A new frog of the Leptodactylus fuscus species group (Anura: Leptodactylidae), endemic from the South American Gran Chaco

A new species of Leptodactylus frog (Anura: Leptodactylidae) from the South American Gran Chaco, morphologically similar and previously confused with the widespread Leptodactylus mystacinus, is described through the use of multiple sources of evidence (molecular, external morphology, coloration, osteology, bioacoustics, and behavior). The phylogenetic analysis with partial sequences of mitochondrial rDNA genes (12S and 16S) recovered the new species within the L. fuscus group, being highly divergent (>3% genetic distance in 16S). The new species was recovered as sister taxa of L. mystacinus, from which it is distinguished by tympanum coloration, cephalic index, dorsum and legs coloration, and some osteological differences in nasals and prevomers. This new frog is characterized by a moderate body size (SVL 46.80–66.21 mm), distinctive color pattern (reddish dorsal surfaces of body with noticeable black stripes in the dorsolateral folds), a circular and dark tympanum with dark tympanic annuli, and behavior of males that call on top of fallen logs and tree branches close to the ground.


INTRODUCTION
The South American Gran Chaco ecoregion includes portions of northern and central Argentina, southeastern Bolivia, western of Paraguay, and a small area in southwestern Brazil. Temperature and rainfall in this region contribute to an increasing gradient of aridity from East to West, which have led to a distinction between Humid and Dry Chaco subregions (Dinerstein et al., 1995;Olson et al., 2001). The Dry Chaco occupies an area of nearly 225,500 km 2 , extending over a large part of the northwestern portion of the morphology, coloration, osteology, behavior, and genetic distances. In addition, we performed a phylogenetic analysis under Maximum Parsimony (MP) and Bayesian approach in order to test its phylogenetic position.

Molecular procedures
We obtained new sequences for 23 specimens previously assigned to L. mystacinus, L. cupreus, and L. bufonius, 20 of which were collected by us, and three obtained from biological collections. Other sequences used in this work come from GenBank (Appendix I).

Phylogenetic analyses and genetic distances
In order to test the distinctiveness and phylogenetic position of the new species, we obtained DNA sequences of two mitochondrial genes (12S-16S » 1,641 bp) from several individuals and localities of this taxon, also of L. mystacinus and two additional species of the L. fuscus group to complete the sampling: L. cupreus and L. bufonius (Appendix I). The phylogenetic analyses included sequences of all species in the L. fuscus group provided by De Sá et al. (2014) available from GenBank (Appendix I). We excluded the 16S sequence of L. fragilis employed by these authors (KM091585) because their identity is doubtful (BLAST searches resulted in 92% identity with sequences of Smilisca baudinii). We used only the available 12S fragment of L. fragilis (KM091469). Additional doubtful or missing sequences were excluded of the analyses, or exchanged for others available in GenBank (see details in Appendix I).
The phylogenetic analyses were performed under MP and Bayesian approach. The first was done with TNT v1.1 (Goloboff, Farris & Nixon, 2008), using gaps as missing data, and New Technology search with 1,000 random addition sequences. The branch supports were tested with 1,000 pseudo-replicates obtaining absolute frequencies of Bootstrap and Jackknifing, the latter with removal probability of 36% (Farris et al., 1996).
To construct trees under Bayesian approach we firstly used PartitionFinder v1.1.0 (Lanfear et al., 2012) to infer the best partition scheme and the evolutionary model that best fitted each partition. We analyzed 12S and 16S genes separately and selected models under the Akaike information criterion (Akaike, 1973) using the "greedy" heuristic search algorithm and linked branch lengths. The best model of evolution for both genes was GTR (General time reversible) + I (proportion of invariable sites) + G (gamma distribution). We performed the analysis in MrBayes v3.2.6 (Ronquist & Huelsenbeck, 2003) on XSEDE in CIPRES Science Gateway Webserver (Miller, Pfeiffer & Schwartz, 2010), implementing the inferred model of nucleotide substitution on two independent runs, each one with four chains sampling every 10,000 generations, for 100 million generations and discarding the first 25,000 trees as burn-in. We verified convergence with TRACER v1.5 (Rambaut & Drummond, 2007) and by examining the standard deviation of split frequencies between independent runs (<0.01).
For the estimation of sequence divergences within the L. fuscus group, including Leptodactylus sp. nov. (Appendix I), uncorrected pairwise distances (p-distances) of a partial fragment of the mitochondrial 16S rDNA gene were calculated in PAUP Ã v4.0b10 (Swofford, 2000). The sequences were aligned using MAFFT, under the strategy G-INS-i with default parameters for gaps opening and extension, obtaining a matrix of 478 bp.

Advertisement calls
We analyzed 241 advertisement calls of five males (IIBP-H 728, LGE 8085, and three unvouchered specimens) of the new species and 390 advertisement calls obtained from 13 males of L. mystacinus (LGE 15209-10, and 11 unvouchered specimens). Advertisement calls were recorded with a Sony WMD6C recorder, and a Sennheiser ME66/K6 directional microphone. The recordings are deposited in the sound collection of the LGE. In addition, we analyzed 850 advertisement calls of 41 males of L. mystacinus provided by Fonoteca Neotropical Jacques Vielliard (UNICAMP), Fonoteca Zoológica of the Museo Nacional de Ciencias Naturales (Madrid), and EcoRegistros (Argentina) (details in Appendix III). All recordings were analyzed employing Sound Forge Pro v11.0 (Magix GmbH & Co, Berlin, Germany) with an FFT of 512 points, at a sampling rate of 44.1 kHz and 16-bit precision. The following traits were measured: call duration (miliseconds, ms), intercall interval (ms), call rate (calls/second), and dominant frequency (hertz, Hz), following Heyer et al. (1990).

Nomenclatural acts
The electronic version of this article in portable document format will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is: urn:lsid:zoobank.org:pub:7AE8062D-2F27-4F99-AC5E-8D2A94874EF3. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central, and CLOCKSS. Specimens belonging to the L. mystacinus species complex were recovered nested in a clade along with L. cupreus, L. troglodytes, and L. bufonius, but forming two independent lineages, strongly supported ( Fig. 1B): lineage (I) specimens from the Dry Chaco and surroundings areas, and lineage (II) specimens from the rest of the distribution currently known for L. mystacinus. Both lineages have an allopatric geographic distribution and correspond to the two species distinguished in addition by morphology and coloration.

Molecular phylogenetic analysis
The Bayesian approach showed some subtle differences in topology with respect to the MP analysis (Appendix IV), but in both analyses Leptodactylus and the L. fuscus group were recovered as monophyletic. However, this relationships of the L. fuscus group and the other species groups remain unresolved with this approach. The relationships within the clade containing the new species were consistent with those recovered under MP.
The new species showed significant genetic divergences (p-distances) in the 16S rDNA sequences, which ranged 3-4.4% with the closely related L. mystacinus, and 5.7-12.5% with the remaining species of the L. fuscus group (Appendix V). Furthermore, the new species showed an intra-specific variability up to 1.3%.

Assignation of available names
Leptodactylus mystacinus (Burmeister, 1861). Type locality: "Rozario" (= Rosario, Santa Fe province, Argentina). Type MLU unnumbered, according to . We examined the type specimen through high quality color photographs (Appendix VI). Although we did not have topotypic specimens to include in our analysis, we examined specimens from nearby localities (Los Nogales and Las Rosas; Santa Fe province; 86 and 95 km, respectively, from Rosario). Their morphology and the location of Rosario, deeply nested within the geographic distribution of lineage II of our phylogenetic analysis, allowed us to assign the name L. mystacinus to this widespread lineage.
Cystignathus labialis Cope, 1877. Type locality: unknown, "the precise habitat of this species is at present uncertain. It is probably a part of Sumichrast's Mexican collection" (Cope, 1877). Type series USNM 31300-5, according to Kellogg (1932). Specimen USNM 31302 considered holotype by Cochran (1961), which is actually a lectotype designation, according to Frost (2019).  studied in detail the type series and considered them conspecific with L. mystacinus based on morphology, despite being very faded juvenile specimens.  also discussed the origin of these specimens, noticing that Cope (1877) published in that same paper the study of a collection of amphibians and reptiles of origin "unknown, but supposed to be the Argentine Confederation," and proposed that the type series of C. labialis could have been part of this collection. High quality photos of four specimens of the type series (USNM 31302, lectotype; and USNM 31303-5, paralectotypes; Appendix VII) of C. labialis were analyzed by us. We agree with  that based on their morphology and Cope's original description, they closely resemble L. mystacinus sensu lato. Based on the diagnostic characters identified by us (see below), we tried to assign these specimens to one of the two lineages recovered in our phylogenetic analysis. Since the specimens are faded and the diagnostic characters cannot be deduced from Cope's original description, coloration pattern is not useful for this task. Fortunately, their morphometric indexes (Cephalic index = CI = HL/HW, and relation between tympanum and eye diameter = TYD/ED) could be calculated from the photographs, and they overlap with juveniles of L. mystacinus, while significantly differ from juveniles of Leptodactylus sp. nov.: C. labialis: CI 0.98-1.01 (0.99 ± 0.01), TYD/ED 0.51-0.61 (0.57 ± 0.04); L. mystacinus: CI 0.95-1.08 (1.01 ± 0.06), TYD/ED 0.49-0.67 (0.59 ± 0.07); Leptodactylus sp. nov.: CI 0.80-0.89 (0.84 ± 0.03), 0.43-0.48 (0.45 ± 0.024). Regarding this evidence, C. labialis still has to be considered a junior synonym of L. mystacinus, leaving lineage I of our phylogenetic analysis without an available name.

Species description
Leptodactylus apepyta sp. nov.   Etymology The specific epithet is an indeclinable noun, constructed from the words of the Guaraní language apé (=back of the neck, dorsum) and pytã (=red), in reference to the intense brick red dorsum of adult and juvenile live specimens.

Definition and diagnosis
Leptodactylus apepyta sp. nov. is assigned to the L. fuscus group (sensu Heyer (1969b)) by its phylogenetic position, and by the presence of the following synaphomorphies (Ponssa, 2008): (1) tectum nasi and alary process of premaxilla at the same level; (2) frontoparietal with posterior margin convex, and (3) cultriform process of parasphenoid sited between neopalatines.
The new species is diagnosed within the L. fuscus group by the following combination of character states: (1) moderate size sensu Heyer & Thompson (2000) (SVL 46.80-61.41 mm in males; 51.67-66.21 mm in females); (2) robust body aspect in dorsal view; (3) head wider than long (CI 0.77-0.95); (4) small, circular, and dark tympanum, with dark tympanic annuli; (5) black broad stripe from tip of snout to the insertion of the forelimb; (6) a distinct light upper lip stripe; (7) one or two pairs of dorsolateral folds, with distinct uninterrupted dark stripes coincident with the upper pair, and interrupted dark stripes in the dorsolateral folds of the flanks; (8) reddish color on dorsal surfaces of body and limbs; (9) dorsum with small dark spots; (10) thigh, tibia, and tarsus with broad, diffuse, and dark bars; and (11) advertisement call composed by a single, short (30-68 ms), and non-pulsed note; call rate of 3.86-7.69 calls/s, without harmonic structure and with dominant frequency between 2,155 and 2,457 Hz; (12) males usually call on the top of fallen logs and low branches of trees.    Heyer, 1983). Leptodactylus apepyta sp. nov. and L. mystacinus cannot be distinguished from each other by their advertisement call, both species have similar structure and overlapping measured variables ( Fig. 4; Table 2). The advertisement calls of L. mystacinus (Figs. 4C and 4D), consist of a series of single non-pulsed notes, produced at a rate of 6.08 calls/s. Call duration ranges from 27 to 66 ms (43.62 ± 0.74 ms), and intercall interval ranges from 84 to 221 ms (124.60 ± 2.31 ms). The advertisement call lacks amplitude and frequency modulation, and dominant frequency ranges from 1,960 to 2,874 Hz (2,368.88 ± 195.05 Hz).

Description of the holotype
Adult male. Robust body aspect, moderate size (SVL 52.30 mm), head wider than long (CI 0.9), HL 34% of SVL. Snout sub-elliptical in dorsal view, and protruding in lateral view. Canthus rostralis indistinct, loreal region oblique. Nostril closer to tip of snout than to the eye. Prominent eye located laterally. Upper lip with a distinct light stripe; black broad stripe from tip of snout to the insertion of the forelimb, passing on the eye and tympanum. Tongue large and free posteriorly, notched from behind. Vomerine and maxillary teeth present. Tympanum evident, circular with dark annuli, largely separated from eye. Tympanum diameter smaller than eye diameter (TYD/ED 0.7). Supratympanic fold developed, from the eye to the forelimb insertion. Commissural gland present. Dorsum with small dark spots. Two pairs of dorsolateral folds. The upper pair accompanied by wide uninterrupted dark stripes; and interrupted dark stripes in the folds of the flanks. Flanks with small-medium dark spots. Skin rough, with small white tubercles on dorsum and flanks. Belly spotted. Dark throats, vocal sacs evident. Cloacal zone without tubercles, femoral zone with smooth texture. Arm robust, stained dorsally. Hand with slender fingers, with rounded tips, without webs or fringes. Relative finger lengths IV > II = V > III. Subarticular tubercles rounded, outer metacarpal tubercle rounded, inner metacarpal tubercle elongate. No thumb asperities or prepollex visible. Legs robust, tibia larger than thigh; broad and diffuse bars in thigh, tibia, and tarsus. Toe tips rounded, without webs or fringes. Relative toe lengths IV > III > V > II > I. Subarticular tubercles rounded, outer metatarsal tubercle small and rounded, inner metatarsal tubercle large. Small white tubercles in the sole of tarsus. Measurements

Coloration in life
Dorsum uniformly reddish (Fig. 5A). A wide black stripe from the tip of snout to the supratympanic fold. Black tympanic membrane, with black tympanic annulus. Near the upper half of the eye brown, and black below. Upper and lower lips dark gray. Dorsum with small black spots. Black stripes accompany the dorsolateral folds, spreading in the dorsum, from the nearest of supratympanic fold to groin. Forelimb reddish above, with a black stripe on anterior and posterior sides of arm and diffuse black stripes on forearm.
Thigh reddish above, with wide, incomplete and diffuse dark bars in thigh and tibia. Belly and ventral surfaces of forelimb and leg clear cream with diffuse spots. Gular region dark. Cloacal zone with distinct white tubercles.

Color in preservative
The reddish coloration becomes gray. Black stripes that accompany dorsolateral fold, lateral head stripe, arm, and leg maintained the dark coloration. The belly continued whitish and the vocal region grayish. Tympanum kept its dark coloration, with dark tympanic membrane and dark tympanic annul. Eyes become complete black.

Advertisement call and natural history notes
Leptodactylus apepyta sp. nov. is a species typical of forests and open areas of the Dry Chaco's environments, characterized by scarce rainfalls and deciduous vegetation. Males were found vocalizing after sunset during the rainy season, from mid-October to February. They usually begin to vocalize near to temporary ponds on the top of fallen logs and low branches of trees, up to 1.5 m high (Fig. 5B). The advertisement call consists of series of non-pulsed notes emitted at a rate of 5.92 notes/s (Figs. 4A and 4B; Table 2). Note duration ranges from 30 to 68 ms (42.2 ± 0.82 ms), and internote interval from 88 to 204 ms (132.5 ± 2.91 ms). The call lacks amplitude and frequency modulation. Dominant frequency = fundamental frequency, ranges from 2,155 to 2,457 Hz (mean 2,266.04 ± 80.41 Hz). Oviposition and tadpoles are unknown.

Geographic distribution
Leptodactylus apepyta sp. nov. occurs in the South American Gran Chaco in Argentina and Paraguay (Fig. 8), almost exclusively inhabiting the Dry Chaco, with some scarce records in surrounding areas of Humid Chaco (i.e., Pirané, Formosa province, Argentina; and Concepción, Paraguay) and the Yungas ecoregion (i.e., Zapla, Jujuy province, Argentina). Additionally, a GenBank sequence (JF789906, voucher not examined by us) of a specimen from the Chiquitanía ecoregion (Ñuflo de Chávez province, Santa Cruz department, Bolivia), was recovered as sister of the all remaining terminals of L. apepyta sp. nov. in the phylogenetic analyses (Fig. 1B), and due to its low genetic distance (0.8-1%) we consider it conspecific with L. apepyta sp. nov. Therefore, the distribution range of L. apepyta sp. nov. would extend from middle Bolivia eastwards to the Paraguay River in Paraguay (Presidente Hayes department as eastern limits), and more southwards reaches the northern half of Chaco and center of Santiago del Estero and southern Tucumán provinces in Argentina.

DISCUSSION
Monophyly of Leptodactylus and the L. fuscus group, and comments on some internal relationships Although our molecular sampling was designed only to determine the phylogenetic position of L. apepyta sp. nov. and we did not include nuclear genes (i.e., rhodopsin), main discrepancies with the previous hypotheses produced by the recent phylogeny of Leptodactylus by De  deserve some discussion. In the total evidence analysis of De Sá et al. (2014) both the genus Leptodactylus and the L. fuscus group appear as monophyletic, but paraphyletic in the molecular-only analysis. The analyses of De Sá et al. (2014) recovered L. fragilis as sister taxon of the clade composed by L. melanonotus + Hydrolaterae + all remaining Leptodactylus. Our dataset recovered Leptodactylus and the L. fuscus group as monophyletic but low supported, with L. fragilis within a clade of this group that included L. longirostris + L. poecilochilus. The non-monophyly of Leptodactylus obtained by De Sá et al. (2014) could be attributed to the sequence of L. fragilis, apparently chimeric with a species of Hylini. Similar discrepancies in topologies attributed to accidental chimeras or contaminated sequences were detected, e.g., in Eupsophus , Cycloramphidae (Fouquet et al., 2013), Ceratophryidae (Faivovich et al., 2014). Leptodactylus mystacinus + L. apepyta sp. nov. were recovered by us in a clade formed also by L. cupreus (not included in De Sá et al., 2014), L. troglodytes, and L. bufonius, in a similar way to the total evidence results by De Sá et al. (2014) where L. ventrimaculatus + L. labrosus are the sister taxa of all other species of the L. fuscus group; with L. bufonius, L. troglodytes, and L. mystacinus forming a clade sister to the remaining species of the group. However, the molecular-only analysis of De  resulted in a very different topology with L. bufonius, L. troglodytes, and L. mystacinus forming an early divergent grade within a clade with all species of L. mystaceus complex and some specimens of L. fuscus. In the description of L. cupreus, it was assigned to the L. mystaceus complex based on morphological and bioacustic characters (Caramaschi, Feio & São-Pedro, 2008). However, Cassini et al. (2013) argued that this relationship was only tentative and presented evidence suggesting that this species would be more similar to L. mystacinus, which is congruent with our results.
The terminals of L. fuscus 1-2 and 4-7 (we excluded sequences of L. fuscus 3, 8, and 9 for low quality or missing data) formed a clade with relatively low genetic distances; while in the analysis by De Sá et al. (2014) the terminals 6-9 were separately nested with L. mystaceus. Camargo, De Sá & Heyer (2006) affirmed that while their molecular data supports a multiple-species hypothesis for L. fuscus, they did not find relevant differences in calls and morphology to strongly clarify the taxonomic status of different lineages. Regarding L. mystaceus, we only used sequences of two lineages (1 and 3 or L. cf. mystaceus), which were sister taxa, contrary to De Sá et al. (2014) who found them as paraphyletic lineages. We do not rule out those discrepancies with our results, it could be due to the inclusion of contaminated and low-quality sequences in the analysis of De Sá et al. (2014).

Taxonomic conclusions
Phylogenetic topology, genetic p-distances, external morphology, coloration, osteology, and behavior support the view that the nominal species L. mystacinus is actually a complex composed by two clearly distinct taxa: L. mystacinus, widely distributed and with strong geographical and genetic structure, and L. apepyta sp. nov., endemic to the South American Gran Chaco. The 3% genetic divergence found in the 16S rDNA between both species is considered as "moderate" (Vences et al., 2005), but enough to identify candidate species among Neotropical anurans. The occurrence of cryptic species complexes is a frequent phenomenon in anurans (Vences & Wake, 2007;Castroviejo-Fisher et al., 2017). However, this phenomenon has been poorly studied in the specious genus Leptodactylus, with only