A new cryptic species of Anolis lizard from northwestern South America (Iguanidae, Dactyloinae)

Abstract A new species of Anolis lizard from the Andean slopes of southwestern Colombia and northwestern Ecuador, from between 1187 and 2353 m in elevation, is described. The new species can be distinguished from other Anolis in squamation, cranial osteology, hemipenial morphology, and nuclear and mitochondrial DNA. The new species is sister to Anolisaequatorialis, and it is suggested that previous records of A.aequatorialis in Colombia correspond to the new species described herein.


Introduction
Anolis lizards (anoles) are members of a diverse clade with 427 recognized species (Losos and Ricklefs 2009, Poe et al. 2017, Uetz and Hošek 2018. The phylogenetic relationships among species of anoles have been controversial for many decades (Guyer and Savage 1986, Cannatella and de Queiroz 1989, Guyer and Savage 1992, Nicholson 2002, Poe 2004, Nicholson et al. 2012) and a phylogenetic hypothesis based on morphology and DNA sequence data including most species of Anolis was not available until recently (Poe et al. 2017).
Herein we describe a new cryptic species of Anolis from northern Ecuador and southern Colombia, similar in morphology to A. aequatorialis Werner, 1894. In addition, we use DNA sequence data to infer its phylogenetic position within the Dactyloa clade (Poe et al. 2017).

Taxon sampling
We examined specimens of Anolis within the Dactyloa clade (Poe et al. 2017) from Colombia and Ecuador, housed in Ecuador: the División de Herpetología del In-stituto Nacional de Biodiversidad (DHMECN), Quito; the Museo de Zoología de la Pontificia Universidad Católica del Ecuador (QCAZ), Quito and in Colombia: the Colección Herpetológica de la Universidad del Valle del Cauca (UVC), Cali; the Museo de Herpetología de la Universidad de Antioquia (MHUA), Antioquia; the Colección de Reptiles, Instituto de Ciencias Naturales (ICN), Bogota; and the Colección de Herpetología del Instituto Alexander von Humboldt (IAvH), Villa de Leyva. All specimens examined are listed in Appendix 1, and we mapped all records using ArcMap 10.5.1 (ESRI, Inc.) with a WGS84 datum and with Universal Transverse Mercator conformal projection. We adopted the unified species concept (de Queiroz 2007), which is operationalized based on substantial and consistent differences between populations, and followed Wiens and Servedio (2000) in recognizing populations as distinct evolutionary lineages, based on the frequency of traits allowing polymorphism. In addition, we considered monophyly as a strong evidence for recognizing populations as new species.

Morphological data
We used the character terminology proposed by Williams et al. (1995) for scale characters and measurements. Specimens were sacrificed by spraying benzocaine directly in the mouth, fixed in 10% formalin, and preserved in 70% ethanol. Tissue samples (liver, muscle tissue, or scales) were extracted before fixing and placed in Eppendorf tubes with 96% ethanol. Sex determination was based on the presence of hemipenes, dewlap size and gonad inspection. Hemipenes were extracted from recently collected adult males, or everted from fixed specimens, using the method described by Pesantes (1994), with the modifications proposed by Betancourt et al. 2018. In the majority of specimens, the left organ was removed with a subcaudal incision, and submerged in sodium dodecyl sulfate for 24 hours. Skulls were prepared using dermestid beetles for two days until the bones were free of muscle tissue, and then degreased with sodium dodecyl sulfate for 24 hours. Data of color in life were obtained from field notes and photographs. All measurements were taken with a digital caliper with a precision of ±0.01 mm.

Stomach contents
From the DHMECN material (n = 18) we removed the stomach contents with a ventrolateral incision to the stomach of the preserved specimens. The stomach contents of each anole were placed in a 4% formaldehyde solution in eppendorf tubes. The identification of samples was accomplished through a stereomicroscope. We determined the stomach content at the order and family level.

DNA sequence data
Total genomic DNA was digested and extracted from liver or muscle tissue using a guanidinium isothiocyanate extraction protocol. Tissue samples were first mixed with Proteinase K and a lysis buffer and digested overnight prior to extraction. DNA samples were quantified using a Nanodrop ND-1000 (NanoDrop Technologies, Inc.), resuspended and diluted to 25 ng/ul in ddH 2 O prior to amplification.
Using primers and amplification protocols from the literature (Folmer et al. 1994, Kumazawa and Nishida 1993, Macey et al. 1997, Schulte and Cartwright 2009, we obtained 2807 nucleotides (nt) representing the nuclear gene recombination-activating gene 1 (RAG1, 811 nt), as well as the mitochondrial genes Cytochrome c oxidase I (CO1, 655 nt) and a continuous fragment including NADH dehydrogenase subunit 2 (ND2, 1038 nt), tRNA Trp , tRNA Ala , tRNA Asn , tRNA Cys (282 nt), and the origin of the light-strand replication (OL, 29 nt). New sequence data were obtained from three individuals of the new species and added to the dataset used by Torres-Carvajal et al. (2018). Gene regions of taxa sequenced in this study along with their GenBank accession numbers are shown in Table 1. Information on other taxa included in the phylogenetic analyses is available in Torres-Carvajal et al. (2018).

Statistical analyses
Given that the new species is very similar to Anolis aequatorialis, we performed a comparison using univariate t-tests for independent samples to evaluate quantitative differences between the two species (for normal data), and a Wilcoxon-Mann Whitney test for differences in squamation (for non-normal data). We conducted the Shapiro-Wilk normality test for the distribution of the data (Table 2). Statistical analyses were conducted in R (R Core Team 2016).

Phylogenetic analyses
Editing, assembly, and alignment of sequences were performed in Geneious ProTM 5.3 (Drummond et al. 2010). Genes were combined into a single dataset with eleven partitions, three per protein coding gene corresponding to each codon position, one with all tRNAs, and one with the OL. The best partition strategy along with the corresponding models of evolution were obtained in PartitionFinder 1.1.1 (Lanfear et al. 2012) under the Bayesian information criterion.
Phylogenetic relationships were assessed under a Bayesian approach in MrBayes 3.2.0 (Ronquist and Huelsenbeck 2003). Four independent analyses were performed to reduce the chance of converging on a local optimum. Each analysis consisted of 20 million generations and four Markov chains with default heating values. Trees were sampled every 1000 generations resulting in 20,000 saved trees per analysis. Station- arity was confirmed by plotting the -ln L per generation in the program Tracer 1.6 (Rambaut et al. 2013). Additionally, the standard deviation of the partition frequencies and the potential scale reduction factor (Gelman and Rubin 1992) were used as convergence diagnostics for the posterior probabilities of bipartitions and branch lengths, respectively. Adequacy of mixing was assessed by examining the acceptance rates for the parameters in MrBayes and the effective sample sizes (ESS) in Tracer. After analyzing convergence and mixing, 2000 trees were discarded as "burn-in" from each run. We then confirmed that the four analyses reached stationarity at a similar likelihood score and that the topologies were similar, and used the resultant 72,000 trees to calculate posterior probabilities (PP) for each bipartition on a 50% majority rule consensus tree. We calculated ND2 uncorrected genetic distances in PAUP* 4.0 (Swofford 2002).  (Poe 2004, Poe et al. 2017) based on the following combination of characters: sexual size dimorphism; large body with high numbers of lamellae; more than 20 scales across the snout; Alpha type caudal vertebrae; prefrontal bone separated from nasal; lengthened dentary and loss of angular.
Among dactyloid species from Ecuador and Colombia, Anolis dracula is similar in color and morphology to A. fitchi and A. podocarpus. However, both species occur east of the Andes in Ecuador and they can be distinguished (character states in parentheses) from A. dracula by the following characters: hemipenis with slightly defined lobules, which means that the outline of the lobules are not clearly distinguishable from the trunk (lobules well defined), and twice as long as hemipenes of A. fitchi and A. podocarpus, hemipenis length in A. dracula 14 mm (A. fitchi 7 mm; A. podocarpus 6 mm; Figure 8); well-developed parietal crests, bowed outwards and projected laterally (irreg-  ular, with curved edges in A. fitchi; relatively straight in A. podocarpus; Figure 9); large and oval pineal foramen (small and rounded in A. fitchi and A. podocarpus); smooth lateral edges of frontal bone (serrated in A. fitchi and A. podocarpus; Figure 9); short nasal bones (elongated in A. fitchi and A. podocarpus; Figure 9); lateral projections on posterolateral edges of parietal crests (no lateral projections in A. fitchi); strongly rugose surface of basioccipital and sphenoccipital tubercles (rugose in A. fitchi and slightly ru- Table 3. Comparison of lepidotic characters, with Wilcoxon-Mann Whitney tests, between Anolis dracula sp. n. and A. aequatorialis, from Ecuador and Colombia. For each character the Z and p values are given, after range and sample size (in parenthesis), and mean/median ± standard deviation for each species. Asterisks indicate the degree of significance, * α = p < 0.05, ** α = p < 0.01.

Character
A. dracula sp. n.
Finally, although the average ND2 genetic distance between A. dracula and its closest relative A. aequatorialis is relatively low (0.049), it is comparable to DNA divergences between other species pairs, such as Anolis heterodermus versus Anolis inderenae (0.042) and Anolis anatoloros versus Anolis jacare (0.041).
Trunk: Middorsal and paravertebral scales small and keeled, slightly larger than flanking scales, which are granular/conical and separated by small skin interspaces; ventral scales smooth, subimbricate, larger than dorsals; groin, axilla and neck covered by granular scales; nuchal and dorsal folds present, reduced in females; two enlarged postanals in males.
Limbs: Fore and hind limbs with keeled scales; hind limbs more robust, 1.8 times longer than forelimbs; lamellae of subdigital pad of fourth toe 19 (18-23; counted in the manner of Williams et al. 1995 Tail: Cylindrical, with keeled scales at the base, others imbricated; 125% longer than snout-vent length. Color in life (holotype and paratypes): Anolis dracula is chromatically variable depending on sex, emotional stress, and perch type ( Figure 5). Males dorsum with dark brown transverse bands delineated by black and separated by greenish brown, or light brown or black bands separated by cream (Figs 6,9); females dorsum varies, from light green with dark green V-shaped transverse bands separated by pinkish cream, turquoise cream or whitish cream lines (Figure 5), to beige or dark brown with darker brown transverse bands separated by whitish coloration (Figs 5, 6); all morphs exhibit a light brown or black hourglass-shaped spot on insertion of forelimbs; tail dark green with bands separated by pink spaces in females and light green or dark brown in males; belly usually cream; throat cream with light green small spots in females and immaculate in males; iris copper; tongue cream; in males naked skin of gular sac dark brown, with bright turquoise to bright green scattered irregular markings, longitudinal rows of sac scales green; in females, naked skin of gular sac brown, with irregular bright turquoise to bright brown scattered markings, longitudinal rows of sac scales turquoise.
Color in preservative (holotype and paratypes preserved between two and ten years): Dorsum in males bluish grey, flanks whitish pale-blue, with light or black hourglassshaped spots, belly grey or bluish cream (Figs 6-7); dorsum in females bluish grey,  separated by light bluish-cream transverse bands on flanks, with black or white spots, belly cream; both sexes with black visceral peritoneum.
Hemipenis (Figure 8): Hemipenis bilobed, 14 mm in length; trunk becoming distinctly wider distally; lobules short and rounded; asulcate side with a semicircular constriction in first quarter of trunk; sulcus spermaticus wide, with thick fringes, branching at base of lobules and extending to base of transversal veil on sulcate side of lobules; apical and asulcate surfaces of lobules covered by calyces; asulcate region of trunk and proximal region of lobules with thin transverse folds; surface of constriction separating stem from apex with small calyces and folds.
Distribution and natural history. Anolis dracula occurs on the foothills of the Andes of southwestern of Colombia and northwestern Ecuador. It has been recorded in the provinces of Carchi and Imbabura, Ecuador, and the department of Nariño, Colombia, between 1187-2353 m in elevation. The known distributional area of A. dracula is relatively small, approximately 1582 km 2 (Figure 13), and all records are from evergreen low montane forest (Cuesta et al. 2009, MAE 2013.
Anolis dracula was the most common species of anole during surveys conducted by the Herpetological Division from Instituto Nacional de Biodiversidad de Ecuador during the June-August period in 2015 and 2016 at Cerro Oscuro in the Dracula Reserve. Specimens were collected in mature and secondary forests, degraded areas with pastures and native vegetation, as well as along the edges of secondary roads. Almost all specimens were found sleeping at night on leaves of Araceae, Arecaceae, and pteridophytes, between 0.6 and 2.3 m above the ground.
Occasional observations during 2016 (March-June) suggest that A. dracula shows sleeping-perch fidelity and is active on the ground. A female was observed sleeping on the same leaf of Araceae for two consecutive nights. In the same field trip, we observed two females in clear and sunny days starting thermoregulatory behavior at 7 am, with slight head movements and small jumps between branches. As the sun rose, the females moved down to the ground. A male was observed foraging on leaf litter at noon. In addition, several specimens were collected in pitfall traps. Some individuals were observed on leaf litter during the day, with cryptic coloration (brown color), whereas at night, most specimens were greenish.
Etymology. The specific epithet dracula it is a noun in apposition that refers to the Dracula Reserve, located within the distribution of the new species and near its type locality. The Dracula Reserve is an initiative of the EcoMinga Foundation, sponsored by the Orchid Conservation Alliance, Rainforest Trust, University of Basel Botanical Garden, and their individual donors. The Reserve protects an area with a high diversity of orchids of the genus Dracula.
Phylogenetic relationships. The Bayesian analysis estimated Anolis dracula to be sister to A. aequatorialis, with strong support (Figure 12). This relationship was expected, as these species are very similar in morphology and coloration (Figs 2-4). The above clade is sister to A. anoriensis, and within a strongly supported clade also containing A. gemmosus, A. otongae, A. poei, A. eulaemus, A. ventrimaculatus, A. peraccae, A. anchicayae, A. fasciatus, A. chloris, A. gorgonae, and A. festae, all representing an important radiation of the Dactyloa clade of Anolis along the western slopes of the Tropical Andes in northwestern South America (Figure 12).

Discussion
Although Ayala-Varela and  reported Anolis aequatorialis from southwestern Colombia, we found that the specimens on which such a distribution was based (UVC 10802, UVC 16001) correspond to specimens of A. dracula. Recently, Cisneros-Heredia (2017) determined that the type locality of A. aequatorialis is located in the western slopes of the Andes in northern Ecuador, in the province of Pichincha. Based on our sampling and phylogenetic analyses (Figure 12), we infer that typical populations of A. aequatorialis are located only in Ecuador.
The hemipenis of Anolis dracula is proportionally larger than that of its sister species, A. aequatorialis. However, further studies are necessary to evaluate whether this sexual morphological difference may have led to the divergence of these two highly cryptic sister species.