A new species of Amphisbaena (Squamata: Amphis baenidae) from the Orinoquian region of Colombia

In northern South America, amphisbaenians are rarely seen among the herpetofauna.Thus, general knowledge about them is very poor. During a herpetological survey in 2012 at Casanare, Colombia, we found two specimens of an unusual Amphisbaena. A third specimen sharing the same morphotype labeled Amphisbaena sp. from Vichada department was found deposided in an Colombian reptile collection. Based on morphological analyses together with phylogenetic analyses of 1029 base pairs of the mitochondrial DNA (mtDNA), we describe a new species of Amphisbaena that inhabits in the Orinoquian region of Colombia. The new species is part of a phylogenetic clade together with A. mertensii and A. cunhai (central-southern Brazil), exhibiting a great genetic distance (26.1–28.9%) between the newly identified lineage versus those taxa, and versus the sympatric taxa A. alba and A. fuliginosa. Morphologically, this new Amphisbaena can be distinguished from their congeners by characters combination of number of preocloacal pores, absence of malar scale, postgenial scales and body and caudal annuli counts. Amphisbaena gracilis is on morphology grounds the most similar species. However, the new species can be distinguished from it by having higher body annuli counts, angulus ories aliegned with the edges of the ocular scales and center of frontal scales, less number of large middorsal segments of the first and second body annulus, and rostral scale visible from above. The description of this new Amphisbaena species points out the urgent need to increase the knowledge of worm lizards in Colombia


Introduction
Amphisbaenians are one of the most enigmatic and unusual squamates. All species have burrowing habits, but some occasionally venture onto the surface or can be found under objects on the ground (Pough et al. 1998). Thus, due to its fossorial habit, cryptic behavior, secretive microhabitats and lower encounter rate, amphis-baenians are considered an elusive research objective. About 102 species of the genus Amphisbaena Linnaeus, 1758 have been described in South America (Gans 2005;Uetz et al. 2020), with Brazil being the country with the highest diversity with over 80 species (Gans 2005;Gomes and Maciel 2012;Teixeira et al. 2014;Uetz et al. 2020).
Colombia is considered to be a megadiverse country in part due to its rich fauna of around 621 species of reptiles (Uetz et al. 2020). However, worm lizards remain poorly represented in the Colombian herpetofauna due to the lack of scientific knowledge. Currently, Colombian worm lizards comprise two genera (Mesobaena Mertens, 1925 andAmphisbaena Linnaeus, 1758) and five species: Mesobaena huebneri Mertens, 1925;Amphisbaena alba, A. fuliginosa, A. medemi and A. spurrelli. Amphisbaena alba and A. fuliginosa (sensu Vanzolini 2002) are the most widely distributed amphisbeanids in the country; A. alba is restricted to the Cis-Andean region while A. fuliginosa is present in both Cis and Trans-Andean regions, ranging from the sea level to 1300 m. a.s.l. Amphisbaena spurrelli was the first amphisbeanid described in Colombia. It is distributed across the Chocoan region to Panáma and its type locality corresponds to corregimiento of Andagoyá, Municipality of San Juan, department of Chocó (Boulenger 1915;Gans and Mathers 1977). Mesobaena huebneri, the second worm lizard species described, is only known from three disjunct and distant localities: Its type locality corresponds to department of Inirida (Amazonian basin, specific locality unknown); the Timbá community, municipality of Mitu, deparment of Vaupes; and Serranía de la Macarena, department of Meta [specific locality unknown (Gans 1971;Cole and Gans 1987)]. Finally, Amphisbaena medemi was erected by Gans and Mathers 33 years ago and is the most recently described worm lizard. This species is distributed across the Caribbean region of Colombia, having as type locality the old Inderena fishing facility at Ciénaga de Amajehuevo, municipality of San Cristobal, Atlántico.
After the early efforts made by Gans and collaborators during the 20 th century, few attempts have been made to carry out a comprehensive taxonomic assessment of the Amphisbaena species distributed in Colombia, as well as in northern South America (Señaris 1999;Costa et al. 2018a). The most recent studies in Colombia have only provided a check list of the already known Amphisbaena species or distributional records obtained from field-work, ignoring the specimens housed in museums that are waiting for a detailed revision (Rangel-Ch et al. 2012;Angarita-Sierra et al. 2013;Aponte-Gutiérrez et al. 2019;Carvajal-Cogollo 2019).
During a herpetological inventory in the department of Casanare, Colombia (Pedroza-Banda et al. 2014), we found two specimens of an unusual Amphisbaena from the municipalities of Paz de Ariporo and Orocué. A third specimen sharing the same morphotype seen in the Amphisbaena specimens from Casanare was found in the reptile collection of the Pontificia Universidad Javeriana, labeled as Amphisbaena sp., from the municipality of Puerto Carreño, department of Vichada. These three specimens shared unique similarities between them and did not match previous descriptions of any recognized species of the genus (Gonzalez-Sponga and Gans 1971; Gans and Mathers 1977;Gans 2005). Hence, it has become clear that these specimens represent an undescribed evolutionary lineage of amphisbaenians. Therefore, the goal of this paper is to recognize this new species and describe it by integrating molecular and morphological analyses.

Ethics statement
Fieldwork was performed under the scientific research permit for collection of wild specimens of biological diversity for non-commercial purpose issued by COR-PORINOQUIA (Research Auto: 500.5712.0380) and the Colombian Ministry of Environment and Sustainable development (MADS) by agreement 083 of 2012. This study was conducted following the Colombian animal welfare law and the collection of wild specimens of the biological diversity acts (Ley 1774(Ley , 2016Decreto 1376Decreto , 2013, as well as considering the Universal Declaration on Animal Welfare (UDAW) endorsed by Colombia in 2007.

Fieldwork and sampling
Fieldwork was carried out in August 2012 in the municipalities of Paz de Ariporo and Orocué, department of Casanare, Colombia. Searches for amphisbaenians were conducted by three researchers from 8:00 to 11:30 and 14:00 to 17:00 for 15 days, with a sampling effort of 97.5 man/hours. We removed covered objects and leaf litter, digging up the ground from 5 to15 cm deep, during three to five minutes for each event. Particularly, we included piles of palm leaves of moriche palm (Mauritia flexuosa L.f., 1782), as part of the microhabitats sampled. Individuals collected were immediately placed into cloth bags for later general procedures of measurement and identification as described by Pedroza-Banda et al.

Molecular data collection and laboratory procedures
Molecular distinctiveness and phylogenetic relationships of the new species of Amphisbaena were assessed by analyzing molecular data corresponding to 1029 bp of the NADH dehydrogenase subunit 2 (ND2) gene, mtDNA. We assembled a data set by aligning the sequence from the new species and colombian individuals of A. alba and A. fu li ginosa, with homologous sequences from the Antillean and South American amphisbaenian species published in Genbank ( Table 1). The homologous ND2 sequence of the lizard species Anolis auratus DQ377355 was used as outgroup. Total genomic DNA was extracted using a standard phenol-chloroform method (Sambrook et al. 1989). We amplified the gene fragment using the primer pairs NADHF/NADH R and L4349/H5540 (Measey and Tolley 2013). We carried out PCRs in a total volume of 30 μl containing one-unit Taq polymerase (Bioline; Randolph, MA), 1 X of buffer (Bioline), a final concentration of 1.5 mM MgCl2 (Bioline), 0.5 μM of each primer, 0.2 mM of each dNTP (Bioline), 0.2 µg of bovine serum albumin (BSA) and approximately 50 ng of total DNA. We purified the PCR products using the ammonium acetate protocol (Bensch et al. 2000), and we sequenced them on an ABI 3130xl genetic analyzer (Applied Biosystems, Foster City, CA, USA) using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) at the Instituto de Genética, Universidad Nacional de Colombia. We stored the remaining DNA extractions at -80°C in the tissue collection of the Instituto de Genética (for voucher numbers see Table 1). We performed the thermocycling conditions as indicated by Measey and Tolley (2013). The GenBank accession numbers of the obtained sequences are: MT433762, MT433763, MT433764, MT433765, MT433766 (Table 1). The sequences were edited and aligned using Chromas 1.51 (http://www.technelysium. com.au/chromas.html) and BioEdit 7.0.5.2 (Hall 1999).

Phylogenetic analyses and genetic divergence
We analyzed the dataset using the unpartitioned and partitioned (i.e., we treated each codon of the protein-coding gene ND2 as distinct partitions) partition schemes. We assessed the optimal partitioning scheme and best-fit evolutionary models using Partitionfinder v1.1.1 and the Bayesian Information Criterion (Lanfear et al. 2012), resulting in the selection of the partitioned scheme. For this scheme we applied the resulting models in a Bayesian analysis with MrBayes v3.2.1 (Ronquist et al. 2012): ND2 1st and 3rd codons -GTR+I+G and ND2 2nd codon -TVM+G. We incorporated these models into a single tree search mixed model partitioning approach (Nylander et al. 2004). For this analysis, we carried out two parallel runs using four Markov chains, each starting from a random tree. We ran the Markov chains for 20 million generations. The burn-in was set to sample only the plateau of the most likely trees that were used for generating a 50% majority-rule consensus. We then used the software TRACER 1.5.4 (Rambaut and Drummond 2007) to assess an acceptable level of the MCMC chain mixing and to estimate effective sample sizes for all parameters. To assess the genetic differentiation between the new lineage and the other related Amphisbaena species (including the sympatric ones A. fuliginosa and A. alba), we calculated uncorrected p genetic distances for the ND2 gene fragment using MEGA 7.0.21 (Kumar et al. 2016).

Morphology
We compared the collected amphisbaenians and the individual found in the collection of the Pontificia Universidad Javeriana to other preserved specimens housed in the following colombian biological collections: reptile collection of the Instituto de Ciencias Naturales, Universidad Nacional de Colombia (ICN-R, Bogotá); Museo de Historia Natural, Universidad de Antioquia (MHUA, Medellin); Museo de la Universidad La Salle (MLS, Bogotá); Pontificia Universidad Javeriana (MUJ, Bogotá); Instituto de Investigación de Recursos Biológicos Alexander von Humboldt (IAvH-R, Villa de Leyva) and the reptile collection of the Universidad Industrial de Santander (UIS-R, Bucaramanga).
We compared the pholidosis of the three specimens analyzed in this study to morphological data available in published references of the 50 nominal four pored Amphisbaena species, as well as to the Amphisbaena species that inhabit the Orinoquian region ( Table 2). The definition and terminology used in the diagnosis, description and comparison sections are, as far as possible, in accordance with the broadly used descriptions of South American amphisbaenians according to Gans (1962Gans ( , 1963Gans ( , 1967; Gans and Mathers (1977); Vanzolini (1994); Vanzolini (2002); Teixeira et al. (2014) as follows: number of precloacal pores (P); supralabial scales (SS); infralabial scales (IS); temporal scales (TS); number of segments of the first postgenial scale row (FPG); number of segments of the second postgenial scale row (SPG); malar scales (M); number of segments of the postmalar scale row (PM); body annuli (BA); caudal annuli (CA); number of dorsal segments per annulus at midbody (DS); number of ventral segments per annulus at midbody (VS); number of segments per annulus at anterior edge of the cloaca (SAC); number of segments per annulus at posterior edge of the cloaca (SPC); number of cloacal annuli (CCA) [Cloaca annuli are those between the anterior and posterior edge of the cloaca]; autotomy sites on caudal annuli (AUC). Likewise, we followed the characters used by Gonzalez-Sponga and Gans (1971), particularly, we added to our analyses the angulus oris (i.e. the lateral limit of the oral fissure formed by junction of upper and lower lips), as well as the presence/ absence or number of the large middorsal segments of the first and second body annulus.

Species n elb alb ful ana ang ano are bah bol bra cae cam car cai cui cun dar has ign kin kra lee leu mer mit mun pre rob sax sil sch uro ven
A. elbakyanae sp. nov.  (3) rostral scale short, subtriangular, ventrally expanded and posteriorly without contact with prefrontal scales; (4) nasal scales in broad contact; (5) six premaxillary teeth; (6) ten maxillary teeth.
Diagnosis. Amphisbaena elbakyanae sp. nov., can be distinguished from all its congeners by the following combination of characters: (1) three supralabial scales; (2) three infralabial scales; (3) second supralabial scale longer than first and third supralabial scales, contacting first and third supralabial, temporal, ocular and prefrontal scales; (4) angulus oris lies in transverse plane passing through the posterior edges of the ocular scales and the center of the frontal scales; (5) second infralabial scale in contact with postmental scales; (6) six premaxillar teeth; (7) ten maxillar teeth; (8) Fig. 3A-B), first body annulus includes one large segment on each side lying immediately posterior to inner parietal scales, abutting onto posterolateral edge of the outer-parietal scales (versus first body annulus including two or three, large segments on each side lying immediately posterior to inner parietal scales, abutting onto posterolateral edge of the outer parietal scales in A. gracilis, Fig. 3A-B); middorsal segments of second and third body annuli non-enlarged (versus three or four middorsal segments of second and third body annuli enlarged in A. gracilis, Fig. 3A-B) and angulus oris lies in trans-   Table 2. (Figs 2-4; Table 4). Male, small body size (SVL = 211 mm; TL = Incomplete tail); slender body (BD = 5.3 mm); head and body slightly differentiated by a small nuchal constriction; head longer than wide (HW/HL 77.7%); snout rounded; six premaxillary teeth beginning with two large, anteromedian teeth that are flanked on either side by a posteriorly directed row of two slightly recurved teeth that gradually diminish in size; ten maxillary slightly recurved teeth that gradually diminish in size arrayed in an oblique row; rostral scale visible from above, subtriangular, ventrally expanded, wider and concave posteriorly, narrowly contacting first supralabial and broadly contacting with nasal scales; nasal, prefrontal, frontal and parietal scales from both sides contacting along the midline of the head forming a longitudinal suture (Figs 2A, 3A); nasal scale quadrangular, contacting the first supralabial, prefrontal and rostral scales; nostrils lateral in the anteroventral part of nasal scale; prefrontal scales roughly pentagonal, wider than long (PFW/PFL 92.9%), broadly contacting nasal, frontal, ocular, first and second supralabial scales, hav-   ing a narrow contact with first supralabial scale and a broad contact with second supralabial scale (Figs 2A,  3A); frontal scales trapezoidal, longer than wide (FW/FL 63.0%), in broad contact with prefrontal, postocular and inner parietal scales and in narrow contact with ocular scale. Four parietal scales roughly pentagonal; inner parietal scales longer than wide (IPW/IPL 91.4%), in broad contact with frontal, postocular, and outer-parietal scales, as well as with the middorsal enlarged segments of the first body annulus; outer parietal scales wider than long (OPL/OPW 91.8% ), in broad contact with inner-parietal and postocular scales; first body annular non-enlarged scales, but in narrow contact with middorsal enlarged segments of the first body annulus; angulus oris lies in transverse plane that passes through posterior edges of the ocular scales and center of frontal scales (Figs 2B, 3E); three supralabial scales, first subtriangular, longer than wide in broad contact with nasal and second supralabial scales, in narrow contact with prefrontal and rostral scales; the second supralabial larger than the first one and third supralabial scales, contacting first and third supralabial, temporal, ocular and prefrontal scales; third supralabial scale smaller than first and second supralabial scales, contacting second supralabial, temporal and in posterior contact with first body annulus; ocular scales rhomboidal, longer than high (OH/OL 62.4%), in broad contact with prefrontal, postocular, temporal and second supralabial scales, in narrow contact with frontal scales; eye slightly visible in the anterior corner of the ocular scale; postocular scales roughly hexagonal, longer than wide (POW/POL 84.8%), broadly contacting frontal, parietal, ocular, temporal and in posterior contact with first body annulus; one temporal scale roughly pentagonal longer than wide (THE/TEL 68.1%) broadly contacting second and third supralabial and ocular scales, as well as the first body annular scales. Mental scales quadrate, smaller and narrower than rostral scale, longer than wide (MW/ML 94.8%), in broad contact with postmental and first infralabial scales; postmental scale oblong, longer than wide (PMW/PML 70.3%), visible longer than and in broad contact with mental scale, first and second infralabials and postgenial scale row; three infralabial scales, first trapezoidal, longer than wide and in broad contact with mental, postmental and second supralabial scales; second infralabial scale larger than first and third infralabial scales, broadly contacting first and third infralabial and postmalar scale rows; third infralabial scale smaller than first and second infralabial scales, in contact with second infralabial scale, postmalar scale row and in posterior contact with first body annulus; malar scales absent; postgenial scale row composed by four segments, in contact with second infralabial, postmental, and in posterior contact with postmalar row of scales; postmalar row of scales composed by seven segments (Figs 2C, 3C). Body annuli demarcated; lateral and middorsal sulci present, beginning from 16 th (left) or 18 th (right) body annulus; 245 body annuli, 13 dorsal segments per annulus at midbody, 16 ventral segments per annulus at midbody; first body annulus with one enlarged middorsal segment on each side contacting with posterior edge of the inner parietals, abutting onto posterolateral edge of the outer parietal scales; middorsal segments of second and third body annulus non-enlarged (Figs 2A, 3A); four precloacal pores rounded; anal flap semicircular; four cloacal annuli, six caudal annuli (incomplete tail), caudal autotomy site between sixth to seventh caudal annuli (Figs 2D, 3G). (Fig. 4). Dorsal and ventral surfaces from dark brown to dark brown-reddish; occipital, parietal, frontal, temporal, third supralabial, third infralabial, postmental scales, as well as postgenial and postmalar scale rows dark brown highly pigmented; rostral, prefrontal, ocular, nasal, first and second supralabial, mental and first infralabial scales dark brown faded.

Color of the holotype in life
Color of the holotype in preservative (Fig. 2). After seven years in preservative, dorsal and ventral surfaces, as well as head scales maintained dark brown coloration having slight differences with color in life, such as a faint grey coloration on dorsal and ventral surfaces, and a few unpigmented scales.
Etymology. We dedicate this species to the Kazakhstani scientist Alexandra Asanovna Elbakyan (Russian: Александра Асановна Элбакян), creator of the web site Sci-Hub, for her colossal contributions for reducing the barriers in the way of science, as well as her reclamation that "everyone has the right to participate and share in scientific advancement and its benefits, freely and without economic constraints".
Distribution and natural history. The known localities of Amphisbaena elbakyanae sp. nov., are distributed in the flooded savanna ecosystem of the Orocué and Ariporo River basin, as well as in the drained savanna ecosystem of the Bita River basin in the department of Vichada (Fig.  5). Amphisbaena elbakyanae sp. nov. seems to be highly associated with the leaf litter of the savanna flood forest dominated by moriche palm (Mauritia flexuosa), which are commonly known as "morichales" or "cananguchales" in Colombia (Fig. 6). The new species was found in sympatry with A. alba and A. fuliginosa.

Discussion
In this research, molecular and morphological evidence allowed us to confirm that Amphisbaena elbakyanae sp. nov. represents a new species of amphisbaenian from northern South America (sensu Eva and Huber 2005). Our phylogenetic analysis suggests that Amphisbaena elbakyanae sp. nov. together with A. cunhai and A. mer-tensii from central-southern Brazil, is part of the same monophyletic clade (Fig. 1). However, great genetic distances for the ND2 gene fragment were revealed between Amphisbaena elbakyanae sp. nov. versus A. cunhai and A. mertensii (28.9% and 26.1%, respectively). Currently, molecular data of several species from northern South America is lacking (e.g. A. medemi, A. spurrelli, A. gracilis, A. vanzolinii, and A. steinegeri), limiting the understanding of the evolutionary relations of northern-South American amphisbaenians. Therefore, it is crucial to include many more taxa, to formulate a complete phylogenetic hypothesis that may reduce spurious phylogenetic relationships, basal polytomies and poorly supported nodes (Teixeira et al. 2014). Despite the scarcity of the molecular data, our analyses revealed that the new taxon is not closely related to the sympatric species A. alba or A. fuliginosa (Fig. 1), confirmed by the great genetic distances between them (Table 3). The morphological evidence analyzed allowed us to clearly diagnose Amphisbaena elbakyanae sp. nov. as a different lineage compared to the 50 nominal four pored Amphisbaena species, demonstrating that it was an undescribed species of worm lizard from Colombia.
Furthermore, both molecular and morphological evidence agreed with Gans and Mathers (1977) group's division of the amphisbaenians from northern South America: The first group included two larger and wide-ranging species (A. alba and A. fuliginosa), and the second group comprised six smaller narrow-ranging species (A. gracilis, A. medemi, A. rozei, A. spurrelli, A. stejnegeri and A. vanzolinii). Based on the morphological characters of Amphisbaena elbakyanae sp. nov., this taxon can be allocated into Gans and Mathersʼ second group. Interestingly, Amphisbaena elbakyanae sp. nov., exhibited a close morphological similarity with both closely distributed taxon (e.g. A. gracilis) and geographically distant taxa (e.g. A. cunhai, A. frontalis, A. talisiae and A. slateri ). Moreover, Amphisbaena elbakyanae sp. nov. and A. gracilis are the continental worm lizards that seem to have the greatest affinity with the Antillean Amphisbaena species by showing a lack of malar scales, four precloacal pores, relatively small size and uniform dorsal and ventral pigmentation. Aditionally, A. elbakyanae sp. nov. together with A. gracilis and A. medemi are the only forms of the northern mainland that have fewer dorsal rather than ventral segments to a midbody annulus closely resembling the Antillean Amphisbaena species (Gans and Alexander 1962;Gonzalez and Gans 1971;Gans and Mathers 1977).
This situation leaves open the question of whether such morphological similarities are due to evolutionary ancestry or could be due to convergent evolution of characters, a product of adaptation to similar habitats (Harmon et al. 2005;Edwards et al. 2012). Some authors have claimed that parallelism, understood as the independent evolution of similar traits, starting from a similar ancestral condition, could be another possibility for morphological similarities between Amphisbaena species (Mott and Vieites 2009). Vidal et al. (2008) dated the split between African and South American Amphisbaenidae at 40 Mya ago (Eocene), proposing that transatlantic dispersal from Africa to South America + West Indies could explain this divergence. According to Gonzalez and Gans (1971), the West Indies species may be the ancestors of the northern South American Amphisbaena species. Consequently, the similarities between some Antillean and South American species may have resulted from the retention of a primitive character pattern in a zone geographically peripheral to the range of the genus. Although we cannot assess directly Gonzalez-Sponga and Gan's hypothesis, the distant evolutionary relationship between Amphisbaena elbakyanae sp. nov. and the Antillean species A. caeca and A. xera revealed by our phylogenetic and genetic distance analyses ( Fig. 1; Table 3), as well as the distant relationships showed by Pyron et al. (2013; Fig 12K) between A. cunhai and A. mertensii (species that form a monophyletic clade together with Amphisbaena elbakyanae sp. nov.) and the Antillean species (i.e. A. bakeri, A. caeca, A. cubana, A. fenestrate, A. manni, A. schmidti and A. xera), suggest that recent evolutionary ancestry may not be the cause of the morphological similarities. Those and many more questions concerning northern South American worm lizards remain open, evidence that the state of knowledge for many fields is still extremely fragmentary.

Conclusions
Amphisbaena elbakyanae sp. nov., described as a new species from the Orinoquia savanna ecosystem of Co-lombia, seems to be related to A. cunhai and A. mertensii from central-southern Brazil. This species of Amphisbaena is one of the several still-unrecognized evolutionary lineages of worm lizards that are deposed in Colombian museum shelves waiting to be described. We think that the lack of worm lizard studies in Colombia is derived from three main factors. First, insufficient funding for field and museum research; second, large areas still lack intensive sampling and third, there are few investigators searching for worm lizards and few experts and trained personnel capable of describing species (Gascon et al. 2007; Ospina-Sarria and Angarita-Sierra 2020). Therefore, the description of this new Amphisbaena species points out the urgent need to generate a research grant program that could support field surveys and research on several disciplines to increase our knowledge of worm lizards, as well as help to train researchers to describe species including the known but yet-undescribed species currently housed in Colombian biological collections. Studies of taxonomy and species descriptions in a megadiverse country like Colombia play a substantial role in the conservation of our natural heritage. Thus, encouraging these activities will allow an evaluation of biodiversity loss and the development of systematic conservation planning and practices, as well as a scientific focus on value judgments that make up environmental policies and laws.