Revision of the Chironius bicarinatus complex (Serpentes: Colubridae): Redefined species boundaries and description of a new species

Currently, the proposed diagnoses for the Chironius bicarinatus complex reflect a wide variation in color pattern and pholidosis. Herein, we review the Chironius bicarinatus complex based on morphological and molecular data from a sample of 485 specimens covering the species distribution. Our results corroborate the recognition of C. bicarinatus and C. gouveai , and diagnose a distinct lineage without an available name. Thus, here we describe this new species restricted to the Baturité Massif, a relictual rainforest isolated in the Caatinga xerophytic domain, in the state of Ceará, northeastern Brazil. The new species can be distinguished from its congeners by its unique combination of qualitative and quantitative morphological characters (scale counts, morphometric, color pattern), and is also supported by molecular and ecological evidence. Additionally, we rectify data on the distribution and morphological variability of C. gouveai to accurately infer the boundaries between this taxon and C. bicarinatus , which was not properly addressed. Finally, we discuss our results in the light of previous studies that suggest diversification hypotheses in the Atlantic Forest already detected for other taxa, highlighting the importance of conserving the areas of “Brejos de Altitude”, in northeastern Brazil, and the southern limit of Serra do Mar up to Serra do Tabuleiro, in southern Brazil.


Introduction
The snake genus Chironius Fitzinger, 1826 currently comprises 23 species of mostly terrestrial and semi-arboreal colubrid snakes that are widespread across Central and South America, ranging from northern Honduras to northeastern Argentina and northwestern Uruguay (Dixon et al. 1993;Uetz et al. 2023).Among all Neotropical colubrid genera, Chironius is unique in its presentation of 10 or 12 dorsal scale rows on the midbody, a feature that, combined with other morphological characters and molecular evidence, supports the monophyly of the genus (Hollis 2006;Klaczko et al. 2014;Hamdan et al. 2017;Torres-Carvajal et al. 2019a).The systematics of Chironius was approached by Dixon et al. (1993), who revised the taxonomy of the genus and Hollis (2006), who performed a phylogenetic study using Dixon's morphological dataset.Hollis (2006) was the first to find support for the monophyly of the genus and also proposed the elevation of all subspecies recognized by Dixon et al. (1993) to full species status.More recently, Klaczko et al. (2014), Hamdan et al. (2017), andTorres-Carvajal et al. (2019a) proposed phylogenetic hypotheses for the genus Chironius, using molecular characters or by combining molecular and morphological characters.Although different authors corroborate the monophyly of the genus, the phylogenetic relationships within Chironius still require investigation (see Hamdan et al. 2017;Torres-Carvajal et al. 2019a).
Chironius bicarinatus is widely distributed throughout the Atlantic Forest and northern Pampas, from northeastern to southern regions of Brazil, northeastern Argentina, eastern Paraguay, as well as some records in northwestern Uruguay; inhabiting the coastal tropical forests, semideciduous forests, grasslands and the Araucaria Forest formations (Dixon et al. 1993).Wied (1820) first described Coluber bicarinatus based on a male specimen collected near a lake, approximately five leagues from the south bank of the Jucú River, state of Espírito Santo, Brazil (Wied 1820: 179).Nonetheless, the species was treated in more detail in later studies, when an illustration and data from four additional specimens became available (Wied 1824: Vol. 8;Wied 1825: 284).In the original description, ventral and subcaudal scale counts, measurements, general body coloration and the presence of two keeled scale rows were reported as diagnostic characters.The latter character led some authors (e.g., Schlegel 1837: 177;Duméril 1853: 452) to include this species in the synonymy of Herpetodryas carinatus.Bailey (1955: 8) later resurrected the taxon, despite suspicions of the weak diagnostic power of this character, in addition to the possible existence of several species under the name of Herpetodryas carinatus (see Cope 1861: 562).Dixon et al. (1993) evaluated the variability of the traditional morphological characters (e.g., color pattern, measurements and scale counts) for Chironius bicarinatus and pointed out some differentiation in color pattern, the sum of ventral and subcaudal scale counts, number of maxillary teeth, and a higher incidence of temporal scales formula 1+1 in the southernmost populations.Neverthe-less, the authors did not propose taxonomic acts related to these variations due to the limited number of samples and evidence of the overlapping of some characters.These authors only indicated the possibility of the existence of two subspecies under the name C. bicarinatus distributed throughout the Atlantic Forest, one in gallery and coastal forests and the other in ombrophilous dense and semideciduous forests.Entiauspe-Neto et al. (2020) reviewed C. bicarinatus based on morphological and molecular data and described the aforementioned southernmost populations of C. bicarinatus as a distinct taxon, C. gouveai.
Currently, the proposed diagnoses for Chironius bicarinatus reflect a wide variability in color pattern and pholidosis (Dixon et al. 1993;Entiauspe-Neto et al. 2020).
Given the long and convoluted taxonomic history of C. bicarinatus and its massive morphological variability, herein, we investigate the variability of phenotypic characters throughout the entire distribution of C. bicarinatus based on geographically representative samples, reporting data on meristic and morphometric characters, color pattern, hemipenial morphology, and cranial osteology in agreement with other lines of evidence (i.e., molecular phylogeny and environmental niche modeling), in order to accurately infer species boundaries (see de Queiroz 2007;Pante et al. 2015).We conclude that C. bicarinatus actually refers to a species complex that requires taxonomic changes to accurately reflect the lineage diversity.Due to taxonomic inconsistencies in C. gouveai found in the course of the present study, we refine and rectify data on its distribution through environmental niche modeling and assess the variability of morphological and molecular characters of this species to infer the boundaries between it and C. bicarinatus.In addition, we diagnose a distinct lineage without an available name and thus describe this new species whose distribution is restricted to the Baturité Massif, a relictual and isolated rainforest in the Caatinga xerophytic domain, in the state of Ceará, northeastern Brazil.

Examined Specimens and Taxonomic Comparisons
We examined 485 specimens from 227 localities currently attributable to the Chironius bicarinatus complex (C.bicarinatus, C. gouveai, and Baturité Massif sample) housed in 35 scientific collections covering the species' distribution to assess the variability in morphological characters (see Appendix 1).The comparisons supporting our taxonomic decisions are based mainly on the examined specimens, but also on the original descriptions of all valid congeners, including information on the holotype of C. gouveai (Entiauspe-Neto et al. 2020), and the diagnoses proposed by Dixon et al. (1993).Regarding the Chironius taxa sympatric to the C. bicarinatus species complex [i.e., C. carinatus (Linnaeus, 1758), C. exoletus (Linnae-us, 1758), C. foveatus Bailey, 1955, C. fuscus (Linnaeus, 1758) and C. laevicollis (Wied, 1824)], only C. exoletus deserves special attention for presenting a coloration pattern and meristic characters which partially overlap those of the C. bicarinatus complex.Thus, we also examined a representative sample of Atlantic Forest specimens of C. exoletus (n = 464) in order to clarify the identities of both species (hereafter represented as Chironius cf.exoletus since it also represents a species complex under ongoing study by the senior author; see Sudré et al. 2014;Hamdan et al. 2017;Torres-Carvajal et al. 2019a).

Geographical Data
We retrieved the geographic coordinates (datum WGS84) based on institutional catalogs or databases (Appendix 1) and, whenever possible, they were refined with the aid of Gazetteers (e.g., Paynter and Traylor 1991) or additional software (e.g., Google Earth).The map and spatial analyses were performed using the software QGIS v.3.30(QGIS Development Team 2023) and the statistical software R (R Core Team 2022) with specific packages indicated below, respectively.

Meristic and Morphometric Characters
The meristic characters mostly follow those used by Dixon et al. (1993), as did the methods for counting scales and teeth, and the standardized notations for reporting data.The main qualitative characters analyzed were: keel intensity in the dorsal rows (weak or strong), body coloration, presence of a postocular stripe, band pattern along the body, presence of apical pits along the body and condition of the cloacal plate (divided or entire).Some specimens had a small scale in the anterior temporal, which we refer to as an accessory scale.The color pattern characteristics mostly reflect those of preserved specimens, but we also provide data obtained from the photographic material of specimens while alive [databases of Museu Argentino de Ciencias Naturales (MACN); Museu Nacional, Universidade Federal do Rio de Janeiro (MNRJ); and iNaturalist].Appendix 2 contains the iNaturalist observation codes used in this study.We performed frequency analyses among operational taxonomic units with respect to the qualitative characters that exhibit some degree of polymorphism that might be informative (see below in the "Determination of Operational Taxonomic Units" section, on how the operational taxonomic units were delimited).Additionally, special attention was given to other potentially informative qualitative characters with respect to the shape and contact of head scales and body scales ornamentation.
The morphometric characters used in this study were: head height (at highest point), head width (at widest point), head length (from snout tip to quadrate mandibular articulation), orbit diameter (on right side only), nostril-orbit distance, snout length (from snout tip to anterior border of right orbit), distance between nostrils, and snout width (at loreals level).All of the aforementioned characters were measured with a dial caliper to the nearest 0.01 mm, except for snout-vent length (SVL) and caudal length (CL) measured using a tapeline to the nearest 1.0 mm.Sex was determined based on the presence/absence of hemipenes.When the organs were not evident, we performed a small incision on the midventral surface of the tail (Pesantes 1994).

Species Concept and Species Delimitation
Herein, the term "species" refers to separately evolving metapopulation lineages (sensu de Queiroz 2007).As species delimitation criteria, we considered the correspondence between the clades recovered in the DNA-based phylogenetic trees and the presence of one or more apparently fixed exclusive diagnostic morphological characters.We used a tolerance of 95% intervals to allow levels of polymorphism (e.g., sensu Wiens and Servedio 2000) that showed statistical confidence in our sample and to allow for a putative taxon to be distinguished from the others in the Chironius bicarinatus complex.Therefore, if an OTU (see "Determination of Operational Taxonomic Units" section) shows correspondence in the aforementioned parameters, we consider that OTU a valid species.In addition to the operational criteria, using morphological and molecular approaches (e.g., geographic variation, multivariate analyses, molecular phylogeny) to infer species' boundaries, we also integrated ecological evidence (e.g., environmental niche modeling) to test whether species occupy distinct environmental niches and thereby, increase the power of species delimitation ( de Queiroz 2007;Padial et al. 2010).

Determination of Operational Taxonomic Units
The delimitation of operational taxonomic units (a definition used prior to data analysis for ordering groups that share a given set of observed characters; Sneath and Sokal 1973) in this study was guided by the results of the phylogenetic analysis and the variations in qualitative and quantitative (discrete and/or continuous) characters reflecting (or not) the search for morphological discontinuity throughout the species' distribution (see Dixon et al. 1993).When sampling the Chironius bicarinatus complex, we found that some quantitative (e.g., number of ventrals, subcaudals, temporals, supralabials in contact with the orbit, and maxillary teeth) and qualitative characters (e.g., presence and number of keeled dorsal rows, presence of apical pits along the body, and general body coloration) were potentially informative, due to variations that may be related to geographical distribution.Thus, a brief biogeographical description of each operational taxonomic group is presented as follow: Chironius bicarinatus.-Comprisedspecimens from coastal forests from the south of the São Francisco River (in the state of Sergipe, northeastern region of Brazil) to the Serra do Tabuleiro (in the state of Santa Catarina, southern region of Brazil).
Chironius gouveai.-Containedspecimens from inland Serra do Mar at higher elevations in Pampas, Mixed and Semideciduous Forests, below the Paranapanema River and bounded to the west by the Paraná River.
Baturité Massif sample.-Includedspecimens restricted to the "Brejo de Altitude" of the Baturité Massif (state of Ceará, northeastern Brazil), above the São Francisco River.

Statistical Analyses
Statistical tests were performed based on meristic and morphometric characters to assess sexual dimorphism (ANOVA test) and other possible differences between operational taxonomic groups (Tukey HSD test).We used Shapiro-Wilk and Levene tests to assess assumptions of normality and homoscedasticity, respectively, for the original variables.Confirming these assumptions (p > 0.05), sexual dimorphism was evaluated using an ANOVA test, applied individually to each operational taxonomic group.In the case of sexual dimorphism, males and females were treated separately in subsequent statistical tests.Then, we applied the Tukey HSD test to report differences between the groups, comparing distinct groups and we posteriorly evaluated which pairs of groups showed differences.When the tests refuted assumptions of normality and/or homoscedasticity, we employed nonparametric procedures to compare means (Mann-Whitney U, Kruskal-Wallis, Games-Howell) (Zar 2010;McDonald 2014).The variables that were investigated regarding sexual dimorphism and differences between groups were: caudal length/total length (CL/TL) ratio, counts of ventral scales (VEN), subcaudal scales (SUB), and maxillary teeth (MT).
In order to avoid analytical noise due to ontogenetic allometry, the analyses involving morphometric characters only included supposedly mature specimens.We established the maturity cut-off based on an ongoing reproductive study of the Chironius bicarinatus complex (G.S. Araújo and S. Almeida-Santos, pers. comm. on August 16, 2022), where the size of the smallest mature individual was 580 mm SVL for males and 630 mm SVL for females considering both, C. bicarinatus and C. gouveai.As the specimens restricted to Baturité Massif are significantly smaller in body size, regarding the first two species, and since we aimed to reach a reasonable maturity cut-off for this sample, we opted to follow the results presented in Pinto et al. (2010), which addressed the reproductive biology of Chironius flavolineatus (Jan, 1863) and C. quadricarinatus (Boie, 1827).Although these taxa are relatively smaller than C. bicarinatus and C. gouveai, the body size of their populations appears to be equivalent to the Baturité Massif sample.Therefore, we established 450 mm SVL as the maturity cut-off for both, males and females from the samples restricted to the Baturité Massif.
Based on the results of the variance analyses, we performed a Discriminant Function Analysis (DFA) using a subset of specimens containing no missing data for the following 11 log-transformed exploratory variables: ventral (VEN), maxillary teeth (MT), snout-vent length (SVL), head height (HH), head width (HW), head length (WL), orbit diameter (OD), nostril-orbit distance (NOD), snout length (SL), distance between nostrils (DN), and snout width (SW).Missing values were only replaced with group mean for maxillary teeth (MT).These variables were selected according to their distribution identified during data collection, using those whose variation proved to be potentially informative for comparisons between the groups.Although subcaudal (SUB) and caudal length (CL) proved to be potentially informative variables for discrimination, they were not incorporated into the multivariate analyses because of the limited num-ber of specimens that still had their entire tail due to the high proportion of excised tails (see Moura et al. 2022).The assumptions of covariance matrix homogeneity and the absence of multicollinearity between the exploratory variables, were verified before the analyses (Manly 2000).We also performed the leave-one-out cross-validation procedure to assess the accuracy of the DFA's group reclassification.Subscripts of our statistics reflect degrees of freedom.All statistical tests were performed using STATISTICA 10.0 (StatSoft 2011) and PAST 4.13 (Hammer et al. 2001) software.Our descriptive data reflect "minimum-maximum ( [95% confidence intervals] ± SD, n = sample size)".

Molecular Data and Phylogenetic Analysis
Aiming to assess the phylogenetic relationships within the Chironius bicarinatus complex and the phylogenetic position of Baturité Massif sample within the genus, we included all representatives of the genus Chironius currently available on GenBank and representatives of the eight other Colubridae genera (Dendrophidion Fitzinger, 1843, Drymarchon Fitzinger, 1843, Drymobius Fitzinger, 1843, Drymoluber Amaral, 1930, Leptophis Bell, 1825, Mastigodryas Amaral, 1934, Oxybelis Wagler, 1830, and Spilotes Wagler, 1830).The tree was rooted with Dipsas catesbyi (Sentzen, 1796).We selected specific outgroup terminals in order to maximize character coverage (i.e., homologous sequences available from GenBank) and phylogenetic structure according to the trees of the four most densely sampled Chironius phylogenies (Klaczko et al. 2014, Hamdan et al. 2017, Torres-Carvajal et al. 2019a, Entiauspe-Neto et al. 2020).We compiled the complete or partial sequences of four molecular markers: mitochondrial genes for 12S rRNA (12S), 16S rRNA (16S), and subunit IV of NADH dehydrogenase (ND4); and nuclear oocyte maturation factor (c-mos).Specimen information and GenBank accession numbers used in this study are provided in Appendix 3. We checked all available vouchers of the Chironius bicarinatus complex, from which the sequences were extracted by first-hand examination, except for LCBB 18, 21, MAP-T 956 (no preserved voucher for examination), CHFURG 4394 and ZFMK 103132.We tentatively identified MAP-T 956 as C. gouveai based on its geographic origin, since there is no record of C. bicarinatus in our sample from this region (see Appendix 1).In addition, the latter two specimens (CHFURG 4394 and ZFMK 103132) were illustrated by Entiauspe-Neto et al. (2020), allowing us to confirm its identification.
We generated new DNA sequences for species of the Chironius bicarinatus complex, including three mitochondrial gene segments (12S rRNA, 16S rRNA, ND4) and one nuclear gene segment (c-mos).In total, 12S rRNA, 16S rRNA and c-mos sequences were generated for nine specimens of C. bicarinatus, three of C. gouveai, and four of Baturité Massif sample, while ND4 sequences were generated for two C. bicarinatus specimens and two specimens from the Baturité Massif sample.Localities and voucher information are provided in Appendix 3. We performed PCRs using a PCR Master Mix and a pair of primers for each segment (Appendix 4).Thermocycling for DNA amplification for the first partition is provided in Appendix 4. The size and quality of the resulting PCR products were confirmed by electrophoresis through a 1% agarose gel, stained and visualized under a UV Image capture and analysis system.PCR products were purified and sent to the Centro de Estudos do Genoma Humano e Células-Tronco from Universidade de São Paulo and Rede de Plataformas Tecnológicas da Fundação Oswaldo Cruz, both in Brazil.The resulting electropherograms for both DNA strands were analyzed using CHROMAS LITE 2.01 (Technelysium Pty Ltd), edited using MEGA X (Kumar et al. 2018) and adjusted manually to generate consensus sequences for each specimen.Sequences were checked using the basic local alignment search tool (Altschul et al. 1997) against the GenBank nucleotide database to ensure that the amplified product was correct and not contaminated.In total, 252 sequences of 81 Chironius specimens were sampled, including 72 sequences of 23 specimens of the C. bicarinatus complex (23 for 12S, 23 for 16S, 22 for c-mos, and 4 for ND4) and 180 sequences of 58 specimens of 18 other Chironius taxa (51 for 12S, 49 for 16S, 54 for c-mos and 26 for ND4).All the aforementioned sequences were aligned using MAFFT 7 (Katoh et al. 2002), with a gap opening penalty of 1.53 and an offset value of 0.0 (Thompson et al. 1994).Parameters for c-mos, and ND4 were left at their default settings (L-INS-i model) and parameters for 12S and 16S were set to the E-INS-i model.We aligned each segment separately, and segments were concatenated using MESQUITE 3.10 (Maddison and Maddison 2016), to a total of 2,284 bp.We partitioned the data set by locus and selected the appropriate models of the DNA sequences using the software jModeltest 2.17 (Posada 2008) based on the Akaike information criterion or AIC (Akaike 1974).The models obtained were HKY + G for c-mos, and GTR + G + I for ND4, 12S, and 16S.
We performed a partitioned Bayesian Inference (BI) using MrBayes 3.2.2(Ronquist and Huelsenbeck 2003) through the portal CIPRES Science Gateway (Miller et al. 2010) with the concatenated dataset and the partition models described.Each analysis included four independent runs of 5,000,000 generations with four chains of Markov Chain Monte Carlo.Parameters and trees were sampled every 5,000 generations and using a random tree as a starting point.We considered the convergence of the runs when the standard deviation of the frequency splits was lower than 0.05 and by observing ESS values above 300 in Tracer 1.4 (Rambaut and Drummond 2007).The first 25% of the samples were discarded as burn-in.Branch support was assessed by posterior probability, and nodes with posterior probabilities > 0.95 were considered strongly supported (Matioli and Fernandes 2012).Additionally, we used the mitochondrial 16S rRNA fragment to calculate the genetic distances (maximum intraspecific and minimum interspecific) between the species of Chironius.This marker has the characteristics for use as a standard DNA barcoding marker for vertebrates to complement COI (see Vences et al. 2005).We estimated the uncorrected pairwise distances using MEGA X (Kumar et al. 2018), with pairwise deletion of missing information.

Niche Modeling and Niche Overlap Environmental variables and occurrence data treatment
We generated model projections for Chironius bicarinatus and C. gouveai to predict species distribution over time, in addition to testing the degree of niche overlap in the environmental space for each pair of species of the C. bicarinatus complex.We downloaded 19 bioclimatic variables for the current climate from the CHELSA database (Karger et al. 2017) and aggregated the data to a resolution of 5 arcmin using the raster package (Hijmans 2023).We also used elevation data (SRTM) (Farr and Kobrick 2000), which we transformed into relief roughness (raster package; Hijmans 2023), that is the maximum difference in elevation between the focal cell and its neighbors.To reduce issues with multicollinearity, we used Variance Inflation Factor (VIF) (usdm R package; Naimi et al. 2014), only selecting variables with VIF < 10.Although we recognize that this selection could remove a variable that is more direct for a species' niche than a highly collinear one that was selected, we discuss the importance of variables, taking into consideration groups of similar variables.From this initial selection, we kept the following non-collinear variables: Bio 10, Bio 13, Bio 14, Bio 18, Bio 19, Bio 2, Bio 3, Bio 8 and relief roughness.
To project the models into past climates, we also obtained the same set of variables estimated for Middle Holocene (6 kyr), the Last Glacial Maximum (21 kyr, LGM), and the Last Interglacial (120 kyr, LIG) from the PaleoClim database (Brown et al. 2018).All variables were cropped to a 5-degree buffer around the species-points, which defined the background and projecting area for the models.
To decrease spatial autocorrelation due to the clustering of occurrence records in areas with denser sampling effort, we performed spatial thinning, keeping unique occurrence records at a distance of at least 15 km apart with the spThin v.3.6 R package (Aiello-Lammens et al. 2015).

Modeling procedure
We used the ENMeval R package (Muscarella et al. 2014) for the modeling procedure below.First, we performed a model selection for the MaxEnt algorithm (Hijmans et al. 2017;Phillips et al. 2006) with several distinct parameters.We used regularization multipliers ranging from 0.5 to 5 (increments of 0.5) and three feature class combinations ("L", "LQ", "LQP"; where L = linear, Q = quadratic, P = product), but avoiding more complex ones (e.g., hinge, threshold) as we were projecting models to a different time.Analyses were performed with the ENMevaluate function with 1,000 points of pseudo-absence (background) and partitioning the data in test and training with "checkerboard2" for the cross-validation bins.This data partition scheme further reduces the effects of spatial sampling bias and spatial autocorrelation (Muscarella et al. 2014).Variable importance and their contributions to the model's performance were obtained through the jackknife tests with the ENMevaluate function (Phillips et al. 2006).Model performance was evaluated with the area under the curve (AUC) and the true skill statistics (TSS).We then selected the models with delta-AICc lower than 10, averaging the selected models by AUC (if more than one) for model projection (both current and past conditions).

Niche overlap
We used the ecospat R package and the functions niche.overlap, niche.equivalency.test,niche.similarity.testfor measuring and testing niche overlap between species (Broennimann et al. 2023).We measured the niche overlap observed for each species-pair with Schoener D metric (Schoener 1968), which ranges from 0 (no overlap/ complete divergence) to 1 (complete/high overlay) and we tested the significance of the metric against two different null distributions (niche equivalence and niche similarity; Warren et al. 2008).The equivalence tests evaluate if there is significant overlap in niches as determined by the environmental conditions in the species' records, whereas similarity tests assess if the niches of one species significantly overlaps with the conditions found across the entire range of the other species (i.e., not only across the species records).Niche equivalence is rejected if observed niche overlap falls within the 95% of the null distribution.Niche similarity is rejected if niche overlap is lower than 95% of null the distribution values.We used the PCAenv approach to estimate the environmental space occupied by each species across the same set of variables used in the niche modeling following Broennimann et al. (2012) and Di Cola et al. (2017).For the Baturité Massif sample, we used a small buffer of 0.1 degrees around the records to gather enough environmental information for the analyses.

Phylogenetic Analysis and Genetic Distances
For the Chironius bicarinatus complex, 48 new sequences for 16 specimens were generated, with 15 sequences for each 12S and 16S fragment, 14 for c-mos and 4 for ND4 (see Appendix 3).The total aligned length of the dataset was 2,166 base pairs (bp).We retrieved a tree topology based on BI recovering at least twelve species within Chironius (C. bicarinatus complex, C. brazili, C. flavolineatus, C. multiventris, C. exoletus, C. flavopictus, C. monticola, C. laevicollis, C. scurrulus, C. fuscus, C. grandisquamis and C. challenger) as monophyletic (pp = 1).Additionally, we recovered Baturité Massif sam-ple, nested within the Chironius bicarinatus clade, strongly supported (pp = 1) as the sister-group of the clade C. bicarinatus + C. gouveai (Fig. 1), which was congruent with the grouping based on morphological characters (see below).All aforementioned relationships are supported by high posterior probabilities (pp > 0.95).However, since checking the identification of most of the sequences of the other Chironius spp.available on GenBank is beyond the aim of the present study, except those of the C. bicarinatus complex, we refrain from commenting on the interrelationships retrieved (see Fig. S1 for the detailed phylogenetic tree).Regarding the genetic differences calculated from the uncorrected pairwise genetic distances for sequences of 16S rRNA, Chironius species differed from other colubrid genera from 5.7-14.0%.Within the C. bicarinatus complex, genetic distances range from 0.4-4.0%,and the complex distance from its congeners from 2.0-10.0%.Chironius bicarinatus is 0.4-3.7%divergent from C. gouveai, while the Baturité Massif sample is 1.9-4.0%divergent from C. gouveai and 1.5-2.8%from C. bicarinatus.Within the four sequences of Baturité Massif sample, distance ranges from 0.2-0.7%.See Table S1 for the detailed genetic differences among species of Chironius for sequences of 16S rRNA.

Quantitative Analyses
Table 1 shows the variability in meristic and morphometric characters for each defined group in the Chironius bicarinatus complex.A detailed report of the morphological data of specimens examined is provided as Supplementary Material (Table S2).We detected sexual dimorphism in all the characters analyzed for at least one of the three operational taxonomic groups.Ventral counts were higher in females of C. bicarinatus (U = 5083.000;median = 158 and 157, for females and males, respectively, hereafter; p < 0.01), C. gouveai (F 1,138 = 76.54;p < 0.0001), and Baturité Massif sample (U = 130.000; median = 156 and 154; p = 0.0001); subcaudal counts and CL/TL ra- tio were only higher in males of Baturité Massif sample (U = 8.000 ; median = 128.5 and 133; p < 0.01, and U =  18.000; median = 0.35 and 0.36; p < 0.05, respectively).
Table 2 reports differences between the groups for ventral and subcaudal counts, CL/TL ratio and number of maxillary teeth.

Cranial Osteology
For skull analyses, we used four specimens that were previously dry prepared (two of Chironius bicarinatus and two of C. gouveai) and three scanned specimens repre- Table 3.Values (%) of the correct reclassification rate and cross validation analysis (in parentheses) for females and males from the three determined groups of the Chironius bicarinatus complex (Group 1: C. bicarinatus, Group 2: C. gouveai, Group 3: Baturité Massif sample).analyses were based on dry preparations of the specimens MCP 4305 (adult male) and MCP 19414; and a scanned specimen ZUFMS 1686 (adult male, SVL 929 mm), as well as the skull information from Chironius gouveai paratype (ZFMK 103132, adult male, SVL 720 mm) in We did not find differences between the skulls of Chironius bicarinatus and Baturité Massif sample.On the other hand, we found four differences comparing Baturité Massif sample with C. gouveai (in parentheses): ventral surface of the septomaxilla smooth (vs.presence of a conspicuous projection); anteroventral surface of prefrontal lacrimal foramen smooth (vs.presence of a conspicuous projection); posterior end of supratemporals straight (vs.slightly laterally curved); and palatine teeth 24 (vs. 16-18).

Ecological Niche Modeling
The model selection of the Maxent parameters for Chironius bicarinatus provided three models with delta-AICc lower than 10 (Table S3).The averaged model presented an AUC = 0.976 and TSS = 0.85, indicating excellent performance.Areas of high suitability for the current climate closely matched the presence records across lowland Atlantic Forest in the Ombrophilous Dense Forests of southern Bahia and Espírito Santo, Rio de Janeiro and São Paulo States.Model projection to the Last Glacial Maximum showed a potential retraction of suitable areas along the coast of southeastern Brazil and a potential expansion of suitable areas across central regions of Bahia and Pernambuco States.Model projection to the Last Interglacial showed a potential retraction of suitable areas across central regions of Bahia and Pernambuco and a potential expansion of suitable areas along the coast of southern Bahia, Espírito Santo, Rio de Janeiro, São Paulo and Paraná States, as well as small areas of the northeastern Brazilian coast (Fig. 5A).The most important variable for C. bicarinatus was Isothermality (Bio 2), followed by Precipitation of Warmest Quarter (Bio 18 -Table S4).
For Chironius gouveai, we obtained four models with delta-AICc lower than 10 (Table S5).The averaged model had AUC = 0.950 and TSS = 0.801, also indicating excellent performance of the modeling procedure.Areas of high suitability in the current climate closely matched the presence records across inland Atlantic Forest in the Mixed Forests of central Paraná and Santa Catarina, and northern Rio Grande do Sul States.Model projection to the Last Glacial Maximum showed practically the same potential distribution and suitable areas in the current climate.Model projection to the Last Interglacial only showed a potential expansion of suitable areas across small areas along São Paulo coast (Fig. 5B).The most important variable for C. gouveai was by far the Precipitation of Driest Month (Bio 14), followed by other precipitation variables (Table S6).
Niche overlap was low or zero between all pairs of groups (D < 0.08) and the null tests showed nonequivalence or niche similarity, indicating that all three species occupy significantly distinctive environmental niches across the exact species occurrences, as well as their respective ranges (Table 4).

Taxonomic Decisions
The data gathered from a geographically representative sample covering most of the variability in the traditional external morphological characters of the Chironius bicarinatus complex, in addition to the clades retrieved in molecular phylogeny and unique niches found by ecological evidence, make it possible to recognize and objectively attribute, two species with available names (C.bicarina- tus and C. gouveai), and diagnose a distinct lineage without an available name (Baturité Massif sample).Herein, we first redefined C. bicarinatus, inferring its morphological limits with C. gouveai and then, we formally describe the Baturité Massif sample as a new species.
On a sand beach of Lagoa, near the Rio Jucú, within "5 legoas" (leagues) of Villa do Espírito Santo (currently Vila Velha), state of Espírito Santo, Brazil.Holotype.Adult male, apparently lost in the American Museum of Natural History (Vanzolini and Myers 2015) from a locality near Lagoa Grande (20°30'08.3″S,40°21'32.4″W;~2 m above sea level; hereafter a.s.l.), municipality of Vila Velha, state of Espírito Santo, Brazil.The data reported by Wied in the original description and in later studies (Wied 1824: color plate, Vol. 8 plate 2; 1825: 284) are adequate for the recognition of the species and its type locality, with the designation of a neotype being deemed unnecessary (ICZN 1999).The Atlantic populations of Chironius cf.exoletus (in parentheses) are the only ones that can present the aforementioned characters and a color pattern similar to C. bicarinatus.Specimens of Chironius cf.exoletus that have 12/12/10 dorsal scale rows are seldom observed (2% in our sample, with 12/12/8 being more frequent).Still, C. bicarinatus can also be differentiated by the number of maxillary teeth 30-38 [33.8-34.4](vs.23-31); outer margins of subcaudals pigmented black (vs.black pigmentation absent); ventral color pattern of tail usually with a black zig-zag line medially between subcaudals, gradually fading to the tip of the tail (vs.black zig-zag line maintaining the same intensity up to the tip of the tail); and hemipenial body elongated covered with more concentrated spines (vs.hemipenial body short with lower concentration of spines; see an illustration of the hemipenes in the supplementary information S1 of Klackzo et al. 2014).
Chironius  Color pattern variation in preservative.Adult specimens with uniformly olive, grayish olive or bluish dorsum and light vertebral and postocular stripes that may or may not be evident, sometimes with black outer margins bordering it, more visible on the anterior part of the body.Predominantly pale yellowish labials, except for the last two or three supralabials, which may present the same dorsal or postocular stripe color.Ventrally, gular region, first third of the belly, and subcaudals pale yellowish; the remainder of the belly pale yellowish or bluish.Most have outer subcaudal margins with a black outline and a medially positioned black zig-zag line between subcaudals (gradually fading to the tip of the tail), which may vary in intensity and may even be absent in some specimens.Juvenile specimens have the same color pattern variation as adults, but they can also have a uniform olive brown dorsum and light crossbands on dorsum.In our sample, the presence of light crossbands on dorsum was found in juvenile specimens with a maximum SVL of 373 mm (CHUFJF 523) and in only one adult specimen (MNRJ 1839, SVL 830 mm), collected in the municipality of Passa Quatro, state of Minas Gerais, Brazil.
Color pattern while alive (Fig. 6A, B).The description of color pattern while alive is based on photographs of six adult specimens (MNRJ 19138, 22017, 23571, 24194, 24877, 26255), and a juvenile specimen (MNRJ 27463), all collected in several localities of the state of Rio de Janeiro, Brazil; as well as on photographic material of other unvouchered specimens (iNaturalist, n = 9; see Appendix 2).Adult specimens have a uniformly green, olive or grayish olive dorsum with a light vertebral, sometimes with black outer margins bordering it, more visible on the anterior part of the body, and a black postocular stripe.Predominantly yellowish supralabials, except for the last two or three, which may present the same dorsal or postocular stripe color.Infralabials and gular region mostly whitish or yellowish; the first third of belly and subcaudals yellowish; the remainder of the belly whitish yellow.As in preserved specimens, while alive juvenile specimens also have a uniform olive brown dorsum and light crossbands on dorsum.
Hemipenial morphology (Fig. 7A).We analyzed the hemipenes of 23 specimens of Chironius bicarinatus, five of which were extracted from the specimens and manually fully everted, rendering almost or virtually maximally expanded organs.et al. (1993), further inland in the state of Bahia, municipality of Andaraí.We verified these records in the Institution's catalog, which contained the following data: "BRA Bahia Andarahy pequeno > Andahary pequeno" donated by "Massini, Hans, Rio de Janeiro".Thus, we can conclude that it is more likely that these records are from a location in the state of Rio de Janeiro, known as Tijuca (formerly known as Andaraí Pequeno), since the donor was a resident of this neighborhood.and Francini (1991: 63), Cei (1993: 538), Dixon et al. (1993: 59, in part), Achaval and Olmos (1997: 68), Giraudo (2001: 42) 27-36 [30.5-31.4] (vs. 30-38 [33.8-34.4]);color pattern in preservative with dorsum usually uniform light brown, grayish olive or bluish, and dorsal and ventral scales with black or brown edges without a zig-zag line (vs.uniformly olive, grayish olive or bluish dorsum and a black zig-zag line medially between subcaudals, gradually fading to the tip of tail; see Entiauspe-Neto et al. 2020: fig. 3 for an illustration of this character); temporal formula usually 1+1 (vs.1+2); ventral surface of septomaxilla with a conspicuous projection (vs.septomaxilla smooth); anteroventral surface of prefrontal lacrimal foramen with a conspicuous projection (vs.anteroventral surface of lacrimal foramen smooth); and posterior end of supratemporals slightly laterally curved (vs.straight).

Chironius gouveai Entiauspe-Neto et al., 2020
Color pattern variation in preservative.Most adult specimens have a uniform light brown, grayish olive or bluish dorsum with a light vertebral stripe, that may or may not be evident, depending on the preserved condition, sometimes with black outer margins bordering it, more visible on the anterior part of the body.It is also possible to notice the presence of a black postocular stripe in better preserved specimens.Dorsals and ventrals with black or brown edges, more visible on the posterior part of the belly.Predominantly pale yellowish or whitish labials, except for the last two or three supralabials, which may present the same dorsal or postocular stripe color.Gular region, first third of belly, and subcaudals (which can also be whitish) pale yellowish; the remainder of the belly pale yellowish, olive or bluish.The outer margins of subcaudals have a black outline, which may vary in intensity and may even be absent in some specimens, and subcaudals have black or brown edges.Some specimens have remnants of black edges on subcaudals and only medially positioned black zig-zag line between subcaudals is evident.
The populations near Serra do Mar, in the states of Paraná and Santa Catarina (e.g., FML 1816, CHUFSC 244, 625, 788-90, 898, 1035, 1105, 2976-77), have a color pattern that is more similar to C. bicarinatus than those that occur in further inland locations.As with the C. bicarinatus specimens, they have a uniform olive or grayish olive dorsum; Gular region, first third of the belly, and subcaudals pale yellowish; the remainder of the belly pale yellowish or bluish; no black or brown edges on dorsals and ventrals; and the only distinguishable feature would be an evident black zig-zag line medially positioned between subcaudals with more intensity compared to C. bicarinatus, also gradually fading to the tip of the tail.
Juvenile specimens have the same color pattern variation as adults and juvenile specimens of C. bicarinatus, differing in the greater intensity of the black zig-zag line medially positioned between subcaudals or in the remnants of black or brown edges on dorsals and ventrals; neither may be evident in some preserved specimens, and it is therefore, only possible to distinguish juveniles of the related species by the combination of some meristic characters (e.g., temporal formula, number of subcaudals or maxillary teeth).
Color pattern while alive (Fig. 6C, D).The description of color pattern while alive is based on photographs of an adult specimen (MACN 38879) collected in the province of Misiones, Argentina; a juvenile specimen (unvouchered specimen) from the municipality of Vacaria, state of Rio Grande do Sul, Brazil, as well as on photographic material of other unvouchered specimens (iNaturalist, n = 9; see Appendix 2).
Adult specimens with uniform light brown, grayish olive or olive dorsum and a light vertebral stripe, sometimes with black outer margins bordering it, more visible on the anterior part of the body.Dorsals and ventrals with black edges, more visible on the posterior part of the belly.Predominantly yellowish or whitish supralabials, except for the last two or three, which may present the same dorsal or postocular stripe color.Infralabials and gular region mostly yellowish or whitish; the first third of the belly (can also be whitish) and subcaudals yellowish or brownish yellow; the remainder of the belly whitish yellow or brownish yellow.For most specimens, the outer margins of subcaudals have a black outline and subcaudals have black edges.
Hemipenial morphology (Fig. 7B).We analyzed the hemipenes of 19 specimens, 13 of which were extracted from the specimens and prepared, rendering a fully everted and almost or virtually maximally expanded organs.The description is based on the fully everted and maximally expanded left organs of the specimens MCP 17282, MHNCI 12369, UFMT 1600, partially expanded organ of the specimen MHNCI 1529, and non-maximally expanded organs of the specimens MHNCI 7134, ZVC 2041, ZVC 3681; and on the fully everted and maximally expanded right organ of the specimen MCP 2423, and partially expanded organs of the specimens MCN 3144, MCN 14090.Retracted organs extend to the level of the twelfth subcaudal (n = 3).
Hemipenis unilobed, unicalyculate, noncapitate, subcylindrical shape, with a simple sulcus spermaticus, running centripetally from the base to the apex (MCN 3144) or slightly more than half of the hemipenis (MHNCI 12369, UFMT 1600); no nude area on the apex, ornamented with papillate calyces in the sulcate side and with spinulate calyces in the middle of the asulcate side; hemipenis eventually showing calyces with few papillae on the apex and medial region (MHNCI 12369, UFMT 1600); each longitudinal row presenting 24-27 calyces; calyces towards hemipenial body replaced by spinulate calyces; hemipenial body with approximately more than half of the total length of the organ, and is covered in spines that gradually increase in size towards the base, reaching maximum size at just over half of the hemipenial body; with each longitudinal row presenting 14-19 spines and 5-6 spines along the sulcus spermaticus; base mostly nude, ornamented by spinules at the upper portion and also laterally distributed on the proximal region of the sulcus spermaticus; a basal naked pocket present on the medial region.
Distribution (Fig. 8).Chironius gouveai is widely distributed across inland Serra do Mar at higher altitudes in the Mixed and Semideciduous Forests, from the Paranapanema River (northernmost record in the municipality of Londrina; 23°18′34.6″S,51°10′26.4″W)to the Uruguayan pampas (southernmost record in the province of Río Negro, M'Bopicua Port; 33°06′38.7″S,58°11′31.6″W),with its western limit by the Paraná River (westernmost record in the province of Corrientes, Itá Ibaté, Argentina; 27°25′43.8″S,57°20′17.0″W),between sea level to ~1030 m a.s.l.. (ZVC 2041, 2956, 3681, 3727, MNHN 78, 1730, 1794, 5707) in the departments of Artigas, Río Negro and Salto, western Uruguay, can be explained by the attempt of specimens to cross the Uruguay River and their transportation by vegetation rafts along the river bed to the La Plata River (see Achaval et al. 1979).The record of C. gouveai (MHNCI 6028) from Montevideo needs to be clarified to confirm the occurrence of the species in southern Uruguay.

Description of the holotype (Figs
Uniform grayish olive dorsum with a light vertebral stripe and two black dorsolateral stripes on each side visible from neck to midbody (until the 75 th ventral); presence of a black postocular stripe; snout (rostral, nasals and internasals) light brown; predominantly whitish labials, except for the last supralabial, which presents the same postocular stripe color.Gular region, the first third of the belly, near cloacal region and subcaudals whitish; the remainder of the belly olive; subcaudals with slightly black edges.
Color pattern variation in preservative.Adult specimens have a uniform bluish dorsum with a light vertebral stripe and two dorsolateral black stripes, laterally visible from neck to midbody, which may or may not be evident, depending on the preserved condition, as well as the presence of a black postocular stripe.The snout can have the same dorsal color; predominantly whitish labials, except for the last two supralabials, which may present the same dorsal or postocular stripe color.Gular region, the first third of the belly, near cloacal region and subcaudals whitish; the remainder of the belly may be bluish.Juvenile specimens have the same color pattern variation present in adults, but they have a uniform olive brown dorsum and light crossbands along the dorsum.In our sample, the presence of light crossbands on the dorsum was found in juvenile specimens with a maximum SVL of 257 mm (CHUFC 3226) from Sítio São José, Pacoti, Ceará.
Color pattern while alive (Fig. 12).The description of color pattern while alive is based on photographs of adult specimens MNRJ 27803 and MHNCE-R 577 from Pacoti and IBSP 76994 or 76995 from Sítio Álvaro, Guaramiranga, both in Ceará state.The latter specimens were destroyed in a fire on May 15, 2010, only leaving photographic material of one of the specimens while alive.Adult specimens of Chironius dracomaris have a uniform green or olive dorsum; snout and first supralabials without contact with the orbit light brown; other labial scales and gular region whitish; the first third of the belly, near cloacal region and subcaudals yellowish; the remainder of the belly greenish.
Hemipenial morphology (Fig. 13A).Based on the fully everted and maximally expanded left organ of the specimens MNRJ 27717 and MNRJ 27804, partially expanded organ of the specimen MNRJ 27718; and on the fully everted and maximally expanded right organ of the specimen MNRJ 27716, and partially everted organ of the specimen CHUFPB 17304 (n = 5).Retracted organs extend to level of eighth or ninth subcaudal (n = 3).Hemipenis unilobed, unicalyculate, noncapitate, subcylindrical shape, with a simple sulcus spermaticus, running centripetally from the base to slightly more than half of the hemipenis, not reaching the apex (MNRJ 27717); no nude area on the apex, which is ornamented with papillate calyces on the sulcate side and with spinulate calyces in the middle of the asulcate side; hemipenis also presenting calyces with few papillae on proximal region of the end of sulcus; each longitudinal row has 23-29 calyces; calyces towards hemipenial body replaced by spinulate calyces; hemipenial body represents approximately more than half of the total length of the organ, and is covered in spines that gradually increase in size toward the base, reaching maximum size at just over half of the body; each longitudinal row has 15-22 spines and 5-8 spines along sulcus spermaticus; base mostly nude, ornamented by spinules on the upper portion and also laterally distributed on the proximal region of sulcus spermaticus; a basal naked pocket present on the medial region.
Cranial osteology (Fig. 3G-I, P-R).The description of the skull of Chironius dracomaris is based on a scanned specimen MNRJ 27717 (paratype, SVL 680 mm).Snout Complex.Premaxilla: subtriangular in frontal view, with the ascending process oriented posterodorsally without contacting anterior end of nasals; in lateral view, anterior edge of the ascending process straight; transverse process elongated and projected laterally, slightly shorter than ascending process (sensu Klaczko et al. 2014); in ventral view, vomerine processes long (sensu Klaczko et al. 2014), posteriorly oriented, does not contact anterior end of vomers; septomaxillae: strongly convex dorsomedially, with long ascending conchal process extending anterolaterally, does not reach the ends of nasals; slightly separated from each other; prominent projection on the posterolateral edge of vomeronasal cupola of the septomaxilla, approaching frontals posteriorly, one side in contact with the frontal and the other slightly separated; posterior portion contacts vomers ventrally; absence of a conspicuous extension in the medial portion of septomaxilla ventrally oriented; nasals: in contact medially, with anterior processes tapered; no contact with premaxilla; oval anterolateral processes (sensu Klaczko et al. 2014); posterior processes almost contact anterior part of frontals; vertical lamina do not contact septomaxillae; vomers: with globular mesoventral portion with rounded opening corresponding to exochoanal fenestra; in this region, vomers approach medially and almost contact anterior region of the palatine laterally; large vertical posteromedial laminae with a large circular fenestra; also presents foramina in vertical laminae in the antero and mesoposterior portions; vomeronasal cupula with a tapered process projected posterodorsally above circular fenestra.Braincase.Prefrontals: in contact with frontals dorsally, and approaching the maxillary process of palatine making little contact on only one side ventrally; in anterodorsal view, they present a process directed medially in basal portion; in lateral view, lacrimal foramen visible in basal portion; broad and rounded anterior process; posterior portion concave, forming anterior border of orbital cavity; large lacrimal foramen without a conspicuous projection on the anteroventral surface; frontals: in dorsal view, frontals in contact, with a straight medial suture; anterior margin rounded; anterolateral margins oblique and in contact with prefrontals; posterolateral margins slightly curved, forming the dorsal margin of orbital cavity; suture contacting parietals oblique, with the presence of foramina in the posterolateral region of frontals; vertical laminae in contact with parabasisphenoid ventrally and septomaxilla anteriorly; parietal: in dorsal view, almost rounded, as long as broad, with depressions on the mesoanterior and lateral margins, forming ridges on the lateral margins; anterior margins oblique in contact with frontals; also form posterodorsal border of orbital cavity; it contacts postorbital anterolaterally; posterior margins rounded in contact with supraoccipital medially; in lateral and ventral views, contacts posterior portion of parasphenoid rostrum anteriorly and basisphenoid lateromedially; posterolaterally contacting anterior margin of prootics and supratemporals; postorbitals: long, slightly curved, forming the posterolateral margin of the orbital cavity; they contact lateral process of parietal dorsomedially, but do not make contact in the basal region; no contact with frontals and broadly separated from ectopterygoid; supraoccipital: in dorsal view, subpentagonal, broader than long, oblique anterior and posterior margins, in contact with parietal anteriorly, with prootics laterally and exoccipitals posteriorly positioned; a constriction in the mesolateral region creates a depression with a foramen in the posterior part; conspicuous medial crests, between these crests emerges a pronounced medial crest perpendicularly in the posterior region, separating two depressions, each one presenting a foramen in basal region; exoccipitals: with a medial constriction and continuing lateral ridges; contact supraoccipital anteriorly, prootics and supratemporals laterally, and basioccipital ventrally; in lateral view, form posterior margin of fenestrae ovalis in contact with prootic anteriorly; presence of foramina ventrally and laterally on posterior margin; also forms the dorsal, lateral, and lateroventral margins of the foramen magnum; columella directed to the inner surface of quadrate; basioccipital: hexagonal, contacting parabasisphenoid complex anteriorly, prootics and exoccipitals, anterolaterally and posterolaterally, respectively; also forms ventral margin of foramen magnum posteriorly; with two conspicuous medial ridges; a larger perpendicular crest emerges between these ridges that extends to the posterior region; prootics: overlain by supratemporals in dorsal region; contacts parietal dorsally and anterolaterally, parabasisphenoid complex and basioccipital ventrally, supraoccipital dorsoposteriorly, and exoccipital posterolaterally; in lateral view, each with two large and three small fo- ramina (two below the large foramina and one above); posteriorly, form anterior margin of fenestrae ovalis in contact with exoccipitals; parabasisphenoid complex: basisphenoid and parasphenoid fused in ventral view, but separate in dorsal view; in dorsal view, basisphenoid trapezoidal shaped, with a large median cavity, and spear shaped parasphenoid, tapering from the level of anterior margin of ectopterygoid, surpassing the palatine choanal process; basisphenoid contacts basioccipital posteriorly, prootic and parietal posterodorsally; parasphenoid contacts frontal dorsoanteriorly, and approaches, but does not contact posterior vomers and choanal process of palatine anteroventrally; in dorsal view, parasphenoid rostrum with lateral groove on each side along its length, with extension on medial part; in ventral view, posterior part of parasphenoid rostrum presents an extension laterally; in ventral view, basisphenoid has two lateral ridges, one extending to suture of prootic and the other to the half where two foramina are present; two further foramina are present on each side of basisphenoid anterior to the ridges.
Palatomaxillary arch.Maxillae: elongated (sensu Klaczko et al. 2014), extends to the level of vomerine premaxilla processes to posterior border of postorbital, surpassing the palatine medially; in dorsal view, arched towards premaxilla in anterior portion, with a constriction at the level of prefrontal, which present a lateral extension approaching palatine; presence of a foramen anteriorly to the lateral extension; posterior portion with a lateral extension with a slight overlap with ectopterygoid, but without contact, forming the inferior margin of orbital cavity; ventral surface with 34-36 subequal, curved, and rear facing teeth, increasing in size towards the last five teeth; ectopterygoids: narrow, slightly curved; semicircle-shaped anterior margin with a rectangular lateral process, that overlaps maxilla; slightly curved posterior process; ventral surface of posterior portion in contact with dorsolateral surface of pterygoid medially; pterygoids: elongated, corresponding approximately to slightly more than half the size of braincase; ventral surface with 30-31 subequal, curved, and rear facing teeth; anterior portion tapered, presenting a slight overlap with posterior portion of palatine and gradually widening from the region of contact with ectopterygoid in anteroposterior direction, tapering again near the rounded posterior end, where both move away in curvature approaching quadrate-compound bone joint laterally; dorsal surface with a lateral longitudinal ridge from the posterior level of contact with ectopterygoid, extending almost to the end of the broad posterior region; palatines: slender, almost straight; ventral surface with 24 subequal, slightly curved, and rear facing teeth; in dorsal view, almost contacting globular portion of vomers in anterior portion; choanal process dorsomedially directed toward parasphenoid rostrum, but without contact; short maxillary process situated at level of anterior border of choanal process on lateral surface of palatine, directed posterolaterally, contacting small portion of prefrontal ventromedial surface, but no contact with palatine process of maxilla; presence of a foramen in the medial region of maxillary process; a short bifurcation before the pterygoid contact zone in posterior portion, with medial branch longer than lateral branch; the medial branch overlaps and contacts the anterior border of pterygoid.
Suspensorium and mandible.Supratemporals: subrectangular, elongated, slightly curved lateromedially; anterior portion does not surpass parietal-prootic suture; overlapping and contacting prootic and exoccipital laterally, with half of supratemporal articulating with quadrate laterally; posterior end straight, surpassing the posterior portion of exoccipital; quadrates: flattened and broad dorsally, tapering and slightly curving dorsoventrally; dorsal portion contacting supratemporal laterally; medial portion with short process, near columella auris; ventral portion articulates with compound bone; ventromedial portion approaching, but not in contact with pterygoid; dentaries: medially curved anteriorly; dorsal surface with 35-36 subequal, slightly curved, rear facing teeth, decreasing in size posteriorly; lateral face convex with a mental foramen medially on one side and ventrally on the other; posterior to this foramen, a bifurcation of dorsal and ventral processes laid over the compound bones in anterior part, with dorsal process longer than ventral; again, dorsal process bifurcates a short medial process, with presence of dentition only in the longer process; ventral process contacts splenial and angular, in addition to compound bone; Meckelian fossa is delimited by the dentary and splenial; splenials: elongated, tapered anteriorly, in the region of contact with dentary; posterior part in contact with anterior region of angular; anterior mylohyoid foramen situated below dorsal extension; angulars: elongated, tapered posteriorly, contacting compound bone laterally, and dentary and splenial anteroventrally; posterior mylohyoid foramen near suture with splenial; in ventral view, angular˗splenial joint visible on one side; compound bones: elongated, approximately two-thirds length of mandible; in lateral view, tapers anteriorly; in contact with dentary posteriorly; presence of a foramen laterally to the posterior tip of angular; prearticular crest prominent, distinctly higher than surangular crest; a rounded depression where it articulates with the quadrate; retroarticular process medially directed.
Etymology.The specific epithet dracomaris is the conjunction of two nouns "draco" (nominative) and "maris" (genitive), used in apposition with the Latinized nickname "Dragão do Mar" (= Dragon of the Sea, in English), as Francisco José do Nascimento (1839Nascimento ( -1914) ) became historically known.He was a popular leader of the harbor pilots in Ceará, who became a symbol of Northeastern resistance against slavery in Brazil.In 1881, "Dragão do Mar" led one of the main port stoppage movements, refusing to transport slaves to the ships, thus preventing interprovincial trafficking.The successive closures of the port accelerated abolitionism in the region, which made Ceará the first Brazilian province to banish slavery, on March 25, 1884, four years before the signing of the Lei Aurea (Morel 1967).The recent worldwide manifestations (Black Lives Matter movement) are proof that we need to change and repair social and historical injustices (see Subbaraman 2020).The fight against modern slavery and structural racism still permeates in Brazil.For example, in the city where "Dragão do Mar" was born (Aracati, Ceará), there are social-environmental conflicts over the traditional use of the territory, as well as in several other areas (Nascimento 2018; Quilombola community of Cumbe).In 2019, the winning Samba of Rio de Janeiro's Carnival-História Para Ninar Gente Grande from the school Estação Primeira de Mangueira-honored "Dragão do Mar", as well as other symbols of black and peripheral resistance against exploitation in Brazil, for example the councilor Marielle Franco, who was brutally murdered by the Rio de Janeiro militia on March 14, 2018.We dedicate this species to Francisco José do Nascimento "Dragão do Mar" and these leaders who inspire the daily struggle of restoring a democratic environment in Brazil.
Distribution (Fig. 8).Based on the current evidence, Chironius dracomaris is restricted to the "Brejo de altitude" from Baturité Massif (known records in the municipalities of Baturité, Guaramiranga, and Pacoti), enclaves of remaining rainforests in the dry Caatinga domain of northeastern Brazil, state of Ceará, between 500-800 m a.s.l.. Remarks.Loebmann and Haddad (2010) recorded a specimen tentatively identified as Chironius bicarinatus (IBSP 77076) from the Ibiapaba Plateau, the westernmost rainforest fragments of the "Brejo de Altitude" in the state of Ceará.Unfortunately, the voucher was destroyed in a fire on May 15, 2010 (Franco 2012), and it was therefore, not possible to check the accuracy of its identification and thus confirm the occurrence of C. dracomaris also in the Ibiapaba Plateau.

Discussion
The main taxonomic reviews that have addressed the name Chironius bicarinatus (Bailey 1955;Dixon et al. 1993), used composite sampling that included specimens, later described by Entiauspe-Neto et al. ( 2020), as C. gouveai.Nonetheless, even Dixon et al. (1993) suspected that the southern populations could represent a new taxon, presenting differences in color pattern, the sum of segmental counts, number of maxillary teeth, frequency of temporals 1+1, and the occurrence into a distinct phytophysiognomy.All these morphological features and ecological preferences were corroborated in the present study, in addition to molecular evidence of a distinct lineage herein described and named as C. dracomaris sp.nov.(Figs 1-5).Chironius dracomaris sp.nov. is distinguished from C. bicarinatus and C. gouveai by the combination of qualitative and quantitative morphological characters (general body coloration, meristic, body proportion, frequency of temporals and apical pits along the body, and cranial osteology with respect to C. gouveai).Table 5 shows comparisons of the most frequent qualitative and osteological characters among the species.Some information (diagnosis, geographical distribution, and material examined) present in the original description of Chironius gouveai (Entiauspe-Neto et al. 2020) should be considered with caution.The frequency of many characters mentioned in the topics of "emended diagnosis" and "description" for C. bicarinatus and "definition and diagnosis" for C. gouveai differ from our own results (Table 6).In addition, most of the characters used by Entiauspe-Neto et al. (2020) in the comparisons between C. gouveai and C. bicarinatus did not have their polymorphism levels tested properly with dense geographic sampling.They distinguish C. gouveai from C. bicarinatus (in parentheses) based on the following characteristics: "white gular coloration in life (vs.yellow), reticulated dorsal pattern in adults (vs.uniform green), dorsal pattern with scattered black blotches in juveniles (vs.black transversal bars), absent or vestigial postocular stripe (vs.present on adults and juveniles, rarely vestigial on juveniles), hemipenis smooth calyculate apex (vs.spinulate calyces)" (Entiauspe-Neto et al. 2020).These features observed from specimens while alive did not have the polymorphism levels tested due to the small sample size, while hemipenial morphology characters are polymorphic in our sample, and are not fixed in any populations of either species.With respect to the osteological characters used in the comparisons by Entiauspe-Neto et al. (2020), three characters (presence of a conspicuous projection on the ventral surface of the septomaxilla; presence of a conspicuous projection on anteroventral surface of prefrontal lacrimal foramen; and posterior portion of supratemporals slightly curved) seem to be fixed for C. gouveai, while the character anterior portion of the supratemporal bones surpassing the parietal-prootic suture is polymorphic in our sample.Despite the small sample size, we found an additional difference related to the number of palatine teeth (23-24 teeth in C. bicarinatus vs. 16-18 in C. gouveai).parapatric, presenting sympatry in several localities.This distribution map differs greatly from the one herein presented (Fig. 8), which was constructed based solely on records of specimens that were examined first-hand, except for ten additional specimens that were examined through photographs (Appendix 1).Although both species present distribution borders that are close to one another, their niche overlap is low, which indicates that niche divergence may be involved in the evolution of the group (Fig. 8, Table 4; see Wiens and Graham 2005).We examined 46 specimens included in the list of material examined by Entiauspe-Neto et al. ( 2020), of which 23 (50%) were redetermined (Table 6).From the geographic distribution (western Paraná State), we estimate that another 39 specimens identified as C. bicarinatus by Entiauspe-Neto et al. ( 2020), which we were not able to examine, may actually be C. gouveai.Therefore, we disagree with these identifications given by Entiauspe-Neto et al. ( 2020) and we believe that they did not examine the number of specimens mentioned, which is clearly perceptible based on the number of specimens used in the comparative table of variation between C. gouveai and C. bicarinatus ("table 1", n = 79), when compared to the list of specimens examined ("Materials and Methods" and "Appendix II -Specimens examined", n = 279).Thus, it seems as though the description of C. gouveai was broadly based on photographs and unreliable information in the literature that was cited as examined material and used for comparisons.Therefore, we emphasize the importance of taxonomic efforts with denser sampling and the verification of the literature records (Willis 2003;Passos et al. 2022).The power of a species' diagnostic characters, delimitation boundaries with its closest related species and the knowledge of its geographical distribution, are more useful than species descriptions, which must be the main aim in taxonomic studies.Many endemic reptile and amphibian species have been described in the "Brejos de Altitude" (i.e., isolated remnants of rainforest restricted to highlands throughout the xerophytic Caatinga domain), mostly from the Baturité Massif in Ceará State (e.g., Hoogmoed et al. 1994 ararype;and, Roberto et al. 2022 for Pristimantis relictus).In fact, there is already a report in the literature of a possible new species of the genus Chironius occurring in a "Brejo de Altitude".Hamdan et al. (2017) highlighted the populations currently assigned to C. flavolineatus from Serra da Ibiapaba, state of Ceará, as a candidate species inferred from a coalescent-based phylogeny covering the entire genus.Thus, further investigations are being made by Hamdan et al. (2017) for the formal recognition (or not) of this lineage.The biota inhabiting these relictual forests in northeastern Brazil is also a result of historical connections between enclaves and the Amazon and/ or Atlantic rainforests, with endemic species serving as witnesses of speciation events which occurred as a result of the geographical isolation of these enclaves across the Caatinga (Vanzolini 1981;Coimbra-Filho and Câmara 1996).Currently, the spatial or temporal scales of these historical connections are under debate and new evidence suggests complex evolutionary histories involving multiple events and processes triggering high-level endemism found in these relictual forests (Batalha-Filho et al. 2013;Peres et al. 2020).

Regarding the geographical distribution presented by
Furthermore, the São Francisco River seems to act as a physical barrier (Carnaval and Moritz 2008;Recoder and Rodrigues 2020;Thomé et al. 2021) to gene flow in some disparate taxa, such as lizards, amphisbaenians and snakes (Rodrigues 1996;Passoni et al. 2008;Bruschi et al. 2019).In fact, the São Francisco River apparently limits the northern distribution of C. bicarinatus and separates it from the area of occurrence of Chironius dracomaris sp.nov.for more than 800 km of airline.Rodrigues (1996) proposed the Paleolacustrine Vicariance Hypothesis in order to explain the recovered pattern of some lizards as sister-species on the opposite sides of the banks of the São Francisco River.This hypothesis is in accordance with estimates of divergence times among species on opposite banks of the river (Recoder and Rodrigues 2020).
The molecular phylogeny recovered Chironius dracomaris sp.nov.as a sister-group of the clade C. bicarinatus + C. gouveai (Fig. 1), which could suggest the existence of a connection between the "Brejos de Altitude" and Atlantic Forest, with the São Francisco River acting as a secondary barrier.However, additional phylogeographic and paleoclimatic studies are necessary to understand the evolutionary history of these lineages.The projection models for C. bicarinatus suggest that the connectivity of the suitable areas in the Atlantic Forest with the Baturité Massif, where C. dracomaris sp.nov.occurs, may be older than the Last Interglacial (120 kyr) (Fig. 5), as proposed by Silveira et al. (2019) with woody plant species.On the other hand, the southern limit of the distribution of C. bicarinatus in Serra do Tabuleiro, state of Santa Catarina, is very close to the area of occurrence of C. gouveai, being separated by Serra do Mar, with C. bicarinatus occurring along the coast in Ombrophilous Dense Forests and C. gouveai occurring in inland Serra do Mar at higher altitudes in Mixed and Semideciduous Forests (Fig. 8).Coincidentally, Serra do Tabuleiro also appears to be the southern limit of the other Atlantic species of the genus (C.foveatus, C. fuscus, and C. laevicollis), except for Chironius cf.exoletus, which extends its distribution to the region of Porto Alegre, state of Rio Grande do Sul, already in a transition area with the Pampas.Despite the low genetic distances between C. gouveai and C. bicari natus (0.4-3.7%; Table S1), we recognize both as valid species because they were recovered as reciprocal monophyletic taxa (Fig. 1).In addition, both species are fully diagnosable on the basis of phenotypic characters with low niche overlap, and ultimately, we do not establish cut-off for genetic distances as a criterion for species delimitation due to gaps in geographic sampling (see Fig. 8B, C).As C. bicarinatus and C. gouveai are broadly distributed across Tropical Atlantic and Subtropical climates, respectively, the bioclimatic variables that were most related are characterized by these climates, with the Tropical Atlantic presenting higher temperatures with precipitation concentrated in winter, and the Subtropical presenting milder and more well-distributed precipitation throughout the year (Tables S3, S5).
We believe that the continental shelf in the LGM, an area currently submerged along the Brazilian coastline, possibly during periods of climatic oscillations and marine regressions (see Leite et al. 2016), could have sheltered an extensive and suitable area for the Chironius bicarinatus lineage.This hypothesis perhaps made it possible for ancestral populations to expand their distribution vertically to the northern limits of the Atlantic Forest.The projection models support the expansion of suitable areas in the LGM with the distention of the continental shelf, suggesting that previous Holocene events may have favored a bottleneck effect related to the consolidation of the speciation process previously initiated between C. bicarinatus and C. gouveai.On the other hand, although some coastal populations may have speciated prior to the LGM, other snake species likely diverged within the last 11,000 years (see Barbo et al. 2022).In this scenario, sea-level transgressions along the coastline (see Castro et al. 2014), may have triggered the interruption of the Atlantic Forest lowland corridor in the case of Chironius bicarinatus dispersion due to the close proximity between the Serra do Mar Mountain range and the Atlantic Ocean.In addition, the Paranapanema River (concordant with the neotectonic barrier of the Guapiara lineament), also appears to act as a potential barrier, limiting the northern distribution of C. gouveai, which is also observed in other taxa (e.g., Grazziotin et al. 2006 with the snakes of the Bothrops jararaca complex; Brunes et al. 2010 with the treefrogs of the Phyllomedusa burmeisteri group).Reinforcing this hypothesis, we noted that populations of Chironius gouveai near Serra do Mar, in the states of Paraná and Santa Catarina, have a color pattern that is more similar to C. bicarinatus than those that occur further inland.We encourage further research on the existence of putative hybrids in this contact zone in Serra do Mar, introducing the possibility of evaluating the effective barriers that currently act on the southern limits between the species.For instance, our data reinforce the need to complement the conservation policy proposed by Álvarez-Presas et al. (2014), which suggested extending the southern limit of the Serra do Mar corridor to embrace the Parque Nacional da Serra do Itajaí and Parque Estadual da Serra do Tabuleiro, as these areas are also part of a relevant biogeographical province in the Serra do Mar Mountain range (Klein 1981;Álvarez-Presas et al. 2014).
Similarly, it would also be extremely important to conserve the areas of "Brejos de Altitude" (Tabarelli and Silva 2003) and essential to elucidate the evolutionary history of the Atlantic Forest and the diversification processes of its biota.To achieve this, more research is needed in the "Brejos de Altitude" to understand the real richness of the biota and develop conservation and political strategies to conserve threatened species, such as Chironius dracomaris sp.nov.whose distribution is restricted to Baturité Massif.Nowadays it is widely acknowledged that the Baturité Massif region is suffering from considerable impacts related to urban expansion, extractive production, agriculture, and cattle ranching, which contribute to forest fragmentation and habitat loss (Coimbra-Filho and Câmara 1996;Borges-Nojosa 2007).Through the newly described species, Chironius dracomaris sp.nov., we are hopeful for a change in social and political awareness in relation to the importance of racial equality and the conservation of crucial areas for understanding biodiversity.

Figure 2 .
Figure 2. Scatterplots of the scores from two main discriminant functions for the groups of the Chironius bicarinatus complex, separating females (A) and males (B).Chironius bicarinatus: blue; C. gouveai: purple; Baturité Massif sample: orange.

Figure 4 .
Figure 4. Micro-computed tomography images of the skull showing in detail the diagnostic characters comparing Chironius bicarinatus (A, B) and C. gouveai (C, D): posterior portion of supratemporal (ST) straight (A) and slightly curved (C); ventral surface of the septomaxilla (SM) smooth (B) and with the presence of a conspicuous projection (D); anteroventral surface of prefrontal lacrimal foramen (PF) smooth (B) and with the presence of a conspicuous projection (D).Scale bar = 5 mm.

Figure 6 .
Figure 6.General view while alive of Chironius bicarinatus and Chironius gouveai: A an adult of C. bicarinatus (MNRJ 26255) from the municipality of Rio Claro, state of Rio de Janeiro, Brazil; B a juvenile specimen of C. bicarinatus (MNRJ 27463), from the municipality of Rio de Janeiro, state of Rio de Janeiro, Brazil; C an adult specimen of C. gouveai (unvouchered specimen) from the municipality of Porto Alegre, state of Rio Grande do Sul, Brazil; D a juvenile specimen of C. gouveai (unvouchered specimen) from the municipality of Vacaria, state of Rio Grande do Sul, Brazil.Photos by M. Bilate (A), F. Dias-Silva (B), M. Borges-Martins (C), and O.L. Balbinot (D).

Figure 8 .
Figure 8. Distribution of the Chironius bicarinatus complex (A), based on first-hand examined and photographic records with locality data: C. bicarinatus: blue circles; C. gouveai: purple circles; C. dracomaris sp.nov.: orange circles.The stars represent their respective type localities; B zoomed map of the Baturité Massif, an area of distribution of Chironius dracomaris sp.nov.; B, C zoomed maps of molecular sampling localities, with the new sequences obtained in this study (highlighted in red) and sequences selected from GenBank.

Figure 12 .
Figure 12.General view while alive of Chironius dracomaris sp.nov.: A, B a specimen (IBSP 76994 or 76995) from Sítio Álvaro, municipality of Guaramiranga, state of Ceará, Brazil.These two vouchers were destroyed in a fire on May 15, 2010, leaving only photographic records of one of the specimens while alive; C-F paratypes of Chironius dracomaris sp.nov.(MNRJ 27803 and MHNCE-R 577, respectively) from Museu de História Natural do Ceará Prof. Dias da Rocha, municipality of Pacoti, state of Ceará, Brazil.Photos by I.J.Roberto (A, B), R.C. Gonzalez (C, D), T. Cavalcante (E, F).
Entiauspe-Neto et al. (2020), the authors mention that Chironius bicarinatus and C. gouveai are allopatric, but they present a distribution map (Entiauspe-Neto et al. 2020: fig.2) with examined specimens (circles) and literature records (squares), where the species are clearly

Table 4 .
Results of niche overlap (D metric) and null tests of equivalence and similarity.The niche overlap was low or zero for all comparisons, and the null tests showed nonequivalence or similarity of the niches.
The description of hemipenes is based on the fully everted and maximally expanded left organs of the specimens MNRJ 4008, 25558, and also on the partially expanded organs of specimens MNRJ 8717, 25434; and the maximally expanded right organ of the specimen MNRJ 18909.Hemipenis unilobed, unicalyculate, noncapitate; subcylindrical shape, with a simple sulcus spermaticus, running centripetally from the base to the apex (MNRJ 4008) or slightly more than half of the hemipenis (MNRJ 25434); apex may present a nude area (MNRJ 25558) or may not (MNRJ 18909), generally with papillate calyces, but can also present calyces with few papillae on the apex and medial region (MNRJ 18909, 25558); each longitudinal row presenting 21-29 calyces; the calyces towards the hemipenial body are replaced by spinulate calyces; hemipenial body represents more than half of the total length of the organ, and is covered in spines that gradually increase in size toward the base, reaching maximum size at just over half of hemipenial body; each longitudinal row presenting 14-22 spines and 5-7 spines along sulcus; base mostly nude, ornamented by spinules at the upper portion and also laterally distributed on the proximal region of sulcus spermaticus; a basal naked pocket is present on the medial region.
(Carreira and Maneyro 2013) Paulo, Brazil(Carreira and Maneyro 2013).Strangely, two records of C. bicarinatus (CHUFC 3604, MZFS 1279) were reported from the municipalities of Mulungu and Juazeiro, respectively in the states of Ceará and Bahia.Similarly, we believe it is more likely that these records are also due to accidental introduction by banana shipments from the Bahia coastal region than that the representation of small, well-estab-lished populations above the São Francisco River.Possible evidence of this would be the northernmost records of C. bicarinatus(NMB 1303(NMB  , 1304) )reported by Dixon

Table 6 .
Comparisons between character frequencies and specimen identifications in the sample examined by Entiauspe-Neto et al. 2020 with the results of this study.