Taxonomic Review of Allobates (Anura, Aromobatidae) from the Atlantic Forest, Brazil

Home , Allobates, Allobates brunneus, Allobates goianus, Aromobatidae

1 VANESSA K. VERDADE AND MIGUEL T. RODRIGUES

Departamento de Zoologia, Instituto de Biocieˆncias, Universidade de Sa˜o Paulo, Caixa Postal 11461, CEP 05422–970, Sa˜o Paulo, Brazil

ABSTRACT.—We present the results of a taxonomic review of the four species of Allobates endemic to the Atlantic Forest Domain in Brazil. A total of 880 preserved specimens from 29 localities covering their range in Atlantic Forest were studied based on external attributes. Characters formerly described as diagnostic in original descriptions and others cited for dendrobatids in the literature were studied along the total geographic range of the populations of these species, including the type localities. We found no discrete characters, qualitative or quantitative, capable of differentiating the species. Most local and geographic variation was limited to snout-vent length and color pattern. Considering the lack of evidence otherwise, we use the results of our morphological analysis to allocate all Atlantic Forest species of Allobates to synonymy with Allobates olfersioides.

Dendrobatids are a monophyletic group of Allobates olfersioides was described as Eupem- frogs that are mostly known for their bright phix olfersioides Lutz, 1925 from the coastal color and poisonous skin. Nevertheless, about region of the state of Rio de Janeiro (RJ). half of the species are cryptically and dull Cochran (1955) considered it a synonym of colored, presumably nontoxic and until recently Phyllobates brunneus Cope, 1887. Bokermann (Grant et al., 2006) placed in the genus Colos- (1966) restricted the type locality of A. olfer- tethus sensu lato. There had been about a hun- sioides to Angra dos Reis (RJ) and, in 1967, dred recognized species in the genus ranging described the other three Atlantic Forest species: from Nicaragua to southeastern Brazil (Frost, Phyllobates alagoanus from Usina Sinimbu, Man- 2006). The systematics of the family has recently gabeiras, state of Alagoas; Phyllobates capixaba gone through significant changes (Grant et al., from Lagoa do Macuco, Refu´gio Sooretama, 2006) supported by a molecular and morpho- Linhares, state of Espı´rito Santo; and Phyllobates logical based phylogeny. The family Dendroba- carioca from Represa Rio Grande, Jacarepagua´, tidae, as previously known, corresponds now to Rio de Janeiro, state of Rio de Janeiro. Boker- Dendrobatoidea, which includes the families mann (1967) also considered P. olfersioides re- Aromobatidae and Dendrobatidae. The four lated to P. alagoanus, P. capixaba and P. carioca. species of Colostethus from Atlantic Forest, Edwards (1974) placed these four species in now allocated to the genus Allobates, are placed Colostethus, as did Silverstone (1975). Grant et al. in the family Aromobatidae. (2006) based on an extensive sampling and The Brazilian species of Allobates are leaf litter molecular plus morphological based phylogeny inhabitants of forests and can be found through- allocated them to the genus Allobates. out the Amazon Basin, Brazilian Savannas, and The four species are very similar in external Atlantic Forest. Most of the 17 currently attributes and were described based on small recognized species are Amazonian and de- and geographically remote samples. The char- scribed recently, probably as a consequence of acters used by Bokermann (1967) to separate a vast and poorly sampled territory along with these species were basically color pattern, increased knowledge of the natural history of length of limbs and fingers, size of carpal and these frogs (e.g., Morales, 2000; Lima and tarsal tubercles, shape of head, and snout-vent Caldwell, 2001; Caldwell and Lima, 2003). There length. All of these characters are presumably are four species in eastern Brazil, endemic to the subject to geographical variation, a possibility that could not be tested at that time. Atlantic Forest Domain (Ab’Saber, 1977): Allo- bates alagoanus (Bokermann, 1967), Allobates Bokermann (1967) also presented information on call variation, pointing out that the calls of A. capixaba (Bokermann, 1967), Allobates carioca capixaba and A. carioca were distinct from that of (Bokermann, 1967), and Allobates olfersioides A. olfersioides, but, except for saying that the call (Lutz, 1925). of A. capixaba was more spaced than those of A. olfersioides, no further description was given. 1 Corresponding Author. E–mail: [email protected] Besides, the putative call differences between TAXONOMIC REVIEW OF ALLOBATES FROM THE ATLANTIC FOREST 567 species were reported by the collector (F. M. Oliveira) and not based on the authors’ field experiences or knowledge. The main characteristics of the advertisement call of Allobates (a trill or an acute buzz) can vary in fundamental frequency, number of pulses and spacing, according to environmental conditions and sometimes, as a function of population density (Coloma, 1995; Junca´, 1998; Grant and Rodri- guez, 2001; Caldwell et al., 2002a). Considering the subjective call differences reported and their variation, we think that the evidence alone is weak to keep the Atlantic Forest species of Allobates separated. Moreover, the observations on vocalization pointed out in the species description refer only to the surroundings of the type locality, and geographical variation of call should be considered. Unfortunately, we were unable to find available calling records in collections, either institutional or personal, and our attempts to get new ones failed. Because this problem will probably persist in the future, FIG. 1. Geographic distribution of Atlantic Forest because the Atlantic Forest species of Allobates Allobates. Open circles represent the type localities of are showing declines (Weygoldt, 1989; Izeck- the four currently admitted species. Numbers are the sohn and Carvalho-e-Silva, 2001), specimens localities sampled (see Appendix 2). The OTUs gathered in collections will be the main source studied are indicated by OTU1, 2, 3, and 4. for studying species limits of this complex in the Atlantic Forest. Indeed, the samples collected (TIL) from the outer edge of the flexed knee to during the last 40 years are a source of new data the heel; foot length (FOL) from the heel to the on morphological variation that could clarify tip of toe IV; head width (HW) between the the similarities and differences among the four angle of jaws; head length (HL) perpendicular species of Allobates from Atlantic Forest. to the imaginary line linking the angle of jaws to We present, herein, a detailed study of the the tip of the snout; internarial distance (IND) morphological variation of these populations between the inner edge of the nares; eye-nostril along a latitudinal gradient that includes the distance (END) from the anterior corner of the type localities of the presently recognized eye to the outer edge of the nostril; eye diameter species. We assume that external morphology (ED) from posterior to anterior corners; tympa- is the strongest available source of data sustain- num diameter (TD) from posterior to anterior ing the Atlantic Forest species of Allobates, and, edges; interorbital distance (IOD) between the if we found no discrete differences in morpho- inner edges of the eyelids; hand length (HAL) logical characters among populations, they dorsally from the junction of radio-ulna and should be considered synonyms until new carpal to the tip of the finger III; thumb length sources of data can be explored. (TBL); second, third, and fourth fingers length (FL2, FL3, FL4) dorsally from the base to the tip of the finger; third finger diameter (TFD) MATERIALS AND METHODS dorsally at the distal edge of the second finger To cover their total range, 880 preserved tubercle; greatest length of the inner and outer specimens of Allobates from 29 localities were carpal tubercles (ICT, OCT); and greatest length studied (Fig. 1, Appendices 1 and 2). All speci- of the inner and outer metatarsal tubercles mens were observed regarding external mor- (IMT, OMT). phology. We measured and classified 396 All measurements were taken on the left side individuals to be used in the statistical analysis. of the specimens. The measurements were This number comprises all individuals from obtained with a stereomicroscope except for localities with N below 100 and 100 randomly SVL, taken by a dial caliper, to the nearest of picked up individuals from localities with N 0.01 mm. The axis of the greatest length at ICT, above 100. Twenty-two measurements were OCT, IMT, and OMT variables was the same for obtained for the specimens included in the all specimens. analysis: snout-vent length (SVL); thigh length Qualitative characters studied were: shape of (THL) from the middle of the cloacal opening to the snout (dorsal, ventral, and lateral views); the outer edge of the flexed knee; tibia length pattern of the lateral dark stripe between the 568 V. K. VERDADE AND M. T. RODRIGUES

Estadual de Saõ Paulo, UNESP); MNRJ (Museu Nacional do Rio de Janeiro); MUFAL (Museu da Universidade Federal de Alagoas); MZUSP (Museu de Zoologia da Universidade de Saõ Paulo, currently including the Werner Boker- mann collection, WCAB); UEFS (Universidade Estadual de Feira de Santana); UFPB (Universi- dade Federal da Paraı´ba); UFRJ (Universidade Federal do Rio de Janeiro); USNM (United States National Museum); ZUEC (Museu de Histo´ria Natural, Universidade de Campinas, FIG. 2. Schematic drawing of an Allobates with the nomenclature used for the study of color pattern of UNICAMP). the specimens. There were no obvious secondary sexual characters to differentiate males and females nares and the eyes; color pattern of the upper among the four species studied. To avoid lip; presence of a median lingual process (Grant damage caused by dissection we checked for et al., 1997, 2006); color pattern of the belly; morphometric differentiation between sexes color pattern on dorsal view; pattern of distri- only in a sample from Tijuca, Rio de Janeiro. bution of tubercles; pattern of dark stripes on Forty-four adult individuals (checked by matu- groin; pattern of the dorsolateral and lateral rity of gonads) from a total of 400 were dissected, and 20 males and 20 females were light stripes; pattern of the lateral dark stripe compared under a Student t-test at a 95% (Fig. 2); cloacal stains; color pattern of the thigh; interval of confidence. presence/absence and extension of metatarsal Three samples of tadpoles identified by the fold; and toe webbing. presence of nursery adults or young meta- To make sure that there were no extralimital morphs were available for study—Ilheús (Ba), specimens belonging to other species in our Una (Ba), and Tijuca (RJ). These samples sample, and that the Atlantic Forest species contained specimens in different developmental were restricted to the study area, we diagnosed stages and were not suitable for geographical them from all other Brazilian species (Sociedade comparison. Tadpoles from Tijuca (RJ) were Brasileira de Herpetologia, 2005) previously used to describe the larvae following Altig and placed in the genus Colostethus (now in the McDiarmid (1999) standard measurements and genera Allobates, Anomaloglossus, and Hyloxalus), descriptive terminology. based on literature and direct examination Statistical Analysis.—Specimens were first when possible: Allobates brunneus (Cope, 1887); observed and compared within localities to Allobates caerulodactylus, (Lima and Caldwell, detect possible cases of sympatry, and then 2001); Allobates conspicuus (Morales, 2000); Allo- observed and compared among localities fol- bates crombiei (Morales, 2000); Allobates fuscellus lowing a north-south transect. In addition, we (Morales, 2000); Allobates gasconi (Morales, ran a preliminary stepwise Discriminant Anal- 2000); Allobates goianus (Bokermann, 1975); ysis to assess differences among localities Allobates granti (Kok et al., 2006); Allobates treated as individual operational taxonomic marchesianus (Melin, 1941); Allobates masniger units (OTUs) and considering the only case of (Morales, 2000); Allobates nidicola (Caldwell and sympatry suggested in the literature (Allobates Lima 2003); Allobates sumtuosus (Morales, 2000); olfersioides and A. carioca; Bokermann, 1967) as Allobates vanzolinius (Morales, 2000); Anomalo- individual units. The observational and statisti- glossus beebei (Noble, 1923); Anomaloglossus cal results were subsequently considered to stepheni (Martins, 1989); Anomaloglossus tama- group the specimens in four OTUs, based on cuarensis (Myers and Donnelly 1997); and overall similarity and geographical distribution. Hyloxalus peruvianus (Melin, 1941). We decided The OTUs are also congruent to the gaps to diagnose them also from Allobates trilineatus observed in the geographical distribution of (Boulenger, ‘‘1883’’ 1884) and Anomaloglossus the four species considered together (Fig. 1). degranvillei (Lescure, 1975), considering their The specimens from the states of Alagoas and distribution close to Brazil (Grant and Rodri- Sergipe were treated as OTU1 (N 5 57), speci- guez, 2001; Frost, 2006). mens from the state of Bahia as OTU2 (N 5 68), The abbreviations for the herpetological col- specimens from the state of Espı´rito Santo as lections consulted for preserved specimens and OTU3 (N 5 11), and specimens from the state of call records are presented throughout the text as Rio de Janeiro as OTU4 (N 5 260; Fig. 1). follows: EI (Eugeˆnio Izechson collection, Uni- Statistics were obtained for each OTU. versidade Federal Rural do Rio de Janeiro, The absence of obvious sexual characters UFRRJ); JJ (Jorge Jim collection, Universidade made it difficult to classify the specimens as TAXONOMIC REVIEW OF ALLOBATES FROM THE ATLANTIC FOREST 569 young or adults. To minimize deviations of the resembles a trapezoid; the body is slender and means, we chose not to use very small speci- elliptical (Fig. 3A, B). The dorsal skin is gener- mens. As a criterion of SVL exclusion, we used ally smooth, but inconspicuous tubercles may the SVL average less the standard deviation of be present at the inguinal region or throughout the OTU 1, which presented the smallest the entire dorsum. individuals. The value obtained was 11.4 mm. Dorsal pattern consists of a reddish-brown All the specimens with SVL equal to or smaller background over which are darker stripes that than 11.4 mm were excluded from the sub- cross once or more at the middle of the dorsum, sequent quantitative analysis, which left us with suggesting intercrossing Xs. Dorsal pattern can 372 specimens to be statistically analyzed: OTU vary from uniform to highly pigmented (Fig. 4; 1(N 5 44), OTU 2 (N 5 66), OTU 3 (N 5 10), for summary, see Table 1). Individuals with and OTU 4 (N 5 252). None of the samples a uniform dorsum are more frequent (about presented normal distributions for all variables, 15%) in samples from the state of Bahia (OTU 2) even with data log-transformed, but deviations but still occur in lower rates in other samples (, were not severe enough to prevent parametric 1% from the State of Rio de Janeiro, OTU 4). tests. Individuals with a single X on the dorsum are As an attempt to express differences in shape more common (74%) in the states of Alagoas among individuals in different samples, some and Sergipe (OTU 1), but specimens with body ratios were used in addition to the raw multiple X also occur in these samples (24%). measurements. They are THL/SVL, HAL/SVL, The multiple-X pattern is the most common in FOL/SVL, HW/HL, END/HL, ED/HL, TD/ OTU 2, 3, and 4, present in about 40, 80, and HL, TD/ED, IND/HW, TBL/FL2, FL2/HAL, 90% of the individuals, respectively (Table 1). and FL4/HAL. There is some variation in this pattern. Some- To classify the specimens, three stepwise times the largest X has longer arms that barely Discriminant Analyses were run separately touch each other. This variation is found in based on the morphological measurements, individuals from samples from the states of body ratios, and qualitative data. The number Espı´rito Santo (OTU 3) and Rio de Janeiro (OTU of steps was unspecified, and all groups were 4) and is more common among the former assigned to equal probabilities before analysis. (45%). The dorsal parts of the thigh and tibia Sexual dimorphism for the Tijuca sample was have two or three dark stripes bordered by assessed by the Student t-test on each variable lighter ones in almost all specimens examined. and body ratio at a confidence interval of 95%. In some individuals, the dark stripes are much Means are given 6 1 SD. narrower, almost a line, in others the stripes are absent. Laterally, all individuals present a dark stripe RESULTS from the tip of the snout to the groin, involving All localities sampled are restricted to the the nares and partially the eyes along the side of coastal region of eastern Brazil. The locality the body (Fig. 5). Apparent north-south varia- farthest north is Novo Dino, state of Alagoas tion exists in the width and definition of this and the farthest south is Angra dos Reis, state stripe. Following the north-south transect, there of Rio de Janeiro. The farthest west locality is is a tendency of the dark lateral stripe to become Baixo Guandu in state of Espı´rito Santo. There wider with a crescent reticulation toward the are two major gaps along the north-south venter and inguinal region (Fig. 6). Individuals transect. One between Areia Branca in the state from the state of Espı´rito Santo (OTU 3) are the of Sergipe and Elı´sio Medrado, state of Bahia, ones in which this characteristic is more pro- and the other between Santa Teresa, state of nounced. To the south, the dark lateral stripe Espı´rito Santo and Tereso´polis, state of Rio de reverses to solid in individuals from northern Janeiro. These areas correspond to drier envir- Rio de Janeiro (OTU 4), but it is still wide. At the onments. In the north, the coastal region is southernmost localities, the lateral stripe is composed by fixed sand dunes, but there are narrower and solid, as observed in individuals many forested regions that could conceal the from the state of Alagoas, Sergipe (OTU 1) and species. The southern region is still covered by Bahia (OTU 2). forests. It seems more probable that the absence The dark lateral stripe can be dorsally of samples of Allobates from these areas is an bordered by a narrow pale stripe that varies in artifact of collection (Heyer, 1988). length and continuity. Most of the individuals Qualitative Data.—All individuals present (97%) from the state of Alagoas (OTU 1) do not a rounded snout tip in dorsal and lateral views present the pale dorsolateral stripe (Table 1). with nares slightly prominent and an upper jaw When it occurs, it is present from the eyes to the projecting slightly above the lower jaw. The arms. The pale dorsolateral stripe is not present general shape of the snout in dorsal view in 57% of the individuals from the state of Bahia 570 V. K. VERDADE AND M. T. RODRIGUES

FIG.3. Allobates olfersioides (MNRJ 5094): (A) dorsal and (B) lateral views of head, ventral views of (C) right hand and (D) left foot.

(OTU 2) and can be observed from the eyes to to the pectoral or inguinal regions in about 30% the arms in 36% of them and from the eyes to of individuals in both states. The general color the groin in 7%. In the state of Espı´rito Santo pattern is highly variable, and we did not find (OTU 3), all individuals present the pale any geographical consistency or discrete local dorsolateral stripe. It is observed from the eyes variation to justify partitioning of the samples. to the arms in 9% of them and from the eyes to Quantitative Data.—It was impossible to di- the groin in 91% of them. To the south, only agnose populations studied and refer them to 16% of the individuals from the state of Rio de any of the recognized species based on body Janeiro (OTU 4) present the pale dorsolateral proportions or the following diagnostic char- stripe. acters pointed out in literature: SVL, ratio of Individuals from the states of Alagoas and length of the thumb and the second finger, Sergipe (OTU 1) commonly (62%) have venter absolute length of the fourth finger, and shape pigmentation restricted to the throat, never of the tarsal tubercle (Bokermann, 1967; Ed- extending from throat to groin (Table 1). Sam- wards, 1974; Rivero, 1988; Morales, 2000). ples of state of Bahia (OTU 2) specimens with an Table 2 presents the descriptive statistics for immaculate venter are more frequent (57%). the main measurements taken for individuals in There are two equally common patterns of the four OTUs. venter pigmentation in individuals from the Individuals from the states of Alagoas and states of Espı´rito Santo (OTU 3) and Rio de Sergipe (OTU 1) are smaller (t df 5 107 526.316, Janeiro (OTU 4). Venter pigmentation extends P 5 0.000) than those from the state of Bahia TAXONOMIC REVIEW OF ALLOBATES FROM THE ATLANTIC FOREST 571

on body ratios are presented in Figure 8. The first and second discriminant functions explain 98% of the differences among the OTUs. The most important variable on function 1 is TBL/ FL2 and on function 2 is FL4/HAL. The variation of the values of the TBL/FL2 and FL4/HAL for each locality from north to south can be seen in Figure 9. The shape of the tarsal tubercle shows variation, and individuals from the same sample can have a lump, or a straight- or comma-shaped ridge. We tried to rearrange the localities into different OTUs, following, for example, the main river barriers (e.g., Sa˜o Francisco, Jequitinhonha, Doce, and Paraı´ba do Sul), but the results were always similar. None of the measurements or proportions studied presented evidence to separate specimens into different groups.

TAXONOMIC CONCLUSION Our study of Allobates from the Atlantic Forest made it clear that the characters currently used to differentiate the recognized species were not useful in our sample (Bokermann, 1967; Edwards, 1974; Rivero, 1988; Morales, 2000). Variation in SVL throughout their distributions is probably a result of restricted sampling in some of the localities (Fig. 7). The data also show that the relative lengths of the thumb, second, and fourth fingers are highly variable, as already found for other species of Allobates (Grant and Castro, 1998). Diagnostic differences between species are probably mis- interpretations based on limited series. The only record of sympatry among the Atlantic Forest Allobates was published by Bokermann (1967) from Represa Rio Grande, FIG. 4. Dorsal view of adult Allobates olfersioides state of Rio de Janeiro, where A. olfersioides and showing variation on dorsal color pattern. A. carioca were found. This sample, presently at MZUSP, includes 19 specimens, three of which were identified by Bokermann as A. carioca (the (OTU 2; Table 2). Individuals from the state of holotype and the two paratypes). At first, these Bahia (OTU2) are larger (t df 5 74 522.324, P 5 specimens look slightly different from the 0.023) than individuals from the state of Espı´rito typical A. olfersioides, presenting a larger size Santo (OTU 3). Mean of SVL of individuals from and uniform dorsum. However, difference in the states of Espı´rito Santo (OTU3) and Rio de size detected visually is not significant when the Janeiro (OTU 4) are not different (t 5 5 df 260 mean SVL is compared (t df 5 16 5 1.62, P 5 20.042, P 5 0.389). Despite statistically signif- 0.12). In addition, although very rare (less than icant OTUs, differences in individuals SVL 1%), the uniform dorsum is present in other among localities are not clear when visualized samples of A. olfersioides from the state of Rio de in a graphic of dispersion (Fig. 7). The smallest Janeiro. Izecksohn and Carvalho-e-Silva (2001) maximum sizes are found among individuals have already suggested the possibility that these from the State of Espı´rito Santo, but it represents two taxa are actually synonyms. When com- the smallest OTU (N 5 10), with localities pared with individuals from other OTUs, these poorly sampled. individuals are very similar to those from the The results of the discriminant analysis based state of Bahia (OTU 2) presenting the same on both raw measurements and body ratios general characteristics in size and color pattern. show overlap among the four operational units. Unfortunately, vocalization, the only other The results for the discriminant analysis based argument used to justify this species could not 572 V. K. VERDADE AND M. T. RODRIGUES

TABLE 1. Frequencies of dorsal and ventral color patterns taken from adults of Atlantic Forest Allobates included in the four operational taxonomic units (OTU; see text for definitions of OTUs).

OTU 1 OTU 2 OTU 3 OTU 4 (N 5 57) (N 5 68) (N 5 11) (N 5 260) Central dorsal pattern Uniform 1 (2%) 9 (13%)0 1(, 1%) Single X 42 (74%)2(3%)0 0 Multiple Xs 14 (24%) 57 (84%) 5 (45%) 240 (92%) Long arm X 0 0 6 (55%)19(7%) Dorsolateral pale stripe Absent 55 (97%) 39 (57%) 0 219 (84%) From eye to arm 2 (3%) 24 (36%)1(9%)15(6%) From eye to inguinal region 0 5 (7%) 10 (91%) 26 (10%) Venter pigmentation Absent 15 (26%) 29 (42%)1(9%) 50 (19%) Restricted to throat 35 (62%) 16 (24%) 2 (19%) 34 (14%) From throat to chest 7 (12%) 19 (28%) 4 (36%) 84 (32%) From throat to inguinal region 0 4 (6%) 4 (36%) 92 (35%) be tested here because there are no available calling records and the attempt to get new records did not succeed. According to the data presented above, there is no external morphological evidence to keep the four species of Allobates known from the Atlantic Forest as separate units. It is possible that new ecological, behavioral, cytogenetic, or molecular evidence could contribute to our understanding of the systematics of this group. Until such time as additional evidence is available, we suggest a more conservative posture and treat A. alagoanus, A. capixaba, and A. carioca as synonyms of Allobates olfersioides. Allobates olfersioides (Lutz, 1925) Figure 3 Eupemphix olfersioides Lutz, 1925: 138. Holotype.—MNRJ 783 (currently exists only as a piece of flesh), collected on the coast of the state of Rio de Janeiro, Brazil; Angra dos Reis, state of Rio de Janeiro, according to Bokermann (1966).

FIG.5. Allobates olfersioides, live specimen from FIG. 6. Lateral view of adult Allobates olfersioides Maceio´, state of Alagoas (Photograph: G. Skuk). showing variation on dark lateral stripe. TAXONOMIC REVIEW OF ALLOBATES FROM THE ATLANTIC FOREST 573

TABLE 2. Descriptives for the main measurements taken from adults of Atlantic Forest Allobates included in the four operational taxonomic units (OTU; see text for definitions of OTUs). Data are presented as mean 6 SD (N)range.

OTU 1 OTU 2 OTU 3 OTU 4 Snout–vent length (SVL) 14.5 6 1.4 (43) 16.5 6 1.7 (66) 15.1 6 1.5 (10) 15.5 6 1.4 (252) 11.5–17.2 11.5–18.8 13.3–17.2 11.5–18.6 Thigh length (THL) 6.5 6 0.6 (43) 7.8 6 0.7 (65) 6.7 6 0.6 (10) 7.1 6 0.6 (249) 5.0–7.7 5.5–8.8 5.7–7.5 3.8–8.5 Tibia length (TIL) 6.8 6 0.5 (42) 7.8 6 0.7 (65) 7.0 6 0.3 (10) 7.2 6 0.6 (249) 5.5–7.5 5.2–8.8 6.5–7.6 4.8–9.0 Hand length (HAL) 2.9 6 0.4 (43) 3.4 6 0.4 (66) 3.0 6 0.2 (10) 3.3 6 0.4 (252) 2.2–3.8 2.0–4.7 2.8–3.3 2.0–4.1 Thumb length (TBL) 1.5 6 0.2 (43) 1.8 6 0.3 (66) 1.5 6 0.2 (10) 1.6 6 0.2 (252) 1.1–1.8 1.1–2.4 1.2–1.8 0.9–2.2 Head length (HL) 3.7 6 0.3 (43) 3.7 6 0.3 (66) 3.5 6 0.3 (10) 3.7 6 0.3 (252) 3.1–4.4 2.8–4.7 3.1–4.1 2.9–4.9 Foot length (FOL) 5.8 6 0.7 (43) 6.9 6 0.7 (66) 6.1 6 0.3 (10) 6.6 6 0.6 (252) 2.9–6.8 4.2–8.1 6.8–7.2 4.2–7.6 Second finger length (FL2) 1.1 6 0.1 (43) 1.4 6 0.2 (66) 1.2 6 0.2 (10) 1.4 6 0.2 (251) 0.8–1.3 0.7–1.9 0.9–1.4 0.8–1.8 Head width (HW) 4.7 6 0.3 (44) 5.3 6 0.5 (66) 4.8 6 0.3 (10) 4.9 6 0.4 (252) 3.7–5.4 3.8–6.2 4.3–5.2 3.9–5.7 Fourth finger length (FL4) 0.6 6 0.1 (43) 0.9 6 0.1 (66) 0.8 6 0.1 (10) 0.9 6 0.1 (252) 0.4–1.0 0.6–1.2 0.7–0.9 0.4–1.2

Phyllobates capixaba Bokermann, 1967: 349. Phyllobates carioca Bokermann, 1967: 352. Holotype.—WCAB 19252 (MZUSP 73752), col- Holotype.—WCAB 38610 (MZUSP 76653), collected at Refu´gio Sooretama, Linhares, state of lected at Represa Rio Grande, Jacarepagua´, Rio Espı´rito Santo, Brazil. de Janeiro, state of Rio de Janeiro, Brazil. Phyllobates alagoanus Bokermann, 1967: 351. Allobates olfersioides Grant, Frost, Caldwell, Holotype.—WCAB 2801 (MZUSP 73707), col- Gagliardo, Haddad, Kok, Means, Noonan, lected at Mangabeiras, state of Alagoas, Brazil. Schargel, and Wheeler, 2006.

FIG. 7. Values of the snout–vent length (SVL) measured in each sample of Atlantic Forest Allobates studied. Dots represent maximum SVL values, the squares are mean, and the bars, the standard error for each sample. Localities are presented with shortened names (see Appendix 2) from north (left) to south (right). The names above the graphic are the states to which the localities belong. 574 V. K. VERDADE AND M. T. RODRIGUES

lateral stripe absent or present; dark lateral stripe continuous extending from tip of snout to groin; ventrolateral stripe present, poorly defined; skin of dorsum smooth or slightly granular in sacral region; skin of venter smooth; paracloacal ‘‘C’’ like light stains present; cloacal tubercles absent; median lingual process absent. Testis unpigmented in adult males. Comparison with Other Species.—Given their overall similarity, in the following section we compare Allobates olfersioides with all species of Allobates, Anomaloglossus, and Hyloxalus re- FIG. 8. Graphic representation of the results of the corded from Brazil, and those likely to be found Discriminant analyses based on the Atlantic Forest in Brazilian territory. Allobates brunneus differs Allobates body ratios divided in four operational by presenting a wider and diffuse dark lateral taxonomic units (OTU). The OTUs are congruent to stripe, vocal sacs visible in males, and a dorsum the political limits of the states of Alagoas/Sergipe, with dark blotches; A. caerulodactylus differs by Bahia, Espı´rito Santo, and Rio de Janeiro, Brazil (for possessing blue coloration on digits and toe detailed distribution see Fig. 1). discs, and basal web between toes II, III and IV; A. conspicuus, A. crombiei, and A. gasconi present Diagnosis.—A small species of dendrobatid reduced fringes on toes. The abdomen is (maximum SVL 5 19 mm in males and granulated in A. conspicuus; A. crombiei have females); thumb slightly longer than second second and fourth fingers of same size (FL2 . finger (FL2 on average 85% of TBL); third finger FL4 in A. olfersioides) and dark blotches on not swollen in adult males (TFD 0.3 mm in dorsum; the males of A. gasconi and A. fuscellus average); throat, chest, and belly light cream, present a swollen third finger and sexual almost unpigmented; toes III–IV basally dimorphism in color with venter darker in webbed; dorsum brown, generally adorned males than females; A. goianus differs by the with a pattern of intercrossing Xs; pale dorso- absence of a conspicuous dark lateral stripe, toes fringed and webbed and dorsal color pattern formed by four reddish-brown blotches; A. granti differs by presenting dorsal color pattern uniform with brown flecks, anal flap, hind limbs with granular skin, reduced webbing between toes II, III, and IV (between III and IV in A. olfersioides), narrow and longer head (HL 88% of HW, HL 30% of SVL; 80% and 24% in A. olfersioides), larger hands (HAL 30% of SVL; 21% in A. olfersioides), and longer legs (TIL 47% of SVL; 44% in A. olfersioides); A. marchesianus differs from A. olfersioides by having a uniform dorsum scattered with darker spots, by the dark lateral stripe which is wider at the inguinal region and can be seen dorsally, by the thumb distinctively longer than the second finger (FL2 about 70% of TBL; 85% in A. olfersioides), and by the gray throat in males; A. masniger has a more robust body, the dorsum is uniformly brown with some dark spots in two parallel lines, and the males have a dark throat; A. nidicola differs from A. olfersioides by its larger maximum size (to 21.4 mm), presence of basal web between toes II, III, and IV, presence of FIG. 9. Ratios of thumb length by length of finger 2 a dark lateral stripe short and diffuse and throat (upper graphic) and length of finger 4 by hand length and chest black to light gray in males; A. (lower graphic) of Atlantic Forest Allobates studied for each locality. Squares are mean and bars are the sumtuosus differs by presenting reduced fringes standard errors for each sample. The localities are on toes II, III, and IV, and swollen third finger in presented with shortened names (see Appendix 2) males; A. trilineatus differs from A. olfersioides from north (left) to south (right). The names above the mainly by males presenting finger III and often graphic are the states to which the localities belong. finger II strongly swollen, venter stippled with TAXONOMIC REVIEW OF ALLOBATES FROM THE ATLANTIC FOREST 575 gray and by the dark lateral stripe along flank of HAL; inner metacarpal tubercle elliptical, that becomes diffuse at the midbody to the slightly smaller than the former; supernumerary inguinal region; A. vanzolinius differs from A. tubercle absent; one subarticular tubercle on olfersioides by its larger maximum size (to fingers I, II and IV and two on finger III; no 22.9 mm) and by males presenting a darker fringes on fingers; finger III not swollen; throat than females; Anomaloglossus beebei by webbing between fingers absent; tips of digits presenting dorsum uniformly brown with scat- expanded (width of finger about 75% of the tered dark dots, by the presence of median width of the disc); discs with distinct dorsal lingual process, swollen third finger, reduced scutes; thighs and tibia brown with two dark web between toes II, III, and IV, and by the stripes that form a continuum when hind limbs presence of a well defined tarsal fold; A. are in natural position; tibia length 44% of SVL; degranvillei has a pigmented and marbled relative length of appressed toes IV . III . V . venter, a more dorsally located lateral dark II . I; discs on toes larger than the width of toes stripe with borders less defined, a dark cloacal (as for fingers); rudimentary webbing present region, the median lingual process is present, only between toes III and IV; inner metatarsal the toes are fringed and a dermal fold is present tubercle elliptical, outer metatarsal tubercle at the outer margin of the feet; A. stepheni differs rounded; one subarticular tubercle on toes I by presenting a wider and robust body shape, and II, two on toes III and V, and three on toe median lingual process, a swollen digit III in IV; toes not fringed; metatarsal fold absent; males, toes II and III webbed and a conspicuous tarsus smooth with an inconspicuous tubercle; light dorsolateral stripe from eyelid to the tarsal fold absent. inguinal region; A. tamacuarensis differs by Measurements of the Specimen Described (mm).— reaching a larger size (to 25 mm), by presenting SVL 16.9; HL 4.1; HW 5.1; ED 2.4; END 1.5; IOD median lingual process, dorsal blotched pattern, 2.2; HAL 3.5; TBL 1.5; FL2 1.3; FL4 1.0; THL 7.8; inner tarsal fold and fringes along side of TIL 7.5; FOL 7.4. fingers; Hyloxalus peruvianus differs by reaching Color Variation in Preservative (N 5 880).—In a larger size (to 21 mm), by presenting finger II most specimens, the dorsum shows a pattern of and toe IV fringed, by the presence of an outer dark brown intercrossing Xs, with branches that tarsal fold, and toes II, III, and IV webbed. can vary in length and width in a lighter Description.—Based on MNRJ 5094 from the background (Fig. 4). It is uniformly light brown type locality Angra dos Reis, state of Rio de in some individuals where intercrossing Xs are Janeiro. Head slightly wider than long (HL 80% absent. A dark lateral stripe extending from the of the HW); head length 24% of the SVL; snout tip of the snout, through the eyes to the inguinal rounded in dorsal and lateral views, with nares slightly prominent located and opening lateral- region is always present. This stripe is narrower ly; upper jaw slightly projected; canthus ros- at the nares level, becoming gradually wider tralis rounded; loreal region slightly concave, through the lateral of the body narrowing again almost flat; internarial distance 39% of the HW; at the inguinal region. The tympanum is eye-nostril distance 62% of the eye diameter covered by the dark stripe dorsally (about (ED); tympanum 42% of the ED with the a third). The lateral stripe can be uniformly anteroventral margin distinct and the poster- dark throughout the length or present scattered odorsal obscured by the depressor musculature; white spots at the inguinal region, being diffuse tongue attached anteriorly; median lingual pro- toward the venter and inguinal region (Fig. 6). cess absent; vocal slits present; dorsal surface At the end of the stripe, there is an oblique smooth with low, inconspicuous granules re- narrow light stripe that divides it from the short stricted to sacral region; dorsal surfaces of thigh dark stripes present at the dorsum. The belly is and shanks smooth; although discolored there is light cream, and some tiny spots of dark brown indication of a pattern of intercrossing exes on pigmentation of some individuals can be seen in dorsum; a dark lateral stripe extends from the the throat and chest or through the entire belly. tip of snout, involving the nares and partially The arms and legs present the same background the eyes, to the groin; ventral surface smooth; color of the dorsum. The arms present some belly light cream; no anal flap above vent; no dark bands mainly at its outer margin, and the enlarged tubercles near anus; forelimb slender, dorsum of the thigh and tibia generally show skin smooth; dark bands present mainly at the two or three dark brown stripes that vary in outer margin of arms; no ulnar fold, hand width and form a continuum when the leg is length 21% of SVL; appressed thumb slightly flexed. The palms of hands and sole of feet are longer than finger II (FL2 87% of TBL); finger II pigmented as the general background color. At longer than IV (FL4 77% of FL2); relative finger the end of the fingers and toes, there are light size when appressed III . II 5 I . IV; outer transversal stripes. There are C-like light cream metacarpal tubercle nearly round, diameter 16% stains at each side of the cloacal aperture. The 576 V. K. VERDADE AND M. T. RODRIGUES color pattern does not seem to vary ontogenet- distance (IOD); eye diameter (ED); distance ically. between nostril and spiracle aperture (DNS); Color in Life.—Based on MZUSP 132193–95 and length of the upper (LUJ) and lower (LLJ) from Una, state of Bahia, and photographs of an jaw sheaths. individual from Porto Seguro, state of Bahia, an Body depressed (GBH 78% of GBW), elliptical individual from Maceio´, state of Alagoas in dorsal view, oval in lateral view; snout (Fig. 5), and an individual from Rio de Janeiro, broadly rounded in dorsal and ventral views, State of Rio de Janeiro (Izecksohn and Carvalho- oval in lateral view; eyes located dorsally, e-Silva, 2001:fig. 78). The iris is bronze. The oriented dorsolaterally; nares dorsolateral with general color pattern follows that described for rounded apertures, located twice as close to the preserved specimens. The light cream parts look tip of snout than to the eyes; spiracle single, more bright and white in live specimens, and the lateral, sinistral, aperture lateral; vent tube contrast between the background color and the medial with lateral displacement, aperture darker pattern is more evident. dextral; tail long (TAL twice as long as BL); tail Sexual Dimorphism.—The data obtained from fins low (DFH plus VFH 47% of MTH); ventral Tijuca, state of Rio de Janeiro suggested no fin near parallel to tail musculature, dorsal fin at marked sexual dimorphism. Females are slight- an angle of 30u to tail musculature; tail tip ly larger (SVL) than males (t df 5 39 522.15, P 5 narrowly rounded, forming an internal angle of 0.04; female 5 16.4 6 0.7 mm [N 5 20]; male 5 approximately 60u; origin of dorsal fin 0.5 mm 16.0 6 0.6 mm [N 5 20]) and present an anterior to the body terminus; origin of ventral immaculate belly more frequently. Fifty percent fin 0.6 mm posterior to the junction of the body of the females examined had an immaculate wall and the ventral margin of the tail muscle; belly, 35% a belly pigmented from throat to oral disc transversally elliptical, positioned chest, and 15% a belly pigmented from throat to anteroventrally, forming an angle of about 45u inguinal region. No males had an immaculate with the midsagittal plan; lateral emargination belly, 15% had a belly pigmented from throat to present; anterior labium with five papillae in % chest, and 85 a belly pigmented from the each side, separated by a large dorsal gap; throat to the inguinal region. Males and females posterior labium with 24 papillae with no gap; did not differ in any other variable. These papillae in a uniserial row, with pointed tip; two results agree with the general pattern of color straight labial ridges at the anterior labium, the dimorphism observed for the genus (Savage, first with a central gap; three straight labial 1968; Edwards, 1974; Coloma, 1995; Kok et al., ridges at the posterior labium, without gaps; 2006). each labial ridge with a single row of labial Advertisement Call.—The vocalization of Allo- teeth; dental formula 2(1)/3. bates olfersioides was described as an acute trill The preserved larvae are faded, but some (Lutz, 1954; Izecksohn and Carvalho-e-Silva, 2001), but unfortunately we were unable to find brown reticulations over the body that become any recordings of their calls at the collections rounded blotches toward the end of the tail are consulted (WCAB, ZUEC, and UFRJ) or with still visible. Izecksohn and Carvalho-e-Silva colleagues herpetologists. (2001) described the life color pattern of larvae Larval Definition (Fig. 10).—Based on 22 tad- as body blackish with a white spot at the end of poles (stages 25 to 41) and six metamorphs the snout. (stages 42 to 46) from Tijuca, state of Rio de Metamorphosis appears to take place with Janeiro (MNRJ 23729), reared in captivity by B. individuals measuring a SVL 7.4 6 0.09 mm on Lutz. Larval definition follows Altig and average (N 5 6), almost half the mean size of McDiarmid (1999), and the stages are identified adults. The hands of metamorphs present following Gosner (1960). All the qualitative supranumerary tubercles, one at the base of characters studied do not vary among the finger II, one at the base of finger III, and one at tadpoles. The characteristics presented belong the base of finger IV. Dorsal color pattern of to a single stage-36 individual. The measure- hands, thighs, and shanks are already visible ments taken were total length (TL); body length but not as contrasting as verified in adults. (BL); greatest body width (GBW), measured Although faded it is possible to see a more behind the eyes in dorsal view; greatest body intense pigmentation laterally on the body from height (GBH), measured at the plane of the the tip of snout to the inguinal region. The center of the coiled intestine; tail length (TAL); tympanum is not visible. tail musculature height (TMH); tail muscle Measurements of the Larvae Described (mm).— width (TMW); maximum tail height (MTH); TL 23.3; BL 7.8; GBW 4.5; GBH 3.5; TAL 16.2; dorsal (DFH) and ventral (VFH) fin height, TMH 2.1; TMW 2.4; MTH 3.2; DFH 0.8; VFH 0.7; measured at 6mm from the body terminus; and IND 1.7; IOD 2.1; ED 0.8; DNS 4.3; LUJ 0.8; LLJ ventral internarial distance (IND); interorbital 0.4. TAXONOMIC REVIEW OF ALLOBATES FROM THE ATLANTIC FOREST 577

FIG. 10. Tadpole of Allobates olfersioides (MNRJ 23729, Tijuca, Rio de Janeiro, state of Rio de Janeiro): (A) lateral and (B) dorsal views, and (C) detail of oral disc, ventral view.

Natural History.—Allobates olfersioides is di- we suggest that the tadpoles from Ilheús urnal and lives on the forest floor. When belonged to two individuals. The presence of disturbed, it immediately jumps and dives into mature eggs and developing follicles together the forest litter (Izecksohn and Carvalho-e-Silva, indicate that females reproduce more than once 2001). The tadpoles hatch in humid terrestrial in the reproductive period, as observed for nests and are carried by their parents to puddles other species of Aromobatidae (Junca´, 1998). of small rivulets on the forest floor where they Gravid females were observed in samples from feed until metamorphosis (Lutz, 1954; Boker- Ilheús (MZUSP 93874–913) in August; from mann, 1967; Izecksohn and Carvalho-e-Silva, Tijuca (MZUSP 93914–94021) in February, Au- 2001). Two nursing frogs captured at Una gust, and September; and from Represa Rio (MZUSP 132191, 132202) and two from Ilheús Grande, State of Rio de Janeiro (MZUSP 93664– (MZUSP 93872–73), Bahia were examined and 665) in April, suggesting reproduction through- found to be male. They carried tadpoles at out the year. Gosner stage 25 agreeing with reports for other Distribution.—Allobates olfersioides occurs in dendrobatids (Coloma, 1995; Junca´, 1998; Cald- coastal forests of the Atlantic Domain from sea well et al., 2002b). Unfortunately, we have level to about 1000 m a.s.l. The species can be information on the number of tadpoles carried found from the north of the state of Alagoas to for only one of the specimens from Una (12). the south of the state of Rio de Janeiro (between The frogs from Ilheús were in a bottle with 22 08u549S and 23u009S latitudinal degrees), with tadpoles. Previous reports on Allobates found a probable occurrence in Parque Estadual do eight to 48 tadpoles being carried by the parent Rio Doce, State of Minas Gerais. (Caldwell et al., 2002b). We found 14 of the 20 females dissected from Tijuca, municipality of Rio de Janeiro, state of Rio de Janeiro with DISCUSSION mature eggs in the ovaries. The females had The lack of information on the natural history a maximum of 11 mature eggs (about 1.5 mm and molecular phylogeography of Allobates from each) and mean of 8 6 1.4. Based on these data, the Atlantic Forest populations are important 578 V. K. VERDADE AND M. T. RODRIGUES barriers to obtaining a better picture of its providing literature, S. P. Carvalho-e-Silva and E. differentiation. Data on vocalization are anecdot- Izecksohn, for sharing their knowledge on al, and reported declines in the states of Espı´rito Allobates populations at the state of Rio de Santo and Rio de Janeiro (Weygoldt, 1989; Janeiro, G. Skuk for the picture of the live Izecksohn and Carvalho-e-Silva, 2001) render it specimen. We also thank the Brazilian environ- difficult to obtain vocalizations and tissue sam- mental organization IBAMA, for collection per- ples, delaying further discussions on possible mits. P. Narvaes, D. Pavan, and T. Grant read and geographical call and molecular variation. criticized the manuscript. This paper is a result As far as we know, the distribution of from the project ‘‘Revisaõ sistema´tica das espe´- Allobates olfersioides seems to be unrelated to cies de Colostethus da Mata Atlaˆntica’’ supported any hydrographic basin, and no river in the by FAPESP-FundaçaõdeAmparoa` Pesquisa do coastal region of Brazil appears to be a geo- Estado de Saõ Paulo process (97/13503–6) as graphic barrier. The presence of continuous fulfillment to the requirement of the master’s character gradients suggests uninterrupted degree in zoology by the Instituto de Biocieˆncias, gene flow between adjacent populations, at Universidade de SaõPaulo,SaõPaulo,Brazil. least until recently. Other amphibian populations widely distributed in the coastal area of Brazil seem to present similar patterns of LITERATURE CITED geographic variation. Pombal and Haddad AB’SABER, A. N. 1977. Domıńios Morfoclima´ticos e (1992); studying populations of Phyllomedusa provıńcias fitogeogra´ficas do Brasil. Orientaçaõ. burmeisteri baihana and P. b. burmeisteri from the Departamento de Geografia da Universidade de states of Rio de Janeiro, Espı´rito Santo, and Saõ Paulo 3:45–48. Bahia; found a continuum in the color pattern of ALTIG, R., AND R. W. MCDIARMID. 1999. Body plan development and morphology. In R. W. McDiar- the posterior surface of the thigh of the mid and R. Altig (eds.), Tadpoles. The Biology of individuals, with more complex character fre- Anuran Larvae, pp. 24–51. University of Chicago quency changes at the state of Espı´rito Santo Press, Chicago. (Silva-Filho and Junca´ [2006] recently attributed BOKERMANN, W. C. A. 1966. Lista Anotada das species status to these subspecies, but without Localidades Tipo de Anfı´bios Brasileiros. RUSP, considering variation in the state of Espı´rito Saõ Paulo, Brasil. Santo). A similar pattern was observed by ———. 1967. Novas espećies de Phyllobates do leste e Jackson (1978) while studying the pattern of sudeste brasileiro (Anura, Dendrobatidae). Revista character variation of lizards of the genus Brasileira de Biologia 27:349–353. Enyalius (Leiosauridae). Again, the populations CALDWELL,J.P.,AND A. P. LIMA.2003.Anew from the state of Espı´rito Santo were those Amazonian species of Colostethus (Anura: Dendro- batidae) with a nidicolous tadpole. Herpetologica presenting the largest amount of variation 59:219–234. (Jackson, 1978) compared to that along a north- CALDWELL, J. P., A. P. LIMA, AND C. KELLER. 2002a. south transect. Redescription of Colostethus marchesianus (Melin, These congruent geographical patterns may 1941) from its type locality. Copeia 2002:157–165. be a hint of historical events that occurred from CALDWELL, J. P., A. P. LIMA, AND G. M. BIAVATI. 2002b. the southern state of Bahia to the northern state Descriptions of tadpoles of Colostethus marchesianus of Rio de Janeiro. It may be caused by the and Colostethus caerulodactylus (Anura: Dendroba- continuity of forest habitats along the coast tidae) from their type localities. Copeia 2002:166– before deforestation or by a residual effect of 172. a complex introgression caused by the flooding COCHRAN, D. M. 1955. Frogs of southeastern Brazil. Bulletin of the United States National Museum of the habitats along the coast prior to the last 206:1–411. marine transgression (Suguio and Nogueira, COLOMA, L. A. 1995. Ecuadorian frogs of the genus 1999). Molecular phylogeographic studies Colostethus (Anura: Dendrobatidae). Miscellaneous might provide much needed evidence for the Publications, Museum of Natural History of the discussion of these hypotheses. University of Kansas 87:1–67. EDWARDS, S. R. 1974. A Phenetic Analysis of the Genus Acknowledgments.—We are indebted to the Colostethus (Anura: Dendrobatidae). Unpubl. Ph.D. following colleagues for the loan of specimens: diss., University of Kansas, Lawrence. E. Freire, E. M. Gonçalves, F. Junca´,F.M.Souza, FROST, D. R. 2006. Amphibian Species of the World: An H. Zaher, J. Jim, J. Pombal, M. Soares, O. L. Online Reference. Version 3.0. [cited 20 September, Peixoto, S. P. Carvalho-e-Silva, U. Caramaschi, 2006]. American Museum of Natural History, New York. Electronic database accessible at http:// and W. R. Heyer. We thank Hussam Zaher and research.amnh.org/herpetology/amphibia/index. C. Castro-Mello for allowing the access and html. logistics at the MZUSP; P. E. Vanzolini for GOSNER, K. L. 1960. A simplified table for staging providing important data from specimens from anuran embryos and larvae with notes on identi- W. C. A. Bokermann collection; V. R. Morales for fication. Herpetologica 16:183–190. TAXONOMIC REVIEW OF ALLOBATES FROM THE ATLANTIC FOREST 579

GRANT, T., AND F. CASTRO. 1998. The cloud forest SILVA-FILHO, I. S. N., AND F. A. JUNCA´ . 2006. Evidence of Colostethus (Anura, Dendrobatidae) of a region of full species status of the neotropical leaf–frog the Cordillera Occidental of Colombia. Journal of Phyllomedusa burmeisteri baihana (A. Lutz, 1925) Herpetology 32:378–392. (Amphibia, Anura, Hylidae). Zootaxa 1113:51–64. GRANT, T., AND L. O. RODRIGUEZ. 2001. Two new species SILVERSTONE, P. A. 1975. A revision of the Poison- of the genus Colostethus (Dendrobatidae) from Arrow Frogs of the genus Phyllobates Bibron in Peru and a redescription of C. trilineatus (Boulen- Sagra (Family Dendrobatidae). Bulletin of the Los ger, 1883). American Museum Novitates 3355:1–24. Angeles County Museum of Natural History GRANT, T., E. C. HUMPHREY, AND C. W. MYERS. 1997. The 27:1–53. median lingual process of frogs: a bizarre character SOCIEDADE BRASILEIRA DE HERPETOLOGIA. 2005. Lista das of old world ranoids discovered in South Amer- espećies de anfı´bios do Brasil. [cited 15 August, 2005]. ican dendrobatids. American Museum Novitates Sociedade Brasileira de Herpetologia, Saõ Paulo, 3212:1–40. Brazil, Electronic database accessible at http//: GRANT, T., D. R. FROST,J.P.CALDWELL,R.GAGLIARDO,C. www.sbherpetologia.org.br/checklist/anfibios.htm. F. B. HADDAD,P.J.R.KOK,D.B.MEANS,B.P. SUGUIO, K., AND A. C. R. NOGUEIRA. 1999. Revisaõ crı´tica NOONAN,W.E.SCHARGEL, AND W. C. WHEELER. 2006. dos conhecimentos geolo´gicos sobre a formaçaõ Phylogenetic systematics of Dart-Poison Frogs and (ou grupo?) Barreiras do Neo´geno e o seu possı´vel their relatives (Amphibia: Athesphatanura: Den- significado como testemunho de alguns eventos drobatidae). Bulletin of the American Museum of geolo´gicos mundiais. Geocieˆncias, SaõPaulo Natural History 299:1–262. 18:461–479. HEYER, W. R. 1988. On frog distribution patterns east WEYGOLDT, P. 1989. Changes in the composition of of the Andes. In P. E. Vanzolini and W. R. Heyer mountain stream frog communities in the Atlantic (eds.), Proceedings of a Workshop on Neotropical mountains of Brazil: frogs as indicators of envi- Distribution Patterns, pp. 245–273. Academia ronmental deteriorations? Studies on Neotropical Brasileira de Cieˆncias, Rio de Janeiro, Brasil. Fauna and Environment 243:249–255. IZECKSOHN,E.,AND S. P CARVALHO-E-SILVA. 2001. Anfı´bios do Municı´pio do Rio de Janeiro. Editora Accepted: 9 May 2007. da UFRJ, Rio de Janeiro, Brasil. JACKSON, J. F. 1978. Differentiation in the genera Enyalius and Strobilurus (Iguanidae): implications APPENDIX 1 for Pleistocene climatic changes in eastern Brazil. Arquivos de Zoologia, Saõ Paulo 30:1–79. Specimens Examined JUNCA´ , F. A. 1998. Reproductive biology of Colostethus Allobates brunneus: Brazil, Mato Grosso, Chapada stepheni and Colostethus marchesianus (Dendrobati- dos Guimara˜es: MZUSP 121875–82. Allobates goianus: dae), with the description of a new anuran mating Brazil, Goia´s, Chapada dos Veadeiros: MZUSP 76652 behavior. Herpetologica 54:377–387. (holotype), 73706, 76651 (paratypes). Allobates march- KOK, P. J. R., R. D. MACCULLOCH,P.GAUCHER,E.H. esianus: Brazil, Amazonas, Reservas INPA–WWF: POELMAN,G.R.BOURNE,A.LATHROP, AND G. L. MZUSP 67823–25. Allobates masniger: Brazil, Para´, LENGLET. 2006. A new species of Colostethus (Anura, Parque Nacional da Amazoˆnia: MZUSP 69166–67 Dendrobatidae) from French Guiana with a re- (paratypes); SW of Itaituba, Rio Tapajo´s: MZUSP description of Colostethus beebei (Noble, 1923) from 69168–71 (paratypes). Allobates olfersioides: BRAZIL, its type locality. Phyllomedusa 5:43–66. Alagoas, Maceio´: MUFAL 98735–36, 98739–40; Man- LIMA,A.P.,AND J. P. CALDWELL. 2001. A new gabeiras: MZUSP 73823–30 (A. alagoanus, paratypes), Amazonian species of Colostethus with sky blue 73707 (A. alagoanus holotype), 78191–78197; Manimbu: digits. Herpetologica 57:180–189. MZUSP 11994, 11997–33; Murici: MZUSP 76432; Novo LUTZ, A. 1925. Batraciens du Bre´sil. Comptes Rendus Dino: MNRJ 9910; Rio Largo: MUFAL 930809–810; Societe Biologie Paris XCIII, 137–139. Bahia, Cumuruxatiba: MZUSP 59433; Elı´sio Medrado: LUTZ, B. 1954. Anfı´bios anuros do Distrito Federal. UEFS 81; Ilheús: MNRJ 23767–70, 23728, MZUSP Memo´rias do Instituto Oswaldo Cruz 52:155– 81133, 93874–913, 93872–73; Porto Seguro: MZUSP 238. 63195–97, 126361–65, UFPB 866150, 866153–54, 866193, MORALES, V. M. 2000. Sistematica y biogeografia del 866196–97, 866212–13; Saõ Jose´: MZUSP 63495; Una: grupo trilineatus (Amphibia, Anura, Dendrobati- MZUSP 63719–21, 125848, 132188–206; Uruçuca: dae, Colostethus), con descripcioń de once nuevas MZUSP 26916–18, 132187; Espı´rito Santo, Baixo especies. Publicaciones de la Asociacion de Ami- Guandu: MZUSP 35674; Linhares: MZUSP 93871; gos de Donana 13:5–59. Refu´gio Sooretama: MZUSP 76628–76629 (A. capixaba, POMBAL, J. P., Jr., AND C. F. B. HADDAD. 1992. Espećies paratype), 73752 (A. capixaba, holotype); Santa Teresa: de Phyllomedusa do grupo burmeisteri do Brasil MZUSP 53559–60; USNM 208141, 208147, 200452–53; oriental, com descriçaõdeumaespećie nova Rio de Janeiro, Angra dos Reis: MNRJ 5094; Araruama: (Amphibia, Hylidae). Revista Brasileira de Biologia MNRJ 4115, 15398–406, 4143, 15431–39; Imbiara: MNRJ 52:217–229. 23716–19; Itaguaı´: EI 4321–56, JJ 7787, 7792–96, ZUEC RIVERO, J. A. 1988. Sobre las relaciones de las especies 2741–42, MNRJ 23724–27; Represa Rio Grande: MZUSP del genero Colostethus (Anura, Dendrobatidae). 73714 (A. carioca, paratype), 73753 (A. carioca, paratype), Memoria da Sociedad de Ciencias Naturales La 76653 (A. carioca, holotype), 93651–65, 93671; Sacra Salle XLVIII, 3–32. Famı´lia do Tingua´: EI 8598–600, JJ 7788–91, ZUEC SAVAGE, J. S. 1968. The dendrobatid frogs of Central 10824; Serra da Piedade: MNRJ 23730–34; MZUSP America. Copeia 1968:745–776. 9801–9803; Tereso´polis: MNRJ 23721–23, MZUSP 580 V. K. VERDADE AND M. T. RODRIGUES

53448–51, USNM 208396–208401; Tijuca: JJ 224–231, Largo: 09u299S, 35u509W; Bahia, 15, Cumuruxatiba: MZUSP 76418, 76728–29, 93624–49, 93666–767, 93914– 17u069S, 39u119W; 8, Elı´sio Medrado: 12u569S, 39u319W; 94377; Tingua´: EI 4290, 4306; Sergipe, Areia Branca: 10, Ilheús:14u489S, 39u039W; 14, Porto Seguro: 16u269S, MZUSP 88954. Allobates sumtuosus: Brazil, Para´, Re- 39u059W; 11, Saõ Jose´:15u099S, 39u189W; 12, Jussari, serva Biolo´gica Rio Trombetas: MZUSP 69157–60 Serra do Teimoso: 15u099S, 39u319W; 13, Una: 15u179S, (paratypes). Anomaloglossus beebei: French Guian, Saut 39u059W; 9, Uruçuca: 14u339S; 39u189W; Espı´rito Santo, Mais: MZUSP 58230. Anomaloglossus degranvillei: 18, Baixo Guandu: 19u319S, 41u029W; 17, Linhares: French Guian, Sau¨ l: MZUSP 58231. Anomaloglossus 19u239S, 40u049W; 16, Refu´gio Sooretama: 19u009S, stepheni: Brazil, Amazonas, Manaus: MZUSP 64569 40u009W; 19, Santa Teresa: 19u569S, 40u379W; Rio de (holotype), 64570 (paratype). Janeiro, 29, Angra dos Reis: 23u009S, 44u199W; 20, Araruama: 22u529S, 42u209W; 22, Imbiara: 22u429S, 42u379W; 26, Itaguaı´:22u489S, 43u419W; Rio de Janeiro, APPENDIX 2 24, Represa Rio Grande: 22u559S, 43u259W; 28, Sacra Gazetteer (State, code number used at Fig. 1, specific Famı´lia do Tingua´:22u309S, 43u379W; Rio de Janeiro, locality, approximate coordinate).—Alagoas, 4, Maceio´: 27, Serra da Piedade: 23u009S, 43u399W; 21, Tereso´po- 09u409S, 35u419W; 6, Mangabeiras: 09u569S, 36u059W; 5, lis: 22u269S, 42u299W; Rio de Janeiro, 25, Tijuca: Manimbu: 09u529S, 36u099W; 2, Murici: 09u189S, 22u489S, 43u329W; 23, Tingua´:22u339S, 43u269W; 35u569W; 1, Novo Dino: 08u549S, 35u389W; 3, Rio Sergipe, 7, Areia Branca: 10u469S, 37u199W.