2000 - Napoli & Caramaschi Dendropsophus elianeae

2000 - Napoli & Caramaschi Dendropsophus elianeae

(Parte 1 de 3)

Description and variation of a new Brazilian species of the Hyla rubicundula group (Anura, Hylidae)

Marcelo Felgueiras NAPOLI & Ulisses CARAMASCHI

Departamento de Vertebrados, Museu Nacional/UFRJ,

Quinta daBoa Vista, 20940-040 Rio de Janeiro,RJ,Brazil E-mail: napoli@acd.ufrj.br

A new member of theHyla rubicundulaspecies group is described fromBelaVista,MatoGrosso do Sul, Brazil. The new species inhabits "cerrado" formations in central and southeastern Brazil, comprising three main geographic morphotypes. Intrapopulation and interpopulation variation is characterized in order to accurately determine the species limits. The analysis was based on external morphology and morphometrics. The new Hylais characterized by being the largest in theHyla rubicundulaspecies group, having a stout body, lateral limits of dorsum under the lower border of the tympanum, immaculate dorsum, and rounded canthus rostralis.

TheHyla rubicundulaspecies group was defined byNAPOLI & CARAMASCHI (1998, 1999) and is known to comprise six species:H. rubicundulaReinhardt & Lütken,1862;H. tritaeniata Bokermann,1965;H. anataliasiasiBokermann,1972;H. araguayaNapoli & Caramaschi, 1998;H. cerradensisNapoli & Caramaschi,1998;andH.cachimboNapoli & Caramaschi, 1 9.

Hyla rubicundulahas the widest geographic distribution of all members of the group, occurring in southeastern, northeastern and central Brazil(NAPOLI & CARAMASCHI,1999). Specimens from the Brazilian states ofMatoGrosso,MatoGrosso do Sul and São Paulo are easily diagnosed from the topotypic samples(LagoaSanta, MinasGerais)by their greater size, robust body, distinct head shape, lateral limits of dorsum under the lower border of tympanum, and immaculate dorsum. These diagnostic characters suggest the presence of a new species.

Herein we describe a new species of theHyla rubicundulagroup and report on intra- and interpopulation variation of this new species, in order to furnish accurate species limits.

Specimens used for descriptions or examined for comparisons are deposited in the following Brazilian collections: Célio F. B. HADDAD Collection, deposited in the Departamento de Zoologia, UniversidadeEstadual Paulista,RioClaro(CFBH); Departamento de Zoologia, UniversidadeEstadual Paulista, São Josédo RioPreto(DZSJRP); Eugenio

IZECKSOHN Collection, deposited in the Departamento de Biologia Animal, Universidade Federal Rural do Rio de Janeiro (EI); Jorge JIM Collection, deposited in the Departamento de Zoologia, UniversidadeEstadual Paulista,Botucatu (J);MuseuNacional, Rio de Janeiro (MNRJ);Museude Zoologia, Universidade deSãoPaulo (MZUSP);Museude História Natural, Universidade Estadual de Campinas (ZUEC).

We examined only adult males because of the rareness of females and juveniles in the samples. Larvae and advertisement calls are unknown. The specimens were sorted into morphotypes by similar morphology and proximity of localities. The measurements used in the account follow NAPOLI & CARAMASCHI (1998, 1999) and are in millimeters: SVL (snoutvent length); HL (head length); HW (head width); ED (eye diameter); UEW (upper eyelid width); IOD (interorbital distance); IND (internarial distance); TD (tympanum diameter); END (eye to nostril distance); NSD (nostril to tip of snout distance); UAR (upper arm); FAR (forearm); HAL (hand length); 3FD (third finger disk diameter); THL (thigh length);

TL (tibia length); FL (foot length); and 4TD (fourth toe disk diameter). Webbing formula notation follows SAVAGE & HEYER (1967) as modified by MYERS & DUELLMAN (1982).

Discriminant function analysis was used to analyze inter-population variation (MARCUS, 1990), but without applying any method to remove the size effect in the groups (Reis et al., 1990). The groups under study were defined a priori, and the eigenvectors and their associated eigenvalues were obtained from a variance-covariance matrix. The degree of contribution of each morphometric character was determined by correlations between the original variables and the scores (loadings). The principal component analysis, "shearing method" (HUMPHRIES et al., 1981) and the multivariate allometric coefficients (JOLICOEUR, 1963) were calculated in order to characterize differences in shape by size. We used a t-test to compare mean values between two morphometric characters of the same species (SOKAl. & ROHLF, 1981).

Hyla elianeaen.sp. (fig. 1-2)

Holotype. -MNRJ 17297, adult male, collected atBelaVista (ca. 22°06'S, 56°31'W), 180 m above sea level, State ofMato GrossodoSul,Brazil, between 20 and 23 January 1967, by S. G.NUNES,S.T. ALBUQUERQUE, J. JIM and E. IZECKSOHN.

Paratopotypes. -Fifty-six adult males and five adult females (MNRJ 17209, 17226-17234; EI 8884-8934), collected with the holotype.

Fig. 1- Hylaelianeae,MNRJ 17297, holotype. (A) Dorsal and (B) lateral views of head; ventral views of (C) hand and (D) foot.

yla elianeae. (A) MNRJ 17204, Botucatu, São Paulo; (B)

MNRJ 17296, Cachoeira do Marimbondo, São Paulo; (C) MNRJ 17297, holotype, Bela Vista, Mato GrossodoSul.

Diagnosis. -Species characterized by the following combination of traits: (1) in life, dorsum consistently green; (2) in preservative, dorsal surfaces pink to violet; (3) snout-vent length

20.0-25.5 m SVL in males and 25.0-26.0 m in females; (4) dorsum imaculate: (5) body robust; (6) lateral limits of dorsum under the lower border of tympanum (fig. 3B1): (7) canthus rostralis rounded (fig. 1A-B).

Fig. 3. - Standards for lateral limits of dorsum. (B1) under the lower border of tympanum; (B2) above the upper border of tympanum. Pattern B1 occurs inH. elianeae,while pattern B2 occurs in the other species of theH. rubicundulagroup.

The absence of a single mid-dorsal sacral stripe inHyla elianeaedistinguishes this species fromH. tritaeniata, H. araguayaandH. cerradensis,which have such a pattern. No specimen of H. elianeaehas two anterior divergent dorsal brown stripes fused with two brown sacral stripes, while many individuals ofH. anataliasiasihave such a pattern. Furthermore, a broad and irregular dorsolateral stripe, with or without an upper white pinkish stripe is present in some specimens ofH. elianeae,distinguishing this species fromH. anataliasiasi,which has thin and well marked stripes. The lateral limitsofdorsum inH. elianeaeare under the lower border of the tympanum (fig. 3B1), whileH. anataliasiasi, H.cachimbo,H. tritaeniata, H. araguaya, H. cerradensisand the specimens ofH. rubicundulafrom State of MinasGerais always have these above the tympanum (fig. 3B2). The absence inH. elianeae ofa thin white stripe on the edges of tibia, above a thin brown stripe, distinguishes it fromH. rubicundula, H. anataliasiasi,H. tritaeniata, H. araguayaandH. cerradensis,which may have these stripes.

The presence inH. elianeaeof a broad white pinkish stripe above a brown canthal stripe distinguishes it fromH. anataliasiasi, H. tritaeniata, H. araguayaandH. cerradensis,which have the canthus well defined by a thin white stripe above a brown stripe. The head ofH. elianeaeiswider than long, its width contained about 3.4 times in the SVL, distinguishing it from H. anataliasiasi,which has the head longer than wide, its width contained about 3.6 times in the SVL. The snout inH. elianeaeis rounded, truncate or slightly acuminate, while in

H.cachimboit is strongly protruding. The canthus rostralis in specimens ofH. elianeaefrom central Brazil and northern State ofSãoPaulo is rounded and not well defined. The canthus rostralis of other species of theH. rubicundulagroup is flat and well defined. The robust body and the maximum SVL inH. elianeae(20.0-25.5 m in males; 25.0-26.0 m in females) distinguish this species fromH.cachimbo(19.8-21.0 m in males; 24.2 m in female),H.

rubicundula(18.0-23.4 m in males; 21.6-25.1 m in females),H. anataliasiasi(16.0-21.8 m in males; 16.6-21.6 m in females),H. tritaeniata(17.6-20.9 m; 20.6-21.5 m in females), H. araguaya(18.9-20.5 m in males) andH. cerradensis(18.9-19.3 m in males).

Description. -This description is based on specimens fromBelaVista,MatoGrosso do Sul only, which is the best sampled and most morphologically homogeneous population. The other geographic samples will be characterized within the geographic variation section. Descriptive statistics for the type series are furnished in tab. 1.

Body robust; head wider than long(t = 2.46, n =6, P <0.0001); internarial distance smaller than eye-nostril distance (t = 7.95,n= 6,P< 0.0001) and eye diameter(t= 45.5,n = 6, P < 0.0001); eye diameter greater than eye-nostril distance(t=36.6,n= 6,P < 0.0001); snout truncate, acuminate or rounded in dorsal outline, and truncate or rounded in lateral outline; nostrils slightly protuberant, directed laterally or slightly forward, sometimes located on small raised areas, making the internarial region appear furrowed; canthus rostralis rounded, not well defined; loreal region oblique; eyes slightly or moderately promi- nent; tympanum distinct, nearly circular; supratympanic fold sometimes present, covering partially the tympanum; vocal sac single, subgular; tongue cordiform or ovoid; vomerine teeth present in two patches between choanae, irregular in shape and position. Forearm more robust than arm, shorter(t =32.86,n= 65,P <0.0001); hands with a distinct palmar tubercle

(fig. 1 C); prepollex distinct; subarticular tubercles rounded; distal tubercle of third and fourth fingers sometimes bifid; supernumerary tubercles present, third finger disk diameter greater than fourth toe disk(t = 4.92, n =65,P <0.00001); modal webbing formula I 2 1/2 - 2 3/4 I 2+

-3 1/4 I 3 - 2 1/4 IV Legs moderately robust; femur and tibia approximately same size and length(t =0.17,n =6,P <0.87); sum of femur and tibia lengths smaller than SVL (t=8.57, n =6, P < 0.0001); foot with subarticular tubercles always rounded (fig. 1D); supernumerary tubercles variable in shape and number; toe disks slender; modal webbing formula, I 1 1/2 - 2 1/4 II1+ -2 1/4 I 1 1/2- 2 1/4 IV 2 1/2 - 1+ V.

Color in preservative. -Dorsum reddish to pinkish brown, immaculate; canthus rostralis, although not well defined, sometimes delimited inferiorly by a brown stripe; loreal region with a variable degree of melanization; a dorsolateral brown stripe sometimes present on the flanks, extending from the posterior corner of orbit to near the groin, sometimes bordered above by a white pinkish stripe; thigh light brown, immaculate; a brown stripe sometimes present on the anterior and posterior edges of the upper surface of tibia, in addition to random dorsal dark brown dots. Ventral surfaces immaculate huff.

Measurements of holotype. -SVL 25.5; HW 7.5; HL 7.3; ED 2.4; UEW 1.6; IOD 2.0; END 1.8; IND 1.7; THL 1.5; TL 12.2; TD 1.1; NSD 1.3; UAR 6.7; FAR 4.3; HAL 7.3; 3FD 0.9; FL 17.9; 4TD 0.8.

Geographic variation.- Three main geographic morphotypes were diagnosed (fig. 2), and were named and coded as follows (each entry in coded series contains the code, the code name, the number of specimens analyzed, and the localities where the samples came from): (1) SSP,

"southern São Paulo", n = 58 ( MINAS GERAIS: Belo Horizonte; SÃO PAULO: Botucatu, Campinas,Corumbatai, Itapeva, Jaguariúna, Paulinia, Pindamonhangaba, Pirassununga andSumaré);(2)NSP, "northernSãoPaulo",n =37 (SÃOPAULO:Icém, Indiaporã and

Pirassununga); (3) CBR, "central Brazil",n =72 (Goiás: Rio Verde;MATOGROSSO:Cáceres; MATOGROSSODOSUL:BelaVista and EstânciaCaiman). Descriptive statistics of each morphotype are furnished in tab. 1-2.

The specimens from southernSãoPaulo (morphotype SSP) differ from the type specimens (morphotype CBR) by the presence of the following characteristics: dorsum sometimes purple and sometimes with additional melanization; absence of a lateral white pinkish stripe; snout profile sometimes protruding; canthus rostralis distinct; dorsal head shape patterns

A1-A3 (fig. 4; tab. 3); modal palmar webbing formula, I 2 1/2- 2 1/2 II2--2 1/4 III3--2+ IV; legs slender; modal plantar webbing formula, I 1 1/4- 2+ II1+- 2 1/4 I 1 1/2- 2+ IV 2+ - l+V.

Fig. 4. - Standards for dorsal head shape patterns. (Al) and (A3) southernSãoPaulo (morphotype

SSP); (A4) and (A5) northernSãoPaulo (morphotype NSP): (A6) central Brazil (morphotype CBR); pattern (A2) occurs among the smallest specimens within the three morphotypes, including adults and juveniles.

The specimens from northernSãoPaulo (morphotype NSP) differ from the type specimens (morphotype CBR) by the presence of the following characteristics: dorsum with additional melanization; head as long as wide (t= 0.62,n =23,P =0.53); snout acuminate or truncate in dorsal outline, rounded or slightly truncate in lateral outline; canthus rostralis distinct and rounded; eyes very prominent; dorsal head shape patterns A4-A5 (fig. 4: tab. 3), prepollex slightly developed; modal webbing formulae, I 2 3/4 -2 3/4 I 2- - 3 1/2 II 3 - 2 1/2 IV; legs slender; femur and tibia approximately with the same size, femur longer than tibia (t = 2.18,n= 2, P =0.03); supernumerary tubercles indistinct; modal plantar webbing formulae, I1 1/4 -2 1/4 I 1+- 2 1/4 II1 1/4 - 2 1/4 IV 2+- 1+V.

Multiple discriminant function analysis was used to analyze morphological variation among morphotypes SSP, NSP and CBR, using 18 measurement variables (fig. 5). There was no statistical difference among mean values of SVL for the three morphotypes analyzed, which implies similar "size" among the samples. Two significant canonical axes (Wilks P <

0.05) resulted from this analysis, representing 100 %, of the total variation. Both canonical vectors alternated positive and negative coefficient values (tab. 4A), which implies variation in "shape" (HUMPHRIES et al., 1981). However, SVL had the fourth largest significant loading in

Table 4. - A, standardized coefficients and loadings from discriminant function analysis for morphometric characters of combined samples frommorphotypes SSP, NSP and CBR (fig 5). B, standardized coefficients from principal component analysis ("shearing method") of combinedsamples areas of morphotype SSP (fig. 8A). C, standardized coefficients and loadings from discriminant function analysis for morphometriccharacters of the combined geographic samples of morphotypes SSP and NSP (fig. 9). D, standardized coefficients from principal componentanalysis ("shearing method") of combined dorsal head shape patterns A2 and A6 of morphotype CBR (fig. l0A).r,Pearson correlation coefficient of the original data with the scores obtained from the discriminant function analysis (loadings); ns, notsignificant; * P< 0.05;

** P< 0.02;

*** P < 0.01.

the first canonical vector (tab. 4A), which implies that this axis is strongly weighted by "size", while SVL is not significant in the second canonical vector. The morphotypeNSP is best discriminated from the others, but morphotypes SSP and CBR are only partially discriminated from each other. This result supports the existence of three main geographic morphotypes, further analysed below.

Variation within SSP-The specimens from southernSãoPaulo (morphotype SSP) were grouped into three areas(Ia, IIaand IIIa) based on geographic proximity and morphological similarity, in order to analyze morphological variation throughout the "cerrado" corridor of SãoPaulo (fig. 6).

Distributions of dorsal head shape patterns were determined for each of the three areas and the observed frequencies were tested against expected frequencies (based on frequency of distribution for entire sample of SSP) with a chi-square test (c2).Areas IIa and IIIahad to be combined in order not to violate minimum cell-size requirements for c2 analyses (SOKAL &

ROLHF, 1981). A statistically significant difference among areas Ia and IIa/IIIa was observed caused by different mean values for SVL between areasIaand IIa/IIIa(Ia= 23.62; IIa/IIIa =

2.01;t=6.09, df = 37,P < 0.001), perhaps representing ontogenetic changes. In order to verify this hypothesis, we analyzed the combined sample areas Ia and IIa/IIIaby principal component analysis, using the "shearing method" ofHUMPHRIESet al. (1981). The areasIa and IIa, which lack differences in SVL means, were combined in order to increase the number of specimens per area. The individual scores were projected in the space of the first two axes (fig. 8A). The coefficients of the second component have both positive and negative values

(Parte 1 de 3)