Description and phylogeny of a new species of Liolaemus (Iguania: Liolaemidae) endemic to the south of the Plurinational State of Bolivia

The Liolaemus montanus group is a diverse group of lizards that ranges from central Peru to southwestern Mendoza, Argentina, including much of the Plurinational State of Bolivia (“Bolivia”) and Chile. The species of this group mainly inhabit high elevation areas with cold temperatures. In the last years, several species of this group have been described, mostly in Argentina and Chile. In Bolivia, there are at least thirteen valid species belonging to the L. montanus group. In this study, we describe a new species of the L. montanus group with a marked endemism in the Cordillera de Sama of the Tarija Department, Bolivia, and a combination of unique character states that allows its formal description as a new species. The phylogenetic relationships based on analysis of 159 morphological characters suggest that it belongs to the L. montanus group, and that it is closest to Liolaemus pulcherrimus, which is found allopatrically in a small area of the Jujuy Province, Argentina. The multivariate analyses of 66 morphological characters support the phylogenetic relationships. Statistical analyses of inter-species comparisons of morphological characters are not considered the only methods due to the non-independence of some characters states among species; thus, a phylogenetic analysis is recommended. The detailed revision of specimens of the L. montanus group held in the collections of Bolivia is filling major geographic gaps and improving our understanding of the phylogenetic and biogeographic relationships of this widely distributed group of South American lizards.


Introduction
The genus Liolaemus includes more than 260 valid species [1,2,3,4,5,6], of small to medium-sized lizards distributed from Tierra del Fuego to the Ancash Region of Peru and is the second most speciose amniote clade at the global level [3,7], surpassed only by Anolis. In the last 15 years, its known species richness has increased considerably [3], with ongoing systematic revisions that have resolved many taxonomic conflicts, in certain cases after more than a century [8,9,10,11,12,13] and have allowed the description of new species within the genus [5,13,14,15,16,17,18]. The subgenus Eulaemus is divided into three large monophyletic groups: the clade comprising the Liolaemus archeforus-kingii and Liolaemus lineomaculatus groups [19]; the Liolaemus boulengeri group (characterized by the presence of a patch of enlarged scales in the posterior mid region of the thigh) [9,20]; and the L. montanus group, characterized by equal sized scales in the mid posterior region of the thigh [21,22]. The latter group is composed of more than 60 species [3], found mainly at high altitudes in the Andes of Argentina, Bolivia, Chile and Peru. The L. montanus group was proposed by Etheridge [21] and is characterized by the presence of a blade-like process of the tibia, associated with hypertrophy of the tibialis anterior muscle. In an informal phylogenetic proposal Lobo et al. [1], identified two groups within the L. montanus group: the Liolaemus andinus and Liolaemus dorbignyi groups. For many years and by many authors, the identity, distribution and diagnosis of several of these species were mistaken [23,24,25,26,27,28].
Recent field studies carried out in the highlands of Argentina, Bolivia, Chile and Peru, as well as revisions of type material and additional specimens from the type localities, have helped clarify the taxonomy of the species related to the L. montanus group [10,12,26,27,28,29]. Despite the wide distribution of Liolaemus in Bolivia, their taxonomy and phylogenetic relationships have remained virtually unknown. This lack of information is likely reflected in the currently known species richness in Bolivia, with only 21 valid species to date, of which 14 belong L. montanus group Table 1: Liolaemus chlorostictus, Liolaemus erguetae, Liolaemus fittkaui, Liolaemus forsteri, Liolaemus islugensis, Liolaemus jamesi, Liolaemus orientalis, Liolaemus pachecoi, Liolaemus pantherinus, Liolaemus pleopholis, Liolaemus puritamensis, Liolaemus schmidti, Liolaemus signifer, and, the new species described in this study. Nine Liolaemus species have been cited from the Department of Tarija: Liolaemus chaltin, Liolaemus puna and Liolaemus variegatus (subgenus Liolaemus sensu stricto); Liolaemus chacoensis, Liolaemus ornatus, and Liolaemus simonsii (L. boulengeri group); and L. islugensis, L. orientalis and L. pantherinus (L. montanus group) [30,31,32,33,34,35]. However, the number of species has varied; for example, L. simonsii is considered to be a junior synonym of L. ornatus [36]. Ongoing revisions of Bolivian material since 2011 by the present authors led to the finding that the lizards from the Tarija Department identified as L. islugensis by Tarifa et al. [32] and Quinteros and Abdala [34] are in fact quite distinct from that species and that a species new to science might be at hand. Table 1. Species of the L. montanus group described or cited from Bolivia; current taxonomic status and distribution by country and first-order administrative divisions, including the known distribution of the species in neighboring countries. The specimens examined are listed in the Appendix with their respective acronyms.
To validate the species described in this study, we used the general or unified definition of species by de Queiroz [37,38], who defines a species as an entity that represents independent historic linages, or divergent linages of meta-populations. The use of an a priori diagnostic criterion to test the boundaries among species as a hypothesis that could be empirically accepted or rejected is recommended by some authors [39,40,41,42,43]; however, with the exception of Aguilar-Kirigin [33], this approach has not been previously applied in taxonomic studies of the L. montanus group in Bolivia. Our operational criteria for inferring species limits are based on phylogenetic trees, morphological characters, and consideration of the geographic isolation of the new species in the Tajzara Basin of Tarija. We analyzed morphological characters traditionally used in taxonomic studies of Liolaemus and performed morphological phylogenetic analyses of 159 characters with 31 terminals to determine the phylogenetic relationships of the new species. We also performed multivariate analyses of 66 characters to evaluate morphological differences among phylogenetically-close species. The phylogenetic analyses indicate that the new species belongs to the L. montanus group and, within this, that it belongs to a subclade including L. fittkaui from central Bolivia and Liolaemus griseus, L. huacahuasicus, L.

Materials and methods
Examined material is listed in the Appendix. All specimens were collected by hand or noose.  [21,48], Abdala [9], Abdala and Juárez [13], and Gutiérrez et al. [6] were studied. Color in life was described based on field observations and photographs of captured specimens. Squamation was examined under a binocular microscope, and body measurements were taken with a ± 0.01 mm precision caliper. Neck fold terminology follows Abdala [9], whereas body color pattern terminology follows Lobo and Espinoza [49], Abdala [9] and, Gutiérrez et al. [6].
Each morphometric variable was measured three times on the same individual, and the mean value for each species was used in the subsequent analyses. Only adult males were used in the multivariate analysis to avoid confounding effects of intraspecific allometric variation [50] and to elude confusions in the multivariate analyses due to possible sexual dimorphism. All bilateral characters were measured on the right side. The measured morphometric traits were: snout-vent length (SVL); minimum distance between the nasal scales (DN); snout width at the edge of the canthal scale (AH); distance from the nose to the back edge of the canthal scale (NC); distance between the posterior edge of the superciliary series (EO); length of the interparietal (LEI); length of the parietal (PA); mental scale width (AM); length of the mental scale (LM); distance from nostril to mouth (NB); rostral height (HR); length of the subocular scale (ES); auditory meatus height (hTy); auditory meatus width (aTy); length of the preocular scale (LPO); preocular width (LPOT); length of the fourth supralabial scale (LCSP); length of the fourth lorilabial scale (LCLB); length between orbits (DEO); length of the first finger of the forelimb, without claw (1D); length of the claw of the fourth finger of the forelimb (G4D); length of the fifth finger of the forelimb without claw (5D); humerus width (AHU); distance from the insertion of the forelimb in the body toward the elbow (LEA1); thigh width (AMU); length of the first toe of the hind limb without claw (1P); length of the claw of the fourth toe of the hind limb (4U); length of the five dorsal scales in a row in the middle of the body (ED); cloacal opening width, measured distance between the corners of the cloaca (PP); body width (AL); width of the base of the tail (WTB); upper width of the pygal area (ASPI); length of the pygal area (LPI).
The following meristic characters were counted: number of scales around the interparietal scale (A11); number of supralabials on the right side (A12); number of supralabials on the left side (A15); number of infralabials on the right side (A13); number of infralabials on the left side (A19); number of scales around the mental scale (A14); number of scales around the rostral scale (A16); number of lorilabials (A17-1); Hellmich index (A18); subdigital lamellae of the first finger of the forelimb (A20-1); subdigital lamellae of the second finger of the forelimb (A20-2); subdigital lamellae of the third finger of the forelimb (A20-3); subdigital lamellae of the fourth finger of the forelimb (A20-4); subdigital lamellae of the fifth finger of the forelimb (A20-5); subdigital lamellae of the first toe of the hind limb (A21-1); subdigital lamellae of the second toe of the hind limb (A21-2); subdigital lamellae of the third toe of the hind limb

Phylogenetic analyses
Phylogenetic analyses were performed with the morphological matrix of Gutiérrez et al. [6], which includes 159 characters and 31 terminals (Ctenoblepharys adspersa and Phymaturus palluma as the outgroup and 29 terminals of the L. montanus group). External morphological data to build the matrix were taken from preserved museum specimens, see supplementary material in http://morphobank.org/permalink/?P3206. In the phylogenetic analysis, the parsimony criterion was used as the optimality criterion, selecting only shorter trees, or those with fewer homoplasies. The software used to search for phylogenetic hypotheses was TNT 1.5 (Tree Analysis Using New Technology, version 1.0) [51]. Discrete characters were classified into binary polymorphic, binary non-polymorphic, multistate polymorphic, and multistate non-polymorphic. The binary and multistate polymorphic characters were treated as given by Wiens [52]. Continuous characters were analyzed using the methodology proposed by Goloboff et al. [53], and these were "standardized" using an associated script (mkstandb.run). Heuristic searches were made to find the most parsimonious trees. Tree bisection and reconnection (TBR) was used for branch permutation. The matrix was analyzed using the "implied weights" method [54]. Twenty-one runs were made with the evidence: matrix one run was made with equally weights (EW) and the other 20 were made with implied weights (IW), with K values ranging from three to 22. One thousand five hundred replicates were run with each K value. Group support was estimated using symmetric resampling, a method that is not distorted by differential costs [51], with 1000 replicates and a 0.33 deletion probability, and the support values go from zero to 100.

Statistical analysis
Normal distributions of the morphometric data were examined using the Kolmogorov-Smirnov test (P � 0.05), and homoscedasticity was evaluated with Levene's test. To reduce the effect of non-normal distributions of the morphological data, all continuous variables were log 10 transformed and meristic variables were square root transformed [55,56,57]. All operational taxonomic units were analyzed by two distinct treatments. A principal component analysis (PCA) was employed to analyze the morphological variation and discriminant function analyses (DFA) were used to verify morphological variation between and within each Liolaemus species employing a jackknife classification matrix [58,59,60,61]. Four species distributed in Argentina (L. griseus, L. huacahuasicus, L. orko, and L. pulcherrimus) and the new species from Bolivia, were used as comparative groups for building the PCA and the DFA. Liolaemus fittkaui was not included in the multivariate analyses due to the availability of only two specimens in the Fundación Miguel Lillo collections; however, it was included in the phylogenetic analyses.
The PCA analysis was performed to evaluate the distribution of individuals corresponding to the five species (L. griseus, L. huacahuasicus, L. orko, L. pulcherrimus and, Liolaemus sp. nov.) in the multivariate space. The PCA was based on the correlation matrices of the morphological variables to reduce dimensionality of the data [60,62]. The PCA and DFA were evaluated separately for continuous and meristic characters, following the recommendations of certain authors to not joint both matrices in multivariate analyses, although there is no mathematical consensus on this approach [63]. Once the PCA was performed, and the lineal combinations that explained the highest variation were extracted, whether the new species exhibited similar or different morphological characters was examined by means of a DFA, with defined Liolaemus groups based on the results of the PCA. This mathematical model allows assessing whether the groups discriminated by the DFA correspond to those established by the PCA. The DFA produces a linear combination of variables that maximizes the probability of correctly assigning observations to predetermined groups; and simultaneously, new observations can be classified in one of the groups, providing likelihood values of such classification [63,64]. All statistical analyses were performed using the Statistica software, version 7.0 [65].

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:author:3528-E2D9-DB5E-4314-8672-F9751BB52FFC. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Phylogenetic analysis
The morphological phylogenetic hypothesis indicates that Liolaemus sp. nov., belongs to the L. montanus group, together with Liolaemus etheridgei, Liolaemus evaristoi, Liolaemus famatinae, L. fittkaui, L. griseus, L. huacahuasicus, L. montanus, L. orko and, L. pulcherrimus (Fig 1). This clade is supported by 10 synapomorphies, of which six are continuous (fewer scales around midbody, dorsals, neck scales, ventrals, pygals, and weaker correlation between tibial length and SVL in respect to other species of L. andinus group) and four are discrete (scales of the body, dorsum, laminar, triangular, keeled, and imbricate). The clade (L. fittkaui (L. orko (L. griseus (L. pulcherrimus + Liolaemus sp. nov.)))) supported by three characters: greater number of temporal scales and presence of parallel light and dark spots from the eye to neck in the temporal region has good support ( Fig 1) and is inferred by the analyses with K = 11 to 22, while the relationship between L. pulcherrimus as sister species of Liolaemus sp. nov., is inferred in all the analyses performed with the implied weights with five synapomorphies: greater number of number of scales separating the fourth chinshields, greater number of gulars, higher ratio of femur length/SVL, suprascapular and gular folds present. In the trees obtained with K = 3 to 10, the (L. pulcherrimus + Liolaemus sp. nov.) clade remains outside the (L. griseus (L. orko (L. montanus + L. huacahuasicus) + (L. fittkaui + L. famatinae))) clade.
Liolaemus sp. nov. is supported by 13 autopomorphies in the tree, two continuous: lower ratio of tail length/SVL, higher ratio of femur length/SVL, two discrete scale characters: slight keel in the dorsal scales of the body, precloacal pores absent or few in females, and nine discrete in terms of coloration: head never darker than body color and absence of dorsolateral bands in males, and in both sexes, paravertebral and lateral blotches presenting dark ocelli with light centers; pattern and colors of the hands and feet brindled.

Statistical analysis
The summary statistics for all the non-transformed continuous and meristic characters taken from five species of Liolaemus are shown in Table 2. Table 2. Morphological characteristics of five species of Liolaemus studied in this work. Range in the first line; mean ± standard deviation (mm) for quantitative characters in the second line.
The homogeneity of variance was not supported for either continuous or meristic characters by the Levene's test in some groups. Therefore, the results of the principal components analyses should be preferred for deriving linear combinations of the variables that summarize    Tables 3 and 4.  Table 3. Principal component (PC) axes loadings of continuous characters for L. griseus (n = 10), L. huacahuasicus (n = 15), L. orko (n = 10), L. pulcherrimus (n = 12), and Liolaemus sp. nov. (n = 15). Eigenvectors, eigenvalues, and percentage of variance explained for the first two principal components from transformed data in the five putative species of Liolaemus.
The first five components of continuous characters explained 74.68% of the variation, and a screen plot test of the PCs indicated that only the two first components contained nontrivial information. The first axis represents morphological variation, loading for most variables negatively and accounting for 33.55% of the variation, with strong loadings for length of the mental scale, length between orbits, distance between the posterior edges of the superciliary series, preocular width, minimum distance between the nasal scales, length of the pygal area, distance from the nose to the back edge of the scale canthal, and thigh width. The second axis represents body size, and accounts for most of the remaining variation, with strong loadings for snoutvent length, and length of the fifth finger of the forelimb without claw. The next axes account for the remaining variation, with important loads for length of the five dorsal scales in a row in the middle of the body, interparietal length, and length of the fourth supralabial scale. The body size effect found in the second component is interesting, since Liolaemus species are phylogenetically related, differing mainly in the morphological variation.  The first five components of meristic characters explained 54.76% of the variation, and a screen plot test of the PCs indicated that only those components contain relevant information. The five axes represent morphological variation, loading strongly for number of scales around midbody, number of dorsal scales between the occiput and the level of the anterior edge of the thigh, number of scales between canthal and nasal scales, subdigital lamellae of the second finger of the forelimb, and the subdigital lamellae of the third, fourth and fifth toe of the hind limb. The five axes account for the remaining variation, albeit with values below 0.70 for Hellmich index, number of lorilabials, subdigital lamellae of the first, third, fourth and fifth finger of the forelimb, subdigital lamellae of the second toe of the hind limb, number of infralabials on the left side, and the right side. The position of species based on their scores of the two morphological principal components axes is illustrated in Figs 2 and 3. The spatial distribution of the continuous characters indicates that morphological variation (PC 1) and body size (PC 2) are sufficient to virtually separate the five Liolaemus species. These species can also be distinguished by their position analyzing meristic characters only. In both analyses, Liolaemus sp. nov., can be differentiated from other phylogenetically related species by its morphological variation and body size.
To further clarify the position of the Liolaemus species in the morphospace of both continuous and meristic characters, a DFA was carried out where the group membership was

Etymology
The scientific name for this new species was assigned in reference to the type locality: the surroundings of the Tajzara Basin lagoons in the Reserva Biológica Cordillera de Sama of the Tarija Department, Plurinational State of Bolivia.

Diagnosis
Liolaemus tajzara sp. nov. belongs to the L. montanus group because it presents a bladelike process on the tibia, associated with the hypertrophy of the tibialis anticus muscle [20,67]. The L. montanus group species differ from those of the L. boulengeri group [9,68] Table 3.   Dorsal surface of the head smooth, with sixteen scales, rostral wider than tall, bordered by six scales. Mental larger than rostral, trapezoidal, bordered by four scales. Nasal not in contact with rostral. Two internasals longer than wide. Nasal surrounded by eight scales, separated from canthal by two scales. Seven scales between frontal and rostral. Frontal divided. Two postrostrals. Interparietal smaller than parietals, in contact with seven scales. Preocular separated from lorilabial row by one scale. Seven superciliaries and fifteen upper ciliaries. Differentiated scales at anterior margin of auditory meatus. Twelve smooth temporals. Four lorilabials in contact with subocular. Seven supralabials, not in contact with subocular. Five supraoculars. Eight lorilabials. Five infralabials. Five chin shields, second pair separated by two scales. Seventy-one scales around midbody. Seventy-eight rounded, juxtaposed and slightly keeled dorsal scales between occiput and hind limbs. Forty scale transverse rows in dorsum. Eighty ventral scales from mental to the cloacal region, following the ventral midline of the body, larger than dorsal scales, flat, imbricate. Twenty-five smooth, imbricate gulars. Thirty-seven scales longitudinal fold of the neck. Six precloacal pores. Antehumeral scales larger and easily distinguishable from the rest. Auricular and longitudinal folds present. Scales on the longitudinal fold granular and smooth. Fourth finger with eighteen subdigital lamellae; fourth toe with twenty-

Color in life
Head light brown, with distinguishable black spots in the supraocular and occipital region. Three dark spots in the temporal region parallel with each other and to the head axis, delimited by thin white stripes. The two lower stripes more evident and continue to the sides of the neck. Supralabial and loreolabial scales yellow, infralabial scales lighter, with black border. Body light brown with several scales and irregular light yellow spots, and dim. Two rows of paravertebral irregular spots in the form of irregular ocelli with black border and brown interior, with few scattered yellow scales, distinguished on the dorsum. Such pairs of ocelli extend throughout the length of the tail, where they merge in a unique central band from the first third. Small spots and black scales irregularly distributed on the vertebral region. Without vertebral line. With dorsolateral stripes, fragmented and discontinuous of an intense yellow. With lateral spots on the flanks of the body, of the same color and irregular form than paravertebral spots. Intense yellow spots distinguished between the lateral spots. Without scapular spots. Forelimbs

Color in life
Liolaemus tajzara sp. nov. shows evident sexual dichromatism (Figs 4-7). In males head varies dorsally from light to dark brown, always with black spots. The central spot is always widest and can contain a brown spot on its center, while the superior spot is generally more marked than the inferior spot. Three bold black stripes on the sides of the head and temporal region, parallel to each other, delimited by thin white stripes. These stripes run from the posterior part of the eye towards the neck, with the two upper stripes integrating to the patterns of the side of the neck. The inferior stripe reaches the auditory meatus. Loreolabial, supralabial and infralabial scales generally yellow or white, always lighter than the rest of the head. In few specimens the infralabial scales present black borders, and the rest of the scale is light-colored. No scapular spots or ante humeral arch observed. Neck dorsally has the same color and pattern design as the rest of the body. Two series of prominent paravertebral spots in the form of irregular ocelli, with brown center and intense black border (Fig 6). These ocelli vary in size, shape and color intensity according to the specimen. Over these ocelli some yellow scales can protrude. These series of ocelli occasionally merge longitudinally but never across the vertebral region. This region is well delimited, without vertebral line and with small specks and black scales irregularly distributed. In most of the observed males, thin light lines, sometimes fragmented, surround the dark ocelli. In some specimens, thin dorsolateral yellow lines are formed, while in others they can be fragmented or present only in the form of irregular spots. On the sides of the body, lateral spots in the form of ocelli protrude, of the same color and irregular shape than the paravertebral spots. Scales and white, black and yellow spots irregularly scattered between the lateral spots. Lateral midline of the body generally yellow, with small dark and light spots that do not form a regular design below it. Fore and hind limbs of the same color as the dorsum of the body, generally with many dark spots. Tail with the same coloration pattern and design as the dorsum of the body, with paravertebral spots merging on the base, forming a single spot on the vertebral region. Single longitudinal spot with black borders and light brown center along the tail in certain specimens. Most males ventrally white, with dark mole-shaped spots. Yellow color in the mental and gular region in certain specimens, and light orange color in the tail and hind limbs in one specimen.
Females generally with the same spot design and pattern than males, but with other combinations of colors (Fig 7). Striking form and design of paravertebral spots, but with less pronounced coloration. When present, yellow or orange scales on the dorsum and sides of the body, almost indistinguishable at first sight. Paravertebral and lateral ocelli reddish brown in the center, with black borders. Ventral color white, with scattered darker spots on the abdomen in some specimens.

Distribution
All known specimens and observations of L. tajzara sp. nov. are from the Reserva Biológica Cordillera de Sama, Yunchara Municipality, Avilez Province, Tarija Department, Plurinational State of Bolivia, mainly in the semi-humid Puna phytogeographic region, at altitudes higher than 3500 m (Figs 11 and 12).

Natural history and conservation status
Liolaemus tajzara sp. nov. is an endemic lizard of Tajzara Basin, where altitudes range from 3600 to 4700 m. This ecoregion is characterized by a cold and arid climate with strong winds and scarce precipitation, including rain and hail; minimum temperature in dry season reaches -18˚C and maximum temperature in wet season reaches 22˚C. Vegetation is composed mainly by bunchgrasses (Festuca orthophylla, Festuca chrysophylla, Stipa leptostachya), "tola" shrubs (Baccharis incarum, Baccharis boliviensis), "kanllar" (Tetraglochin cristatum), "yareta" cushion plants (Azorella compacta) and patches of queñoa forests (Polylepis tomentella). In the basin there are several saline lagoons, both temporal and permanent, inhabited by flamingos (Phoenicoparrus andinus, Phoenicoparrus jamesi, Phoenicopterus chilensis), many duck species (Anatidae spp.), Andean geese (Chloephaga melanoptera), horned coots (Fulica cornuta) and other aquatic bird species. In the highlands there are condors (Vultur gryphus), and towards the eastern slopes there are taruca (Hippocamelus antisensis), which move to the lowlands during the dry season months. Other native fauna of the area include tuco-tucos (Ctenomys lewisi) and mice (Akodon spp.). Liolaemus tajzara sp. nov. protect themselves from the extreme weather conditions by building underground burrows. The entire known of population of L. tajzara sp. nov. is found within the Reserva Biológica Cordillera de Sama. Given that its estimated area of occupancy is much less than 2000 km 2 , that it is known from less than 10 locations, and that we believe that projected climate change for the high Andes is likely to lead to a decline in the area of suitable habitats and thus put this high-altitude viviparous species at risk [69,70,71], we recommend that L. tajzara sp. nov., be listed as Vulnerable B.2ab following the IUCN Red List criteria.

Discussion
Until a few years ago, little was known about the taxonomy, systematics, and phylogenetics of the Liolaemus species present in the Plurinational State of Bolivia. The work by Langstroth [12] on L. stolzmanni, L. reichei, and Liolaemus jamesi pachecoi points out that the type material of the majority of the species of this genus described in the past is kept largely in European and North American museums, which for many decades limited herpetological research in Bolivia. However, the taxonomic revision of the "alticolor-bibronii" group in the scientific collections of the Colección Boliviana de Fauna allowed for the description of Liolaemus aparicioi from the dry inter-Andean valleys just below the city of La Paz [72]. Similarly, exhaust revisions of specimens housed in the major museums of Bolivia during the last few years has led to the establishment of new records for various species: L. chacoensis, Reserva Natural "El Corbalán", Gran Chaco Province, Tarija [73], L. chaltin, Reserva Biológica Cordillera de Sama, Avilez Province, Tarija [74], L. puna, Nor Chichas Province, Potosí and in the Reserva Biológica Cordillera de Sama, at Laguna Grande, Méndez Province [75], L. puritamensis, Sur Lipez Province, Potosí [29], and L. pleopholis, Sajama Province, Oruro [76]. In addition, range extensions in Bolivia were proposed for L. chacoensis [77] and L. variegatus [35]. Finally, the revision of the range of L. signifer [78] highlights the need for a historical bibliographic review of the species, as well as a taxonomic study of the supposed rediscovered holotype, the populations around Lake Titicaca, and morphologically similar taxa such as L. pleopholis.
Despite these advances, our understanding of the Bolivian members of the L. montanus group was nearly stagnant since Pellegrin [79] until the pioneering work of Raymond Laurent from the early 1980s through the late 1990s, followed again by another period of stagnation. Prior to L. tajzara sp. nov. the last species of the L. montanus group described from Bolivia were L. erguetae and L. pachecoi, both described by Laurent in 1995 [80]. The classic morphological and morphometric approach of Laurent using informative characters lead to the identification and description of the following species from Bolivia: L. erguetae [80], L. fittkaui [81], L. forsteri [82], L. pachecoi [80], and L. variegatus [83], all of which are currently recognized as valid taxa. Following Laurent's studies, the morphological-morphometric approach continued to develop, in particular in regards to the origins of characters, for example, the work of Aguilar-Kirigin [33] which correlated continuous characters in a PCA to delimit species as function of body size. The findings of Aguilar-Kirigin [33] were in line with the hypothesis of Laurent [84], which was subsequently confirmed by the macromolecular studies of Schulte et al. [68]. These studies have contributed to a better understanding of the species richness of a region of Bolivia, which still remains underexplored.
The description of the new Liolaemus species from southernmost Bolivia increases our understanding of the herpetological richness of the region, as did the morphological and taxonomic studies of Harvey and Gutberlet [85] in the isolated Huanchaca Range of the Santa Cruz Department of Bolivia, which allowed them to identify and describe Tropidurus callathelys, Tropidurus chromatops, and Tropidurus xanthochilus as new species. Likewise, the studies of the genus Stenocercus by Torres-Carvajal [86,87] which include the descriptions and redescriptions of many new or poorly understood species. As noted by Torres-Carvajal [86], approximately a quarter of the species of the genus were described after the year 1990, now more than 180 years after the description of the type species Stenocercus roseiventris [88] from Bolivia. According to Torres-Carvajal [87], one of the principal causes for the discovery of the many of these new species in recent decades has been the exhaustive morphological review of material in museums, as many collections often hold undescribed material that represent species new to science. In recognition of these findings, we are faced with the urgent need to complete and support the herpetological collections in Bolivia, in agreement with the conclusions of Langstroth [12,89], who noted that many species remained to be described from Bolivia, especially in Liolaemus, Stenocercus, and Tropidurus in the underexplored regions of the country. In the case of L. tajzara sp. nov., the first specimens were collected in 1995 and deposited in the Colección Boliviana de Fauna and Fundación Miguel Lillo, where they remained unstudied until our work on the Bolivian members of the L. montanus group began in 2010. These specimens are representative of the many of others in Bolivian collections which await study.
Abdala et al. [22] noted that from 1998 to 2007 an average of five new species of Liolaemus were described annually, but from 2008 to date, 66 new species have been described for an average of 6.5 new Liolaemus per year. The principal evidence used in support of these taxa has been morphological [1,6,15,16,90,91,92,93,94], molecular [14,66,95,96,97,98,99], morphological and molecular [5,19,100,101,102], morphological, molecular, and cytogenetic [18], and phylogenetic [2,103]. This taxonomic research demonstrates that Liolaemus have morphological characters that are informative for the delimitation of species and that many of these characters are likely to be adaptive, allowing these lizards to exploit a wide range of habitats and macroenvironments, as expressed by the high species richness of the genus [104].
To evaluate our hypothesis that L. tajzara sp. nov. is a new species to science we adopt a comparative approach, evidencing that this population of lizards inhabiting the isolated Tajzara Basin in the Cordillera de Sama of the Tarija Department can be distinguished by morphological characters, which we have supported with multivariate analyses of continuous and meristic characters. Given that species are interrelated through ancestor-descendent lineages, species can share similar character states that have been inherited from a common ancestor (i.e., plesiomorphic characters), which means that characters shared by species may not be independent; as such, we analyze the relationships among species through a phylogenetic perspective.
Given that our hypothesis is supported by our morphological analyses, we consider the molecular studies of Aguilar-Puntriano et al. [66] and the thermal ecology studies of Jiménez-Robles and De la Riva [105] to provide independent evidence that supports our hypothesis. In their research on convergence within the L. montanus group, Aguilar-Puntriano et al. [66] recovered a lineage they identified as "sp2 Sama" and "sp2 Torohuaico" (from the Tajzara Basin) based on molecular evidence. Their phylogenetic estimate includes specimens representing the populations we describe here as L. tajzara sp. nov., which they identify as Liolaemus sp. 2 in the following phylogenetic position: (L. sp. 2 + ((L. islugensis + L. pleopholis) (L. orientalis (L. sp. 1 (L. multicolor (L. sp. 4 + L. cf. schmidti)))))). Although somewhat different from our phylogenetic proposal (Fig 1), due in part to a different set of species utilized, the findings of Aguilar-Puntriano et al. [66] validate the hypothesis that the lizards of the Tajzara Basin described in this present work represent a new species of the L. montanus group. The work of Jiménez-Robles and De la Riva [105] on an assemblage of four species of Liolaemus over a topographic and altitudinal gradients in the Tajzara Basin demonstrates that the lizards they identified as Liolaemus sp. which we here describe as L. tajzara sp. nov., is differentiated in its thermal ecology and use of microhabitats in respect to the three other sympatric species: L. puna (alticolor-bibronii group), L. ornatus (boulengeri group), and L. orientalis (montanus group).
This evidence from three independent lines of research-morphological, molecular, and ecological-firmly support the hypothesis that the lizards described as L. tajzara sp. nov. belong to a new species for Bolivia and for the L. montanus group and which in the past had been confused with L. islugensis and L. signifer.
Agencia de Investigación Científica y Tecnológica (PICT 2015-1398) and to 1799 performance agreement of Tarapacá University, Chile. We would like to thank Omar Rocha and Octavio Jiménez-Robles for allowing the use of their photographs of the Liolaemus erguetae and Liolaemus fittkaui.