possible to use suggested by comparisons in the meaningful these data for phylogenetic purposes. However, range for allozmes(0.46-1.66), they are likely O. ígnea is approximately equidistant from to be genuine. boththe poelzí and uniformis dades. If it were The distances from each of the species in anywhere outside of the clade including the the cye/ocauda clade to O. poelzi on the one poelzi and unifo rmisclades, the genetic distance hand and to O. alfarai and O. pacificensis(the to the species of the uniformís dade would be species of the uniformis c1ade that have not expected to be higher than the distance to the undergone apparent rate changes) on the other poelzi clade because of the hypothesized rate are expected to be equal if rates are constant íncrease in the branch defining the uniformís according to the aboye argumento Distances dade. Therefore, the most parsimonious from O. poelzi to O. altura, O. cyclocauda, and placement for O. ígnea is as the sister taxon to O. pseudouniformis (0.49-0.89) are the unifo rmis dade, as suggested by Aat-l. A consistently lower than the distances from O. potential confounding factor for this hypothesis pacificensis and O. alfaroi to the same three is that the genetic distances between the poelzi species (0.89-1.17), suggesting either that there and unifo rmísclades areconsistently lower than has been arate increase in the common ancestor the distances between O. ígnea and the of the uniformis dade or arate decrease in O. uniformis dade, eyen though O. ignea is poelzi. Distances aregre ater from the parvipes hypothesized to be more dosely related lo the clade to O. alfaroi and O. pacificensis (1.84- uniformis clade than to the poelzi clade. We 2.67) than to either O. poelzi (1.07-1.95) or the hypothesize an increase in the eyolutionary rate cyclocauda clade (1.14-1.99), suggesting that in O. ígnea, which also accounts for the higher Good& Wake: Mesoamericansalainander genus Oedipina 1191

genetic distances· from the cyclocauda clade to geographic distance between the latter units is O. ignea than from the cyclocauda dade to the only about 40 km while that between the poelzi dade or to the uniformis dade. Honduran population and Costa Rican O. Two other populations of Oedipina were cyclocaudais greater than 640 km.Isolation by studied, but wereexcluded from the discussions distance within a single species could account above because the individuals available were for trus level of divergence. it is far less than unscorable for several proteins. One of these would be predicted for populations at a similar consisted of a single individual of the O. geographic distance within either O. poelzi or cyclocauda morphotype from Honduras. At 13 O. uniformis (see Figs. 3 and 4). of the 18 proteins scorable this individual was Following completion of the main identical to the Costa Rican O. cyclocauda. At laboratory work tor this study we obtained the other five (Icdhl, Ldh, Mdhl, Mdh2, aro specimens from a single population of O. Pepl), it was fixed for a difierent allele than grandis. Unfortunately, four proteins were that found in the Costa Rican O. cyclocauda. In unscorable (Ada, Gdh, Hadh, and Pk). Of the two of these five cases (Icdhl and Ldh), the proteins that remained, three (Ldh, Pep1, and Honduran population had an allele that was Pep2) showed a unique allele in O. grandis. The unique among the Oedipina sampled. For two alleles present at the remaining 15 proteins other proteins (Mdhl and Mdh2) it had the were(compare with Table 3): Aat-c, Aconl-a, postulated ancestral allele for the genus. At Acon2-a, Adh-a, Estl-a, Est2-b, Gd-a, Gpi-a, Pepl, the Honduran population shared an allele Icdhl-a(0.875)/d(0.125), Icdh2-a, Mdhl-b, with O. altura but at the same locus the Costa Mdh2-a, Mpi-c, Nadhdh-c, and Pgm-d. Of Rican O. cyclocauda possessed a unique allele these, three present phylogenetic ínformation: and O. pseudouniformis had aneles among the Aat places O. grandison the lineage containing ancestral seto On the basis of allele distribution, the cyclocauda, poelzi, and unifo rmis dades, therefore, the Honduran population appears to Mdhl places it in the cyclocauda dade, and be the sister taxon to the Costa Rican O. Mdh2 suggests that it falls outside of the cyclocauda. The genetic distance between the lineage . containing O. cyclocauda and O.

two of 0.31 is a level of genetic divergence pseudouniformis (Fig. 6). The species IS similar to that between the main body of O. similar in coloration and proportions to O. poelzi andthe Moravia deChirripo population altura. (a possible distinct species. see above), but the

TABLEl

Voucherspecimens far population samples usedin theanalysís of allozymevariation in Oedipina. Loca/ityabbreviations in parentheses correspandto those used in Tables 3 and 4.

O. alfaroi morphotype: 1. O. alfaroi(n=l): MVZ181228, Moravia de Chirripo region, moss bankon Rd. 232, 2-3 km W of junction with road to ElSeis, Prov. Cartago, Costa Rica, 850 m elev., 9°49'N, 83°15W.

O. alturamorphotype: 2. O. altura (n=1): MVZ 190849, 0.8 km W El Empalme, Prov. San José, Costa Rica, 2320 m elev., 9°43'N, 83°57'W.

O. cyclocaudamorphotype: 3. O. cyclocauda (LAS) (n=2): MVZ 203747-48, Estación Biología La Selva, Puerto Viejo de Sarapiquí, Prov. Heredia, Costa Rica, 100 m elev., 10° 26'N, 84° 01'W. 4. O. cyclocauda (RIO) (n=2): MVZ 210397, 210399, S Río Frío, 11.4 km (rd.) N San José-Guapiles Rd. at Río Sucio, Prov. Heredia, Costa Rica, 140 m elev., 10019'N, 83°57'W. 1192 REVISTA DE BIOLOGIA TROPICAL

O. ígnea morphotype: 5. O. ignea(n=5): MVZ 138918,143947, 160917, 163648-49, Finca Santa Julia,1.25 km E & 0.75 km S San Rafael Pie de la Cuesta,Depto. San Marcos,Guatemala, 1100 m elev.,14°56'N, 91°54'W.

O. parvipesmorphotype: 6. O. allení (COR) (n=2): MVZ 190856-57,Sirena, Corcovado NationaJ Park,Prov. Puntarenas,Costa Rica,8 m elev.,8°29'N, 83°34'W. 7. O. alleni (QUE) (n=3): MVZ210401-03, 1.7 km (rd.) NE Parrita-Quepos Hwy. at point 3.7 km NW Damas, Prov. Puntarenas,Costa Rica,SO m elev.,9°30'N, 84°13'W. 8. O. pampes (CZ) (n=l): MVZ 210405,Near RíoFrijoles, W side Pipeline Rd.,4.8 km (air) NW Gamboa,Canal Zone,Panama, SO m elev.,9°09'N, 79°44'W. 9. O. parvipes(NUS) (n=1): MVZ210404, Nusagundi, Kuna Yala,Prov. San Blas� Panama,280 m elev.,9°16'N, 78°58'W.

O. poelzi morphotype: 10. O. poe/zi (BAR) (n=9): CRE 7432,744 1,MVZ 206396-99,UCR ZP-566,ZP-616, 2050 m camp,Braulio Carilla National Park,Volcán Barba,Prov. Heredia,Costa Rica,20SO m elev.lOo08'N,84°06'W. 11. O. poe/zi (MON) (n=5): MVZ 207127, Peñas Blancas Trail below (E) continental divide, Monteverde Reserve,Prov. Alajuela,Costa Rica,1535 m elev.,10019'N, 84°48'W;MVZ 207128,Carril Bosque Eterno,N of Pantanosa Trail,Monteverde Reserve,Prov. Alajuela,Costa Rica,1595 m elev., 10019'N, 84°48'W;MVZ 207129�30,207132, La Ventana,Monteverde Reserve,Prov. Alajuela,Costa Rica,1560 m elev.,1O°19'N, 84°48'W. 12. O. poe/zi (MOR) (n=1): MVZ 194873,Quebrada Platanillo,5.7 km E Moravia de Chirripo,Prov. Cartago, Costa Rica,1200 ro elev.,9°49'N, 83°24'W. 13. O. poe/zi (NUB) (n=10): MVZ 163702,Las Nubes,pastures near Castillo's farro,Prov. San José,Costa Rica, 10002'N,83°56'W; MVZ 181347,Castillo's farro,Las Nubes area,12.6 km ENE San Isidro de Coronado on Rd. 216,Prov. San José,Costa Rica,10002'N, 83°56'W; MVZ 190831-36,190896-97, Las Nubes de Coronado,ca. 0-2 km NE RíoCascajal, Prov. San José,Costa Rica,1700 m elev.,10002'N, 83°56'W.

O. pseudouníformismo rphotype: 14. O. pseudounifonnis(MOR) (n=I): MVZ181229, Moravia de Chirriporegion, moss bank on Rd. 232,2-3 km W of junction with road to El Seis,Prov. Cartago,Costa Rica, 850 m elev.,9°49'N, 83°27'W. 15. O. pseudounifonnis(TUR) (n=I): MVZ 203749,Los Espaveles,above RíoReventazon, C.A.T.I.E., ca. 2 km ESE Turrialba,Prov. Cartago,Costa Rica,9°53'N, 83°39'W. 16. O. pseudounifonnis(VIN) (n=4): MVZ 181230,Colorado Swamp,6.9 km W Turrialba (centro), Prov. Cartago, Costa Rica,1140 m elev.,9°54'N, 83°43'W; MVZ 190850-52,La Cienega de Colorado,ca. 3 km ENE Juan Viñas,Prov. Cartago,Costa Rica,1035 m elev.,9°54'N, 83°43'W.

O. uniformismorphotype: 17. O. gracílis(n=3): MVZ203752-54, Estacion BiologiaLa Selva,Puerto Viejo de Sarapiqui,Prov. Heredia,Costa Rica,100 m elev.,10° 26'N,84° Ol 'W. 18. O. pacificensis (COR) (n=2): MVZ 190858-59,Sirena, Corcovado National Park,Prov. Puntarenas,Costa Rica, 8 m elev.,8 °29'N,83°34'W. 19. O. pacificensis(QUE) (n=1): MVZ 210396,1.7 km (rd.) NE Parrita-Quepos Hwy.at point 3.7 km NW Damas, Prov. Puntarenas,Costa Rica,SO m elev.,9°30'N, 84°13'W. 20. O. unifonnis (ANG) (n=5): MVZ 181247,Varablanca region,ca. 4 km E national Hwy. 9 on regional Hwy. 120,Prov. Alajuela,Costa Rica,1950 m elev.,lOo10'N, 83°45'W; MVZ 190854-55,Salto El Angel,Prov. Alajuela,Costa Rica,1340 m elevo 1001O'N,83°45'W; MVZ 207133,2.4 km WSW Poasito junction on Hwy. 120, Prov. Alajuela,Costa Rica,21oo ro elevo 1001O'N, 83°45'W; MVZ 207139, 0.6 km W Varablanca junction on Hwy. 120,Prov. Alajuela,Costa Rica,1925 m elevo 1001O'N,83°45'W. 21. O. unifonnis (CAP) (n=2): MVZ194871-72, Río Turrialba,5.1 km NE Capelladas,Prov. Cartago,Costa Rica, . 1640 m elev.,9°47'N, 83°46'W. 22. O. unifonnis (MON) (n=5): MVZ 207134,207136-37, Peñas Blancas Trail below (E) continental divide, Monteverde Reserve,Prov. Alajuela,Costa Rica,1535 m elev.,10019'N, 84°48'W; MVZ 207138,Trail from Peñas Blancas Rd. to Ventana,Monteverde Reserve,Prov. Alajuela,Costa Rica, 1560 m elev.; MVZ 207140,Pantanosa Trail, Monteverde Reserve,Prov. Alajuela,Costa Rica,1590 m elev.,10019'N, 84°48'W. 23. O. unifonnis(NUB) (n=lO): MVZ 190837-46,Las Nubes de Coronado,0-2 km NE RíoCascajal, Prov. Cartago, Costa Rica,1700 m elevo 10002'N,83°56'W. 24. O. unifonnis(TAP) (n=l): MVZ 2037SO,Río Quiri, 0.3 mi NE junction Tapanti Rd. and Tausito Rd.,Prov. Cartago,Costa Rica,1300 m elev.,9°46'N, 83°47'W. 25. O. unifonnis (TUR) (n=5): MVZ 194861-65,Volcán Turrialba,5.0-5.5 km above Pastora,Prov. Cartago,Costa Rica,2150 m elev.,1000l'N, 83°45'W. Good & Wake: Mesoamerican salamander genilS Oedipina 1193

26. O. uniformis (VIN) (n=l): MVZ190893, La Cienega de Colorado,3 km ENEJuan Viñas, Prov. Cartago,Costa Rica,1035 m elev.,9°54'N, 83°43W.

Additional populations (see text): 27. O. cyclocauda(n=l): MVZ 167772,32.0 km (rd.)W Yoro on rd. to Morazan, Depto. Yoro,Honduras, 1000 m elev.,15°16'N, 87"38W. 28. O. grandis (n=5): MVZ 219593,4 uncatalogued MVZspecimens, Woods beside Río Coton below "La Casita," ca. 2 cm ESE Las Tablas,Prov. Puntarenas,Costa Rica,1880-1910 m elev.,8°56'N, 82°45W.

TABLE 2

Buffersystems and loei scored in tlleanalysis 01 relationships among tllespecies olOedipina. Buffertype abbreviations are as Jollows: A = Poulilk (pH 8.7); B = PGI Phosphate (pH 6.7); C = UOH (pH 8.2), D = Tris-Citrate II (pH 8.0), E = Tris-CitrateIII (pH 7.0).

Enzyme Enzyme Commission Locus Buffer system number

3-hydroxyacylCoA dehydrogenase 1.1.1.35 Hadh B Aconitase (2 lod) 4.2.1.3 Acon-1,2 O Adenosine deaminase 3.5.4.4 Ada E Alcohol dehydrógenase 1.1.1.1 Adh B Aspartate aminotransferase 2.6.1.1 Aat A Esterase (2 lod) 3.1.1.1 Est-1,2 C Glucose dehydrogenase 1.1.1.47 Gdh B Glycosephoshateisomerase 5.3.1.9 Gpi B,E Glocose-6-phosphate dehydrogenase 1.1.1.49 Gd O Isodtrate dehydrogenase (210d) 1.1.1.42 Icdh-1,2 O L-Iactate dehydrogenase 1.1.1.27 Ldh C Malate dehydrogenase (2 loci) 1.1.1.37 Mdh-1,2 .E Mannose-6-phosphate isomerase 5.3.1.8 Mpi B Nicotinamide adenine dinucleotide dehydrogenase 126.99.3 Nadh-dh C (reduced) Peptidase (2100) 3.4.13.9 Pep-1,2 A Phosphogluconate dehydrogenase 1.1.1.44 Pgdh O Phosphoglucomutase 2.7.5.1 Pgm B,E Pyruvate kinase 2.7.1.40 Pk B,E - Table 3. Allele distributions at the 22 lociexamined in the analysis of variationin Oedipina. Locality abbreviations are explained in Table l. '" .¡:.. aJfaroi. alt!!!.:í! ¡;�¡;lQQal.!í;!a ígnea parvipes morphotype � morphotype morpho- morpho- morphotype morpho- type type type

Locus outgroup alfawi .l!l.i:!!!:/! ¡;�clQ¡;ª!Jºª � ªlleni pª,:yips.:S �

LAS RIO COR QUE COR QUE BAR MON MOR NUB

Aat a b e e e b (0.70) a a a a e e e e (0.95) d (0.30) e (0.05) Aconl a a a a a b b e e a (0.89) d a a d (0.11) Acon2 a b a a e d d a (0.50) a b a a b d (0.50) (il <: Ada a o b b e e e (0.50) e e b b b b ¡¡; ....¡ d (0.50) >- Adh a a a a b e e d d a a a a O Estl a a b e e d e e e f a a a a tI1 I:l:l Est2 a a b e e b d d e e f a f f a Gd a/b e d a a e a (0.50) b b a a (0.78) a (0.50) b a (0.05) t"' O b (0.50) b(0.22) b (0.50) b (0.95) O G,dh a b a a a e a a a a a a a a :;:: Gpi a a a a a a b (0.75) b b (0.50) b a a a a :id e (0.25) d (0.50) O '"el Hadh a b e e d e e e e e e e e ñ Icdhl a a a a a a a a a a a a a a ¡:: Icdh2 a a a a a a a a a a a a a o Ldh a b e e d e e f f g '" h g Mdhl a b b b a a a e e a a d a Mdh2 a/b e a d d b e e f f b (0.56) b b b.(0.25) g (0.44) g (0.75) Mpi a b e e d e e e e a a a a Nadhdh a b e e e d a a e f a a g a (0.95) e (0.05) Pepl a b e d d e a a f f a (0.28) a (0.50) g b (0.65) b (0.28) b (0.50) g (0.35) Pep2 a a a a a b e e e e a a d a (0.95) e (0.05) Pgm a b e e d e e e e d d d d Pk a a a a a b b b b a a a a Table 3. Continued.

12S¡:l.lQQYnifQrmiíl unifQrmis morphotype morphotype

Locus ps¡:ugmmifQlmis gracilis pacific¡:nsis uniformis

MOR TUR VlN COR QUE ANO CAP MON NUB TAP TUR VIN

(l Aat e e e f f f f b (0.50) f f (0.90) " f (0.13) g � Q g (0.50) g (0.10) g (0.87) e Acon-l a a a e a a a a a a a a a Q. � Acon-2 a a a e a a a a (0.50) a a a a (0.37) b b (0.50) b (0.63) �'" e e b b b b b b b :or Ada b b b b (0.17) � d (0.83) a � Adh e e e a a a a a a a a a '" Estl a a a a g g a a a a a a a o a Est2 g g g h a a a a i a a a � Od a a a e e e e a (0.50) e e e e e '"::l. e (0.50) § b b b b b '" Odh a a a b b b b (0.70) b (0.90) e. >-' d (0.30) d (0.10) 3 a a a a Gpi a a a e a a a a a § Q, e e a f f a a a a a a a Hadh e "'" lcdh1 a a a a a a a a a (0.70) a a a a (0.50) b (0.30) e (0.50) g a '" Icdh2 a a a a a (0.75) a a a a a a a '" b (0.25) "a i (0.40) !:>. Ldh j .¡;j' (0.60) ;:¡. j ::, Mdh1 b b b a a a a a a a a a a Mdh2 d d d e b b b b b b b b b Mpi f f f a a a a a a a a a a Nadhdh a a a (0.87) h e e i i i i e (0.13) Pep1 b b b (0.50) b (0.20) h h (0.05) h (0.50) h (0.60) j (0.95) j (0.20) Pep2 a a a f a a a a a a a a a Pgm d (a.50) e d (0.12) e e e e (0.90) e d (0.60) e e e e e (0.50) e (0.88) f (0.10) f (0.40) �. a a a a a a a a a a ::o ..... a a a V> Table4. Nei (197 8 ) geneticdistances among 26 populatiollllof � Locality abbreviatiollllare explained inTable 1. ::o 0\

2 4 5 6 7 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 2S 26

1. Q. JIiiimi ---.- 1.15 1.01 1.01 1.18 1.97 1.98 2.15 1.990.690.620.890.700.890.890.930.690.53 0.53 0.36 0.37 0.44 0.37 0.38 0.410.48

2. Q . ilI.!JJm 1.15 ----- 0.610.611.2 9 1.97 1.98 1.97 1.990.67 0.87 0.890.67 0.78 0.790.75 1.66 1.03 1.01 0.98 0.90 1.01 1.01 1.01 0.92 0.94

3. º.�LAS 1.01 0.61----- 0.00 1.47 1.56 1.691.36 1.15 0.690.710.79 0.76 .044 0.45 0.421.44 0.800.790.77 0.74 0.890.78 0.790.84 0.94

4. Q.�RlO 1.01 0.610.0 0----- 1.47 1.56 1.691.36 1.15 0.690.71 0.790.76 0.44 0.45 0.421.44 0.800.79 0.77 0.74 0.890.78 0.79 0.84 0.94

5. Q. j¡¡m;¡¡ 1.18 1.291.47 1.47 ----- 1.67 1.821.9 6 1.98 1.04 1.111.1 3 1.07 1.36 1.47 1.421.69 1.1 6 1.13 1.10 1.04 1.05 1.13 1.13 1.12 1.20

6. Q. JIlkniCOR 1.97 1.97 1.56 1.56 1.67 ----- 0.020.61 0.74 1.27 1.25 1.811.34 1.26 1.19 1.211.9 6 2.04 1.97 1.94 1.84 2.03 1.96 1.97 1.95 2.14

7. Q.�QUE 1.98 1.98 1.691.69 1.820.0 2 ---- 0.610.83 1.34 1.26 1.691.2 8 1.36 1.29 1.311.84 2.06 1.98 1.95 1.93 2.05 1.97 1.98 1.96 2.15 8. Q.�C2 2.15 1.97 1.36 1.36 1.96 0.610.61·--- 0.18 1.86 1.66 1.56 1.66 1.45 1.36 1.36 2.37 2.25 2.152.122.212.24 2.15 2.152 _272.65 � t;; 9. º.�NUS 1.991.9 9 1.15 1.15 1.980.74 0.83 0.18 --.-- 1.73 1.56 1.711.9 5 1.21 1.15 1.142.39 2.07 1.991.9 6 1.952.0 6 1.991.9 92.21 2.67 ;¡; 0.690.67 0.690.691.0 4 1.27 1.34 1.86 1.73 ____o 0.160.37 0,030.490.54 0.54 1.080.83 0.810.58 0.50 10 . Q. Il!!clzi BAR 0.55 0.61 0.620.55 0.56 g 11 . Q. MON 0.620.87 0.710.71 1.1 1 1.25 1.26 1.66 1.56 0.16 ----- 0.39 0.21 0.53 0.58 0.58 1.090.68 0.67 0.46 0.47 0.51 0.490.500.53 0.61 Il!!clzi t:xI 12 . Q . Il!!clzi MOR 0.890.89 0.79 0.79 1.1 3 1.81 1.691.56 1.710.37 0.39----- 0.320.73 0.790.75 1.26 0.91 0.890.66 0.690.63 0.690.69 0.74 0.83 13 . Q.l2Qm.iNUB 0.700.67 0.76 0.76 1.07 1.34 1.28 1.66 1.95 0.03 0.210.32 ----- 0.56 0.620.62 1.08 0.85 0.84 0.600.55 0.58 0.64 0.65 0.57 0.59 §

14 . Q.l'seudounifQrmis MOR 0.890.78 0.44 0.44 1.36 1.26 1.36 1.45 1.21 0.490.53 0.73 0.56 ---'- 0.01 0.02 1.43 J.l6 1.13 0.73 0.73 0.800.77 0.78 0.83 0.93 S; 15 . Q. pseudouniformis TUR 0.890.790.45 0.45 1.47 1.19 1.291.36 1.150.54 0.58 0.790.620.01----- 0.01 1.44 1.17 1.150.74 0.74 0.84 0.78 0.790.84 0.94 � 16 . Q. pseudounifQrrnis VIN 0.930.75 0.420.421.42 1.2 1 1.311.36 1.140.54 0.58 0.75 0.620.020.01 ----- 1.42 1.13 1.100.690.72 0.76 0.75 0.76 0.810.9 2

17 . º.� 0.691.66 1.44 1.44 1.69 1.96 1.84 2.37 2.391.08 1.091.26 1.08 1.43 1.44 1.42--.- 0.700.690.510.58 0.63 0.510.510.56 0.62 �

18 . Q . � COR 0.53 1.03 0.800.80 I.!G 2.04 2.06 2.25 2.07 0.83 0.68 0.91 0.85 1.16 1.I7 1.130.70··--- 0.000.320.41 0.400.320.320.41 0.48

19 . Q . �QUE 0.53 1.010.790.791.13 1.97 1.98 2.15 1.990.810.67 0.890.84 1.13 1.151.100.690.00----- 0.310.40 OAO0.32 0.3 20.40 0.48

20 . Q _ l!!I.i.f.2rnlj§ ANG 0.36 0.980.77 0.77 1.10 1.94 1.952.121.960.580.46 0.66 0.600,73 0.74 0.690.510.320.31 ----- 0.080.11 0.020.09 0.08 0.14

21. Q.llIlifllnl:lL1.CAP 0.37 0.900.74 0_74 1.04 1.84 1.93 2.21 1.950.500.47 0.690.55 0.73 0_74 0.720.58 0.410.400.0 8 ----- 0.23 0.04 0.07 0.01 0.04

22 . Q . � MON 0.44 1.100.890.891.052.0 3 2.05 2.242.060.55 0.510.63 0.58 0.800.84 0.76 0.63 0.4000400.11 0.23 ----- 0.16 0.220.23 0.29

23 _ Q . !JIlÍÍllI1!li:; NUB 0.37 1.01 0.78 0.78 1.13 1.96 1.97 2.15 1.990.610.490.690.64 0.77 0.78 0.75 0.510.320.320.0 20.040.16 ----- 0.08 0.05 O.JO

24 . Q.lIIli.fQonis.TAP 0.38 1.01 0.790.79 1.13 1.97 1.98 2.15 1.990.620.500.690.65 0.78 0.790.76 0.51 0.320.320.090.310.2 20.08 ----- 0.07 0.11

25 . Q.ll!lÍ!2llIli!TUR 0.410.9 2 0.84 0.84 1.12 1.95 1.96 2.27 2.21 0.55 0.53 0.74 0.57 0.83 0.84 0.8l 0.56 0.410.400.0 8 0.01 0.23 0.05 0.07 -.--- 0.02

26 . Q.�VIN· 0.48 0.94 0.94 0.94 1.202.14 2.15 2.65 2.67 0.56 0.610.83 0.590.9 3 0.94 0.920.62 0.48 0.48 0.14 0.04 0.290.1 0 0.11 0.02-----

1198 REVISTA DE BIOLOGIA TROPICAL

Fig. 1. Species of Oedipina displaying different morphotypes. A. The alfaroi morphotype: O. alfaroi. The remaining specimensrepresent theuniformis morphotype: B. O. gracilis; C. O. uniformis; D. O. pacijicensis. The scale is 2 cm long. Oedipina Good &Wake: Mesoarnerican salarnander genus 1199 1200 REVISTA DE BIOLOGlA TROPICAL

Fig.2. Species of Oedípina displaying different morphotypes. A. The ígnea morphotype: O. ígnea. B. The parvipes morphotype: O. anení. C. The poelzi morphotype: O. poelzi. D. The altura morphotype: O. altura. E. The pseudouniformísmorphotype: O. pseudouniformi5. F. The cyclocauda morphotype: O. cyclocauda. The scale is 2 cm long. Good &Wake: Mesoamerican salamander genus Oedipina 1201 1202 REVISTA DE BIOLOGIA TROPICAL

0.7 O. A • O O D"

o" o 3 • • .. O� .. .. O U I .. � A e "O • � • ti A ti A A ¡ '6 ¡u u � • :1:1.. 8. lA A a. A e 0.1 A • .. ti' • O) :,; 0.2 .. • lO 0.1 •• .1 .. l . . Iib ,,'o 1� . .ao '.ro Geographlcdlstance (km) Geographic dilltl\nce(km) . .7'-['-----;.;--:======--, Fig. 4. The relatíonship of Nei (1978) genetíc distance to U. IIIM101"..fIIHIII. U. . geographic distance in the poelzi morphotype. Closed .. circles represent comparisons among populatíons of O . poe/zi exclusive of the Moravia population, and open circles represent comparisons of the Moravia population withall remainingpopulatíons of O. poelzi. O• .,,_y .. O. _,.

O. _'_.... O.IIfI/(onnI •

...

0.1

Wlograpnlcdl81ance (km)

Fig.3 . . Therelatíonship of Nei (1978) genetíc distance to geographic distance in the unifonnis morphotype. In upper graph closed circles represent comparisons within O. unformis, the open circle represents the comparison of O. alfaroiand O. graci/is, triangles represent comparisons between O. pacijicensis and O. uniformis, closed diamonds represent comparisons between O. alfaroi and O. unifonnis, open diamonds represent comparisons between O. gracilis and O. uniformis, closed squared represen! comparisons between O. alfaroi and O. pacijicensis. and open squares represent comparisons between O. gracilis andO. pacificensis

Fig. 5. The two moS! parsimonious phylogenetíc hypotheses of the 11 species of Oedipina for which all loci were scored. Consistency index (CI) = 0.974, rescaled consistency index (RC) = 0.925, retentíon index (RI) = 0.950.

1204 REVISTA DE BIOLOGIA TROPICAL

The Moravia de Chirripo region also Fig.s Costa Rica. There is substantial genetic in the case of O. poelzi, in which genetic difference between fuis sample and nearby, differentiation from populations in fue higher elevation samples(20, 22). The genetic Cordillera Central is of similar magnitude to distance is 0.51, and there are seven fixed alIele fuat found in the uniformis dade. However, differences separating fue samples. In contrast, morphological differentiation is not as great as the greatest genetic distances in montane O. in fue case of O. alfaroi. The population from uniformis. between geographically distant theregion of Moravia de Cbirripo may warrant samples (22, 26): is less than 0.3. We assign species status, but we refrain from describing population 17 to O. gracilis Taylor 1952, the species at tbis time (see aboye). which Brame (1968) synonymized with O. Thesítuation in Nototriton also is relevant uniformis. This lowland species differs from O. to fue discussion of the Moravia de Chirripo uniformis by its substantially smaller size and isolate. Nototriton is more ecologically more attenuate fonu, but in no other specialized and less vagile fuan Oedipina. In the morphological details that we can discem(see case of fue N. abscondens complex, the sample also Brame, p. 39-40). The coinbination of from fue Moravia de Cbirripo region is genetic and size differences together with morphologicallydistinct and was described as a differences in distribution and ecology (one in new species, N. major, by Good and Wake fue hot lowlands and the other in the cool (1993). Anofuer instructive comparison uplands in geographically adjoining areas) between Nototriton and Oedipina involves support our decision to recognize O. gracilis. samples from the Cordillera Central and from The type locality for O. gracilis is near the valley of the Río Orosi, at fue extreme Guapiles, a locality relatively near to La Selva, northem end of fue Cordillera de Talamanca, at a similar elevation and with similar Costa Rica. In this case O. uniformis samples ecological conditions at the time of capture in fue two areas are only slightly differentiated. (a1though the locality has changed dramatically The genetic distance between relatively nearby in the past 40 years). Based on this result, we samples from the Cordillera de Talamanca assign aU populations of fue uniformis clade (sample 24) and fue Cordillera Central(sample from the Atlantic lowlands of Costa Rica 23) is only 0.084. In contrast, fuere are two (except O. alfaroi) to O. gracilis. Curiously, instances of species-level differences in fue uniformis dade has not yet been recorded in Nototriton between the same localities, and the Atlantic lowlands of Nicaragua or Panama. genetic distances are 0.30 and 0.50 (Good and Sympatry is unknown among species in the Wake 1993). These examples suggest that past dade, but it is to be expected between O. vicariant events have had varying effects on uniformis and O. gracilis along fue northem different táxones, depending on fueir vagility slopes of the Cordillera Central in fue anddispersal propensity. In fue case of Moravia Provincia de Heredia, Costa Rica. de Chirripo, Nototriton major and O. alfaroi The final case involving the uniformis clearly have diverged from fueir "stem" group relates to two samples (l8, 19) that also lineages, wbile such divergence is less apparent occur in lowland sites, but in fue Pacific in O. poelzi. tbis contrast between táxones is lowlands rather fuan the Atlantic lowlands. even more apparent when comparing fue Thesesamples aredifferentiated from upland O. Cordillera Central and Talamancan populations uniformis by genetic distances of 0.32-0.48 of Nototriton and O. uniformis. (four fIxed differences) and fromO. gracilis by We identified two additional cases of distances of 0.69-0.70 (11 fixed differences). allopatric differentiation relating to the Because of the combination of substantial uniformis group. One involves sample 17, genetic distance, geographic distinctiveness, and from the Organization for Tropical Studies' La (re1ative to O. uniformis) differences in Selva field station in the Atlantic lowlands of elevational distribution and ecology, we assign Good &. Wake: Mesoamerican salamander genus Oedipina 1205 fuese Pacific lowland populatiolls to O. slopes of fue Cordillera de Talamanca, four pacificensis Taylor 1952, a taxon fuat was other uniformis group species, O. collarL�, O. reduced lo synonoymy with O. uniformis by poelzi, O. pseudouniformis, and O. uniformis, Brame(19 68). The type locality is near San occur in general proximity to each other, but Isidro del General, at 646 m elevarion, and the al! are now uncommon (O. collaris is known species may extend upward to about 1000 m. in only from a single specimen in this area). In this respect the distribution is similar to that of the parvipes group, O. parvipes and O. O. alfaroi of the Atlantic slope. The species compex, both uncommon, occur together on characteristically has whitish to tannish Barró Colorado Island,. Panama. Species in the postocular stripes on fue head, and the head parvipes group occur in sympaL.]' with other occasional1yappears almost wrute. Trus feature species of Oedipina in both the Atlantic and is not unique to fue taxon, however, and the Pacific lowlands of Costa Rica (O. carablmu.u color pattern may at times be obscure. We are wÍth O. cyclocauda and O. gracilis at Guapiles. unaware of any other morphologically and O. alleni with O. pacificensis from the distinctive features. fue species apparently is vicinity of Parrita to fue Osa Peninsula on the somewhat larger than O. gracilis but noí as Pacific coast). large as O. uniformis (see also Brame 1968, p. Speeimens of O. cyclocauda and O. 40). We assign aU populations of the uniformis pseudouniformis are difficult to separate on group from fue Pacific lowlands a..'1d adjacent morphological grounds, although O. cyclocauda middle-elevation areas (as in fue area around has slightly shorter legs, and the two have not San Isidro del General) of Costa Rica and been taken in sympatry, so we have sorne Pan amato O. pacificensis; question conceming correcí alloeation of In contrast to O. uniformis, we find no preserved specimens of these genetical1y very reasons for separating fue populations of O. distinctmon es. pseudouniformis from each other. Our samples Despite the relatively weak morphological were takenfrom relatively nearby areas, and we differentiation of the many species in Brame's have no evidence of range disruption, as we do (1968) unifomtis group (induding our for the O. uniformís complexo Furthermore, O. cyclocauda, poelzi, and uniformis dades), our pseudouniformis appears to have a broader demonstration of deep genetic divergence distribution than do either of fue other two suggests that it is an oId lineage ihat has bren táxones, perhaps relating to higher vagility. differentiated formillions of years. Members of No more than two species in either the fue uniformis group are widespread in the parvipes or unifo rmis group (the latter uplands and lowlands, especiaUy in southem includingfue cyclocauda, poelzi, and uniformis Mesoamerica, and they can be loeally abundant. dades) have been taken in sympatry, although Aecordingly, severa! species in the group are Brame (1968) Usted O. alfaroi, O. cyclocauda, relatively wel1 known. 00 the other band, O. gracilis, and O. pseudouniformis as allimals in Brame'sparvipes group (identieal to occurring in close proximity. These species our parvipes dade) are restrieted to relatively occur in various combinations of pairwise low elevations and are diffieult to colleet sympatry in Limón Province in the eastem (Bramehad fe wer than100 preserved specimens Iowlands of Costa Rica (see also Wake 1987, of all species for rus revision), and as a whole Fig. 15). Our results suggest that O. gracilis trus group is relatively poorly known. Brame and O. alfaroi are the most genetically similar diseussed the possibility of raising his parvipes species of Oedipina thatare sympatric. Hábitats and uniformis groups to fue rank of genera, but at low elevation in eastern Costa Rica have decided against it because he wanted to been changed severely by deforestation, and any emphasize the close relationship between them. salamanders now are difficult to fm d there. At Our results corroborate a sister group somewhat higher elevations along the Atlantic re1ationship of the two groups. 1206 REVISTADE BIOLOGlA TROPICAL

We analyzed only two species of Brame's disclosed that they have unusual meiosis and parvipes group, and thus cannot elucidate strongly heteromorphic XIY sex chromosomes phylogenetic relationships within that group. (Kezer et al. 1989). They share this Within Brame's uniformis group, however, we heteromorphic system with Nototriton, and to had ten of the 16 species available for analysis. some degree with Thorius and Dendrotriton Brameprovided theonly previous discussion of (Sessions and Kezer 1991), but cytological relatíonships among the species in his studies of theparvipes group have not yet been uniformis group, and his hypothesis is undertaken. 1t is critical that cytological illustrated in Fig. 7. His results are similar to information be obtaiped, because at present the ours in that, of the species available to us, he XIY chromosome system constitutes an grouped O. uniformis(including O. gracilis and important synapomorphy that may be O. pacificensis), O. alfaroi, and O. ígnea into a significant in sorting out the patterns of monophyletic unit. This corresponds exactly to phylogenetíc relationships in the supergenus our uniformis clade. His hypothesis differs Bolitoglossa. from ours, however, in two ways. He did not Members of the genus Oedipina are, with recognize our cyclocauda dade, but ínstead some Bolitoglossa, the most "tropical" of suggested that O. cyclocauda, O. salarnanders. Only members of the parvipes pseudouniformis,and O. altura form sequential group in Oedipina and species in the alpha outgroups to the uniformis dade. He also group of Bolitoglossa(Wake and Lynch 1976), suggested that O. poelzi is more distantly of all salamanders, reach South AmerÍCa. The related to the uniformis clade than are the majority of species of Oedipina occur in what members of our cyclocauda dade. This conflicts Wake and Lynch referred to as Talamancan with our hypothesis that O. poelzi is the sister Central America, and only one species (O. taxon to the uniformis clade. Brame's evidence elongata) extends northward as fa r as MexÍCo. for the "stair-step" pattern of relationships of AH species of the parvipes group occur in the O. poelzi, O. altura, O. pseudouniformis, O. tropical lowlands (areas below 500 m cyclocauda, andthe uniformis clade was that elevation), but at least some of these species eachrepresents a sequential step in an apparent occur as high as 1250 m. The species of the morphocline from the long-limbed, broad­ parvípes group have a number of extreme footed, broad-headedparvipes group species to specializations, notably in head shape and in the short-limbed, narrow-footed, narrow-headed the shape of the hands andfeet. Species of the uniformis clade species. Oedipina poelzi does, unifo rmis group also occur in tropical in fact, approach the parvipes group more lowlands, from Guatemala at least to western dosely than the otherspecies in leg length and Panama. Presumably, it is the successful foot width, but Brame's Fig. 25 (p. 53) does invasion of the tropical lowlands that has not corroborate his statement about facilitated dispersal of salamanders into South intennediacy in head width. In this character, America and into Nuclear Central America. the cyclocauda clade appears to mos� dosely Some species of the uniformis group occur approach the parvipes group, while O. p'oelzi exclusively in the uplands, where they live at and theunif ormisclade have become derived in least as high as 2300 m. opposite directions (wider heads in O. poelzi and narrower in the uniformis clade). Our phylogeny would require either a reversa! in O. ACKNOWLEDGMENTS poelzi to more robust límbs or a parallel reduction of the limbs in the cyclocauda and Manyindíviduals have assistedin collecting uniformisclades. specimens used in the present study, and we Cytological studíes of two species of especially appreciate the help of F. Bolaños, J. Oedipina (O. poelzi and O. uniformis) have Cadle, D. Cannatella, K. de Queiroz, R. GoocI &; Wake:Mesoamerican salamander genus Oedipina 1207

Gamez, A. Graybeal, T. Jackman, P. León; K. Brame, A. H., Jr. 1963. A new Costa Rican salamander Lips, F. Muñoz, G. Roth, S. Sessions, N. (genus Oedipina) with a re-examination of O. collarls Staub, and T. Wake.We especially thank Pedro and O. serpens. Contrib. Sci. Nat. Hist. Mus. Los Angeles Co. 65: 3-12. León and bis colleagues in the Center for Cellular and Molecular Biology, University of Brame, A. H., Jr. 1968. Systematics and evolution of the Mesoamerican salamander genus Oedipina. J. Costa Rica, and Federico Bolaños of the Herpetol. 2: 1-64. Museum of Zoology of the University of Costa Brame, A. H., Jr. , &; W. E. Duellman. 1970. A new Rica, for much assistance, cooperation and salamander (genus Oedipina) of the uniformis group friendship during our woric in C�sta Rica. We from westem Panama. Contrib. Sci. Nat. Hist. Mus. Los Angeles Ce.20h-l-8. thank the governments oí CoSta Rica, Guatemala, Honduras, and Panama for Brodie, E. D., k, &; J. A. Campbell. 1993. A new salamander of the genus Oedipina (Caudata: perqrission to study and collect salamanders in Plethodontidae) from the Pacific versant of Guatemala. Herpetologica 49: 259-265. their countries. Aspects of this research were supported by the National Science Foundation Buth, D. G. 1984. The application of electrophoretic data (NSF BSR 90-19810), the National Geograpbic in systematic studies. Ann. Rev. Ecol. Syst. 15: 501- 522. Society (Zona ProtectoraLa Selva Expedition), and the Museum ofVertebrate Zoology. Dunn, E. R. 1921. Two new Central American salamanders. Proc. Biol. Soco Washington 34: 143-146.

Frost, D. R., &; D. M. Hillis. 1990. Species in concept and RESUMEN practice: applications in herpetology. Herpetologica 46: 87-104. Se analizó la variación proteica entre muestras de 28 poblaciones de salamandras del género Oedipino, de Good, D. A., &; D. B. Wake. 1992. Geographic variation Guatemala, Honduras, Costa Rica, y Panama, para and speciation in the torrent salamanders of the genus evaluar la validez de las especies actualmente Rhyacotrlton (Caudata: Rhyacotritonidae). Univ. reconocidas así como sus relaciones IDogenéticas. Se usó CaliforniaPubl. Zool. 126: 1-91. como grupo de referencia ocho especies del género emparentado Nototriton . Hay cuatro clades princiipales Good, D. A., &; D. B. Wake. 1993. Systematic studies of con distancias genéticas .de Nei de 0.46 a 2.67. Los the diminutive salamanders of the genus Nototriton . patrones de variación geográfica dentro de la O. from Costa Rica, with descriptions of three new uniformis as actualmente reconocida indican que en species. Herp. Monogr. 7: 131-159. realidad contiene poblaciones que deberían ubicarse en cuatro especies: O. a1faroi es una especie válida, y O. Harris, H., &; D. A. Hopkinson. 1976. Handbook of gracilisy O. pacificensis se reviven de la sinomnimia de Enzyme Electrophoresis and Human Genetics. North­ O. uniformis, mientras O. bonitaensis, O. inusitata, O. Holland Publ. Co., Amsterdam. Unpaginated. longissima, y O. syndactyla continóan en sinonimia. La Hillis, D. M., &; C. Moritz. 1990. Molecular Systematics. variación geográfica también es aparente en O. Sinauer Assoc., Sunderland, MA. 588 p. cyclocauda, O. poelzi, y O. pseudouniformis, pero no se justifican cambios taxonómicos en este momento. Se Kezer, J., S. K. Sessions, &; P. Leon. 1989. The meiotic apoya el estado específico de O. alturay O. grandis . Las structure and behavior of the strongly heteromorphic cuatro muestras que representan al grupo parvipes group X/Y sex chromosomes of neotropical plethodontid son las más lejanamente relacionadas con el resto de las salamanders of the genus Oedipina. Chromosoma 98: muestras, y la variación genética dentro del grupo es 433-442. suficiente para justificar la reinstalación de las O. alleni costarricenses de la sinonimia con O. parvipes de Panamá Larson, A. 1984. Neontological inferences of y Colombia Todas las muestras morfológicamente evolutionarypattem and process in the salamanders of diferenciadas también son diferentes genéticamente. the famüy Plethodontidae. Evol. Biol. 17: 119-217. Nuestros resultados indican que se muestrearon 12 especies (O. alfaroi, O. alIeni, O. altura, O. cyclocauda, Mabee, P. M., &; J. Humphries. 1993. Coding polymorphic O. gracilis, O. grandis, O. ignea, O. pacificensis, O. data: examples from allozymes and ontogeny. Syst. parvipes, O. poelzi, O. pseudouniformis, y O. uniformis). Biol. 42: 166-181. no dispusimos de muestras de O. carablanca, O. collaris, O. complex, O. elongata, O. gephyra, O. paucidentata, O. McCranie, J. R., L. D. Wilson, &; K. L. Williams. 1993. A stenopodia, O. stuarti, u O. taylorl, pero todas son new species of Oedipina (Amphibia: Caudata: reconocibles como especies diferentes segtln la Plethodontidae) from northern Honduras. Proc. Biol. morfolog{a. Soco Washington 106: 385-389.

Murphy, R. W. 1993. The phylogenetic anaIysis of REFERENCES allozyme data: invalidity of coding eles all by presence/absence and recommended procedures. Biochem. Syst. Ecol. 21: 25-38. Brame, A. H., Jr. 1960. The salamander genus Oedipina in northem Central America. Bull. So. California Acad. Sci. 59: 53-162. 1208 REVISTA DE BIOLOGlA TROPICAL

Nei, M. 1972. Genetic distance between populations. Am. Taylor, E. H. 1948. New Costa Rican salamanders. Proc. Natur. 106: 283-292. Biol. Soco Washington 61: 177-1 80.

Nei, M. 1978. Estirnation of average heterozygosity and Taylor, E. H. 1949. New salamanders from Costa Rica. genetic distance from a small number of individuals. Univ. Kansas Sci. Bull. 33: 279-288. Genetics 89: 583-590. Taylor, E. H. 1952. The salamanders and caecilians of Richardson, B. J., P. R. Baverstock, & M. Adams. 1986. Costa Rica. Univ. Kansas Sci. BuI!. 34, pt. 2: 695-791. Allozyme E1ectrophoresis: A Handbook for Systematics and Population Structure. Academic Press, Tay!or, E. H. 1954. Additions to the known herpetological Sydney. 410 p. fauna of Costa Rica with comments on other species. No. !. Univ. Kansas Sci. Bull. 36, pt. 1: 597-639. Selander, R. K., M. H. Smith, S. Y. Yang, W. E. Johnson, & 1. B. Gentry. 1971. Biochemical polymorphism and Wake,D. B. 1987. Adaptive radiation of salamanders in systematics in the genus Peromyscus. 1. Variation in Middle American cloud forests. Ann. Missouri Bot. the old_field mouse (Peromyscus polionotus). Stud. Gard. 74: 242-264. Genet. VI. Univ. Texas Pub!. 7103: 49-90. Wake, D. B., & P. Elias. 1983. New genera and a new Sessions, S. K., & J. Kezer. 1991. Evolutionary species of Central American salamanders, with a cytogenetics of bolitoglossine salamanders (fami1y review of the tropical genera (Amphibia, Caudata, Plethodontidae). P. 89-130. In D. M. Green and S. K. Plethodontidae). Contrib. Sci. Nat. Hist. Mus. Los Sessions (eds.). Cytogenetics and Angeles Co. 345: 1-19. Evolution. Academic Press, Nueva York. Wake, D. B., & 1. F. Lynch. 1976. The distribution, Swofford, D. L. 1985. PAUP, Phylogenetic Analysis Using ecology, and evolutionary history of plethodontid Parsimony. User's Manual. IIlinoís Natrual History salamanders in tropical America. Sci. Bull. Nat. Hist. Survey, Champaign. Unpaginated. Mus. Los Angeles Co. 25: 1-65.

Swofford, D. L., & R. K. Selander. 1981. BIOSYS-1: a Wake, D. B., T. J. Papenfuss, & J. F. Lynch. 1992. FORTRAN program for the comprehensive analysis of Distribution of salamanders along elevational transects electrophoretic data in population genetics and in Mexico and Guatemala. Tulane Stud. Zool. Bot., systematics. J. Hered. 72: 281-283. Suppl. Pub!. 1: 303-319.