Revealing the cryptic diversity of the widespread and poorly known South American blind snake genus Amerotyphlops (Typhlopidae: Scolecophidia) through integrative taxonomy

Abstract

Morphological stasis is generally associated with relative constancy in ecological pressures throughout time, producing strong stabilizing selection that retains similar shared morphology. Although climate and vegetation are commonly the main key factors driving diversity and phenotypic diversification in terrestrial vertebrates, fossorial organisms have their morphology mostly defined by their fossorial lifestyle. Among these secretive fossorial organisms, blind snakes of the South American genus Amerotyphlops are considered poorly studied when compared to other taxa. Here, we evaluate the cryptic diversity of Amerotyphlops using phylogenetic and multivariate approaches. We based our phylogenetic analysis on a molecular dataset composed of 12 gene fragments (eight nuclear and four mitochondrial) for 109 species of Typhlopidae. The multivariate analysis was implemented using 36 morphological variables for 377 specimens of Amerotyphlops. Additionally, we contrast our phylogenetic result with the morphological variation found in cranial, external and hemipenial traits. Our phylogenetic results recovered with strong support the following monophyletic groups within Amerotyphlops: (1) a clade formed by A. tasymicris and A. minuisquamus; (2) a clade composed of A. reticulatus; (3) a north-eastern Brazilian clade including A. yonenagae, A. arenensis, A. paucisquamus and A. amoipira; and (4) a clade composed of A. brongersmianus and a complex of cryptic species. Based on these results we describe four new species of Amerotyphlops from north-eastern and south-eastern Brazil, which can be distinguished from the morphologically similar species, A. brongersmianus and A. arenensis.

Atlantic Rain Forest, new species, phylogenetics, Serpentes

INTRODUCTION

In the last two decades, cryptic species have posed a challenge for biologists and taxonomists (Winker, 2005). Cryptic species are considered morphologically indistinguishable evolutionary units, which are usually classified as a single species, even when other evidence indicates that they have unique evolutionary trajectories (Bickford et al., 2007; Fišer et al., 2018; Struck et al., 2018; Struck & Cerca, 2019). Despite distinct evolutionary trajectories, usually the phenotypes of cryptic species are driven by equalizing evolutionary forces causing morphological stasis, convergence, parallelism and phylogenetic niche conservatism (Bickford et al., 2007; Fišer et al., 2018; Struck et al., 2018). Among these evolutionary forces, morphological stasis can be properly tested through molecular phylogenetic analysis (Bickford et al., 2007). Groups of cryptic species under morphological stasis, although retaining a high degree of phenotypic similarity, usually present deep phylogenetic structure (Struck et al., 2018; Struck & Cerca, 2019). Stasis is generally associated with relative constancy in ecological pressures throughout time, producing strong stabilizing selection that retains a similar shared morphology (Swift et al., 2016; Struck et al., 2018; Struck & Cerca, 2019).

Although climate and vegetation are commonly the main key factors driving diversity and phenotypic diversification in terrestrial vertebrates (Stayton, 2005), fossorial organisms have their morphology mostly defined by their burrowing lifestyle and the characteristics of the soils in which they burrow (Daly & Patton, 1990; Reig et al., 1990; Steinberg et al., 2000; Massarini et al., 2002; Mulvaney et al., 2005; Daniels et al., 2009; Parham & Papenfuss, 2009; Marcy et al., 2016).

Vertebrates with a fossorial and subterranean lifestyle are difficult to obtain and observe due to their secretive behaviours, and many aspects of their evolutionary biology remain poorly known, resulting in an underestimation of their diversity (Massarini et al., 2002; Fernández-Stolz et al., 2007; Thomas & Hedges, 2007; Gonçalves & de Freitas, 2009; Correia et al., 2018; Maddock et al., 2020; O’Connell et al., 2021). Among these secretive subterranean organisms, Scolecophidia (Typhlopoidea, Leptotyphlopidae and Anomalepididae) represents one of the most understudied groups. Despite recent advances in higher level systematics of scolecophidians (Adalsteinsson et al., 2009; Hedges et al., 2014; Pyron & Wallach, 2014; Graboski et al., 2018), the diversity in the species-rich South American radiation has not been explored so far through integrative approaches.

AmerotyphlopsHedges et al. (2014) was proposed for a clade of Neotropical typhlopids that currently comprises 15 species, 14 of which are distributed on the continent, from Mexico to northern Argentina, and one occurring in the West Indies (Hedges et al., 2014; Pyron & Wallach, 2014). Of the 14 mainland species, eight are restricted to South America (see the Supporting Information, Table S1 for detailed distribution): (1) Amerotyphlops reticulatus (Linnaeus, 1758), (2) A. lehneri (Roux, 1926), (3) A. brongersmianus (Vanzolini, 1972, 1976), (4) A. minuisquamus (Dixon & Hendricks, 1979), (5) A. paucisquamus (Dixon & Hendricks, 1979), (6) A. yonenagae (Rodrigues, 1991), (7) A. amoipira (Rodrigues & Juncá, 2002) and (8) A. arenensis (Graboski et al., 2015). Despite the wide geographic distribution of South American blind snakes, they are often poorly sampled in phylogenetic studies, due to their secretive habits and the lack of comprehensive taxonomic studies.

Here, we evaluate the genetic and morphological diversity of Amerotyphlops, aiming to test the existence of cryptic species in the genus. Our results provide evidence of the existence of four new species from north-eastern and south-eastern Brazil that can be distinguished from the morphologically similar species, A. brongersmianus and A. arenensis.

MATERIAL AND METHODS

Taxon and gene sampling

Our data matrix comprises 149 terminals (including 109 described species) sequenced for 12 genes: four mitochondrial (12S, 16S, cytb and cox1) and eight nuclear (bdnf, rag1, bmp2, nt3, prlr, jun, dnh3 and amel). We sequenced 124 new DNA fragments (GenBank accession numbers: OP093574–OP093604; OP094006–OP094029; OP177794–OP177861) for seven described species of Amerotyphlops (A. arenensis, A. amoipira, A. brongersmianus, A. minuisquamus, A. paucisquamus, A. reticulatus and A. yonenagae) and four new species described in this study (Supporting Information, Table S2).

We also included sequences from GenBank for four subfamilies of Typhlopidae: Afrotyphlopinae (19 species), Asiatyphlopinae (43 species), Madatyphlopinae (six species) and Typhlopinae (39 species) (Supporting Information, Table S3). Additionally, we also included sequences for four other families of scolecophidians: Anomalepididae [Liotyphlops albirostris (Peters, 1858)], Leptotyphlopidae [Epictia columbi (Klauber, 1939) and Rena dulcisBaird & Girard, 1853], Gerrhopilidae [Gerrhopilus mirus (Jan, 1860) and Gerrhopilus hedraeus (Savage, 1950)] and Xenotyphlopidae [Xenotyphlops grandidieri (Mocquard, 1905)] (Supporting Information, Table S3). All sequences of scolecophidians from GenBank included in our analysis were mainly generated by the following previous taxonomic studies: Vidal et al. (2010), Hedges et al. (2014), Marin et al. (2013a, b) and Graboski et al. (2018) (see Supporting Information, Table S3 for accession numbers). We rooted our phylogenetic tree with Anomalepididae and Leptotyphlopidae.

DNA sequencing

DNA was extracted following the protocol described by Hillis et al. (1996). Sequences were amplified via polymerase chain reaction (PCR) using the primers for: 12S and 16S, as described by Zaher et al. (2009); cytb, as described by Grazziotin et al. (2012); bdnf and cox1, as described by Graboski et al. (2018); jun, as described by Zaher et al. (2009); and dnh3, as described by Townsend et al. (2008).

PCRs were performed using standard protocols, with adjustments to increase the efficiency of amplification, as following: the addition of 10% of Trehalose for 12S, 16S, cytb and cox1, or 0.4% of Triton 100 for bdnf, dnh3 and jun. We used the annealing temperature of 54 °C for 12S and 16S; 56 °C for bdnf, jun and dnh3; and a touch-down cycle of 60–50 °C with final annealing of 54 °C for cytb and cox1. Amplified fragments were purified with shrimp alkaline phosphatase and exonuclease I (GE healthcare, Piscataway, NJ) and both strands were processed using the DYEnamic ET Dye Terminator Cycle Sequencing Kit in a MegaBACE 1000 automated sequencer (GE healthcare), following the manufacturer’s protocols. Both strands were quality checked and, when necessary, manually edited. The consensus of both strands was generated using Geneious Prime® 2022.1.1 (https://www.geneious.com, Kearse et al., 2012).

Molecular analyses

Sequences were aligned using MAFFT v.1.3.6 (Katoh & Standley, 2013) as implemented in Geneious Prime (see our concatenated alignment in the Supporting Information, Appendix S1). The 12S and 16S were aligned under the E-INS-i algorithm, while cox1, cytb, and the nuclear genes 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 Prime to verify if all sequences follow the correct reading frame.

We used PartitionFinder 2 (Lanfear et al., 2016) to identify the combined sets of partitioning schemes and models of molecular evolution. We divided our matrix into 32 partitions (coding genes were partitioned by codon positions and each rRNA was analysed as a separate partition) and performed a search using the greedy option. We used the Akaike information criterion with correction (AICc) to select the best fit model of evolution. We only allowed the selection of models implemented in RAxML (i.e., GTR and GTR+GAMMA). To avoid overcorrection regarding values of alpha and P-Invar (proportion of invariant sites), we did not allow PartitionFinder to select models with P-Invar, as suggested in the RAxML manual.

We performed a maximum likelihood (ML) analysis using RAxML v.8.2.3 (Stamatakis, 2014). The ML tree was estimated using the RAxML algorithm that conducts a rapid bootstrap analysis and searches for best-scoring ML tree in the same run (option -f a). We ran 1000 bootstrap replicates, and the best-scoring ML tree was estimated 200 times using as starting tree each fifth bootstrap tree. Additionally, we calculated the patristic distance among Typhlopinae (Amerotyphlops, AntillotyphlopsHedges et al., 2014, CubatyphlopsHedges et al., 2014 and Typhlops Oppel, 1811) using the R package ape (Paradis & Schliep, 2019)

Specimens and characters examined for morphological analysis

We examined a total of 377 specimens (347 specimens for pholidosis; 176 specimens for morphometrics) representing all mainland South American species of Amerotyphlops (Supporting Information, Appendices S2, S3). To correctly identify all examined specimens, we used the available information in the literature, as previous taxonomic revisions (Dixon & Hendricks, 1979; Graboski et al., 2018), the information available in the original description (Linnaeus, 1758; Vanzolini, 1972, 1976; Dixon & Hendricks, 1979; Rodrigues, 1991; Rodrigues & Juncá, 2002; Graboski et al., 2015) and by reviewing the type series of species (see Supporting Information, Appendices S2, S3). Additionally, for Central America congeneric species (i.e. Amerotyphlops costaricensis (Jiménez & Savage, 1962), A. lehneri (Roze, 1956), A. microstomus (Cope, 1866), A. stadelmani (Schmidt, 1936), A. tasymicris (Thomas, 1974), A. tenuis (Salvin, 1860), A. trinitatus (Richmond, 1965) and A. tycherus (Townsend et al., 2008) we used the information available in the original description and previous taxonomic revisions.

We used our phylogenetic tree as a framework to evaluate both external and internal morphology (hemipenis and skull), searching for diagnostic characters for the retrieved phylogenetic lineages. We measured the total length (TTL) and snout–vent length (SVL) to the nearest 1 mm by carefully stretching the specimen along a graduated ruler, and for other measurements we used a digital calliper to the nearest 0.1 mm. Our definitions of external morphological characters followed Dixon & Hendricks (1979) and Thomas & Hedges (2007) or were defined in the present study. We examined a total of 41 characters from external morphology: 12 categorical, 15 continuous and 14 ratios between continuous variables. In the descriptions, a slash (/) is used for rows counts from the anterior/median/posterior part of the body (Supporting Information, Table S4). We determined the sex of the individuals by making a small incision at the base of the tail to assess the presence (male) or absence (female) of hemipenes.

We examined four hemipenes of the candidate species Amerotyphlops sp. 2 from South America (see Supporting Information, Appendix S2). We everted hemipenes from fresh specimens or, alternatively, from fixed specimens following the protocols described by Zaher (1999) and Zaher & Prudente (2003). Hemipenial terminology follows Branch (1986), Peters & Orejas-Miranda (1970), Graboski et al. (2018) and Montingelli et al. (2022). We photographed the hemipenes using a Leica DFC425 digital camera attached to a Leica M205a stereoscopic microscope and made the combination and montage of multifocal photographs using the Leica Application Suite software (LAS core v.3.8, Leica Microsystems).

All specimens examined (Supporting Information, Appendices S2, S3) are deposited in the following institutions in Brazil (acronyms in parentheses): Laboratório de Anfíbios e Répteis, Universidade Federal do Rio Grande do Norte, Natal, Rio Grande do Norte, under the care of Adrian Antônio Garda (AAGARDA); Laboratório de Herpetologia do Instituto de Biociências, Universidade de São Paulo, São Paulo, under the care of Miguel Trefaut Rodrigues (MTR) and Gindomar Gomes Santana (GGS). Other museum acronyms follow Sabaj (2019) and Frost (2018).

Maps of species distribution

We also built a geographical dataset of 456 distribution records for the species of Amerotyphlops that have sympatric distribution with the new species described here (i.e. A. amoipira, A. arenensis, A. brongersmianus, A. paucisquamus and A. yonenagae). Maps were generated through the software ArcGIS v.10.2.2 (ESRI 1999). Initially, we obtained the geographical coordinates from the Species Link online database (https://splink.cria.org.br), which is based on institutions’ databases. We evaluated these registers through the contrast with our current knowledge about the group, and we included additional localities compiled from the literature – previous taxonomic revisions, information available in the original description and lists of species distribution (Dixon & Hendricks, 1979; Rodrigues, 1991; Rodrigues & Juncá, 2002; Van-Silva et al., 2007; Santana et al., 2008; Martins et al., 2010; de Arruda et al., 2011; Caicedo-Portilla, 2011; de Brito & Freire, 2012; de França et al., 2012; Guedes et al., 2014; Wallach et al., 2014; Graboski et al., 2015, 2018; Roberto et al., 2015, 2017; de Freitas et al., 2019; Nogueira et al., 2019). Additionally, we included in the geographical dataset registers taken directly from the localities of specimens examined in collections (see Supporting Information, Appendices S2, S3)

High-resolution X-ray computed tomography and osteological comparisons of cranial anatomy

We obtained data from skulls by using non-destructive high-resolution X-ray computed tomography (HRXCT) from seven specimens of the genus Amerotyphlops from South America (see Supporting Information, Appendix 2). All HRXCT data were generated using a GE phoenix v|tome|x m system at the Museu de Zoologia da Universidade de São Paulo, São Paulo, Brazil. The raw data were imported to VG studio Max 2.1 and exported for analysis, segmentation and visualization. We segmented each bone slide by slide by considering the different densities of bones, applying threshold tools. We isolated or removed cranial (lower jaws and quadrate) and postcranial elements, and the final volume was converted into an isosurface with triangular mesh (.ply files) in VG studio Max. The segmentation was performed at the Museu de Zoologia da Universidade de São Paulo. Our cranial descriptions and comparisons followed the terminology employed by List (1966), Cundall & Irish (2008), Chretien et al. (2019) and Lira & Martins (2021). We measured the skull length (SL) from the posterior margin of the otoccipital to the mid-dorsal region of the premaxilla; the skull width (SW) between the frontal processes of the parietal; the length of the ventral portion of the pterygoid process of the palatine (VPPL) from the base of the process to its distal end; and the palatine foramen diameter (PFD) measuring the larger distance between two opposite point at the internal margin of the foramen. We measured all morphometric osteological characters (Supporting Information, Table S5) to the nearest 0.01 mm directly on to the .ply files with the measurement tool implemented in the open source system MeshLab v.2020.06 (Cignoni et al., 2008).

Meristic and morphometric comparisons

To test the existence of cryptic species using morphological evidence, we defined nine OTUs (operational taxonomic units) for South American diversity of Amerotyphlops. We based the definition of OTUs on previously described species and on the lineages recovered in our molecular phylogenetic analysis, as follows: OTU 1, A. arenensis; OTU 2, A. yonenagae; OTU 3, A. amoipira; OTU 4, A. paucisquamus; OTU 5, A. sp. 1, from Parque Nacional da Chapada Diamantina, on BR 144 Road, municipality Lençóis, Bahia state; OTU 6, A. sp. 2, from the Parque Nacional Serra das Lontras, municipality of Arataca, state of Bahia; OTU 7, A. sp. 3, from the Praia das Neves, municipality of Presidente Kennedy, state of Espírito Santo; and OTU 8, A. sp. 4, from municipality of Trancoso, state of Bahia; and OTU 9, A. brongersmianus.

Exploratory descriptive analysis was performed to summarize the character variations and to evaluate outliers through visual inspection of histograms and boxplots (see Supporting Information, Appendix S4) using the software R v.3.6.3 (R Core Team, 2017). We calculated mean, standard deviations and ranges (maximums and minimums) of the variables for all taxa and sexes. Normality was investigated through visual inspection of quantile–quantile plots (which also helps in the identification of outliers) and through Kolmogorov–Smirnov using nortest R package (Gross & Ligges, 2015). The homoscedasticity assumptions were tested through Levene using car R package (Fox & Weisberg, 2019). Sexual dimorphism was evaluated for all characters through t-tests with Bonferroni P-corrections (for normally distributed data) and a Wilcoxon test (for non-normally distributed data) within each species using ‘t.test’ and ‘wilcox.test’ functions implemented in R package stats (R Core Team, 2017).

For an exploratory multivariate analysis, we created a new pipeline based on the R script used by Barbo et al. (2022b). We implemented the following three different approaches: (1) principal component analysis (PCA) based on our original dataset; (2) partial least square discriminant analysis (PLS-DA); and (3) linear discriminant analysis (LDA) based on a simulated equalized dataset. Because the sample size for some OTUs was not sufficient to perform PLS-DA and LDA, we implemented a pipeline to simulate an equalized dataset with 100 individuals per OTUs, regarding the original sample size for each OTU.

The pipeline to generate the simulated dataset was guided by the following rules: values for discrete and categorical variables for each OTU were drawn from (1) their binomial distribution – when only two ‘states’ were present for the variable – with a probability equal to the proportion of the OTUs more frequent state; or (2) from its Poisson distribution – when more than two ‘states’ were present – with a lambda equal to the OTUs median. Values from continuous variables were drawn from (3) their normal distribution with mean equal to the OTU’s mean and standard deviation (SD) equal to the OTU’s SD. When the OTU was represented by only one sample in the original dataset the values were draw from (4) the binomial distributions set to a probability of 0.95; (5) the Poisson distribution with lambda set as equal to the OTUs variable value; and (6) the normal distribution with mean set as equal to the OTUs variable value and the SD set as equal to the global SD (considering all OTUs).

The PCA was implemented based on the complete dataset, while the PLS-DA only used discrete characters and ratios, and the LDA was based only on the continuous characters. The LDA and PLS-DA analyses were performed in MASS (Vanables & Ripley, 2002) and caret (Kuhn, 2008) R packages, respectively. The dataset was scaled before implementing the PCA (correlation PCA), and before analysing the data in LDA, we removed variables presenting values of collinearity greater than 0.7. The training metric for PLS-DA was set to ‘Accuracy’ and scaled predictors were pre-processed by centring and scaling. The repeated cross-validation method was applied, and the training dataset was defined using three repeated ten-fold cross-validations to render a total of 30 accuracy estimation. We applied the ‘one standard error’ rule to select the optimal least complex model. The multivariate results were plotted using the R packages ggplot2 (Wickham, 2016) and ggbiplot (Vu, 2011). All R scripts used in these analyses are available on Figshare (https://doi.org/10.6084/m9.figshare.20092721).

Species delimitation

Our understanding of species is based on the general lineage concept (i.e. species with separately evolving metapopulation lineages; de Queiroz, 2005), and we accessed the distinct evolutionary history of lineages of Amerotyphlops based on the analyses of genotypic and phenotypic diversity. Additionally, based on the evaluated morphological variation and the results of the statistical analyses, we considered populations as representing a new diagnosable species when they present a ‘unique combination of characters states’ (sensuDavis & Nixon, 1992).

RESULTS

Phylogenetic analysis

Our concatenated alignment totalized 9107 base pairs (957 bp for 12S, 1414 bp for 16S, 1134 bp for cytb, 1205 bp for cox1, 709 bp for bdnf, 525 bp for rag1, 594 bp for bmp2, 639 bp for nt3, 486 bp for prlr, 378 bp for amel, 739 bp for dnh3 and 327 bp for jun). The proportion of gaps and undetermined characters in the concatenated alignment was 65.3%. PartitionFinder selected a scheme of partitions composed of 20 partitions with GTR+G and GTR models for the ML analysis (Supporting Information, Table S6).

The resulting ML topology (Fig. 1; Supporting Information, Fig. S1) for higher level affinities was similar to those presented by Vidal et al. (2010), Hedges et al. (2014), Pyron & Wallach (2014), Nagy et al. (2015), and Graboski et al. (2018). American typhlopids were recovered as a well-supported clade (95%) (Fig. 1; Supporting Information, Fig. S1) with two main clades within typhlopines. The first clade was formed by species from the Greater Antillean radiation (100%), and the second by species from South America and the Lesser Antilles (92%). In the latter clade, we found four well-supported monophyletic groups, as follows: (1) a clade formed by A. tasymicris and A. minuisquamus (75%); (2) a clade composed of A. reticulatus as the sister-group all the other South American species (87%); (3) a clade containing the north-eastern Brazilian species A. arenensis, A. yonenagae, A. amoipira and A. paucisquamus (clade A; 99%); and (4) a clade composed of the A. brongersmianus species complex (clade B; 87%).

Figure 1.

Maximum likelihood tree of Typhlopoidea zoomed into the South American radiation of Amerotyphlops. The numbers on the branch represent bootstrap values > 75%. Letter A indicates the species of northern Brazil (Clade A), and letter B indicates species of the Amerotyphlops brongersmianus species complex (Clade B). Colour on branch represents the cryptic species Amerotyphlops sp. 1 (light blue), Amerotyphlops sp. 2 (dark pink), Amerotyphlops sp. 3 (green), and Amerotyphlops sp. 4 (red).

Within clade B, we recovered five subclades: (1) a clade composed of one individual from the municipality of Lençóis, Bahia state, Brazil (A. sp. 1); (2) a clade composed of two individuals from Parque Nacional Serra das Lontras and Reserva Particular do Patrimônio Natural Serra do Teimoso, municipalities of Arataca and Jussari, respectively, in Bahia state, Brazil (A. sp. 2); (3) a clade composed of two individuals from Praia das Neves, Presidente Kennedy municipality, Espírito Santo state, Brazil (A. sp. 3); (4) a clade composed of one individual from the municipality of Trancoso, Bahia state, Brazil (A. sp. 4); and (5) a clade representing the species Amerotyphlops brongersmianus.

Patristic distances among species in the genera of Typhlopinae indicate a mean value of 0.13 considering all genera (standard deviation of 0.06). The genus Amerotyphlops shows the highest mean (0.19), while Cubatyphlops has the lowest mean (0.05). Patristic distances between the species of Cubatyphlops range from 0.014 to 0.96, with the lowest distance shown between C. contorhinus (Thomas & Hedges, 2007) and C. notorachius (Thomas & Hedges, 2007), and the highest distance between C. biminiensis (Richmond, 1955) and C. caymanensis (Sackett, 1940) (Supporting Information, Table S7). Patristic distances within Antillotyphlops range from 0.017 to 0.182, being the lowest distance between A. naugus (Thomas, 1966) and A. richardi (Duméril & Bibron, 1844), and the highest between A. guadeloupensis (Richmond, 1966) and A. naugus (Supporting Information, Table S8). Patristic distances within Typhlops range from 0.019 to 0.181, being the lowest distance between T. hectusThomas, 1974 and T. sylleptorThomas & Hedges, 2007, and the highest between T. proancylopsThomas & Hedges, 2007 and T. pusillusBarbour, 1914 (Supporting Information, Table S9).

Patristic distances among South American species range from 0.048 to 0.289, with the lowest distance being between Amerotyphlops brongersmianus and A. sp. 4, while the highest one was between A. minuisquamus and two species (A. paucisquamus and A. sp. 4). The genetic distance between A. brongersmianus and the four phylogenetically related clades (A. sp. 1, A. sp. 2, A. sp. 3 and A. sp. 4) ranges from 4% to 11.6% of divergence (Supporting Information, Table S10).

Morphological analysis

Our exploratory analysis found a highly homogeneous pattern of cephalic scales for the genus Amerotyphlops throughout its geographic distribution. Thus, these characters were not informative for determining our OTUs as potential evolutionary units. The most useful morphological characters were the number of scales along the body (mid-dorsal, ventral and dorsal rows) and the measures and proportions of body traits (see Supporting Information, Appendix S4).

In the principal component analysis using 36 variables (discrete, continuous and ratios), based on the raw dataset, the first two components explained 35% (20% and 15%, respectively) of the total variance for males and females (Fig. 2A, B). The results obtained by PCA recognized two higher groups based on external morphology (Fig. 2A, B). The first group (GR1) is easily distinguished by having 18/18/18 dorsal row scales around the body, which correspond to species distributed exclusively in north-eastern Brazil, as follows: A. arenensis, A. yonenagae, A. amoipira, A. paucisquamus and A. sp. 1. The second group (GR2) is distinguished by having 20/20/20 or 20/20/18 dorsal row scales around the body, which correspond to OTUs distributed in north-eastern Brazil (A. sp. 2 and A. sp. 4), south-eastern Brazil (A. sp. 3) and the broadly distributed species A. brongersmianus.

Figure 2.

Results of principal component analysis (PCA) using the raw dataset based on 36 characters for nine OTUs. A, PCA of females. B, PCA of males. Species are colour-coded according to OTUs (see Key-colour OTUs, on the bottom).

In the PCA based on the simulated dataset, the first two components explained 34% (18% and 16%, respectively) and 35% (20% and 15%, respectively) of the total variance for GR1 and GR2, respectively (Supporting Information, Fig. S2A, B). The results obtained by PCA based on both raw and simulated datasets showed a small overlap of variables between A. amoipira, A. arenenesis and A. paucisquamus within the GR1 (Supporting Information, Fig. S2A), while in GR2 we observed a large overlap of variables only between A. brongersmianus and A. sp. 4 (Supporting Information, Fig. S2B).

Results obtained by PLS-DA and LDA, based on a simulated dataset, also recognize the two higher groups mentioned above (GR1 and GR2), separated by the first and second components (Fig. 3A–D). In the PLS-DA using 22 variables (discrete and ratios), the first two components explained 83.3% (43.5% and 39.8%, respectively) of the total variance for GR1 (Fig. 3A), and the first two components explained 63.9% (36.6% and 27.3%, respectively) of the total variance for GR2 (Fig. 3B). In the LDA using 14 variables (continuous), based on the simulated dataset, the first two linear discriminants explained 81% (60% and 21%, respectively) of the total variance for GR 1 (Fig. 3C), and the first two linear discriminants explained 89% (61% and 28%, respectively) of the total variance for GR2 (Fig. 3D). The results obtained by PLS-DA and LDA also showed a small overlap of variables between A. amoipira, A. arenensis and A. paucisquamus within the GR1, while in GR2 we also observed a large overlap of variables only between A. brongersmianus and A. sp. 4 (Fig. 3A, B, D). However, the results obtained by LDA showed a large overlap of A. arenensis with all OTUs within GR1 (Fig. 3C).

Figure 3.

Results of partial least square discriminant analysis (PLS-DA) and linear discriminant analysis (LDA). A, PLS-DA of the simulated dataset based on seven discrete (pholidosis) and 14 rates characters for five OTUs (Group 1). B, PLS-DA of the simulated dataset based on seven discrete (pholidosis) and 14 rates characters for four OTUs (Group 2). C, LDA of the simulated dataset based on 15 continuous (linear morphometrics) characters for five OTUs (Group 1). D, LDA of the simulated dataset based on 15 continuous (linear morphometrics) characters for four OTUs (Group 2). Species are colour-coded according to OTUs (see Key-colour OTUs, on the bottom).

Additionally, the results obtained in an exploratory MANOVA-NP (we used PerMANOVA as implemented in the R package vegan) based on the simulated dataset indicated that the morphological variation among OTUs is significantly different (see Supporting Information, Table S11). The P-values for the pair-wise difference among OTUs were highly significant for most comparisons (P > 0.01 and P > 0.05), except between A. brongersmianus and A. sp. 4 (P = 0.86; Supporting Information, Table S11).

Species accounts

Based on the results of our phylogenetic tree, in combination with estimated patristic distances and multivariate morphological analyses, we were able to recognize four new species of South American Amerotyphlops. We formally describe them below.

Amerotyphlops caetanoi sp. nov.
(Fig. 4; Supporting Information, Fig. S3)

Zoobank registration:

Figure 4.

Holotype of Amerotyphlops caetanoi sp. nov. (MZUSP S-023380). Head in left lateral (A), dorsal (B) and ventral (C) views. TTL = 176 mm. Scale bar equal to 5 mm.

Holotype:

An adult female, MZUSP S-023380, (field number MTR 19921), collected by Ana C. Q. Carnaval, José C. Silva, Marco A. Sena, Mauro Teixeira Jr., Miguel T. Rodrigues, Renata C. Amaro and Renato Recoder on 15 December 2010 from Parque Nacional da Chapada Diamantina, on BR 144 Road, municipality Lençóis (12° 32′ 44.682′′ S, 41° 21′ 50.364′′ W; c. 493 m a.s.l.), state of Bahia, Brazil (Fig. 4; Supporting Information, Fig. S3).

Diagnosis:

This species is distinguished from all other congeneric species by the unique combination of the following of characters: (1) nasal suture incomplete; (2) rostral scale oval; (3) supralabial scales four; (4) infralabial scales three; (5) rows scales around the body 18/18/18; (6) mid-dorsal scales 212; (7) ventral scales 202; (8) rows of dorsal scales dark brown 13; (9) rows of ventral scales yellowish cream and immaculate 5; (10) caudal spine dark brown; (11) subcaudal scales 9; (12) TTL 176 mm; (13) TL 4.33 mm; (14) broad contact between the lamina of the premaxilla and the vertical laminae of the nasals, forming a continuous bony septum separating the olfactory chambers; (15) large palatine fossa on the lateral side of the maxilla; (16) maxilla with a straight medial border; (17) ventral pterygoid process of palatine straight; (18) ratio between length of ventral pterygoid process of palatine and skull length 0.25; (19) angle between mandibular condyle articulation and the retroarticular process of the compound bone close to 90°; and (20) dorsal surface of dentary bone with two evident foramina.

The new species differs from Amerotyphlops costaricensis, A. lehneri, A. microstomus, A. stadelmani, A. tasymicris, A. tenuis, A. trinitatus and A. tycherus, by having an incomplete nasal suture (vs. complete nasal suture); from A. brongersmianus, A. reticulatus and A. minuisquamus by having 18/18/18 rows scales around the body (vs. 18/16/14, 18/18/14, 20/18/14 or 20/18/15 in A. minuisquamus; 20/20/18 or 20/20/20 in A. brongersmianus and A. reticulatus); from A. brongersmianus by having an angle close to 90° between mandibular condyle articulation and the retroarticular process of the compound bone (vs. an angle of 135°); from A. yonenagae by having less than 250 mid-dorsal scales (vs. more than 250 mid-dorsal); from A. amoipira by having highly pigmented cephalic scales with a dark brown dorsum (vs. few pigmented cephalic scales, creamy brown dorsum with a fine darker brown paravertebral line concentrated in the anterior part of the body); from A. paucisquamus by having a largest number of mid-dorsal, 212 (vs. fewer number of mid-dorsal, between 162 and 209); and from A. arenensis by having a smaller rostral width (RW1) at dorsal portion, 1.29 mm (vs. larger rostral width at dorsal portion (RW1), between 1.44 and 2.13 mm). Table 1 shows additional morphometric characters and scale patterns found in A. caetanoi and morphologically similar species distributed in southern and north-eastern Brazil.

Table 1.

Variation for selected characters of morphologically similar species of Amerotyphlops from Brazil. Values displayed in the table represent (in this sequence) range (mm), mean, standard deviations and available sample size.

Character	Sex	A. arenensis	A. brongersmianus	A. martis	A. montanum	A. caetanoi	A. illusorium
ED	♂	0.9 ± 0.13 (0.61–1.05) 12	0.99 ± 0.27 (0.40–1.69) 39	0.76 ± 0.21 (0.63–1.16) 4	1.03 ± na (1.02) 1	-	0.72 ± 0.01 (0.71–0.72) 2
	♀	0.94 ± 0.14 (0.78–1.21) 9	1.03 ± 0.26 (0.61–1.72) 63	-	-	0.56 ± na (0.56) 1	0.77 ± na (0.77) 1
HR	♂	2.98 ± 0.13 (2.73–3.13) 12	3.56 ± 0.55 (2.77–4.78) 39	2.30 ± 0.17 (2.61–3.05) 4	3.88 ± na (3.88) 1	-	3.03 ± 0.57 (2.63–3.43) 2
	♀	3.24 ± 0.18 (2.97–3.50) 9	3.80 ± 0.56 (2.87–5.39) 63	-	-	2.86 ± na (2.86) 1	3.78 ± na (3.78) 1
HWE	♂	5.04 ± 0.41 (4.37–5.54) 12	5.75 ± 0.99 (4.10–8.17) 39	3.24 ± 0.12 (3.84–4.10) 4	6.64 ± na (6.64) 1	-	4.66 ± 0.65 (4.20–5.12) 2
	♀	5.52 ± 0.43 (4.80–6.06) 9	6.12 ± 1.02 (4.45–8.62) 63	-	-	4.44 ± na (4.44) 1	5.75 ± na (5.75) 1
IN	♂	2.41 ± 0.19 (2.08–2.65) 12	2.77 ± 0.45 (1.84–3.59) 39	1.68 ± 0.13 (1.89–2.22) 4	2.63 ± na (2.63) 1	-	2.32 ± 0.18 (2.19–2.45) 2
	♀	2.63 ± 0.22 (2.35–3.00) 9	2.92 ± 0.45 (2.16–4.14) 63	-	-	2.16 ± na (2.16) 1	2.82 ± na (2.82) 1
INORB	♂	3.25 ± 0.17 (2.87–3.45) 12	3.55 ± 0.6 (2.73–4.97) 39	2.08 ± 0.17 (2.31–2.68) 4	4.09 ± na (4.09) 1	-	3.00 ± 0.57 (2.60–3.40) 2
	♀	3.45 ± 0.25 (3.00–3.83) 9	3.68 ± 0.57 (2.72–5.11) 63	-	-	2.90 ± na (2.90) 1	3.90 ± na (3.9) 1
RL	♂	2.76 ± 0.23 (2.26–3.09) 12	3.22 ± 0.54 (2.27–4.37) 39	1.95 ± 0.19 (2.23–2.66) 4	3.15 ± na (3.15) 1	-	2.61 ± 0.25 (2.44–2.79) 2
	♀	3.11 ± 0.26 (2.71–3.54) 9	3.37 ± 0.53 (2.47–4.80) 63	-	-	2.51 ± na (2.51) 1	3.12 ± na (3.12) 1
RW1	♂	1.63 ± 0.09 (1.44–1.77) 12	1.76 ± 0.27 (1.27–2.37) 39	1.12 ± 0.09 (1.23–1.44) 4	1.88 ± na (1.88) 1	-	1.33 ± 0.16 (1.22–1.44) 2
	♀	1.77 ± 0.18 (1.56–2.13) 9	1.89 ± 0.30 (1.34–1.2.70) 63	-	-	1.29 ± na (1.29) 1	1.54 ± na (1.54) 1
MBD	♂	6.82 ± 1.11 (4.81–7.86) 12	8.14 ± 1.99 (5.03–14.47) 39	3.83 ± 0.44 (4.09–5.13) 4	7.12 ± na (7.12) 1	-	5.01 ± 0.14 (4.92–5.11) 2
	♀	8.03 ± 1.03 (5.93–8.98) 9	8.87 ± 2.15 (5.82–14.76) 63	-	-	6.21 ± na (6.21) 1	8.29 ± na (8.29) 1
INORB/HWE	♂	0.65 ± 0.03 (0.61–0.71) 12	0.62 ± 0.04 (0.55–0.71) 39	0.51 ± 0.02 (0.60–0.66) 4	0.72 ± na (0.72) 1	-	0.64 ± 0.03 (0.62–0.66) 2
	♀	0.62 ± 0.02 (0.58–0.66) 9	0.60 ± 0.04 (0.53–0.69) 63	-	-	0.66 ± na (0.66) 1	0.68 ± na (0.68) 1
NE/HWE	♂	0.41 ± 0.02 (0.38–0.44) 12	0.38 ± 0.02 (0.35–0.41) 39	0.33 ± 0.04 (0.36–0.46) 4	0.49 ± na (0.49) 1	-	0.39 ± 0.01 (0.39–0.40) 2
	♀	0.41 ± 0.02 (0.39–0.44) 9	0.39 ± 0.03 (0.33–0.47) 63	-	-	0.39 ± na (0.39) 1	0.42 ± na (0.42) 1

Abbreviations: ED, eye diameter; HR, head radius; HWE, head width; IN, internasal distance; INORB, interorbital distance; RL, rostral length; RW1, rostral width; MBD, midbody diameter.

Description of the holotype:

Adult female, TTL 176 mm, TL 4.33 mm, MBD/(SVL-HR) 0.036 mm, and TL/SVL 39.72 mm. Head slightly depressed dorsoventrally, not wider than ‘neck’. Snout round in dorsal and ventral views. Rostral oval, longer than wide, narrow at anteroposterior region and wider at medial region; visible in dorsal view, extending ventrodorsally without reaching the imaginary transverse line between anterior borders of eyes. Rostral contacting nasal (anterior and posterior) dorsolaterally, and first supralabial and anterior nasal scales ventrally. Nasal suture incomplete, only partially dividing the anterior and posterior portions of nasal scale. Suture begins in the upper edge of second supralabial, passes through nostril, but fails to reach rostral. Anterior nasal in contact with first infralabial and upper edge of second infralabial. Posterior nasal longer than wide, contacting upper margin of second supralabial and preocular. Supralabials four, fourth twice longer than third. Infralabials three, third largest. Eye diameter 0.56 mm; eyes not visible in ventral view, located dorsolaterally, close to suture between preocular and ocular scales, completely covered by ocular scale. Ocular scales contacting frontal. Body cylindrical and robust. Midbody diameter 6.21 mm. Dorsal and ventral scales cycloid, wider than long, highly imbricated and arranged in diagonal series; scale rows around the body 18/18/18. Mid-dorsal scales 212. Ventral scales 202. Cloacal plate rounded, bordered anteriorly by four rows of scales and posteriorly by five rows of scales. Subcaudal scales nine, excluding the terminal spine. Terminal spine large, stout base and dark brown.

Skull osteology (N = 1; MZUSP S-023380):

The length of the skull is 6.15 mm, the width is 2.92 mm. The snout region has a globular enlarged-shape and highly consolidated. The snout articulates with the braincase by the nasal and prefrontal sutures and with the frontal bone. The anteroventral region of the premaxilla has a short backward process. The midsagittal lamina separates both sides of the premaxilla (Fig. 5). The lamina of the premaxilla is confluent with the mid-dorsal laminae of the nasals and with the mid-dorsal ridges of the vomeronasal cupola of the septomaxillae (Fig. 5). The lamina of the premaxilla and the nasal laminae are in contact, forming a continuous bony septum separating the olfactory chambers (Fig. 5B). The medial side of the maxilla has a shallow depression (or fossa), where lodges the maxillary process of the palatine. The palatine fossa is on the lateral side of the maxilla, in the region of the articular fossa. The palatine fossa is large with a diameter of 0.25 mm long (Fig. 6A). The medial border of the maxilla is straight (Fig. 6A). The ventral pterygoid process of the palatine is straight and ventrally directed (Fig. 7A). The retroarticular process projects in parallel to the horizontal plane of the articular. The angle between mandibular condyle articulation and the retroarticular process of the compound bone is close to 90° (Fig. 8A). The edentulous dentary is restricted to the distal end of the mandible, articulating mainly with the splenial. The dorsal side of the dentary is flat and pierced by two foramina (Fig. 9A).

Figure 5.

Three-dimensional cutaway views along the sagittal axis of the nasal cavity in Amerotyphlops species based on HRXCT data. A, Amerotyphlops brongersmianus (MZUSP 14689); B, Amerotyphlops caetanoi sp. nov. (MZUSP S-023380). Insert show a lateral profile of the skull of A. brongersmianus presenting in red the position of the detailed region. Scales bars equal to 5 mm. Abbreviations: pmx, premaxilla; smx, septomaxilla; vm, vomer; na, nasal.

Figure 6.

Three-dimensional reconstruction of the lateral region of the skull in Amerotyphlops species based on HRXCT data. A, Amerotyphlops caetanoi sp. nov. (MZUSP S-023380); B, Amerotyphlops montanum sp. nov. (MZUSP 20065); C, Amerotyphlops brongersmianus (MZUSP 14674). Insert shows a lateral profile of the skull of A. brongersmianus presenting in red the position of the detailed region. Scales bars equal to 5 mm. Abbreviations: mx, maxilla; mxpalf, palatine articulation fossa of the maxilla.

Figure 7.

Three-dimensional reconstruction of the latero-ventral region of the skull in Amerotyphlops species based on HRXCT data. A, Amerotyphlops caetanoi sp. nov. (MZUSP S-023380); B, Amerotyphlops brongersmianus (MZUSP 14674). Insert shows a lateral profile of the skull of A. brongersmianus presenting in red the position of the detailed region. Scales bars equal to 5 mm. Abbreviations: mx, maxilla; palmx, palatine maxillary process; palptv, palatine pterygoid ventral process; pt, pterygoid.

Figure 8.

Three-dimensional reconstruction of the posterior region of the left mandible in Amerotyphlops species based on HRXCT data. A, Amerotyphlops caetanoi sp. nov. (MZUSP S-023380); B, Amerotyphlops brongersmianus (MZUSP 14674). Insert shows a lateral profile of the skull of A. brongersmianus presenting in red the position of the detailed region. Scales bars equal to 5 mm. Abbreviations: cb, compound bone; q, quadrate; rp, retroarticular process.

Figure 9.

Three-dimensional reconstruction of the anterior region of the left mandible in Amerotyphlops species based on HRXCT data. A, Amerotyphlops caetanoi sp. nov. (MZUSP S-023380); B, Amerotyphlops montanum sp. nov. (MZUSP 20065). Insert shows a lateral profile of the skull of A. brongersmianus presenting in red the position of the detailed region. Scales bars equal to 5 mm. Abbreviations: cb, compound bone; co, coronoid; d, dentary; sp, splenial.

Coloration of the holotype in preservative:

Dorsum (13/13/13 row scales) dark brown (Supporting Information, Fig. S3A), venter (5/5/5 rows scales) yellowish cream (Supporting Information, Fig. S3B). Dorsal portions of snout yellowish cream, with a dark brown spot, covering both rostral and nasal scales (two-thirds of snout) (Fig. 4A, B). Ventral portions of snout yellowish cream and few pigmented (Fig. 4C). Symphysial region yellowish cream and immaculate (Fig. 4C). Dorsal head scales (supraoculars, frontal, postfrontal, parietals and occipitals) dark brown. Dorsal portions of lateral head scales (ocular, nasal and lower nasal) and ventral portions yellowish cream with dark brown spots. Cloacal plate pale yellowish cream and terminal spine dark brown (Fig. 4A–C).

Etymology:

The name is a homage to Brazilian composer, singer and political activist Caetano Emanuel Viana Telles Veloso, better known as Caetano Veloso. Caetano is one of the most famous Brazilians born in the state of Bahia (in 1942), the same state in which the new species occurs. He became known for his participation in the Brazilian musical movement ‘Tropicalismo’ that encompassed theatre, poetry and music in the 1960s, at the beginning of the Brazilian military dictatorship. Veloso is also a well-known conservationist, acting to give voice to the preservation of the Brazilian natural environments and to indigenous resilience.

Distribution and habitat:

Amerotyphlops caetanoi is known only from Parque Nacional da Chapada Diamantina, in the BR 144 Road, situated at 2 km from the municipality of Lençóis, state of Bahia, Brazil (Fig. 10B). This region is one of the largest upland Atlantic dry forest enclaves of the east-central Bahia state (Veloso et al., 1991). The phytophysiognomy corresponds to a submontane seasonal semi-deciduous forest (Couto et al., 2011; Braz et al., 2013), ranging from 400 to 600 m a.s.l., with an annual average of temperature and rainfall of 20 °C and 100 mm, respectively (Funch et al., 2009). This area presents a non-continuous canopy, consisting of tall trees (approximately 10–16 m), and a subcanopy, consisting of medium-height trees (approximately 6–9 m), with a well-established and preserved understory (Couto et al., 2011).

Figure 10.

Geographical distribution of nine species belonging to the genus Amerotyphlops from South America. A, zoomed map of species distributed in the north-eastern Brazil; B, zoomed map of four new species described in this work. Symbols: A. arenensis (black pentagons); A. amoipira (black diamonds); A. brongersmianus (black circles); A. yonenagae (black squares); A. pauciquamus (black inverted triangles); A. martis sp. nov. (black star); A. montanum sp. nov. (black triangles); A. illusorium sp. nov. (black cross); Amerotyphlops caetanoi sp. nov. (black asterisk).

Amerotyphlops montanum sp. nov.
(Fig. 11; Supporting Information, Fig. S4)

Zoobank registration:

Figure 11.

Holotype of Amerotyphlops montanum sp. nov. (MZUSP 20065). Head in left lateral (A), dorsal (B) and ventral (C) views. TTL = 216 mm. Scale bar equal to 5 mm.

Holotype:

An adult female, MZUSP 20065, (field number MTR 16379), collected by Augustín Camacho, José Cassimiro, Mauro Teixeira Jr., Miguel T. Rodrigues, Renata C. Amaro and Renato Recoder 6 March 2009 from Parque Nacional Serra das Lontras (15° 11′ 46.32′′ S, 39° 20′ 54.24′′ W; c. 234 m a.s.l.), municipality of Arataca, state of Bahia, Brazil (Fig. 11; Supporting Information, Fig. S4).

Diagnosis:

This species is distinguished from all other South American congeneric species by a unique combination of the following of characters: (1) nasal suture incomplete; (2) rostral scale oval; (3) supralabial scales four; (4) infralabial scales three; (5) rows scales around the body 20/20/18; (6) mid-dorsal scales 220; (7) ventral scales 217; (8) rows of dorsal scales dark brown 11; (9) rows of ventral scales yellowish cream and immaculate 7–9; (10) caudal spine dark brown; (11) subcaudal scales 11; (12) TTL 216 mm; (13) TL 5.32 mm; (14) contact between the nasal process of premaxilla and vertical laminae of the nasals restricted to the anterodorsal portion, with the central and posteroventral portions not in contact, leaving a large canal between the olfactory chambers; (15) small-sized palatine fossa on the lateral side of the maxilla; (16) maxilla with a straight medial border; (17) ventral pterygoid process of palatine straight; (18) ratio between length of ventral pterygoid process of palatine and skull length 0.06; (19) angle between mandibular condyle articulation and the retroarticular process of the compound bone close to 90°; and (20) dorsal surface of dentary bone without evident foramina.

Amerotyphlops montanum differs from A. costaricensis, A. lehneri, A. microstomus, A. stadelmani, A. tasymicris, A. tenuis, A. trinitatus and A. tycherus by having an incomplete nasal suture (vs. complete nasal suture); from A. arenensis, A. caetanoi, A. amoipira, A. minuisquamus, A. paucisquamus and A. yonenagae by having 20/20/18 rows scales around the body (vs. 18/16/14, 18/18/14, 20/18/14 or 20/18/15 in A. minuisquamus and 18/18/18 in A. arenensis, A. caetanoi, A. amoipira, A. paucisquamus and A. yonenagae); from A. reticulatus by having highly pigmented cephalic scales with a dark brown dorsum and dorsum tail brown (vs. yellow and few pigmented cephalic scales, dorsum brown or black and dorsum tail black with cream or yellow spot); and from A. brongermianus by having a larger interorbital relative width (INORB/HWE) 0.725 mm (vs. smaller interorbital relative width, between 0.526–0.705 mm). Table 1 shows additional morphometric characters and scale patterns found in A. montanum and in a morphologically similar species distributed in south-eastern Brazil (A. brongermianus).

Description of the holotype:

Adult female, TTL 216 mm, TL 5.32 mm, MBD/(SVL-HR) 0.034 mm and TL/SVL 39.60 mm. Head slightly depressed dorsoventrally, not wider than ‘neck’. Snout round in dorsal and ventral views. Rostral oval, longer than wide, narrow at anteroposterior region and wider at medial region; visible in dorsal view, extending ventrodorsally without reaching the imaginary transverse line between anterior borders of eyes. Rostral contacting nasal (anterior and posterior) dorsolaterally, and first supralabial and anterior nasal scales ventrally. Nasal suture incomplete, only partially dividing the anterior and posterior portions of nasal scale. Suture begins in the upper edge of second supralabial, passes through nostril, but fails to reach rostral. Anterior nasal in contact with first infralabial and upper edge of second infralabial. Posterior nasal longer than wide, contacting upper margin of second supralabial and preocular. Supralabials four, fourth twice longer than third. Infralabials three, third largest. Eye diameter 1.03 mm; eyes not visible in ventral view, located dorsolaterally, close to suture between preocular and ocular scales, completely covered by ocular scale. Ocular scales contacting frontal. Body cylindrical and robust. Midbody diameter 7.12 mm. Dorsal and ventral scales cycloid, wider than long, highly imbricated and arranged in diagonal series; scale rows around the body 20/20/18. Mid-dorsal scales 220. Ventral scales 217. Cloacal plate rounded, bordered anteriorly by four rows of scales and posteriorly by five rows of scales. Subcaudal scales 11, excluding the terminal spine. Terminal spine large, stout base and dark brown.

Skull osteology (N = 1; MZUSP 20065):

The length of the skull is 8.08 mm and the width is 4.14 mm. The snout region has a globular enlarged-shape and highly consolidated. The snout articulates with the braincase by the nasal and prefrontal sutures and with the frontal bone. The anteroventral region of the premaxilla has a short backward process. The midsagittal lamina separates both sides of the premaxilla (Fig. 5). The lamina of the premaxilla is confluent with the mid-dorsal laminae of the nasals and with the mid-dorsal ridges of the vomeronasal cupola of the septomaxillae (Fig. 5). The lamina of the premaxilla and the nasal laminae are restricted to the anterodorsal portion, with the central and posteroventral portions not in contact, leaving a large canal between the olfactory chambers (Fig. 5A). The medial side of the maxilla has a shallow depression (or fossa), where lodges the maxillary process of the palatine. The palatine fossa is on the lateral side of the maxilla, in the region of the articular fossa. The palatine fossa is small, with a diameter of 0.17 mm long (Fig. 5B). The medial border of the maxilla is straight (Fig. 6B). The ventral pterygoid process of the palatine is straight-shaped and ventrally directed (Fig. 7A). The retroarticular process projects in parallel to the horizontal plane of the articular. The angle between mandibular condyle articulation and the retroarticular process of the compound bone is close to 90° (Fig. 8A). The edentulous dentary is restricted to the distal end of the mandible, articulating mainly with the splenial. The dorsal side of the dentary is flat and without evident foramina (Fig. 9B).

Coloration of the holotype in preservative:

Dorsum (11/11/11 rows scales) dark brown, venter (9/9/7 rows scales) yellowish cream (Supporting Information, Fig. S4A, B). Dorsal portions of snout yellowish cream, with a dark brown spot, covering totally both rostral and nasal scales (Fig. 11A, B). Ventral portion of snout yellowish cream and few pigmented (Fig. 11C). Symphysial region yellowish cream and immaculate (Fig. 11C). Dorsal head scales (supraoculars, frontal, postfrontal, parietals and occipitals) and dorsal portions of lateral head scales (ocular, nasal and lower nasal) predominantly dark brown (Fig. 11A, B). Ventral portion of head scales (nasal and lower nasal) yellowish cream (Fig. 11A, C). Cloacal plate pale yellowish cream and terminal spine dark brown (Supporting Information, Fig. S4B).

Etymology:

The specific epithet is derived from the neutral form of Latin adjective ‘montanus’. It is a reference to the type locality, a high elevational forest, located on the slopes of a hill summit in the Brazilian state of Bahia.

Distribution and habitat:

Amerotyphlops montanum is known from the Parque Nacional Serra das Lontras, situated at 10 km from the municipality of Arataca, in state of Bahia, Brazil, and from Reserva Particular do Patrimônio Natural Serra do Teimoso, situated at 5 km from the municipality of Jussari, in state of Bahia, Brazil (Fig. 10B).

These regions are known as part of the Serra das Lontras montane complex (Nacif et al., 2009), belonging to the Atlantic Forest morphoclimatic domain, in the Bahia Coastal forest ecoregion (Olson et al., 2001). The prevalent phytophysiognomy correspond to ombrophilous dense and semi-deciduous forests (Veloso et al., 1991; Amorim & Matos, 2009; Reis & Fontoura, 2009), with an elevational gradient ranging from sea level to more than 1000 m a.s.l., with an annual average of temperature between 23.3 and 23.6 °C and rainfall between 1300 and 1600 mm (Silveira et al., 2005; Amorim & Matos, 2009; Nacif et al., 2009; Reis & Fontoura, 2009). These areas have different levels of successional forests according to elevational gradients. Areas between 400 and 800 m, presenting a high and dense forest, with large height trees (taller than 30 m), a well-defined canopy structure, with abundant epiphytes, with a well-established and preserved understory. The forest changes dramatically above 800 m, presenting stunted trees (10–15 m tall), covered with small bromeliads, heavy bryophyte and lichen growth (Veloso et al., 1991; Amorim & Matos, 2009; Reis & Fontoura, 2009). Additional data on the habitat of Serra das Lontras and Serra do Teimoso are, respectively, in Recoder et al. (2010) and Rodrigues et al. (2002).

Remarks:

In our phylogenetic analysis we included a sample from a specimen from the Reserva Particular do Patrimônio Natural Serra do Teimoso, from the municipality of Jussari, in the state of Bahia, Brazil. Unfortunately, we did not have access to review this specimen, but we have recognized its molecular relationships with A. montanum. The tissue sample is currently stored in the genetic resource collection of Instituto de Biociências da Universidade de São Paulo University (see Supporting Information, Table S2).

Amerotyphlops martis sp. nov.
(Fig. 12; Supporting Information, Figs S5, S6)

Zoobank registration:

Figure 12.

Holotype of Amerotyphlops martis sp. nov. (MNRJ 18744). Head in left lateral (A), dorsal (B) and ventral (C) views. TTL = 157 mm. Scale bar equal to 5 mm.

Holotype:

An adult male, MNRJ 18744, collected by Ana C. C. Lourenço and Délio Baêta between 2 and 8 September 2009 from Praia das Neves (21° 16′ 45.59′′ S, 40° 57′ 47.86′′ W), municipality of Presidente Kennedy, state of Espírito Santo, Brazil (Fig. 12; Supporting Information, Fig. S5).

Paratypes:

Three male specimens, MNRJ 18743, MNRJ 18745 and MNRJ 18747, collected in the same locality of the holotype by Ana C. C. Lourenço and Délio Baêta between 2 and 8 September 2009 (Supporting Information, Fig. S6A–F).

Diagnosis:

This species is distinguished from all other South American congeneric species by the unique combination of the following of characters: (1) nasal suture incomplete; (2) rostral scale oval; (3) supralabial scales four; (4) infralabial scales three; (5) rows scales around the body 20/20/18–20; (6) mid-dorsal scales 208–217; (7) ventral scales 195–211; (8) rows of dorsal scales pale brown 12–13; (9) rows of ventral scales yellowish cream and immaculate four to five; (10) caudal spine pale brown; (11) subcaudal scales ten to 12; (12) maximum TTL 170 mm; (13) maximum TL 6.13 mm; (14) nasal process of premaxilla contacting the vertical laminae of the nasals in the anterodorsal and posteroventral portions, with the central portion not in contact, leaving a large round canal between the olfactory chambers; (15) large palatine fossa on the lateral side of the maxilla; (16) maxilla with a concave medial border; (17) ventral pterygoid process of palatine straight; (18) ratio between length of ventral pterygoid process of palatine and skull length 0.06; (19) angle between mandibular condyle articulation and the retroarticular process of the compound bone close to 90°; (20) Dorsal surface of dentary bone with two evident foramina; and (21) hemipenis single, with an additional structure in the apical cup, with a tissue projection in the form of a curved papilla.

Amerotyphlops martis differs from A. costaricensis, A. lehneri, A. microstomus, A. stadelmani, A. tasymicris, A. tenuis, A. trinitatus and A. tycherus by having an incomplete nasal suture (vs. complete nasal suture); from A. arenensis, A. caetanoi, A. amoipira, A. minuisquamus, A. paucisquamus and A. yonenagae by having 20/20/18 or 20/20/20 rows scales around the body (vs. 18/16/14, 18/18/14, 20/18/14 or 20/18/15 in A. minuisquamus and 18/18/18 in A.arenensis, A. caetanoi, A. amoipira, A. paucisquamus and A. yonenagae); from A. reticulatus by having pigmented cephalic scales with a pale brown dorsum and tail (vs. a yellowish and few pigmented cephalic scales, dorsum brown or black and dorsum tail black with cream or yellow spot); from A. montanus by having a smaller total length (TTL), between 130 and 170 mm (vs. larger total length 216 mm); and from A. brongermianus by having a small midbody diameter (MBD), between 4.090 and 5.133 mm and a single hemipenis with an additional structure in the apical cup, a large papillae projected laterally from the tip that extends horizontally over the proximal portion of the apical cup (vs. robust midbody diameter, between 5.03 and 14.76 mm and a single hemipenis with an unornamented apical cup). Table 1 shows additional morphometric characters and scale patterns found in A. martis and morphologically similar species distributed in southern and north-eastern Brazil.

Description of the holotype:

Adult male, TTL 157 mm, TL 6.13 mm, MBD/(SVL-HR) 0.032 mm and TL/SVL 24.61 mm. Head slightly depressed dorsoventrally, not wider than ‘neck’. Snout round in dorsal and ventral views. Rostral oval, longer than wide, narrow at anteroposterior region and wider at medial region; visible in dorsal view, extending ventrodorsally without reaching the imaginary transverse line between anterior borders of eyes. Rostral contacting nasal (anterior and posterior) dorsolaterally and first supralabial and anterior nasal scales ventrally. Nasal suture incomplete, only partially dividing the anterior and posterior portions of nasal scale. Suture begins in the upper edge of second supralabial, passes through nostril, but fails to reach rostral. Anterior nasal in contact with first infralabial and upper edge of second infralabial. Posterior nasal longer than wide, contacting upper margin of second supralabial and preocular. Supralabials four, fourth twice longer than third. Infralabials three, third largest. Eye diameter 0.90 mm; eyes not visible in ventral view, located dorsolaterally, close to suture between preocular and ocular scales, completely covered by ocular scale. Ocular scales contacting frontal. Body cylindrical and robust. Midbody diameter 4.86 mm. Dorsal and ventral scales cycloid, wider than long, highly imbricated and arranged in diagonal series; scale rows around the body 20/20/18. Mid-dorsal scales 215. Ventral scales 211. Cloacal plate rounded, bordered anteriorly by three rows of scales and posteriorly by five rows of scales. Subcaudal scales 11, excluding the terminal spine.

Skull osteology (N = 1; MNRJ 18743):

The length of the skull is 6.52 mm, the width is 2.96 mm. The snout region has a globular enlarged-shape and highly consolidated. The snout articulates with the braincase by the nasal and prefrontal sutures and with the frontal bone. The anteroventral region of the premaxilla has a short backward process. The midsagittal lamina separates both sides of the premaxilla (Fig. 5). The lamina of the premaxilla is confluent with the mid-dorsal laminae of the nasals and with the mid-dorsal ridges of the vomeronasal cupola of the septomaxillae (Fig. 5). The nasal process of premaxilla contacts the vertical laminae of the nasals in the anterodorsal and posteroventral portions, with the central portion not in contact, leaving a large round canal between the olfactory chambers (Fig. 5A). The medial side of the maxilla has a shallow depression (or fossa), where lodges the maxillary process of the palatine. The palatine fossa is on the lateral side of the maxilla, in the region of the articular fossa. The palatine fossa is large (Fig. 6A), with a diameter of 0.27 mm. The medial border of the maxilla is concave-shaped (Fig. 6C). The ventral pterygoid process of the palatine is straight-shaped and ventrally directed (Fig. 7A). The retroarticular process projects in parallel to the horizontal plane of the articular. The angle between mandibular condyle articulation and the retroarticular process of the compound bone is close to 90° (Fig. 8A). The edentulous dentary is restricted to the distal end of the mandible, articulating mainly with the splenial. The dorsal side of the dentary is flat and pierced by two foramina (Fig. 9A).

Hemipenial morphology

(N = 4; organs fully everted and inflated): Hemipenis single, with a long cylindrical body and conical in the distal region (apical cup) (Fig. 13A–D); a tissue sheet extends from the lateral surface of the apical cup, it folds and runs transversely forming a curved papilla (Fig. 13A); the region between this flounce and the lateral sheet is deeper, forming a pocket on the sulcate side (Fig. 13A); internal surfaces of the flounce and the conical termination covered with smooth and shallow striations (Fig. 13A, B); sulcus spermaticus single, protruding over the surface of the hemipenial body, originating on the medial surface of the basal region of the hemipenis and running distally sinuously, reaching the flounce and draining to the pocket of the sulcate side (Fig. 13C); proximal region of the asulcate side of the hemipenial body with a transversal groove (Fig. 13D); medial region of the sulcate and asulcate sides (including the sulcus walls) covered with smooth and shallow striations (Fig. 13C, D).

Figure 13.

Hemipenis of Amerotyphlops martis sp. nov. (MNRJ 18744), detail of the apical region on the sulcate (A) and asulcate views (B), sulcate (C) and asulcate sides (D). Scale bar equal to 1 mm.

Coloration of the holotype in preservative:

Dorsum (13/11/13 rows scales) pale brown. In the dorsal part of the body up to the tail, a fine darker brown reticulum, particularly concentrated in the central part of dorsal scales (Supporting Information, Fig. S5A). Venter (7/9/5 rows scales) pale cream (Supporting Information, Fig. S5B). Dorsal portions of snout pale cream, with a few light brown spots, covering partially both rostral and nasal scales (Fig. 12A, B). The ventral portion of snout pale cream and immaculate (Fig. 12C). Symphysial region pale cream and immaculate (Fig. 12C). Dorsal head scales (supraoculars, frontal, postfrontal, parietals and occipitals) and dorsal portions of lateral head scales (ocular, nasal and lower nasal) predominantly pale cream with few pale brown spots (Fig. 12A, B) and ventral portions pale cream (Fig. 12C). Cloacal plate pale cream and terminal spine creamy pale brown (Supporting Information, Fig. S5B).

Variation of paratypes:

Number of subcaudal scales ten to 12 (mean = 11, SD = 1, N = 3). Tail length 2.95–3.20% of TTL (N = 3). Largest male with 170 mm TTL. MBD 4.09–5.13 mm (mean = 4.61, SD = 0.52, N = 3); number of mid-dorsal scales 208–217 (mean = 212.3, SD = 4.50, N = 3); number of ventral scales 195–208 (mean = 201.0, SD = 6.55, N = 3); and number of scale rows around the body 20/20/18 (N = 2) or 20/20/20 (N = 1). The colour patterns of the paratypes are similar to that found in the holotype (Supporting Information, Fig. S6A–F).

Etymology:

The specific epithet is derived from the Latin name ‘Mars’, in allusion to the Mars symbol, used to represent the male gender. The choice of the name is a reference to the distinct hemipenial morphology of this species that differs from all other species of Amerotyphlops.

Distribution and habitat:

Amerotyphlops martis is only known from Praia das Neves, situated at 20 km from the municipality of Presidente Kennedy, state of Espírito Santo, Brazil (Fig. 10B). This region is considered as part of the Atlantic Forest morphoclimatic domain, in the Atlantic Coast Restingas ecoregion, a sandy plain located along to the coast of Brazil (Olson et al., 2001).

The prevalent phytophysiognomy in Praia das Neves is heterogeneous, presenting an herbaceous, shrubs and arboreous forest usually distributed in parallels ridges to the shoreline (Braz et al., 2013), with an annual average of temperature and rainfall of 20 °C and 1561 mm, respectively (Peel et al., 2007). Praia das Neves has several different levels of successional vegetation, with areas close to the beach line presenting herbaceous and shrubby vegetation, with stretches of graminoid beach communities, changing the vegetation in the interior, where presenting a fragmented Ridge forest, with medium height trees (between 15 and 20 m), with lianas, epiphytes and herbaceous understory (Braz et al., 2013).

Amerotyphlops illusorium sp. nov.
(Fig. 14; Supporting Information, Figs S7, S8)

Zoobank registration:

urn:lsid:zoobank.org:act:AF7FA356-4412-4C70-AA6D-8C4D24D81966

Figure 14.

Holotype of Amerotyphlops illusorium sp. nov. (MZUSP 18787). Head in left lateral (A), dorsal (B) and ventral (C) views. TTL = 229 mm. Scale bar equal to 5 mm.

Holotype:

An adult female, MZUSP 18787, (field number MTR 13542), collected by Miguel T. Rodrigues and collaborators on 26 March 2007 from Fazenda Nova Alegria (16° 31′ 50.7′′ S, 39° 07′ 06.7′′ W, c. 30 m a.s.l.), municipality of Trancoso, state of Bahia, Brazil (Fig. 14; Supporting Information, Fig. S7).

Paratypes:

Two male specimens, MNRJ 19614 and MNRJ 19613, collected by Tiago S. Soares between 6 and 15 June 2010 from Praia de Porto Seguro, municipality of Trancoso, state of Bahia, Brazil (Supporting Information, Fig. S8A–D).

Diagnosis:

This species is distinguished from all other South American congeneric species by the unique combination of the following of characters: (1) nasal suture incomplete; (2) rostral scale oval; (3) supralabial scales four; (4) infralabial scales three; (5) rows scales around the body 20/20/18; (6) mid-dorsal scales 221–230; (7) ventral scales 210–219; (8) rows of dorsal scales dark brown 13–15; (9) rows of ventral scales yellowish cream and immaculate five to seven; (10) caudal spine dark brown; (11) subcaudal scales ten to 12; (12) maximum TTL 229 mm; (13) maximum TL 6 mm; (14) nasal process of premaxilla contacting the vertical laminae of the nasals in the anterodorsal and posteroventral portions, with the central portion not in contact, leaving a large round canal between the olfactory chambers; (15) large palatine fossa on the lateral side of the maxilla; (16) maxilla with a straight medial border; (17) ventral pterygoid process of palatine straight; (18) ratio between length of ventral pterygoid process of palatine and skull length 0.06; (19) angle between mandibular condyle articulation and the retroarticular process of the compound bone close to 90°; and (20) dorsal surface of dentary bone with one to two evident foramina.

Amerotyphlops illusorium differs from A. costaricensis, A. lehneri, A. microstomus, A. stadelmani, A. tasymicris, A. tenuis, A. trinitatus and A. tycherus by having an incomplete nasal suture (vs. complete nasal suture); from A. arenensis, A. caetanoi, A. amoipira, A. minuisquamus, A. paucisquamus and A. yonenagae by having 20/20/18 rows scales around the body (vs. 18/16/14, 18/18/14, 20/18/14 or 20/18/15 in A. minuisquamus and 18/18/18 in A.arenensis, A. caetanoi, A. amoipira, A. paucisquamus and A. yonenagae); from A. reticulatus by having highly pigmented cephalic scales with a dark brown dorsum and dorsum tail brown (vs. yellow and few pigmented cephalic scales, dorsum brown or black and dorsum tail black with cream or yellow spot); from A. montanum by having a smaller rostral width (RW1) at the dorsal portion, between 1.22 and 1.54 mm (vs. a larger rostral width at the dorsal portion 1.88 mm); from A. martis by having a largest number of mid-dorsal scales, between 221 and 230 (vs. fewer number of mid-dorsal scales, between 208 and 217); and from A. brongermianus by having the ventral portion of the pterygoid process of palatine straight (vs. ventral pterygoid process of palatine curved). Table 1 shows additional morphometric characters and scale patterns found in A. illusorium and morphologically similar species distributed in southern and north-eastern Brazil.

Description of the holotype:

Adult female, TTL 229 mm, TL 6 mm, MBD/(SVL-HR) 0.037 mm and TL/SVL 37.16 mm. Head slightly depressed dorsoventrally, not wider than ‘neck’. Snout round in dorsal and ventral views. Rostral oval, longer than wide, narrow at anteroposterior region and wider at medial region; visible in dorsal view, extending ventrodorsally without reaching the imaginary transverse line between anterior borders of eyes. Rostral contacting nasal (anterior and posterior) dorsolaterally and first supralabial and anterior nasal scales ventrally. Nasal suture incomplete, only partially dividing the anterior and posterior portions of nasal scale. Suture begins in the upper edge of second supralabial, passes through nostril, but fails to reach rostral. Anterior nasal in contact with first infralabial and upper edge of second infralabial. Posterior nasal longer than wide, contacting upper margin of second supralabial and preocular. Supralabials four, fourth twice longer than third. Infralabials three, third largest. Eye diameter 0.77 mm; eyes not visible in ventral view, located dorsolaterally, close to suture between preocular and ocular scales, completely covered by ocular scale. Ocular scales contacting frontal. Body cylindrical and robust. Midbody diameter 8.28 mm. Dorsal and ventral scales cycloid, wider than long, highly imbricated and arranged in diagonal series; scale rows around the body 20/20/18. Mid-dorsal scales 221. Ventral scales 218. Cloacal plate rounded, bordered anteriorly by three rows of scales and posteriorly by five rows of scales. Subcaudal scales ten, excluding the terminal spine. Terminal spine large, stout base and dark brown.

Skull osteology (N = 2; MZUSP 18787 and MNRJ 19613):

Length of the skull 7.03–8.22 mm, and skull width 3.46–3.99 mm. The snout region has a globular enlarged-shape and highly consolidated. The snout articulates with the braincase by the nasal and prefrontal sutures and with the frontal bone. The anteroventral region of the premaxilla has a short backward process. The midsagittal lamina separates both sides of the premaxilla (Fig. 5). The lamina of the premaxilla is confluent with the mid-dorsal laminae of the nasals and with the mid-dorsal ridges of the vomeronasal cupola of the septomaxillae (Fig. 5). The nasal process of premaxilla contacting the vertical laminae of the nasals in the anterodorsal and posteroventral portions, with the central portion not in contact, leaving a large round canal between the olfactory chambers (Fig. 5A). The medial side of the maxilla has a shallow depression (or fossa), where lodges the maxillary process of the palatine. The palatine fossa is on the lateral side of the maxilla, in the region of the articular fossa. The palatine fossa is large (Fig. 6A), with diameters between 0.24 and 0.29 mm long. The medial border of the maxilla is straight-shaped (Fig. 6A). The pterygoid process of the palatine is straight-shaped and ventrally directed (Fig. 7A). The retroarticular process projects in parallel to the horizontal plane of the articular. The angle between mandibular condyle articulation and the retroarticular process of the compound bone is close to 90° (Fig. 8A). The edentulous dentary is restricted to the distal end of the mandible, articulating mainly with the splenial. The dorsal side of the dentary is flat and pierced by one or two foramina (Fig. 9A).

Coloration of the holotype in preservative:

Dorsum (13/13/13 rows scales) dark brown, venter (7/7/5 rows scales) yellowish cream (Fig. S7A, B). Dorsal portions of snout yellowish cream, with a dark brown spot, covering both rostral and nasal scales (Fig. 14A, B). Ventral portion of snout yellowish cream and immaculate (Fig. 14C). Symphysial region yellowish cream and immaculate (Fig. 14C). Dorsal head scales (supraoculars, frontal, postfrontal, parietals and occipitals) and dorsal portions of lateral head scales (ocular, nasal, and lower nasal) predominantly dark brown (Fig. 14A, B) and ventral portions yellowish cream (Fig. 14C). Cloacal plate pale yellowish cream and terminal spine dark brown (Fig. S7B).

Variation of paratypes:

Number of subcaudal scales 11–12 (mean = 11.5, SD = 0.7, N = 2). Tail length 2.58–3.0 % of TTL (N = 2). Largest male with 207 mm TTL. MBD 4.91–5.11 mm (mean = 5.01, SD = 0.13, N = 2); number of mid-dorsal scales 226–230 (mean = 228.0, SD = 2.82, N = 2); number of ventral scales 210–219 (mean = 214.0, SD = 6.36, N = 2); and number of scale rows around the body 20/20/18. The colour patterns of the paratypes are similar to that found in the holotype (Supporting Information, Fig. S8A–D).

Etymology:

The specific epithet is derived from the Latin adjective illusorius, illusory or ironic, in reference to the external morphology that challenges its identification when compared to Amerotyphlops brongersmianus.

Distribution and habitat:

Amerotyphlops illusorium is known from Fazenda Nova Alegria, situated at 6 km from the municipality of Trancoso, in the state of Bahia, Brazil (Fig. 10B). This region is considered as part of the Atlantic Forest morphoclimatic domain, in the Bahia Coastal forest ecoregion, along the north-eastern coast of Brazil (Olson et al., 2001)

The prevalent phytophysiognomy in Trancoso presenting a restinga and ombrophilous dense lowland forests (Veloso et al., 1991; Rizzini, 1997; Assis et al., 2011), with an elevational gradient ranging from sea level to 50 m a.s.l., with an annual average of temperature of 24.4 °C and rainfall between 1300 and 1600 mm, respectively (Veloso et al., 1991; Rizzini, 1997; Assis et al., 2011; Santos, 2013). This region has several different levels of successional vegetation, with areas close to the sandplain with herbaceous, shrubs and arboreous forest and areas at the beginning of piedmont, with lowland forests, with medium height trees (approximately 20 m), with epiphytes and herbaceous understory (Santos, 2013).

DISCUSSION

Diversity of South American AmerotyphlopsHedges et al., 2014

Regardless of the recent advances in our knowledge about the diversity of South American blind snakes the genus Amerotyphlops remains elusive, with several species still known only from their type series (Graboski et al., 2015, 2018). Our molecular and morphological analyses revealed four new species of Amerotyphlops, distributed throughout the north-eastern and south-eastern parts of the Brazilian Atlantic rain forest (ARF), representing an increase of 50% in the diversity of this group. Amerotyphlops illusorium, A. martis and A. montanum occur sympatrically and are morphologically similar to A. brongersmianus, while A. caetanoi is morphologically similar to A. arenensis and is sympatric with A. amoipira, A. arenensis and A. yonenagae (Fig. 10A, B).

The ARF is among the most biodiverse regions in the world, containing a high number of endemic and threatened species (Myers et al., 2000). This biome comprises different types of vegetation or forest formations distributed in both tropical and subtropical areas (Scarano, 2006; Marques et al., 2011; IBGE, 2012). In the last decade, many studies supported the existence of biogeographical barriers (e.g. rivers or mountains) defining areas with high levels of biodiversity and diversification within the ARF (Thomas et al., 1998; Costa, 2003; Pellegrino et al., 2005; Cabanne et al., 2007; Carnaval et al., 2009, 2014; Lirio et al., 2015; Barbo et al., 2022a). However, several studies have suggested that ecological factors (e.g. climate, soil and disturbance) may also be responsible for these geographically restricted areas of high endemism (Rocha et al., 2005; Grazziotin et al., 2006; Rocha & Sluys, 2007; Thomé et al., 2010; Carnaval et al., 2014; Saiter et al., 2016).

Saiter et al. (2016) divided the ARF into three major subregions based on floristic composition: (1) Bahia interior forests, encompassing moist and dry forests throughout the interior lands of the state of Bahia and north-eastern Minas Gerais; (2) Bahia coastal forests, encompassing the wet forests of the northern extreme of the state of Espírito Santo and most of the coastal region of Bahia; and (3) Krenák-Waitaká, which encompasses the wet forests of southern Espírito Santo. Furthermore, Rocha et al. (2005) analysed the status of endemic and endangered terrestrial vertebrates inhabiting the coastal sandplains (Restingas) along the eastern coast of Brazil, and identified three main areas of high endemism: (1) the coastal Restingas stretching from northern Espírito Santo to southern Bahia; (2) the coastal Restingas encompassing the regions of Maricá and Jurubatiba in the state of Rio de Janeiro state; and (3) Praia das Neves in the southern part of the state of Espírito Santo.

The Restingas and Bahia coastal and interior forests have been shown as highly biodiverse areas in the ARF, with the incidence of endemic and threatened species for several groups of organisms (Rocha et al., 2005, 2007; Silva & Satyamurty, 2006; Carnaval et al., 2009; Mattos et al., 2009; Telles et al., 2012; Cosendey et al., 2016; Saiter et al., 2016; Medina et al., 2020). All new species of Amerotyphlops described in the present study are endemic to these areas: Amerotyphlops illusorium and A. martis are restricted to ombrophilous dense lowland coastal forests of Bahia and the Restingas of Espírito Santo, while A. montanum and A. caetanoi occur in the ombrophilous dense and semi-deciduous coastal forests and submontane seasonal semi-deciduous forests of the interior of Bahia, respectively (Rodrigues et al., 2002, 2013; Recoder et al., 2010).

Comparative hemipenial morphology of South American Amerotyphlops

Hemipenes of South American Amerotyphlops follow the general pattern observed in scolecophidians, with single organs with an undivided sulcus spermaticus (Branch, 1986; Wallach, 1998; Graboski et al., 2018). Recently, Montingelli et al. (2022) demonstrated that the hemipenes of scolecophidians retain a plesiomorphic lizard condition. Amerotyphlops martis also conforms to this general morphology, with a sulcus spermaticus that flares near the tip of an expanded apical cup (Montingelli et al., 2022). The organs of A. minuisqua- mus and A. reticulatus are unique in presenting large and calcified spines, while this character is absent in other species of Amerotyphlops (Dixon & Hendricks, 1979; Graboski et al., 2018; Montingelli et al., 2022). The hemipenis of Amerotyphlops martis is similar to A. brongersmianus, differing only by the presence of an additional structure in the apical cup. According to our results and previous studies (Graboski et al., 2018), the micro- and macro-ornamentation was shown to be highly diverse and taxonomically informative, representing a potential source of characters that can be used to differentiate and describe hidden diversity.

Comparative osteological morphology of South American Amerotyphlops

Traditionally in taxonomic and systematics studies, osteological characters have been obtained from a variety of invasive methodologies, which considerably reduced the availability of specimens and their representation in these studies (Cundall & Irish, 2008; Bell & Mead, 2014; Bell et al., 2021). Additionally, the interpretation of the osteological character and its variation in scolecophidians has been controversial in taxonomic studies, since osteological descriptions for most taxa have been based on single-specimen preparations (List, 1966; Bell & Mead, 2014). The sample size of the four species described here is a direct product of the scarcity of specimens collected during fieldwork expeditions and their rarity in the natural history collection, which is typical in most studies involving South American scolecophidians (Rodrigues, 1991; Rodrigues & Juncá, 2002; Graboski et al., 2015).

The use of non-invasive methodologies (e.g. HRXCT) in the description and evaluation of osteology has granted access to data from taxa represented by a single or just a few specimens stored in natural history museums (Bell & Mead, 2014; Bell et al., 2021). The lack of large samples of osteological preparations can hinder the recognition of the unique patterns of variation required in taxonomic studies. However, the skull osteology of the four new species described here was obtained by HRXCT, which allowed us to recognize discrete and continuous characters from concealed portions of the internal anatomy of the skull (see characters 14 to 20 in each species diagnosis and Table S4).

Postnatal allometric growth is a common trait through most bone structures in squamates (Rossman, 1980; Scanferla & Bhullar, 2014; Palci et al., 2016), and its variation (mainly in size and sturdiness) in skulls of scolecophidians has been attributed to differences in age and body size (Laver et al., 2021). Based on our osteological samples, we did not recognize size-dependent variation associated with ontogeny. Also, our set of osteological diagnostic characters exhibited no variation associated with skull size (see diagnostic characters for each species and Supporting Information, Table S2), except for the continuous diagnostic characters (15, 18 and 19), which showed distinctive allometric patterns for each species. For example, Amerotyphlops montanum has a medium-sized skull (SL = 8.08 mm) but a small palatine fossa (PFD = 0.17 mm), while A. caetanoi has a small skull (SL = 6.15 mm) and a large palatine fossa (PFD = 0.25 mm) (Fig. 6).

In scolecophidians, females usually have a larger body and head size than males (Cox et al., 2007). However, few studies have explored the relationship between skeletal anatomy variation and sexual dimorphism. Pinto et al. (2015) studied the sexual dimorphism in Trilepida salgueroi (Amaral, 1955) and found differences between males and females in the size and number of thoracolumbar and caudal vertebrae. In our osteological analysis we did not recognize sexual-dependent variation associated with dimorphism (e.g. A. brongersmianus and A. illusorium; see Supporting Information, Table S5).

After six decades of List’s (1966) primal study – one of the most complete studies on typhlopoid and leptotyphlopoid skull osteology – the amount of knowledge on osteology of scolecophidians is still elusive. However, our current results agree with McDowell’s (1967) suggestion that osteological characters are valuable sources of evidence to evaluate the taxonomy of scolecophidians and to support the definition of natural groups. Although we acknowledge that the anatomy of the skull of Amerotyphlops remains poorly studied (List, 1966; Lira & Martins, 2021), the set of information presented here indicates a broad osteological variation for the genus. We expect that further studies on the skull morphology of Amerotyphlops will shed light on different evolutionary aspects of anatomy, ecology, behaviour and diversification patterns of this elusive lineage of fossorial snakes.

Cryptic species and morphological variation

Studies about genetic variability have been shown to be essential in assessing and inferring the cryptic diversity of all sorts of organisms (Daly & Patton, 1990; O’Connell et al., 2021). Studies based on DNA evidence have shown that some fossorial vertebrates have highly diversified modes of speciation, ranging from rapid and almost sympatric chromosomal speciation (Reig et al., 1990; Massarini et al., 2002) to allopatric speciation by geographic isolation (Daly & Patton, 1990; Steinberg et al., 2000; Maddock et al., 2020; O’Connell et al., 2021). The study of Thomas & Hedges (2007) represents a noteworthy example with scolecophidians, in which genetic information helped them to focus their attention on morphological traits capable of diagnosing hidden diversity. Based on genetic and morphological evidence, they described 11 cryptic species of Typhlops from Hispaniola, some of which were sympatric and almost indistinct from other species (Thomas & Hedges 2007).

Our results corroborate the validity of the Thomas & Hedges (2007) approach for scolecophidians. The phylogenetic information generated based on DNA sequences provided us with a solid framework on which to base our search for variability in morphological characters. Most of the external morphology (e.g. pholidosis) does not differentiate the new species described here from the widespread species A. brongersmianus, indicating their cryptic nature. However, we were able to include quantitative morphometric traits and qualitative hemipenial and skull osteology characters that enabled us to delimit these four highly divergent molecular lineages of Amerotyphlops as new species.

Extreme environmental conditions can impose stabilization on phenotypic characteristics (Lefébure et al., 2006; Bickford et al., 2007). These environments are characterized by the presence of physicochemical stressors that are lethal to most organisms (Wharton, 2002; Bell, 2012; Riesch et al., 2015; Tobler et al., 2018). Fossorial organisms, like blind snakes and rodents, must be adapted to the extreme hypoxia and hypercapnia found in their burrows as a consequence of oxygen consumption and low gas-permeability of the soil (McNab, 1966). An emblematic case is the subterranean amphipod Niphagus virei Chevreux, 1896 complex that occupies habitats marked by a complete lack of light and long periods of anoxia and food shortages (Malard & Hervant, 1999; Hüppop, 2000; Lefébure et al., 2006). In this case, molecular analyses indicated three highly divergent clades (over 13 million years of divergence), but no specific morphological character can differentiate individuals belonging to these highly divergent clades. These studies reinforce the idea that an organism with an underground lifestyle may have selective restrictions for adapting to extreme conditions, which certainly limit morphological variation (Nevo, 1979). The results of our morphological analyses support these studies, indicating, overall, a low level of morphological dissimilarity among species of Amerotyphlops. Particularly, the distribution of the morphological diversity in the multivariate space indicates extensive overlap between A. illusorium and A. brongersmianus, and a slight overlap between A. montanum and A. brongersmianus. However, our molecular phylogenetic analysis showed that all four new species represent independent lineages and exhibit high levels of genetic distances among them. Although A. illusorium and A. brongersmianus showed the lowest patristic distances within the genus (Fig. 15), the amplitude of the interspecific distance was higher than that found among species of other genera of Typhlopinae (e.g. Antillotyphlops and Cubatyphlops) (Fig. 15). Our study strongly indicates that an integrative approach is essential to evaluate the diversity of poorly studied groups of fossorial organisms, such as blind snakes, which tend to present highly conservative morphologies.

Figure 15.

Boxplot of patristic distance among Typhlopinae (Amerotyphlops, Typhlops, Antilotyphlops and Cubatyphlops).

Conservation status

Although the description of four new species increases the diversity and our knowledge of South American Amerotyphlops, these snakes are still considered rare and elusive. The most recent list of Brazilian endangered species ((https://www.icmbio.gov.br/portal_antigo/biodiversidade/fauna-brasileira/lista-de-especies.html) included A. amoipira and A. yonenagae in the category of endangered (EN), while A. paucisquamus is in the category of vulnerable (VU). Therefore, these species are currently considered under threat due to their restricted distribution and scarcity.

The four new species described here are also known only from their type locality in the Brazilian Atlantic rain forest, one of the most impacted Brazilian biomes (SOS Mata Atlântica & Instituto Nacional de Pesquisas Espaciais, 2000; Silva et al., 2007; Ribeiro et al., 2009). The description of these species reinforces the importance of conservation policies towards the north-eastern remnants of the ARF, as well as their coastal sandplain ecoregion known as Restingas.

CONCLUSIONS

Despite recent advances in the higher level systematics of scolecophidians, the diversity of this elusive group of snakes remains poorly evaluated in South America. Our study explored, through integrative analytic approaches based on an expanded taxon sampling, the cryptic nature of the South American blind snakes, genus Amerotyphlops, showing that their diversity has been largely underestimated.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article at the publisher’s web-site.

Appendix S1. Concatenated alignment used in this study to perform molecular analysis.

Appendix S2. List of specimens examined (pholidosis). Information on the museum in which they are stored, voucher and locality is given. (*) Hemipenis from specimens examined and (#) high resolution x-ray computed tomography (HRXCT) of cranial osteology examined.

Appendix S3. List of specimens examined (morphometrics). Information on the museum in which they are stored, the voucher and the locality is given.

Appendix S4. Boxplot of 36 characters for nine OTUs (separated by sex when sexual dimorphism is significant). Continuous characters – Fig. a, total length (TTL); Fig. b, snout–vent length (SVL); Fig. c, tail length (TL); Fig. d, head width at the level of the eyes (HWE); Fig. e, standard maximum width of the dorsal portion of the rostral to the internasal (RW1); Fig. f, standard length of the rostral from the internasal sutures (RL1); Fig. g, internarial distance (IN); Fig. h, midbody diameter (MBD); Fig. i, naris–eye distance (NE); Fig. j, mid-tail width (MTW); Fig. k, head radius (HR); Fig. l, eye diameter (ED); Fig. m, interorbital distance (INORB); Fig. n, width of the rostral on the apex of the snout (RW2); Fig. o, width of the rostral at the labial border (RW3); Ratios – Fig. p, ratio between tail length and snout-vent length (TL/SVL); Fig. q, ratio between midbody diameter and snout–vent length minus head radius (MBD/(SVL-HR)); Fig. r, ratio between head width at the level of the eyes and head radius (HWE/HR); Fig. s, ratio between rostral width and rostral length (RW1/RL1); Fig. t, ratio between interorbital distance and head width (INORB/HWE); Fig. u, ratio between the width of the rostral on the apex of the snout and rostral length (RW2/RL1); Fig. v, ratio between rostral length and head width (RL1/HWE); Fig. w, ratio between eye diameter and head width (ED/HWE); Fig. x, ratio between the width of the rostral at the labial border and rostral length (RW3/RL1); Fig. y, ratio between naris–eye distance and head radius (NE/HR); Fig. z, ratio between mid-tail width and snout–vent length (MTW/SVL); Fig. aa, ratio between mid-tail width and tail length (MTW/TL); Fig. ab, robustness of the head (HR/(SVL-HR)); Fig. ac, ratio between width of the rostral on the apex of the snout and rostral width (RW2/RW1); Discrete characters – Fig. ad, number of ventral scales (V); Fig. ae, number of dorsal scales (D); Fig. af, number of subcaudal scales (SbC); Fig. ag, number of scale rows around the anterior body (ROW-a); Fig. ah, number of scale rows around the midbody (ROW-m); Fig. ai, number of scale rows around the posterior body (ROW-p); and Fig. aj, number of dorso-cloacal scales (DC).

Figure S1. Maximum likelihood tree generated by RAxML. The numbers on the branch represent bootstrap values > 75%.

Figure S2. Results of principal component analysis (PCA) using the raw dataset based on 36 characters for nine OTUs. A, Group 1 (GR1). B, PCA of Group 2 (GR2). Species are colour-coded according to OTUs (see Key-colour OTUs, on the bottom).

Figure S3. Holotype of Amerotyphlops caetanoi (MZUSP S-023380). Specimen in dorsal (A) and ventral (B) views. TTL = 176 mm.

Figure S4. Holotype of Amerotyphlops montanum (MZUSP 20065). Specimen in dorsal (A) and ventral (B) views. TTL = 216 mm.

Figure S5. Holotype of Amerotyphlops martis (MNRJ 18744). Specimen in dorsal (A) and ventral (B) views. TTL = 157 mm.

Figure S6. Paratypes of Amerotyphlops martis: (MNRJ 18743) specimen in dorsal (A) and ventral (B) views, TTL = 170 mm. (MNRJ 18745) specimen in dorsal (C) and ventral (D) views, TTL = 130 mm. (MNRJ 18747) specimen in dorsal (E) and ventral (F) views, TTL = 145 mm.

Figure S7. Holotype of Amerotyphlops illusorium (MZUSP 18787). Specimen in dorsal (A) and ventral (B) views. TTL = 229 mm

Figure S8. Paratypes of Amerotyphlops illusorium: (MNRJ 19613) specimen in dorsal (A) and ventral (B) views, TTL = 207 mm. (MNRJ 19614) specimen in dorsal (C) and ventral (D) views, TTL = 151 mm.

Table S1. Detailed distribution of South American species of Amerotyphlops.

Table S2. List of taxa sequenced using Sanger methods used in this study. Information on the museum in which they are stored, the voucher and the locality is given.

Table S3. List of taxa and GenBank accession number of specimens used Sanger sequencing methods for this study. Codes in bold represent the sequences generated in this study.

Table S4. List of 50 taxonomic characters collected in this study for South American species of Amerotyphlops.

Table S5. Variation of osteological measures. Values displayed in the table are in mm. Abbreviations are as following: SL, skull length; SW, skull width; VPPL, length of the ventral pterygoid process of the palatine; and PFD, palatine foramen diameter.

Table S6. Best-fit models for 20 data partitions as determined by PartitionFinder analyses using the AIC optimality criterion to run RAxML.

Table S7. Patristic distance among species of Cubatyphlops.

Table S8. Patristic distance among species of Antillotyphlops.

Table S9. Patristic distance among species of Typhlops.

Table S10. Patristic distance among species of Amerotyphlops.

Table S11. Results of MANOVA non-parametric analysis (MANOVA-NP) using the simulated dataset based on 36 characters for nine OTUs.

[Version of record, published online 12 November 2022; http://zoobank.org/ urn:lsid:zoobank.org:pub: 9E6031A3-0186-415B-86D3-24F6F267DD10]

ACKNOWLEDGEMENTS

The authors thank the following colleagues who supplied tissue samples and specimens loans for the present study: Antônio Argôlo (UESC), Paulo Roberto Manzani and Luis Felipe de Toledo Ramos Pereira (ZUEC), Paulo Passos (MNRJ), Santiago Castroviejo-Fisher and Glaucia F. Pontes (MCP-PUCRS), Frederico França and Gustavo Vieira (UFPB), Renato Feio (MZUFV), Felipe Franco Curcio (UFMT), Fernanda Werneck (INPA), Daniel Loebman (FURG), Selma Torquato (MUFAL), Gilda Andrade (UFMA), Róbson Ávila (URCA), Rejâne Lira da Silva (UFBA), Robert Murphy (ROM), Adrian Garda (UFRN), João Tonini, Gindomar Gomes Santana, Gentil Pereira Filho and Marco Antônio de Freitas. We also thank Rosely Rodrigues da Silva, Alberto Barbosa de Carvalho and Jaqueline Battilana from MZUSP for technical support.

RG was supported by a scholarship from Fundação de Amparo à Pesquisa do Estado de São Paulo scholarship (FAPESP, 2008/52285-0) and Programa de Capacitação Institucional (MPEG-MCTI, 301346/2020-8). JCA was supported by a scholarship from the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). FGG was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, 2012/08661-3 and 2016/13469-5) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, 312016/2021-2). RAGF was supported by a grant from Apoio Estratégico a Projetos Emergentes da Universidade Federal do Pará (PAEPE-UFPA; 23073.020915/2020-35). ALCP thanks Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, 30.2611/2018-5 and 44.462/2020). SLB thanks Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, 310472/2017-2) for support. MTR thanks Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, 2003/10335-8 and 2011/501-46), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and the students of his lab for help in the field. This research was supported through grants from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, 2011/50206-9 and 2016/50127-5) to HZ.

CONFLICTS OF INTEREST

The authors declare no conflict of interest.

DATA AVAILABILITY

DNA sequences generated in this study are deposited in the GenBank Nucleotide Database and can be accessed with unique accession numbers that are given within this publication and in Supporting Information, Table S3. All other data generated or analyzed during this study are included in this published article and its supplementary information files. Scripts used in our morphological analyses are available from the Figshare repository (https://doi.org/10.6084/m9.figshare.20092721).