A taxonomic mystery for more than 180 years : the identity and systematic position of Brachysaura minor ( Hardwicke & Gray , 1827 )

Brachysaura is a monotypic genus of agamid lizard found in the Indian subcontinent; the identity and systematic position of B. minor has been long debated, and it has at times been subsumed into Agama, Charasia and Laudakia, with some authors suggesting affinities to Calotes. We constructed nuclear and mitochondrial phylogenetic trees including Brachysaura and allied agamid genera to resolve its phylogenetic position. We also compared osteology and external morphology with the genera Agama, Calotes and Laudakia. Hemipenial morphology was compared with Calotes and some other agamids from South Asia. Both nuclear and mitochondrial phylogenies demonstrate that Brachysaura is nested within the widespread South and Southeast Asian genus Calotes, with which it also shares certain external morphological, osteological and hemipenial characters. Adaptations to ground dwelling in Brachysaura minor has resulted in unique modifications to its body plan, which is likely why generic allocation has been long confused. This study also highlights the need for an integrated systematic approach to resolve taxonomic ambiguity in Asian agamids.


Introduction
Hardwicke & Gray (1827) described an interesting agamid with a very short tail from Chittagong, Bangladesh, as Agama minor.Their description was based on unpublished colour sketch by Hardwicke (Hardwicke NHM archives No. 82), which was in accordance with the figure (Smith 1935).Later Blyth (1856) described a new genus and species, Brachysaura ornata, based on specimen collected by Jerdon from Ságur, Central India.He considered this species as 'Calotes with enormous head, short and thick body, the tail not exceeding the body in length, and the toes short and strong'.He also mentioned a weak nuchal crest, medial ridge with row of high-keeled scales and two detached tufts of sincipital spines.He distinguished this new genus based on the transverse arrangement of body scales, which are less obliquely oriented than in Calotes.Although Blyth's B. ornata appeared very similar to Agama minor, he did not refer to hardwicke & Gray (1827).Later, Günther (1864) doubted the generic allocation of this species and noted that it was impossible to characterize the new genus Brachysaura, or fix its position in the family Agamidae based on the descriptions given by Blyth (1856).Subsequently Stoliczka (1872) reported this species from Kutch (then Kachha) as B. ornata.The specific epithet ornata was followed by BoulenGer (1885), who also did not refer to hardwicke & Gray (1827).Based on characters such as depressed body, very feeble dorsal crest, shoulder pits, a transverse throat fold, and a distinct tympanum, BoulenGer (1885) placed it in a new genus Charasia, along with Psammophilus spp.from the Indian subcontinent.
Smith (1935) compared Hardwicke's painting (Hardwicke NHM archives No. 82) with the type of Bra chy saura ornata and synonymized them, placing them in the genus Agama.Smith's (1935) generic allocation was based on the exposed tympanum and presence of callose preanal scales and therefore called it Agama minor.However he noted that the "the absence of callose preanal scales should not deter one from placing it under Agama" and gave an example of other species (A. mutabulis, A. inermis and A. sinaita) in the genus, which had reduced callose scales (Smith 1935).These three species are presently placed in the genera Trapelus and Pseudotrapelus (uetz & hošek 2015).wermuth (1967) also followed Smith (1935) and retained the name Agama minor.daS (1994) erroneously placed this species in the genus Laudakia, mentioning that he followed nomenclature of moody (1980), though Moody had treated it as the monotypic genus Brachysaura.manthey & SchuSter (1999) placed this species back into the genus Brachysaura.Subsequently, most authors have followed the genus name Brachysaura (daS 2003(daS , inGle et al. 2012(daS , khan 2006(daS , khan & kumar 2010)).Although murthy (2010) without any explanation moved this species to the genus Laudakia.Furthermore khan & kumar (2010) wrongly list Acanthosaura minor as one of the synonyms of Brachysaura minor although the former is a synonym of Oriocalotes paulus (Smith 1935).
The taxonomic affinities of Brachysaura minor thus remain unclear.This species has been assigned to various genera of which Charasia is not valid any more.Most previous studies suggest that Brachysaura minor could be related to either Calotes (Blyth 1856), Agama (Smith 1935) or Laudakia (daS 1994).We combined molecular and morphological data in an effort to resolve the chequered taxonomic history of Brachysaura minor and shed light on its evolutionary affinities.

Materials and methods
Sampling for this study is restricted to Gujarat state in India.Tissue sample were collected from two specimens, CESG 162 from Kutch and NCBS AQ035 from Chotila in Saurashtra.Remaining three specimens, CESG 466 is from Narayan Sarovar, BNHS 2307 is from Velavar and NCBS AQ036 is from Wardhwan.Genomic DNA was extracted from tissue samples that were stored in 99.9% ethanol and refrigerated at -20 o C. DNeasy® (Qiagen™) blood and tissue kit was used to extract DNA.Partial sequences of mitochondrial gene nicotinamide adenine di-nucleotide dehydrogenase subunit 2 (ND2; 1059 bp) and nuclear recombinant activating gene (RAG1; 940 bp) were PCR amplified and sequenced using published primers (macey et al. 1997, macey et al. 2000, Groth & BarrowclouGh 1999).Additionally, ND2 and RAG1 sequences for agamids were downloaded from GenBank (Table S1).These sequences were aligned using ClustalW in Mega 5. 1 (tamura et al. 2011).Uncorrected genetic distances between taxa were also calculated using MEGA 5.1.We used the GTR+I+G model for all phylogenetic analyses based on results from JModelTest 2. 1.2 (darriBa et al. 2012).Trees were generated using both maximum likelihood (ML) and Bayesian approaches (BI).The ML tree was generated using the program RaXML GUI (SilveStro & michalak 2012, StamatakiS et al. 2005) with bootstrap 500 reps in ML + rapid bootstrap settings.The program MrBayes 3.2 (ronquiSt et al. 2012) was used to generate the Bayesian tree with default prior settings.Markov chains were sampled every 1000 generations from a total of 1,000,000 generations, and the first 25% of the trees were discarded as "burn-in".Based on the findings by Pyron et al. (2013), Brookesia brygooi was used as an outgroup.To further test the relationship of Brachysaura minor with different genera the best ML tree was compared with alternative trees where Brachysaura minor was constrained to be sister to either Laudakia or Agama.These three tree topologies were compared using Shimodaira-Hasegawa (SH) test in PAUP 4.0b10 (Swofford 2002) Voucher Specimens were fixed in 4% formaldehyde for 24 hours and subsequently preserved in 70% ethanol for long-term storage.All morphometric measurements were taken using Mitutoyu TM digital vernier calipers (to the nearest 0.01mm).Most morphological characters were measured following zuG et al. (2006), which are explained below.The following measurements were taken: snout-vent length (SVL, from tip of snout to anterior border of cloaca), head length (HL, distance from anterior edge of tympanum to tip of snout) head width (HW, distance from left to right outer edge of temporal at their widest), head depth (HH, at a point behind orbit), snouteye length (SE, from snout tip to anterior border of orbit), eye to tympanum (EE, from posterior border of orbit to anterior border of tympanum), jaw length (JL, from rostrum to end of jaw), interorbital width (IO, transverse distance between anterodorsal corners of left and right orbits), naris to eye (NE, distance from the anterior edge of orbit to posterior edge of naris), snout width/internasal distance (IN, transverse distance between left and right nares), tympanum diameter (TD, greatest diameter of tympanum), orbit diameter (OD, distance between anterior and posterior margins of orbit), lower arm length (LAL, distance between elbow to upper side of forefoot), upper arm length (UAL, distance from anterior insertion of forelimb to elbow), finger lengths (F1, F2, F3, F4, F5), femur length (FL, length of femur from groin to knee), crus length (CL, length of crus (tibia) from knee to heel), hind foot length (HFL, distance from proximal end (heel) of hindfoot to distal most point of fourth toe excluding claw), toe lengths (T1, T2, T3, T4, T5) (e.g.T4 = Distance from juncture of 3rd and 4th digits to distal end of 4 th digit on hindfoot excluding claw), torso length (TrL, from arm pit to groin), torso height (TrH, at mid-torso), torso width (TrW, at mid-torso), tail length (TL, from posterior border of cloacal opening to tip of tail), tail height (TH) and tail width (TW) at tail base.Following meristic characters were counted: mid-body scale rows (ABS, number of scale rows around trunk at mid-body), ventral scales (VEN, number of scales from below mental to cloacal opening for females, number of scales after the end of dewlap to cloacal opening for males), Number of subdigital lamellae (from 1 st lamella at digits' cleft to most distal lamella, head scales (number of scales in transverse line between posteriormost left and right supraciliary scales), supralabials (SL, posterior end defined by posteriormost enlarged scales that touches infralabials at rear corner of mouth), Infralabial (IL, posterior end defined by posteriormost enlarged scales that touches with supralabials at rear corner of mouth).

Phylogenetic relationship
In both ML and Bayesian trees built using mitochondrial sequences Brachysaura minor was nested within the wide spread Asian agamid genus Calotes with high bootstrap support and posterior probability (Fig. 1).Within the sampled Calotes clade, it is sister to Calotes jerdoni, an arboreal species found in Northeast India.Furthermore, the best tree, where Brachysaura minor is nested within Calotes, had a significantly higher likelihood score than the alternative trees (SH test, P < 0.05).The nuclear gene tree also had strong bootstrap support for Calotes minor being sister to Calotes calotes and not with Laudakia or Agama (Fig. S1).(Smith 1935, 'n' not provided).The following characters, although not mentioned in other literature but are consistent with the type description: two tufts of spines over the tym panum (Fig. 2), broad scales on the dorsum and a relatively short SVL to tail length ratio (Fig. 3).

Morphology
To avoid the ambiguity there is a need to provide a revised morphological description for Brachysaura mi nor, based on published literature and the present study which is given as below.scales around the body 48 -60, supralabial 10 -15 and infralabials 10 -15; dorsal and lateral scales large, moderately keeled, and larger than ven trals (Fig. 3); subdigital lamallae under the fourth toe 16 -18, and subdigital lamallae under the fifth toe 8 -9; tufts of spine above tympanum 2 -6 and tufts of spine on the occiput region 5 -6.They have 9 -10 scales on the dorsal eyelids; 5 -7 scales on the snout between the nostrils and 12 -14 scales on the head between the posterior most supraciliary scales.
Brachysaura minor is a medium sized agamid with stout body and relatively short tail (Fig. 2; Table 1).This species have the shortest tail compared to 19 out of the 25 known species of Calotes (Fig. 4A, Table 1).Bra chy

Hemipenial morphology
Hemipenis is single, clavate (divided less than half of its length), length of the entire organ greater than its width.Base naked, sulcus spermaticus single, proximal two third of sulcus spermaticus deep, dorsal half shallow.Lips of sulcus spermaticus smooth, widely open at the apex.A fleshy cardiod structure present at the base of ventral sulcus.Calyculate ornamentation present on each lobe.Thick walled smooth calyces forming deep oval pits.Apex capitate and are divided into four segments (Fig. 5).Hemipenis of B. minor matches with all the 14 characters reported from six other Calotes spp.from Sri Lanka and 13 out of the 14 characters with Calotes cey lonensis (Table 2).

Osteology
The skull of Brachysaura minor is subpentagonal in outline and raised in the parietal region and slopes steeply towards the nasal Fig. S2.The hyoid apparatus of B. mi nor is composed of six parts viz. 1.The entoglossal process; 2. The paired ceratohyals; 3. Ceratobranchials I; 4. Ceratobranchials II; 5.The paired epibranchials articulated posterior to the certobranchials I; 6.All the other five structures originate from a central body called the hyoid body.The Alzarin red stain suggest that most of the hyoid is bony, except the epibranchial I and II which is the tip of the respective ceratobranchials which are cartilaginous.Ceratobranchial II is more than half

A B
the length of ceratobranchial I (Fig. 6).Brachysaura mi nor has 21 vertebrae excluding atlas and axis and 29 caudal vertebrae (Fig. 6).Sternum with a central foramen which is oval and elongated.Ten of the vertebrae had ribs and seven vertebrae has small ribs instead.The shoulder girdle is made of a broad clavicle; interclavicle long and rod like, suprascapula wedge shaped.The forelimbs is composed of a humerus with well-developed proximal and distal ends, ulna relatively thicker than radius.The phalange formula of the manus is 2 : 3 : 4 : 5 : 3.
The hindlimb is formed by femur, tibia and fibula.The phalange formula on pes is 2 : 3 : 4 : 5 : 4. Thirteen out of sixteen characters examined in the skeleton of B. minor match with those of Calotes (Table 3).Only eight out of the sixteen characters compared matched with the genus Laudakia and only five out of the sixteen characters matched with the genus Agama (Table 3).Overall characters of B. minor are more similar to Calotes and are different from Laudakia and Agama (Table 3).   the Zoological Survey of India, Calcutta.We identified this specimen as Psammophilus cf.blanfordianus based on the small scales on the body and strong fold in the shoulder region.Additionally, Brachysaura minor have plate like scales between the eye and above tympanum, which is missing in this specimen.The only record of B. minor from eastern part of the subcontinent is the type locality (Chittagong).Previous authors have raised their about the occurrence of B. minor in Chittagong (see khan & kumar 2010).We also suspect this record is erroneous as the habitat at Chittagong is forested with high rainfall compared to open grassland in semi-arid and arid regions of western and central India.While rest of the areas where they occur have mean annual rainfall between 200 -1500 mm, Chittagong has more than 2500 mm mean annual rainfall (Fig. 7).Interestingly, the closest sister species to Brachysaura minor, Calotes jer doni is distributed in North Eastern India, Myanmar and Bhutan (Smith 1935(Smith , uetz & hošek 2015)).

Discussion
Phylogenetic relationship and morphology    Brachysaura minor have the shortest tail length and fifth toe compared to many other Calotes, these morphological difference probably lead to the confusion in the allocation of the genus.The relative length of tail to SVL is remarkably distinct and was used as key for this species (Smith 1935).Most members of genus Calotes are arboreal except Calotes chincollium, which are reported to be primarily on ground during feeding (vindum et al. 2003).The reduced fifth toe in Calotes minor is most likely an adaptation to exclusive ground dwelling habits.In the Indian sub-continent there are other ground dwelling genera in the subfamily Draconinae, which have reduced or lost its fifth toe.In Otocryptis, for example, fifth toe is reduced or rudimentary; while in Sitana it is completely lost.Although members of both these genera have relatively long tails.The reduction of tail length is not known in any other members in the subfamily Draconinae.However, in the sister subfamily Agaminae, the following genera Xenagama, Bufoniceps and a few species of Phrynocephalus, which are exclusively ground dwelling also have reduced tail length.Thus, the characters unique to Brachysaura minor, such as short tail and toes, appear to be an adaptation to ground dwelling habit and these characters have evolved convergently in multiple unrelated lineages of agamids.
Overall B. minor have greater similarity in osteological characters with the genus Calotes than with Laudakia or Agama (Table 3).Calotes have 46 -65 caudal vertebrae and the genus Agama have 35 -45 except Agama hispida, which have 20 -34 (moody 1980).Two out of the three species of Agama studied by moody (1980) are saxicolous in habit and one of them "Agama hispida" is ground dwelling.Brachysaura minor is another ground dwelling agamid, which has reduced number of caudal vertebrae (Table 3).In case of agamids the number of vertebrae does not always correlate with length of the tail instead they are modified long or short and thick (moody 1980).However, in species like B. minor and A. hispida, which have relatively short tail, there is evident loss of caudal vertebrae.The genus Psammophilus which was earlier considered as Charasia have 13 out of 16 osteological characters (included in this study) similar to Calotes (moody 1980) and 11 out of the 16 characters matched with B. minor.

Taxonomic implications
Subsuming Brachysaura into Calotes warrants reassessment of diagnostic characters for the genus Calotes.This is because many of the characters used to differentiate be-tween Pseudocalotes, Calotes and Bronchocela are body ratios (hallermann & Bӧhme 2000).However B. minor is unique among Calotes in that it has body ratios that are very different from other Calotes spp.This study shows that there are unique hemipenial characters that can be used to diagnose the genus Calotes.Though the osteology of agamids is well documented in moody (1980), it requires extensive revision in the light of new advances in agamid taxonomy.Further studies are required on both hemipenis and osteology to resolve taxonomic ambiguities and for a better understanding of the evolutionary relationships within Draconinae.
The inclusion of genus Brachysaura to Calotes will not have much taxonomic implications for Brachysaura minor, which becomes Calotes minor (hardwicke & Gray 1827).Although Gray (1845) listed Agama minor in the synonymy of Calotes minor but the latter is a different agamid, presently considered as Oriocalotes pau lus (Günther 1864, Smith 1935, ananjeva et al. 2011).The specimens of O. paulus referred by Gray (1845) under C. minor were misidentified and they were not types of the new species.The interpretation by Smith (1935) was wrong, which was corrected by wermuth (1967).Thus the name Calotes minor is not preoccupied and is available for Brachysaura minor.

Conclusions
Convergence in morphological characters driven by its ground dwelling habits probably caused confusion in the genus allocation of Brachysaura minor.This highlights the need to revise diagnostic characters for the genus Calotes.The genus Trapelus and Bufoniceps are the only Agaminae found in Western India.Therefore our finding also limits the distribution of members of Agaminae to the arid regions ( < 500 mm rainfall) of Western India.S1.Sequences used in this study and their GenBank and voucher number.Table S2.Source for morphometric data on the genus Calotes used in figure 4.

Fig. 1 .
Fig. 1.Bayesian tree inferred from mtDNA data in MrBayes 3.2.The values assigned on the internodes indicate maximum likelihood bootstrap values and dark circles indicate posterior probability support above 95% and light circles indicate less than 95% probability.

Fig. 2 .
Fig. 2. Close up image of the head of Brachysaura minor (NCBS AQ036) collected from Wadhwan, Gujarat.Note the tuft of spines on the head and the row of enlarged scales/plate like between eye and tympanum.Scale bar: 10 mm.

Fig. 4 .
Fig. 4. Plot of body measurements of Brachysaura minor (blue circles (males), red circles (females)) compared to other known Calotes spp.(Blue triangles (males), red triangles (females)).Black circle Calotes bhutanensis sub-adult male.A) SVL versus tail length of 17 out of the 25 described species of Calotes.B) SVL versus fifth toe length of 9 out of the 25 described species of Calotes.The extremely short fifth toe and tail reflect its adaptation to terrestrial habits.

Fig. 6 .
Fig. 6.Dorsal and ventral view of clear stained Brachysaura minor (CES G466).Note.The extreme tip of tail was broken and are not shown in this image.Scale bar = 15 mm.
Brachysaura minor was considered to be one of the widespread species of agamid in the Indian subcontinent.However, most recent and confirmed records on the distribution of B.minor are from western and central India (see inGle et al. 2012; khan & kumar 2010).The record of this species from Angul, Odisha by chakraBorty & GuPta (2009) is that of Psammophilus cf.blanfordi anus.The voucher specimen (ZSI 25833) is deposited at macey et al. (2000) described six major clades of agamids, which are now considered as subfamilies (Pyron et al. 2013).The genus Calotes, which belongs to subfamily Draconinae is largely Indian radiation.Our study reveals that the genus Brachysaura should be formally placed in the genus Calotes (subfamily Draconinae) as B. minor is nested within the Calotes radiation.Furthermore, B. mi nor also share some prominent morphological characters present in the genus Calotes and notable among them is the presence of cardioid structure on the hemipenis, which

Fig. 7 .
Fig. 7. Rainfall pattern in the India sub-continent and the distribution of Brachysaura minor.Red triangle: type locality of B. minor, black triangle: location where Blyth's specimens were collected; black circles are place where we collected the samples, green circles previous records.

Fig. S1 .
Fig. S1.Maximum likelihood tree built using RAG1 sequences.Fig. S2.Skull of Brachysaura minor (CESG466) used to study cranial osteology TableS1.Sequences used in this study and their GenBank and voucher number.TableS2.Source for morphometric data on the genus Calotes used in figure 4.

Table 1 .
Morphometric measurements of the four specimens used in this study.

Table 3 .
Comparisons of select osteological characters of B. minor and the three genera with which it was placed earlier.* In some species one or more phalange lost in 4 th or 5 th toe/finger.