Molecular phylogeny of the tribe Philodryadini Cope, 1886 (Dipsadidae: Xenodontinae): Rediscovering the diversity of the South American Racers

South American racers of the tribe Philodryadini are a widespread and diverse group of Neotropical snakes with a complex taxonomic and systematic history. Recent studies failed to present a robust phylogenetic hypothesis for the tribe, mainly due to incomplete taxon sampling. Here we provide the most extensive molecular phylogenetic analysis of Philodryadini available so far, including 20 species (83% of the known diversity) from which six were not sampled previously. Our results reveal that Philodryadini is not monophyletic, but instead includes a central Andean clade formed by Philodryas simonsii, P. tachymenoides, and P. amaru, and a southern and cis-Andean clade including all remaining philodryadines. This discovery requires resurrection of two genera as well as erection of a new tribe of Xenodontinae for the central Andean clade. Within the southern and cis-Andean radiation, our analyses resolve a basal dichotomy separating two main lineages: Clade A, containing the Common Green Racers P. laticeps and P. viridissima and the South American Vine snakes P. georgeboulengeri and P. argentea; and Clade B, including the remaining species of Philodryas sensu stricto. We resurrect the genera Chlorosoma and Xenoxybelis to better represent the monophyly of lineages within the southern and cis-Andean clade. Key-Words. Philodryas; Chlorosoma; Xenoxybelis; Andean Endemism; New Tribe; Hemipenial morphology.


INTRODUCTION
The South American racers of the tribe Philodryadini Cope 1886 are a diverse and widespread group of snakes with a wide variety of morphologies and ecological roles, ranging from large, semiarboreal, and generalist species to small, secretive, and diet specialist species (Greene & Jaksic, 1992;Hartmann & Marques, 2005;Marques et al., 2006). Most of this diversity occurs along the cis-Andean portion of South America (Table 1). Twenty species are distributed from Colombia to Argentina, while only four species inhabit the trans-Andean parts of Ecuador, Peru, and Chile (Cacciali et al., 2016;Grazziotin et al., 2012;Zaher et al., 2014). Because of their diversity, broad distribution, and variety in ecomorphological traits, the species have a long history of taxonomic instability and uncertainty (Thomas, 1976;Zaher et al., 2008;Wallach et al., 2014).
Several molecular phylogenetic studies incorporated a limited number of representatives of Philodryadini, and obtained conflicting hypotheses of relationships for the tribe (Cadle, 1984a, b, c;Grazziotin et al., 2012;Jenner & Dowling, 1985;Machado, 1993;Maglio, 1970;Zaher, 1999;Zaher et al., 2009). Cadle (1984a) found that Philodryas was not related to Alsophis, as previously stated by Maglio (1970), but rather was more closely related to the South American genus Xenodon. Machado (1993), based on hemipenial evidence, allocated Oxybelis argenteus (Daudin, 1803) to the new xenodontine genus Xenoxybelis which, along with Pseudablabes agassizii (Jan, 1863), were considered by Zaher (1999) to share hemipenial characteristics with Philodryas. Lobo & Scrocchi (1994) provided the first osteological phylogenetic analysis of Philodryas, based on representatives of 11 species. Thomas & Fernandes (1996) further revised the morphological definition of Philodryas by including monotypic Platyinion Amaral, 1923 in the synonymy of Philodryas. According to Zaher (1999), Philodryas was paraphyletic and could be divided in two distinct lineages defined by their hemipenial morphology: the olfersii-group including Xenoxybelis; and the chamisso-nis-group. More recently, a series of molecularly oriented studies allowed a better understanding of supra-generic relationships within New World colubroid snakes, corroborating many of Zaher's (1999) previous morphological conclusions for that group (Vidal et al., 2000(Vidal et al., , 2010Zaher et al., 2009). Subsequent molecular assessments that included species of the tribe Philodryadini invariably recovered the tribe as a monophyletic assemblage Pyron et al., 2011Pyron et al., , 2013Vidal et al., 2010;Zaher et al., 2018Zaher et al., , 2019. Nevertheless, the sampled diversity of Philodryadini in these studies never exceeded slightly more than half of the known species of the tribe (Table 2).
Here, we provide a phylogenetic analysis based on a multilocus molecular dataset that incorporates representatives of 83% of all known species of Philodryadini. Our assessment requires a taxonomic revision to maintain monophyletic lineages.
We used standard PCR protocols with modifications to improve the efficiency as follows: adding 10% of Trehalose 100 for 12S, cytb, and cox1, and 0.4% of Triton 100 for cmos, bdnf, and nt3. We amplified both strands and employed an annealing temperature of 54°C for 12S, 56°C for bdnf and cmos, a touch down cycle of 58-46°C with final annealing of 48°C for nt3, and a touch down cycle of 60-50°C with final annealing of 54°C for cytb and cox1. All PCR products were purified with the Exo-Sap (exonuclease and shrimp alkaline phosphatase) protocol and the sequences were processed at the Laboratório de Biologia Genômica e Molecular from the Pontifícia Universidade Católica do Rio Grande do Sul (PUCRS), Laboratório de Toxinologia Aplicada (LETA) from Instituto Butantan, and Macrogen (Seul, Korea). We sequenced and checked both strands, and when necessary the chromatographs were edited manually. We performed the assembly and generated the consensus sequences using Geneious 6.1.8 (http://www.geneious.com, Kearse et al., 2012).

Phylogenetic analyses
We generated 179 new sequences and obtained other 300 sequences from GenBank to build our molecular matrices (Extended Table S1). Our concatenated dataset comprised a total of 4,433 base pairs of aligned sequences, including 510 from 12S, 1,107 from cox1, 997 from cytb, 710 from bdnf, 586 from cmos, and 523 from nt3 (Extended Data S1).
We used MAFFT 1.3.6 (Katoh et al., 2005), as implemented in Geneious, to align all sequences. The rRNA gene 12S was aligned under the E-INS-i algorithm, while the protein-coding genes bdnf, c-mos, cox1, cytb, and nt3 were aligned under the G-INS-i algorithm. We used default parameters for gap opening and extension. All protein-coding genes were visually checked using Geneious to verify if all sequences follow the correct reading frame. We concatenated our sequences using Sequence Matrix 1.8 (Vaidya et al., 2011).
We used PartitionFinder 2 (Lanfear et al., 2016) to choose the models of molecular evolution for our database and employed two different analyses. In the Table 2. Species of Philodryadini (sensu lato) employed by the most recent molecular phylogenetic studies. X A = Species sampled for the first time in a molecular study.

Species
Zaher et al. first analysis, we used the Akaike Information Criterion with correction (AICc), allowing the selection of models of molecular evolution implemented in RAxML 8.2.3 (Stamatakis, 2014), using unlinked branch lengths and the greedy search option. We only allowed PartitionFinder to select GTR model with or without GAMMA, preventing models with correction for proportion of invariant sites, as suggested in the RAxML manual. In the second analysis, we used the Bayesian Information Criterion (BIC), allowing selection of all models of molecular evolution implemented in MrBayes 3.2.5 (Ronquist et al., 2012). We preliminarily defined 16 partitions for our concatenated matrix, treating the rRNA gene (12S) as a separate partition and partitioning all five protein-coding genes by their codon positions. The results of PartitionFinder are provided in Extended Table S2. We conducted a maximum likelihood (ML) analysis in RAxML using the algorithm that conducts a rapid bootstrap analysis and searches for best scoring ML tree in the same run (option -f a), and defined 1,000 bootstrap iterations to estimate branch support (BS). We also conducted a Bayesian Inference phylogenetic analysis (BI) in MrBayes 3.2.5 with four independent runs, 20 million generations, sampling trees every 1,000 generations. We used Tracer 1.7.1 (Rambaut et al., 2018) to check the effective sample size (ESS) and trace convergence to set the burn-in. We estimated Bayesian posterior probability (PP) after burn-in. Both ML and BI phylogenetic analyses were carried out on the CIPRES Science Gateway (http://www.phylo.org, Miller et al., 2010). Robustness of clades were checked using both BS and PP support values.
When defining and naming evolutionary lineages retrieved in our analyses, we followed the Taxon Naming Criteria (TNCs) suggested by Vences et al. (2013), avoiding taxonomic instability and inadequate classification schemes of non-monophyletic groups.

RESULTS
The ML and BI tree topologies showed very similar higher-level relationships, with Colubridae, Natricidae, Sibynophiidae, and Dipsadidae being all recovered as monophyletic lineages with robust to unambiguous support values ( Fig. 1; Extended Figs. S1-S2). Only the position of Grayiidae varied, being retrieved with ambiguous support as the sister group of Colubridae in the ML tree and as the sister group of a clade formed by Colubridae and Sibynophiidae in the BI tree ( Fig. 1; Extended Fig. S2). Within Dipsadidae, the Asian Thermophis baileyi and Sticophanes ningshaanensis were retrieved as two successive sister taxa to the New World dipsadids in both ML and BI topologies (Fig. 1). Within the New World radiation, North American Carphophiinae was retrieved as monophyletic only in the ML tree, with no statistical support. Subfamilies Dipsadinae and Xenodontinae as well as tribes Imantodini and Dipsadini were retrieved as robustly supported clades in both ML and BI topologies ( Fig. 1; Extended Fig. S2).
Considering both ML and BI analyses, only the tribes Philodryadini and Echinantherini within Xenodontinae were not recovered as monophyletic groups. All other tribes represented by more than one terminal were recovered with strong to unambiguous support values in both analyses, but relationships between them were not supported by high values of support, with the BI tree resulting in a large polytomy ( Fig. 1; Extended Fig. S2).
The traditionally recognized tribe Philodryadini was split into two monophyletic lineages in both analyses-a central Andean clade and a southern and cis-Andean clade-with Tropidodryadini nested between them, robustly positioned in both analyses as the sister group of the southern and cis-Andean radiation of Philodryadini ( Fig. 1; Extended Figs. S1-S2). The central Andean clade was retrieved as the sister group of the clade formed by Tropidodryadini + southern and cis-Andean Philodryadini only in the ML analysis, with low values of BS, while in the BI analysis it was recovered in a polytomy along with several other clades (Fig. 2).
The central Andean clade was composed by the strictly Andean species Philodryas simonsii and Philodryas tachymenoides. These represented a distinct evolutionary lineage from the southern and cis-Andean clade of Philodryadini. Phylogenetic affinities exhibited within southern and cis-Andean radiation were the same in both ML and BI topologies (Extended Figs. S1-S2).
Phylogenetic relationship within the southern and cis-Andean radiation revealed the presence of two main evolutionary lineages: Clades A and B in Fig. 2. Clade A was recovered with low values of support, and contained the Common Green Racer P. viridissima and the rare P. laticeps forming Subclade A1, and the South American Vine Snakes P. argentea and P. georgeboulengeri composing Subclade A2, both with robust and unambiguous support values (Fig. 2). The remaining species of Philodryas grouped in the robustly supported Clade B, with P. nattereri placed as the sister group of five successive, robustly to unambiguously supported subclades, as follow: Subclade B2 composed by P. mattogrossensis and P. erlandi; Subclade B3 including the Southern Andean P. chamissonis as sister of the Argentinian species P. baroni and P. trilineata; Subclade B4 formed by the South American Green Racers P. olfersii and P. arnaldoi; Subclade B5 composed only by the Brazilian Green Racer P. aestiva; and Subclade B6 formed by the "patagoniensis group" composed by P. agassizii, P. livida, P. patagoniensis, P. psammophidea, and P. varia (Fig. 2). Within the "patagoniensis group", P. varia and P. patagoniensis on the one hand, and P. psammophidea nested with P. livida + P. agassizii, on the other hand, grouped together as sister clades, the former lacking statistical support values while the latter was robustly supported.
Our results indicate that the traditional taxonomic arrangement for Philodryadini does not reflect the topology of the phylogenetic tree. To determine a more accurate classification based on monophyletic groups, we propose below a revised classification for the tribe based on our phylogenetic results and on a diagnostic set of morphological characters. Arredondo, J.C. et al.: Molecular Phylogeny of Philodryadini Pap. Avulsos Zool., 2020;   Diagnosis: Philodryas can be distinguished from the other Philodryadini genera by the following combination of characters: Hemipenial body much longer than the lobes (more than twice the length), with the asulcate side of the hemipenial body covered with two parallel rows of enlarged body calyces on most or all its surface (Figs. 3-4); dentary teeth equal in size (significantly enlarged dentary teeth in Ditaxodon); ventral scales smooth (keeled in Chlorosoma); buccal epithelium cream or white.
Geographical distribution: Same as the tribe.
Etymology: Donnelly & Myers (1991: 46) argued convincingly that the meaning of the generic name Philodryas is "friendly tree nymph, " being a feminine gender resulting from the combination of the Greek words Philos-(noun, φῐλος, meaning "friend or friendly") and -Dryas (noun, Δρῠᾰς, meaning "tree nymph").

Etymology:
The generic name is neuter, being a combination of the Greek words Chloros-(adjective, χλωρός, meaning "bright green") and -soma (noun, σῶμα, meaning "body"), in reference to the overall bright green coloration of the body in these species. The gender of the name Chlorosoma is grammatically neuter (Amaral 1929a(Amaral , b, c, 1932 since the meaning given by Wagler (1830) was of a "green snake. " Therefore, the correct spelling for the type species of the genus is Chlorosoma viridissimum.  Amaral (1929aAmaral ( , b, 1932 decided to apply the Principle of Priority and assigned all species of Philodryas known at the time to the genus Chlorosoma (i.e., C. aestivum, C. arnaldoi, C. burmeisteri (= trilineata), C. mattogrossense, C. olfersii, C. schottii (= patagoniensis), C. psammophideum, and C. viridissimum). However, Parker (1932)  Diagnosis: Xenoxybelis can be distinguished from the other genera of Philodryadini by the following combination of characters: markedly elongated snout, forming an acuminated and sharp head shape; large number of prediastemal maxillary teeth (16-21), followed by one or two grooved postdiastemal teeth; short heart-shaped hemipenes with a well-defined papillate longitudinal crest, formed by the confluence of the body calyces, that runs medially in the asulcate surface; lateral surfaces of hemipenes covered with two to four rows of well-developed enlarged lateral spines (Fig. 4).
Geographical distribution: Both species occur in the Amazonian region, from Colombia and Guianas to Bolivia and Paraguay (Cunha & Nascimento, 1978;Prudente et al., 2008).

Etymology:
The generic name of the South American Vine snake Xenoxybelis is formed by the Greek words Xenos-(adjective, ξένος, meaning "different"), -oxy-(adjective, οξύς, meaning "sharp"), and -belos (noun, βέλος, meaning "dart"), in reference to their external similarity with the unrelated Neotropical Vine snake genus Oxybelis Wagler, 1830.  , resulting in the proposition of the new replacement name georgeboulengeri for the latter species. According to our phylogenetic results, returning Philodryas georgeboulengeri (Procter, 1923) to the revalidated genus Xenoxybelis eliminates the homonymy. Therefore, we revalidate the species name Xenoxybelis boulengeri (Procter, 1923).

Diagnosis:
Members of the tribe Incaspidini can be distinguished from the other genera of Xenodontinae by the following combination of characters: maxillae diacranterian and opisthodont, with ungrooved postdiastemal teeth; hemipenes significantly longer than wide (up to 3.5 times), with large shallow body calyces on upper half of the asulcate region, numerous small-sized lateral enlarged spines, and a well-defined constriction in the proximal region of the hemipenial body (Fig. 3).

Geographical distribution: Andean region in Southern
Ecuador, Peru and possibly northern Chile.

Etymology:
The singular of the genitive case of the second part of the tribe name (aspis) is aspides. Therefore, the correct spelling of the new tribe is Incaspidini.  Thomas (1977). Although no tissue samples were available from Incaspis amaru, we allocate this species in the genus Incaspis since it shares all the diagnostic features listed above for the genus and the tribe. The occurrence of I. amaru in the same biogeographical region as I. simonsii and I. tachymenoides also supports its allocation in the tribe, which seems to be an endemic component of Central Andes.

DISCUSSION
As in most recent systematic studies, our analyses retrieve the highly diverse family Dipsadidae with strong support (Figueroa et al., 2016;Grazziotin et al., 2012;Pyron et al., 2011;Vidal et al., 2010;Zaher et al., 2009Zaher et al., , 2018Zaher et al., , 2019. Notwithstanding, our results from both ML and BI support the monophyly of most previously hypothesized xenodontine tribes (Zaher et al., 2009;Grazziotin et al., 2012), but do not recover well-established relationships between them (Fig. 1). One notable exception is the robustly supported clade formed by the Tropidodryadini and the redefined tribe Philodryadini (Figs. 1-2). Historically, a close relationship between members of the tribes Philodryadini and Tropidodryadini has already been advocated by several authors (Amaral, 1937;Dowling & Duellman, 1978;Ferrarezzi, 1994;Günther, 1858;Lobo & Scrocchi, 1994;Pyron et al., 2011;Vidal et al., 2010;Zaher et al., 2012Zaher et al., , 2018Zaher et al., , 2019; but see Grazziotin et al., 2012 andPyron et al., 2013). However, none of these studies had adequate taxon sampling within Philodryas, reaching only a maximum of 58% of the known diversity of the tribe Philodryadini and lacking representation of the Central Andean species (e.g., Pyron et al., 2011;Figueroa et al., 2016) (Table 2). Consequently, Pyron et al. (2011) retrieved a non-monophyletic Philodryadini, with P. viridissima clustering with Tomodon and Hydropsini instead of the other species of Philodryas. In contrast, Figueroa et al. (2016) recovered a highly supported clade composed by P. chamissonis, P. trilineata, and the genera Xenopholis, Hydrodynastes, and Caaeteboia, while the remaining species of Philodryas grouped with Sordellina punctata and the Echinantherini. As pointed out by these authors, their topology most likely resulted from the absence of homologous molecular markers for these lineages, as they only used sequences for the mitochondrial gene nd4 for P. chamissonis and P. trilineata, and a combination of 12S, 16S, bdnf, cmos, and cytb for the remaining species of Philodryas. Pyron et al.'s (2011) andFigueroa et al.'s (2016) unexpected phylogenetic results highlight the necessity of a comprehensive taxon sampling to resolve the phylogenetic affinities within this group. Here, we expand both gene and taxon samplings for the group, reaching 83% of the known diversity of the former composition of Philodryadini. We also added for the first time critical central and Southern Andean lineages of Philodryas and the rare species C. laticeps (Table 2). Surprisingly, our results reveal that the central Andean species-represented by Incaspis simonsii, I. tachymenoides, and I. amaru-are a distinct lineage of Andean endemics while the genus Tropidodryas appears in a robustly supported clade as the sister group of all other southern and cis-Andean species of Philodryas (Fig. 2). Resurrection of Incaspis for the central Andean species, with its allocation in the new tribe Incaspidini, reflects their distinctiveness and uncertainties regarding their phylogenetic relationships within Xenodontinae, while highlighting the closer affinities revealed by the Tropidodryadini and the southern and cis-andean Philodryas ( Fig. 1; Extended Fig. S2). Therefore, our results restrict Philodryadini to a widespread lineage of mostly lowland cis-Andean species, except for the trans-Andean P. chamissonis, which occurs in both low and high elevations throughout southern Andean region (Sallaberry-Pincheira et al., 2011). Further, Incaspidini represents a distinct lineage of endemic species mainly found in highlands of the central Andes of Ecuador and Peru.
Historically, a certain confusion has existed around the systematics of the species here allocated in the tribe Incaspidini, mainly because their characteristic ungrooved postdiastemal maxillary teeth led several authors to associate them with the opisthodont genera Alsophis, Dromicus, and Leimadophis (Amaral, 1929a, b, c;Parker, 1932;Peters, 1960;Thomas, 1977). Additionally, P. chamissonis also retains ungrooved postdiastemal (opisthodont) maxillary teeth, although its hemipenial morphology characterized by a very long hemipenial body and moderately long lobes closely approaches the one of the larger mainland cis-andean species P. nattereri, P. psammophidea, P. baroni, and P. trilineata (Zaher, 1999). Philodryas chamissonis was viewed accordingly as a morphological intermediate between the central Andean opisthodont and the cis-Andean opisthoglyph species of Philodryas (Thomas, 1976;Lobo & Scrocchi, 1994;Zaher, 1999). However, our results show that the opisthodont condition found in the Incaspidini and in P. chamissonis owes to convergence rather than being a shared derived characteristic (Zaher et al., 2014).
The southern and cis-Andean radiation is recovered in our analyses as two sister Clades A and B (Fig. 2). The topology provides the basis for the resurrection of Chlorosoma and Xenoxybelis, for the Common Green Racers and South American Vine snakes, respectively. Xenoxybelis corresponds to a genus of highly specialized arboreal snakes that have been traditionally considered to be morphologically distinct from all the other Philodryadini. However, their uncertain position within the group led to its synonymization with the latter as a way to reinstate a monophyletic definition for the group . Our expanded taxonomic coverage recovers a more stable phylogenetic hypothesis that allows the recognition of its morphological differences. Similarly, Chlorosoma corresponds to a vestigial clade of arboreal snakes that also harbors morphological differences from Philodryas (Clade B), including a short hemipenes and laterally keeled ventral scales.
The redefined genus Philodryas includes now 14 species that are retrieved in our analyses in six strongly supported subclades (Subclades B1 to B6 in Fig. 2). Subclades B1 and B2 include Philodryas nattereri and P. erlandi + P. mattogrossensis, respectively, two successive sister groups to the remaining Subclades B3-B6. Subclades B1 and B2 include species that occur throughout the open biomes of Brazil, Bolivia, Paraguay, and Argentina (Nogueira et al., 2019), reinforcing the view that cis-Andean Racers of the genus Philodryas originated in open areas in South America and only later occupied forested biomes, such as the Atlantic and Amazonian forests. Species from Subclade B2 have a unique coloration pattern of gradual change from anterior green or yellow to reddish brown in the rest of the body and tail (Cacciali et al., 2016;Thomas, 1976). This appears to represent a synapomorphic trait for this subclade.
Subclade B3 includes P. chamissonis, P. baroni, and P. trilineata, and it corresponds to a group of racers restricted to the southern part of the continent, with P. chamissonis reaching as a single dispersal event the eastern part of southern Andes. According to Thomas (1976) and Lobo & Scrocchi (1994), P. baroni, and P. trilineata are sister species, a hypothesis corroborated in our analysis.
Subclades B4, B5, and B6 include the commonest species of the genus, and these occur mostly in open and forested areas of the continent, east of the Andes and south of the Amazonian basin. The only exception is P. olfersii, whose range extends to parts of the Amazonian biome. Subclades B4 and B5 include the southern and southeastern Atlantic forest endemics P. aestiva and P. arnaldoi, and widespread P. olfersii; all three occur mainly in forested areas in South America.

Species in Subclade B6 inhabit both forested and open biomes. In this subclade, P. varia occurs in northern
Argentina and Bolivia, while P. patagoniensis inhabits the southern and eastern parts of the continent, from northern Pará and Tocantins to Chubut in southern Argentina. In contrast, P. psammophidea, P. livida, and P. agassizi occur in the open biomes throughout the central and southern parts of the continent.
Redefined Philodryadini contains species with high relevance in public health (Sánchez et al., 2014;Weinstein et al., 2011). Records of envenomation exist for several species, including lethal cases for humans (Da Rocha et al., 2006;Weinstein et al., 2013). Philodryas has opisthoglyphous dentition, a venom delivery apparatus, and gland secretions containing venomous proteins and toxins (Modahl et al., 2016;Urra et al., 2015). This genus of South American opisthoglyphous snakes is responsible for most of the non-front fanged envenomations (Oliveira et al., 2017). Because most studies focusing on venom and envenomation of Philodryas were developed without a robust evolutionary context (Acosta et al., 2003;Da Rocha et al., 2006), they likely bypassed issues on intra and inter-specific variation and phylogenetic structure within species groups. Our phylogenetic hypothesis provides a more accurate evolutionary framework that will lead to a better understanding of venom variation and diversification in that group.

AUTHORS' CONTRIBUTIONS
JCA and HZ conceived the research project. JCA gathered the molecular data and performed the analyses; FGG contributed to the molecular analyses; JCA, HZ, and FGG wrote the manuscript; GJS, MTR, and SLB contributed intellectually to the project, revised and discussed the text; HZ funded and supervised the research project.

ACKNOWLEDGMENTS
We are in dept with the curators and institutions staff who granted us access to the samples for obtention of molecular evidence. We thank J. Battilana for her help during acquisition of molecular data. We thank F. Curcio and L.J. Vitt for providing photographs of P. nattereri, X. boulengeri and P. olfersii used in Fig. 2. We are grateful for the comments from two reviewers whose suggestions improved this manuscript. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES) -Finance Code 001. FGG was supported by FAPESP (grants 2012/08661-3 and 2016/13469-5). HZ, MTR, and SLB are grateful to Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for continuing financial support. SLB and MTR were supported by FAPERGS andFAPESP (grants 2003/10335-8 and2011/50146-6), respectively. This study was funded by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP grants 2011/50206-9 and 2016/50127-5) to HZ. Figure S1. Best scoring maximum likelihood (ML) tree estimated using RAxML. Numbers on nodes correspond to Bootstrap values.

Extended
Extended Figure S2. Maximum Clade Credibility Tree estimated using MrBayes. Numbers on nodes correspond to posterior probability values.