HERPETOLOGICA VOL. 38 JUNE 1982 NO. 2 Herpetologica, 38(2), 1982, 257-272 ? 1982 by The Herpetologists' League, Inc. EVOLUTIONARY RELATIONSHIPS AMONG CENTRAL AMERICAN SALAMANDERS OF THE BOLITOGLOSSA FRANKLINI GROUP, WITH A DESCRIPTION OF A NEW SPECIES FROM GUATEMALA DAVID B. WAKE AND JAMES F. LYNCH ABSTRACT: Bolitoglossa meliana, a new species from the mountains of central Guatemala, is described and assigned to the B. franklini group. Electrophoretic comparisons with the other species in the group confirm the distinctiveness of B. meliana, but cause us to question the validity of B. brevipes and B. nigroflavescens, which are hereby synonymized with B. franklini. Plio-Pleistocene tectonic and climatic events may have led to secondary contact and hybridiza- tion between previously isolated inland and Pacific coastal populations, thereby complicating patterns of genie similarity. Key words: Amphibia; Caudata; Bolitoglossa; Electrophoresis; Proteins; Biogeography NUCLEAR Central America, the tecton- ically and topographically complex re- gion that lies between the Isthmus of Te- huantepec and the Nicaraguan depression (Schuchert, 1935), is the major center for species diversity in tropical plethodontid salamanders (Wake and Lynch, 1976). The present paper considers evolution- ary relationships within the Bolitoglossa franklini species-group (Stuart, 1943) of the highlands of Guatemala and adja- cent Chiapas, Mexico. This complex also has been called the "Uncolni subgroup" of the Bolitoglossa rostrata species- group (Wake and Lynch, 1976), but Stuart's name has priority. We first de- scribe a newly discovered species, and then examine electrophoretic data and selected morphological information for all species in the group. Finally, we pre- sent an evolutionary scenario that is con- sistent with available data. MATERIALS AND METHODS Electrophoretic Analysis Samples used for electrophoretic anal- ysis (Table 1) were obtained from three populations representing the new species of Bolitoglossa and six populations pre- viously referred to B. brevipes (1), B. franklini (2), B. nigroflavescens (1), B. Uncolni (1) and B. resplendens (1). Sam- ple sizes per population ranged from 1 to 14 (Table 2 below). Use of small samples in interspecific comparisons has been as- sessed by Gorman and Renzi (1979). All animals were brought to the labo- ratory alive. Freshly sacrificed specimens were dissected and samples of liver, kid- ney, spleen, heart and intestine were ex- tracted, frozen, and stored at ?76 C for future use. Combined tissue extracts were subjected to horizontal starch-gel electrophoresis (Selander et al., 1971). 257 258 HERPETOLOGICA [Vol. 38, No. 2 TABLE 1.?Collecting localities for samples used in electrophoretic analysis. Numbers refer to popula- tions in Fig. 2. 1. Bolitoglossa meliana. Santa Rosa Pass, 9 km NE Santa Cruz del Quiche, El Quiche, Guatemala (2520 m). 2. B. meliana. La Bella crest, 20 km NNW San Au- gustin Acasaguastlan, El Progreso, Guatemala (2725 m). 3. B. meliana. 4-6 km (by road) S Purulha, Baja Verapaz, Guatemala (1650-1800 m). 4. B.franklini nigroftavescens (previously B. brev- ipes). Sierra Madre, 6.7 km (by road) W Motozin- tla, Chiapas, Mexico (1780-1860 m). 5. B. franklini franklini. Above Colonia Talquian, Volcan Tacana, Chiapas, Mexico (2400-2510 m). 6. B. franklini franklini. Volcan Chicabal, 5 km SE San Martin Sacatepequez, Quezaltenango, Gua- temala (2250 m). 7. B. franklini nigroftavescens. Cerro Ovando, Chiapas, Mexico (1650-1680 m). 8. B. lincolni. Sierra de los Cuchumatanes, 3.5 km (air line) NNW Uspantan, El Quiche, Guatemala (2260-2640 m). 9. B. resplendens. Montanas de Cuilco, 1.5 km NE Pena Blanca, Huehuetenango, Guatemala (2800 m). The following gel buffer systems were used: 1.?Tris Maleate EDTA, pH 7.4 (0.01 M Tris/0.1 M maleic acid/0.01 M EDTA) for malate dehydrogenase (Mdh; 2 loci) and malic enzyme (Me); 2.?Tris Citrate II, pH 8.0 (0.687 M Tris/0.15 M citric acid) for glutamate oxalate trans- aminases (Got; 2 loci), sorbitol dehydro- genase (Sod), mannose phosphate isom- erase (Mpi) and 6-phosphogluconate dehydrogenase (Pgd), for which 1.0 cc NADP was added to 240 cc gel buffer; 3.?Tris Citrate III, pH 7.0 (0.135 M Tris/ 0.043 M citric acid) for glucophosphate isomerase (Gpi), phosphoglucomutase (Pgm), alpha-glycero-phosphate dehy- drogenase (a Gpd) and isocitrate dehy- drogenase (led; 2 loci); 4.?Poulik, pH 8.7 (gel: 0.076 M Tris/0.005 M citric acid; trays: 0.30 M boric acid, pH 8.2) for lac- tate dehydrogenase (Ldh; 2 loci), leucine aminopeptidase (Lap), and leucyl ala- nine peptidase (Pep) (Ayala et al., 1972; Selander et al., 1971). Genetic interpre- tations of allozymic data are based on cri- teria elaborated by Selander et al. (1971). Alleles are designated by letter, with "a" being the fastest migrant. Morphological Data Most meristic and proportional char- acters used in salamander systematics vary with overall size or sex, or both. With sufficiently large samples, regres- sion methods can be used to adjust mea- surements for body size and sex (e.g., Lynch, 1981; Lynch and Wake, 1975, 1978), but in the present study this pro- cedure was possible only for B.franklini, B. resplendens and the new species. For* the remaining forms, only general com- ments on color pattern and external mor- phology are offered. External measurements and tooth counts are based on examination of ani- mals that had been fixed in dilute form- aldehyde and stored in 70% ethanol. Os- teological data were obtained from radiographs and from specimens that had been cleared in KOH and stained with Alizarin Red. Except as noted below, all specimens are deposited in the collection of the Museum of Vertebrate Zoology (MVZ), Berkeley, California. DESCRIPTION OF NEW SPECIES In 1972, we collected specimens of an unknown species of Bolitoglossa in a newly accessible region of Baja Verapaz, Guatemala. Subsequent collecting has revealed this large, all black species to be widespread in the Chuacus-Minas mountain system of central Guatemala. In allusion to its characteristic pigmen- tation, this species shall be known as: Bolitoglossa meliana sp. nov. Holotype.?MVZ 160736. An adult fe- male from the vicinity of Santa Rosa Pass, 9 km NE Santa Cruz del Quiche, El Quiche, Guatemala (elevation 2520 m), collected 16 July 1978 by P. Elias, E. J. Koford, D. B. Wake, and T. A. Wake (Fig. 1). Paratypes.?MVZ 160361-72, 160385- 97, 160373-84,160398-99,160737-71(74 June 1982] HERPETOLOGICA 259 FIG. 1.?An adult specimen of Bolitoglossa meliana from Santa Rosa Pass, El Quiche, Guatemala. Scale is 25 mm. specimens), all collected at the type lo- cality; MVZ 108854, 113160-61, 150519, 169069 (5 specimens), 4-6 km (by road) S Purulha, Baja Verapaz, Guatemala (elev. 1650-1800 m); MVZ 150804-07, 150813 (5 specimens), San Antonio, 8 km (by road) ESE Chilasco, Baja Verapaz, Guatemala (elev. 1850 m); MVZ 150789, 150812 (2 specimens), Finca Planada, 15 km NNE Rio Hondo, Zacapa, Guatemala (elev. 1700 m); MVZ 150808-11, 160772 (5 specimens), Sierra de las Minas, 12 km N Santa Cruz, Zacapa, Guatemala (elev. 2200 m); MVZ 169038-39 (2 specimens), La Bella crest, 20 km NNW San Augustin Acasaguastlan, El Progreso, Guatemala (elev. 2725 m); KU 18613839 (2 speci- mens), El Volcancito, 3.5 km SE Purulha, Baja Verapaz, Guatemala. Diagnosis.?Distinguished from all other Bolitoglossa of Mexico, Guatemala, Honduras and El Salvador by its combi- nation of uniform black coloration, large adult size (males to 66 mm standard length [SL], females to 88 mm SE), and truncated tips of toes with terminal pha- langes free of webbing (Fig. 1). Resem- bles other species of the franklini com- plex in body proportions, overall size, and in possession of slightly webbed "Magnadigita"-type (Taylor, 1944) hands and feet; however, other members of the franklini complex are marked with bright patches of white, silvery, brassy, or red- orange pigment. In addition, B. meliana is strongly differentiated genetically from the other species (see below). The only other Guatemalan salaman- der with which B. meliana might possi- bly be confused is the sympatric B. mo- rio, a much smaller, stouter species with shorter tail and limbs, and somewhat more webbing of hands and feet. As op- posed to the featureless black pigmenta- tion of B. meliana, B, morio has white flecks on the venter and irregular cream or pink spots dorsolaterally. Variation.?In B. meliana and other 260 HERPETOLOGICA [Vol. 38, No. 2 members of the franklini complex, on- togenetic and sexual differences exceed interspecific variation for most external features except coloration. The following comments are based on measurements of 17 juveniles (SL < 41 mm), ten subadults (SL 41-49 mm), 17 adult and near-adult males (SL > 49 mm) and 18 adult and near-adult females (SL > 49 mm). Linear regression methods were used to analyze bivariate trends in variation separately for juveniles, near-adult plus adult males, and near-adult plus adult females. For purposes of comparison, the value of each morphological trait was projected to a common mean value (CMV) at an SL of 30 mm for juveniles and 60 mm for near- adults and adults (Lynch and Wake, 1975, 1978; Lynch, 1981). Males and females of B. meliana differ in minimum size at sexual maturity (males ca. 55 mm, females ca. 65 mm), maximum size (males 66 mm SL, females 80 mm SL), relative tail length in adults (0.84-1.10 SL [CMV = 0.94 SL] in males >49 mm SL; 0.79-0.98 SL [CMV = 0.83 SL] in females >49 mm SL), and rel- ative limb length in adults (hind limb length 0.22-0.28 SL [CMV = 0.26 SL] in males; 0.22-0.25 SL [CMV = 0.24 SL] in females). There is no statistically signif- icant sexual variation in relative head size or in numbers of maxillary, premax- illary, or vomerine teeth. Relative to juveniles, adult and near- adult individuals (SL > 49 mm) have pro- portionately longer tails (tail length H- SL = 0.79-1.10 [CMV = 0.89] in adults and near-adults; 0.38-0.69 [CMV = 0.58] in juveniles), and more maxillary-pre- maxillary teeth (43-76 [CMV = 60] ver- sus 11-62 [CMV = 43] in juveniles and subadults). Relative limb length and head width are similar in juveniles and adults. Comparative osteology.?Groups within the large genus Bolitoglossa are not well differentiated in osteology. The only character that segregates substantial numbers of species is the complex tail base that distinguishes the "beta" group of the genus from the "alpha" group (Wake and Brame, 1969; Wake and Dres- ner, 1967; Wake and Lynch, 1976). The following observations are derived from x-rays of 12 B. meliana (MVZ 108854, 113160, 150789, 105804-812) and one cleared and stained individual (MVZ 160740). We have examined cleared and stained specimens of B. f franklini, B. f. nigroflavescens, B. re- splendens and other less closely related Bolitoglossa but have detected no skel- etal features that distinguish B. meliana from the other members of the franklini group. The skeleton of B. meliana is well de- veloped and, apart from the complex tail base, is about as generalized as is found in the genus. The single premaxillary bone bears slightly to moderately ex- panded frontal processes. The relatively large nasal bones are strongly protuber- ant and extend forward well in advance of the premaxillaries. The prefrontal bones, though discrete and well formed, are much smaller than the large nasals. Septomaxillaries are absent. Maxillaries, frontals, parietals, otic-occipitals, squa- mosals, quadrates, and parasphenoid bones are generalized in form. The vo- mers also are generalized and have long and relatively stout preorbital processes that bear teeth to their midpoint. The operculum has at most a very small stilus. The limbs are stout. The tibia bears a dis- tinct proximal crest in larger specimens, and it is partially free (i.e., attached only at the distal end) to form a tibial spur in the two largest specimens examined. The digits are well formed, and the terminal phalanges of the longer digits are greatly expanded terminally to a "T" or "Y" shape. Phalangeal formulae are 1-2-3-2, 1-2-3-3-2. There are two caudosacral ver- tebrae. The first caudal vertebra has very long transverse processes which are bifid in most individuals, forming the complex tail base typical of the beta group of the genus. The long transverse processes of caudal vertebrae arise at the anterior ends of the vertebrae and extend antero- June 1982] HERPETOLOGICA 261 laterally. Unregenerated tails have 27-34 trunk vertebrae in adults. Measurements of the holotype (in mm).?Head width 10.3; snout to gular fold (head length) 14.7; head depth at posterior angle of jaw 5.4; eyelid length 3.8; anterior rim of orbit to snout 3.8; hor- izontal orbit diameter 2.9; interorbital distance 3.9; snout to insertion of fore- limb 19.8; distance separating internal nares 2.9; distance separating external nares 3.5; projection of snout past man- dible 0.7; snout to posterior angle of vent (SL) 68.7; posterior to anterior angle of vent 5.1; axilla to groin 37.1; posterior angle of vent to tip of tail (tail length) 54.1; tail width at base 4.8; tail depth at base 4.2; axilla to tip of outstretched fore- limb (forelimb length) 15.5; groin to tip of outstretched hind limb (hindlimb length) 17.5; width of right hand 5.6; width of right foot 7.2. The holotype has 59 maxillary teeth, 7 premaxillary teeth, and 26 anterior vomerine teeth. Habitat.?B. meliana inhabits humid forest and forest margins at moderate to high elevations (1650-2725 m). Vegeta- tion associations at the six known locali- ties range from extremely wet cloud for- est in the east (Baja Verapaz, El Progreso, Zacapa) to less humid oak-pine forest in the west (El Quiche). A generalized map of rainfall (Institute Geografico Nacional de Guatemala, 1966) indicates that an- nual rainfall in areas where B. meliana occurs ranges from approximately 150- 400 cm. Like B. resplendens and B. lin- colni (Wake and Lynch, 1976), B. me- liana is relatively generalized in its mi- crohabitat requirements. Most individuals have been collected on the ground be- neath logs or other cover objects, but ar- boreal bromeliads and the loose bark of fallen logs are also used as retreats. Sympatric salamanders.?In the east- ern portion of its range (Baja Verepaz, El Progresso, Zacapa), B. meliana co-occurs with two other salamanders, a medium- sized species (B. helmrichi) and a dimin- utive form (Chiropterotriton veraepacis). Both tend to be more arboreal than B. meliana, but all three species inhabit bromeliads and crevices beneath loose bark. In El Quiche, where B. meliana reaches its greatest local abundance (over 70 specimens collected), Chiropterotri- ton is absent but B. morio and an un- named salamander possibly referable to B. hartwegi (Elias, 1982) are present. Range.?Known from several localities that span approximately 160 km along the predominantly east-west axis of the Chuacus-Minas mountain systems of cen- tral Guatemala (Fig. 2). Some popula- tions are undoubtedly isolated by inter- vening low mountain passes. The species should be sought in the higher mountains of Alta Verapaz, where both B. helmrichi and C. verapaecis occur (Lynch and Wake, 1978). B. meliana is the first sala- mander reported from the departamentos of El Progreso and Zacapa. RELATIONSHIPS AMONG POPULATIONS OF THE FRANKLINI GROUP Systematic background.?The first member of the franklini group to be de- scribed (Schmidt, 1936) was Bolitoglossa franklini, a common inhabitant of bro- meliads in cloud forests at moderate el- evations (1700-2600 m) along the Pacific escarpment of SW Guatemala. Taylor (1941) described B. nigroflaves- cens from a large series of salamanders collected by H. Smith on Cerro Ovando, a peak in the Sierra Madre del Sur of Chiapas some 100 km NW of the nearest of Schmidt's Guatemalan localities for B. franklini (Fig. 2; locality 7). The two species are similar in size and habitus and in their use of bromeliads as the pri- mary microhabitat. Both species possess a black background coloration that is marked with light-colored patches. In B. nigroflavescens, the iridophore patches tend to be dense and laterally concen- trated, and are cream to yellow, orange- yellow, or brassy in color; in B. franklini, iridophores tend to be more diffuse and silver-gray to bronze in color, and they are concentrated dorsally (Schmidt, 1936; Taylor, 1941; Wake and Lynch, 1976). 262 HERPETOLOGICA [Vol. 38, No. 2 16' 92' ~r~ 90? 88? ^0//7(7^/055J MEXICO N. GUATEMALA \ ? B.franklini ? B.resplendens %/ SALVADOR J^= O- 18? -14? 94? 92? 90? 88" FIG. 2.?Known localities of the B. franklini group in Guatemala and in Chiapas, Mexico. Numbered localities are the sites sampled for the electrophoretic analysis (see Table 1). Area above 1500 m is stippled. Although the Sierra Madre of Chiapas is topographically continuous with the Pa- cific escarpment of Guatemala, Taylor (1941) did not compare B. nigroflaves- cens with B. franklini; instead, he sug- gested that relationships of the former species were with B. engelhardti, a much smaller and differently patterned Guate- malan species, and with B. macrinii of southern Oaxaca and Guerrero. In 1978, one of us (JFL) obtained the first topo- typic B. nigroflavescens since the date of original collection by Smith in 1941. A poorly preserved series of Bolitoglossa in the MVZ collection, from near Mapasta- pec (NW of Cerro Ovando), is morpho- logically similar to topotypic B. nigrofla- vescens. Bolitoglossa brevipes was described by Bumzahem and Smith (1955) from a single specimen that was present in a se- ries of preserved amphibians and reptiles obtained from E. Matuda of Chiapas, Mexico. The color of the badly faded ho- lotype, a large and robust female, is brown, with yellow to yellow-green lat- eral blotches. The only locality data for the specimen are "Region de Soconusco, Chiapas," a local name for the eastern section of the Sierra Madre del Sur be- tween the Guatemalan frontier and the vicinity of Cerro Ovando, where Matu- da's finca was located. Today, as when the Matuda collection was assembled in the 1940's, the only road that penetrates the high elevations of this section of the Sierra Madre del Sur extends northward from the coastal plain town of Huixtla to the interior village of Motozintia. It cross- es the continental divide just south of Cerro Mozotal (elevation ca. 2800 m), the highest peak in the Sierra Madre. This area is almost certainly the source of the two high elevation species of Pseudoeu- June 1982] HERPETOLOGICA 263 rycea (P. brunnata and P. goebeli) re- ported by Bumzahem and Smith (1955) from the Matuda collection. Between 1972 and 1980, we collected several se- ries of a large, dark Bolitoglossa marked with cream to green-yellow lateral spots at three localities near the continental divide (elevation 1900-2300 m) above Motozintla. The salamanders occurred both in bromeliads and under surface ob- jects. On the nearby flanks of Cerro Mo- zotal, we collected two species of Pseu- doeurycea (apparently the same forms reported by Bumzahem and Smith) at el- evations of 2600-2800 m. Our Bolitoglos- sa specimens strongly resemble the ho- lotype of B. brevipes, and we suggest that they represent the same population from which it was collected. Specimens of Bolitoglossa collected by Norman Hart- weg a few km NW of Motozintla (UMMZ collection) also appear to be conspecific with our material from the mountains just above Motozintla. Stuart (1943) described B. lincolni from a series of three red-and-black sal- amanders that he collected in the Sierra de los Cuchumatanes of NW Guatemala. Stuart compared the new species with B. franklini, but not B. nigroflavescens. Ex- cept for two additional specimens col- lected by Stuart in the 1950's, B. lincolni remained unknown until 1978, when Paul Elias discovered the species near the village of Uspantan in the eastern Cuchumatanes. McCoy and Walker (1966) described B. resplendens, a large red-and-black sala- mander from the Mesa Central of Chia- pas. They differentiated the new species from B. franklini on the basis of color differences and the supposed larger size of B. resplendens. McCoy and Walker further stated that B. resplendens dif- fered from B, lincolni in body size and relative limb length. However, the larg- est specimen of B. lincolni available to McCoy and Walker measured only 57 mm SL, and showed no external signs of sexual maturity. Comparisons using re- cently collected, fully adult specimens have invalidated both of the diagnostic characters that previously were thought to distinguish B. lincolni from B. resplen- dens (Elias, 1982). In 1970, we discovered a disjunct pop- ulation of red-and-black Bolitoglossa at high elevations (2600-2850 m) along the Pacific escarpment near San Marcos in SW Guatemala. Typical B. franklini was common in cloud forest lower on the same slope (1800-2650 m), and the two forms were locally sympatric over an el- evational interval of approximately 50 vertical meters. Electrophoretic analysis indicated the existence of hybridization between B. franklini and the red-and- black form, which we tentatively referred to B. resplendens (Wake and Lynch, 1976; Wake et al., 1981). Another isolated population of red- and-black Bolitoglossa occurs in the Montanas de Cuilco, an isolated massif that lies between the Sierra de los Cu- chumatanes and the Pacific escarpment of Guatemala (Fig. 2). This population (indicated as locality no. 9 on Fig. 2) also was tentatively assigned to B. resplen- dens (Wake and Lynch, 1976). Finally, we note that an undescribed member of the franklini group occurs on Cerro Tres Picos, an isolated peak (2000 m) at the western end of the Sierra Madre, some 150 km NW Cerro Ovando. We have examined two preserved speci- mens of this large dark salamander, but final determination of its taxonomic sta- tus awaits the availability of living ma- terial. Summarizing, prior to our discovery of B. meliana, the franklini complex com- prised (1) two morphologically similar red-and-black species (B. lincolni and B. resplendens) found in at least four geo- graphically isolated areas within the in- terior highlands of eastern Chiapas and western Guatemala, and (2) three Pacific slope species (B. brevipes, B. franklini, B, nigroflavescens) marked with silvery, bronze, or yellowish iridophore patches on a black background. Recently, Elias (1982) suggested that B. resplendens is 264 HERPETOLOGICA [Vol. 38, No. 2 TABLE 2.- ?Genie variation in the Bolitoglossa franklini group. For species and localities see Table 1. Sample sizes are in parentheses. Population Locus 1 234 5 678 9 (10) (2) (1) (13) (11) (14) (1) (7) (7) Mpi a (0.7) a (0.5) a b (0.08) b(0.18) c a (0.5) b (0.21) c b (0.3) b (0.5) c (0.92) c (0.82) b (0.5) c (0.79) a Gpd a (0.2) c (0.4) d (0.4) a (0.5) b (0.5) a (0.5) b (0.5) d c (0.22) d (0.78) d d d d Me b b (0.75) c (0.25) b a (0.31) c (0.69) a a a a a Pgd a (0.06) a (0.25) a (0.5) b (0.38) b (0.82) b (0.96) c b c b (0.94) b (0.75) b (0.5) c (0.62) c(0.18) c (0.04) led 1 b b b b a (0.09) b (0.91) a a b b led 2 b a (0.25) b a(0.15) b (0.86) b (0.88) a (0.5) b (0.93) c b (0.75) b (0.85) c(0.14) c (0.12) b (0.5) c (0.07) Mdh 1 a a a b a (0.09) a (0.91) b b b c Mdh2 a (0.7) b (0.2) d(0.1) a b a (0.92) d (0.08) a (0.9) c (0.05) e (0.05) a a a (0.71) d (0.29) a Gpi d d e a (0.04) b (0.71) d (0.25) c (0.09) d (0.91) b (0.04) d (0.96) b b b (0.5) d (0.5) Pgm a(0.1) a (0.5) b (0.5) a (0.08) b (0.09) a b a (0.29) a b (0.9) b (0.5) c (0.5) b (0.92) c (0.91) b (0.71) Ldh 1 c c (0.75) f a (0.27) a (0.68) a a (0.5) b (0.71) a (0.5) f(0.25) d (0.69) e (0.04) d (0.32) d (0.5) d (0.29) b (0.5) Ldh 2 b b (0.75) c (0.25) b b (0.88) c(0.12) a (0.05) b (0.95) b b b a (0.43) b (0.57) Got 1 a (0.06) b b b (0.92) a (0.64) b (0.62) a (0.5) b b b (0.94) c (0.08) b (0.36) c (0.38) b (0.5) Got 2 a (0.17) b (0.83) a a a (0.58) b (0.42) b b b b b Lap b b b a a a a a a Sod b b b a a a a a a Pep b b b b b b b b a an invalid species, and that all of the red- and-black populations should be referred to B. lincolni. Results of electrophoretic compari- sons.?Table 2 summarizes the distribu- tion of allozymes at the 17 electromor- phic loci we were able to score consistently. For each between-popula- tion comparison, two genetic distances, DN (Nei, 1972) and DR (Rogers, 1972), were computed (Table 3). Genetic diver- gence within the group is relatively great, with DN values of 0.12-1.80 (t = 0.67) and DR values of 0.18-0.72 (x = 0.41) sepa- rating the six putative species (Table 3). The three populations of B. meliana form a cluster that is well separated from the other taxa (minimum DN = 0.54; mean DN = 0.96). There is, however, con- siderable genetic differentiation among the three samples of B. meliana. Al- though two of our three samples are very small, they nevertheless contain six al- leles that do not appear in the larger re- June 1982] HERPETOLOGICA 265 TABLE 3.?Rogers' (above diagonal) and Nei's (below diagonal) genetic distances, based on 17 variable electromorphic loci, for populations of the Bolitoglossa franklini group. See Table 1 for localities. Population 1 ? 0.185 0.285 0.414 0.434 0.490 0.541 0.417 0.609 2 0.117 ? 0.250 0.432 0.496 0.534 0.601 0.498 0.608 3 0.314 0.256 ? 0.509 0.587 0.668 0.663 0.556 0.724 4 0.545 0.615 0.796 ? 0.268 0.310 0.247 0.192 0.368 5 0.604 0.788 1.064 0.233 ? 0.190 0.276 0.215 0.373 6 0.880 0.871 1.381 0.319 0.162 ? 0.254 0.238 0.348 7 0.743 1.156 1.338 0.204 0.299 0.275 ? 0.243 0.384 8 0.595 0.796 0.979 0.130 0.197 0.251 0.239 ? 0.342 9 1.100 1.189 1.797 0.448 0.444 0.425 0.493 0.404 ? maining sample. As a result, computed genetic distances within B. meliana are relatively large (maximum DN = 0.31; maximum DR = 0.28). The greatest computed genetic dis- tance in the entire assemblage separates the Purulha populations of B. meliana from the Cuilco population assigned to B. resplendens (DN = 1.80; DR = 0.72). Populations presently assigned to brev- ipes, franklini, lincolni, nigroflavescens and resplendens show a complex pattern of genie relationships. Except for the samples of meliana, the resplendens sample from the Cuilco is the most di- vergent (mean DN = 0.79; mean DR = 0.47). The Uspantan sample of lincolni, although morphologically similar to the Cuilco sample (Elias, 1982), is closer ge- netically to franklini, brevipes, and ni- groflavescens. Indeed, the most surpris- ing result of the electrophoretic survey is the discovery that the Uspantan and Mo- tozintla populations (representing B. lin- colni and B. brevipes) are each other's closest genetic relatives (DN = 0.13; DR = 0.19). Other results were more in line with current taxonomic assignments. Thus, the two populations presently assigned to B. franklini (Chicabal and Tacana) are quite similar genetically (DN = 0.16; DR = 0.19). However, the mean distance of both populations to lincolni from Us- pantan is unexpectedly small (mean DN = 0.22; mean DR = 0.23). The Ovando population (topotypic B. nigroflavescens) is only slightly more similar to the nearby Motozintla popula- tion of B. brevipes (DN = 0.20; DR = 0.25) than it is to the relatively distant Uspantan population of B. lincolni (DN = 0.24; DR = 0.24). Fixed differences at two loci (Sod and Lap) separate B. meliana from the re- maining populations of the franklini group, and unique alleles for Pgd and a Gpd are found at relatively high frequen- cies in B. meliana. The remaining species of the franklini group all share two unique fixed differences (Sod and Lap). Unique alleles for Me and Mpi are wide- spread in the species other than B. me- liana. Thus, B. meliana appears to qual- ify as a sister group for the remaining species of the franklini group. DISCUSSION Bolitoglossa meliana is electrophoret- ically and morphologically well differ- entiated from other members of the franklini group, and it clearly merits sta- tus as a full species. However, the valid- ity of the other five nominal species in the complex must be reassessed. Based on morphological similarities and geo- graphic relationships, we initially pre- dicted that the four disjunct populations of red-and-black Bolitoglossa would prove to be close genetic relatives, pos- sibly conspecific (this is also the view of Elias, 1982). Similarly, we suspected that the Pacific versant populations previ- ously assigned to B. brevipes, B. frank- 266 HERPETOLOGICA [Vol. 38, No. 2 lini, and B. nigroflavescens also would show very close relationships. Although the results of our study do not confirm the first prediction, the second is indeed supported. Genetic differentiation among the three supposed species-level taxa found in cloud forests along the Pacific versant of Chiapas and Guatemala is about the same as is seen within B. me- liana. The Pacific populations are similar in ecology and morphology, and they oc- cur along a topographically continuous mountain system. We propose that they be considered members of a single species, for which B.franklini (Schmidt, 1936) is the earliest available name. The Chiapan populations of B. franklini that occur west of the border between Gua- temala and Chiapas (Volcan Tacana) are somewhat differentiated in color pattern, and we propose that they be recognized as a distinct geographic race, B.franklini nigroflavescens. We consider B. brevipes to be a subjective junior synonym of B. franklini nigroflavescens. This subspe- cies is distinguished by the presence of dense, laterally concentrated cream, or- ange-yellow, or green-yellow iridophore patches. In B.f. franklini, the iridophore patches are more diffuse, and are often tattered or fragmented; they are usually gray, silvery, or brassy in color, and are more evenly distributed across the dor- sum. Theodore Papenfuss has drawn our attention to a subtle behavioral differ- ence between the two populations: B. f. nigroflavescens from the Motozintla area tends to be noticeably more sluggish than B. f. franklini. Populations from Volcan Tacana to be somewhat intermediate in color pattern, but are more similar to typ- ical B. f. franklini. The deeply incised, relatively xeric canyon immediately west of Volcan Tacana may correspond to the zone of transition from B. f. franklini to B. f. nigroflavescens. While we do not advocate the unrestricted use of trinomi- als in salamander systematics, we believe that in this instance the subspecies des- ignation conveys useful geographic and morphological information. Our initial prediction that all four in- terior populations of red-and-black sala- manders would prove to be close genetic relatives was not suppported by our elec- trophoretic data. Bolitoglossa lincolni from U span tan is more distant genetical- ly from the morphologically similar Cuil- co population than it is from morpholog- ically distinct populations of B.franklini. The Cuilco population differs from all others that we have sampled except B. meliana in having two unique fixed al- leles (Mdh-l, Pep). We think that the rel- atively great genetic distance between the Cuilco population and other popula- tions reflects a long history of isolation (see below). Given our present ignorance of the ge- netic relationships of topotypic Chiapan resplendens to the remainder of the franklini group, we are hesitant to revise taxonomic status of the red-and-black populations. Future comparisons might show that Chiapan resplendens is genet- ically closer to lincolni from the Cuchu- matanes than to "resplendens" from the Cuilco area. This result might justify (1) including Cuilco, San Marcos, the Chia- pan and Cuchumatanes populations within B. lincolni, as has been suggested by Elias (1982), or (2) erecting an entirely new species to encompass the Cuilco and San Marcos populations. A practical dif- ficulty in the latter course of action is posed by the lack of external morpholog- ical characters that reliably differentiate the various red-and-black populations (Elias, 1982). An even more serious ob- jection to precipitous lumping of all of the red-and-black populations into a sin- gle species is that various populations of B.franklini (including those formerly re- ferred to B. brevipes and B. nigroflaves- cens) would be closer genetically to some red-and-black populations than the latter are to supposed conspecific populations. Therefore, we tentatively continue to re- fer the San Marcos, Cuilco, and Chiapan populations of red-and-black Bolitoglos- sa to B. resplendens, and include only the Cuchumatanes population within B. June 1982] HERPETOLOGICA 267 lincolni. This is done with the full knowl- edge that some future taxonomic reas- sessment of the red-and-black popula- tions in the franklini group probably will be necessary in the future. A basic difficulty in assessing the sys- tematic status of populations in the franklini group is the strong likelihood that hybridization has occurred between previously isolated populations (Wake et al., 1981). For example, the surprisingly close genetic similarity between B. lin- colni and some populations of B. frank- lini may reflect secondary contact, hy- bridization and introgression between coastal and interior populations. We think it likely that Bolitoglossa of the franklini group originally colonized the Pacific versant of Chiapas and Guatemala from the Cuchumatanes, rather than the Cuilco area. However, present-day para- patry and hybridization in the San Mar- cos area apparently is a result of very re- cent secondary contact between local B. franklini and B. resplendens from the Cuilco area (Wake et al., 1981). The featureless black coloration of B. meliana, which appears to be the earliest offshoot from the ancestral stock that gave rise to the franklini group, may be the primitive color pattern, or may be de- rived from a still earlier red-and-black pattern that is preserved in the Cuilco population of B. resplendens (see Fig. 3 below). In either event, the red-and- black coloration of populations that in- habit the ancient interior highlands of Nuclear Central America appears ances- tral to the silver-, bronze-, or yellow-and- black patterns of salamanders associated with the younger mountains along the Pacific versant (see below). Despite the close electromorphic similarity of the Us- pantan population to B. franklini, we suggest that the close similarity in color pattern between the Uspantan and Cuil- co populations reflects their retention of a primitive red-and-black pattern. Historical development of the frank- lini group.?A brief review of the geolog- ic history of Nuclear Central America is a necessary prelude to our attempt to re- construct the evolutionary development of the franklini group. The oldest exposed rocks in the area occur in the Sierra Madre del Sur of Chia- pas and the Chuacus-Minas system of central Guatemala (Dengo, 1968; Mc- Birney, 1963), which consist mainly of Paleozoic metamorphics and sediments. The characteristic ridge-and-valley to- pography in this area is controlled by fault zones that mark the boundary be- tween the North American and Caribbe- an plates (Malfait and Dinkelman, 1972; Plafker, 1976). The plateaus of the Sierra de los Cu- chumatanes and the Mesa Central of Chiapas are composed mostly of some- what younger Mesozoic sediments (An- derson et al., 1973; Dengo, 1968) that are thought to have been uplifted during the late Cretaceous or earliest Tertiary. The Paleozoic and Mesozoic sediments of the Cuilco massif are isolated from the Cu- chumatanes by fault-controlled valleys (Anderson et al., 1973). Most of the area south of the Cuilco and Chuacus-Minas mountains and east of the Chiapan bor- der is overlain by thick deposits of late Tertiary volcanics that arose from wide- spread fissure-type eruptions (Dengo, 1968; Williams, 1960). The spectacular strato-volcanos along the Pacific-facing slope of the Guatemalan plateau are of Pleistocene to Recent age. The focus of intense volcanic activity appears to have shifted progressively southeast from the Chiapas border to the vicinity of Guate- mala City, where several presently active cones (e.g., Santa Maria, Fuego, Pacaya) exist. Although the entire area inhabited by the franklini group has been above sea level since earliest Tertiary, the major uplift that produced the present extreme elevations in the region (the mountain systems discussed above all approach or exceed 3000 m) probably occurred after the late Miocene (Dengo, 1968; Wil- liams, 1960). There is good evidence for the existence of small Pleistocene gla- 268 HERPETOLOGICA [Vol. 38, No. 2 j^ , 1 1 t-> ? ,