Resurrection of Bronchocela burmana Blanford, 1878 for the Green Crested Lizard (Squamata, Agamidae) of southern Myanmar

Abstract Recent fieldwork in southern Tanintharyi revealed the presence of a small Green Crested Lizard in the wet evergreen forest. We generated mtDNA sequence data (ND2) that demonstrates that this population’s nearest relative is Bronchocela rayaensis Grismer et al., 2015 of Pulau Langkawi, northwestern Peninsular Malaysia and Phuket Island. Morphologically the Burmese Bronchocela shares many features with Bronchocela rayaensis, which potentially would make this recently described Thai-Malay species a synonym of Bronchocela burmana Blanford, 1878; however, we interpret the genetic and morphological differences to reflect evolutionary divergence and recommend the recognition of both species.


Introduction
Bronchocela is a light weight among agamid lizards, rivaling Draco in the lightness and slenderness of its body and limbs. Draco's slenderness is associated with its gliding locomotion. The slenderness and extremely long tail (~ 3× body length) of Bronchocela seem to be an adaptation for moving on the outer edge of the branches of shrubs and trees.
The first member of the Bronchocela clade to be recognized was Agama cristatella Kuhl, 1820 (now B. cristatella). This species continues to be the most commonly recognized Bronchocela owing to its broad distribution from Peninsular Malaysia to the Philippines and into the Lesser Sunda Islands (Grismer et al. 2015). The first mention of a Bronchocela in Burma (Myanmar) was in 1878 by Blanford. He recognized that a Burmese specimen differed strikingly from B. cristatella and the other species described to that time and named this taxon B. burmana. One major difference was the height, shape, and number of nuchal crest scales. Boulenger (1880) did not believe that this Burmese species was unique and synonymized B. burmana with Calotes cristatellus, a name used then because Boulenger also did not consider Bronchocela Kaup, 1827 to be a unique group even though Bronchocela had more than 50 years of recognition by other major authorities in herpetology. Malkmus (1992) appears to be the first author to use Bronchocela in a modern publication; however, his use was limited to B. cristatella presence in the Mount Kinabalu (Sabah, Borneo) area. Diong and Lim's (1998) review of B. cristatella is the first formal use of Bronchocela with a full explanation of their decision to revive the generic name. Their decision was based on a confirmation of Scott Moody's data in his unpublished dissertation demonstrating that cristatella was not closely related to other members of the Calotes clade.
Bronchocela burmana has remained a forgotten name until the present, in part because of Boulenger's nomenclatural authority, but also because few Bronchocela have been collected in Myanmar owing to its limited occurrence to southern Tanintharyi ( Fig. 1) and the failure to examine closely the few specimens from southern Burma. One of us (DGM) has been participating in rapid assessment surveys of proposed national parks in southern Tanintharyi. He first discovered two specimens of Bronchocela in the Lenya NP area in 2015 and one in the Lenya NP Extension in 2016. We likely would have labeled these specimens as B. cristatella without examining them closely if part of the survey did not also include mtDNA barcoding of all collected specimens. When additional sequence data (ND2) indicated that the Lenya specimens were not closely related to B. cristatella, instead to the recently described Bronchocela rayaensis (Grismer et al., 2015) from Pulau Langkawi, northwestern Peninsular Malaysia and Phuket Island, Thailand (Grismer et al. 2016), we decided to make a morphological comparison as well.
We present below our analysis of the molecular and morphological data for the southern Burmese Bronchocela specimens. As our title indicates, these data demonstrate that the Burmese population is unique.  Taylor's (1963) two localities for B. cristatella (open squares) in southern Thailand, and B. rayaensis type locality (star) in northwestern Peninsular Malaysia and its newly reported localities (open circles) in Thailand (Grismer et al. 2016). A solid diamond denotes the type locality of B. burmana. The red dashed lines depict the political boundaries between Myanmar-Thailand, Cambodia-Thailand, and Malaysia-Thailand. Base map from: CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=166887.

Molecular analyses
We sequenced eight specimens of Bronchocela, five from the California Academy of Sciences (CAS) from the area near Kawthung, and three National Museum of Natural History-Smithsonian (USNM) specimens from the Lenya NP area (Table 1). Liver and muscle tissue samples were collected in the field during specimen preparation and pre-served separately in EtOH (CAS specimens) or a DMSO/EDTA salt-saturated buffer (USNM specimens;Mulcahy et al. 2016). Specimens were subsequently fixed in 10% formalin and transferred to 70% ethanol for long-term storage at the USNM collection, housed at the National Museum of Natural History (NMNH) and at CAS. Extractions of genomic DNA were conducted on small pieces of liver or muscle tissue and run on an Auto-Genprep 965 (2011 AutoGen, Inc.), using standard phenol manufacturer protocols. Genomic DNA was eluted in 100 µl of AutoGen R9 re-suspension buffer. We sequenced the mtDNA ND2 gene because that is the locus used by Grismer et al. (2015) in their descriptions of two new species of Bronchocela. For our PCR reactions, we used the primers L4437-H5934 (Macey et al. 1997). Cycle-sequence reactions were performed in both directions, using the PCR primers using BigDye Terminator v3.1 Cycle Sequencing Kit's in 0.25 × 10 µl reactions run on and ABI3730 Sequencer (2011 Life Technologies) using the 950 chemistry. Raw trace files were edited in Geneious 9.1.5 (Bio matters Ltd 2005-2016), complementary strands were aligned, edited, and inspected for translation. All sequences were deposited in GenBank under the accession numbers KY366308-KY366315. Our sequences were aligned with the ND2 samples (including the outgroups) from Grismer et al. (2015), the single new individual from Phuket Island, Thailand (Grismer et al. 2016), and another specimen in GenBank (AF128497) initially identified as B. cristatella (Macey et al. 2000a) and AF128495, initially identified as Aphaniotis fusca because these specimens were reported as being switched in GenBank (Zug et al. 2006). We used Geneious and the MUSCLE Alignment with default settings and secondarily inspected for codon alignment and translation. Grismer et al. (2016) did not deposit the ND2 sequence of their new specimen (ZMKUR0017/LSUHC12347) from Phuket Island in GenBank, but did provide us with a copy. We deposited their sequence in GenBank under the sequence number KY498356. The tRNA region of the ND2 locus required some adjustments to insure the tRNA's were properly aligned based on secondary stem and loop structure (sensu Macey et al. 1997). The tRNA secondary structure was determined using ARWEN (Laslett and Canback 2008). We performed maximum-likelihood (ML) analyses using RAxML (v8.2.9, Stamatakis 2014) with the rapid bootstrap inferences (1000 replicates) and subsequent thorough ML search, with ND2 partitioned by codon and the tRNAs as a separate partition.

Morphological analyses
Hallermann began his systematic studies of Bronchocela in 2004 and, in a series of studies (Hallermann 2004(Hallermann , 2005(Hallermann , 2009, developed a set of characters for morphological analysis. Grismer et al. (2015) used Hallermann's characters and introduced several new ones in his molecular and morphological study of Peninsular Malaysian populations of Bronchocela. Our study of morphological variation used most of the previous proposed characters and includes the following: Measurements-snout-vent length (SVL), distance from the tip of snout to the vent; trunk length (TrnkL), distance from posterior edge of forelimb insertion to anterior edge of hindlimb insertion; tail length (TailL), distance from middle of vent to tip of tail; forelimb length (ForelL), distance (dorsal) from trunk between forelimb's insertion to tip of third finger, not including claw (Grismer et al. 2015 listed fourth toe; we suspect that mentioned digit was incorrect); forefoot length (ForefL), distance from base (middle) of palm to tip of third finger, not including claw; 4 th finger lengths (4FingL), distance from the juncture of the third and fourth finger to the end of the ultimate lamellae of the fourth finger, hence excluding the claw; 4 th toe lengths (4ToeL), distance from the juncture of the third and fourth toes to the end of the ultimate lamellae of fourth toe; hindfoot length (HindfL), distance from base (middle) of sole to tip of fourth toe, not including claw; hindlimb length (HindlL), distance (dorsal) from trunk between hindlimb's insertion to tip of fourth toe, not including claw; head length (HeadL), distance from posterior edge of tympanum to tip of snout; head width (HeadW), transverse width of the head at posterior angles of jaws; head depth (HeadD), distance from top of head between orbits to lower surface of jaw (dentary); orbit diameter (OrbD), maximum horizontal distance from anterior and posterior edges of orbit (not eyeball); tympanum diameter (TypmD), greatest horizontal distance regardless tympanum height; body scale size dorsally (BSC.dors), width of parasagittal dorsal scale at midbody; body scale size ventrally (BSC.vntl), width of median ventral scale at midbody; nuchal crest length (NucCrstL), length of nuchal crest from anterior edge of anteriormost crest scale to posterior edge of posteriormost crest scale; nuchal crest height (NucCrstH), height of the highest/largest nuchal crest scale.
Scalation-supralabial scales (Suplab), number of scales between rostral scale and angle of jaw; infralabial scales (Inflab) numbers of scales between mental scale and angle of the jaw; loreal scales (Loreal = canthal scales in Hallermann [2009]), number of scales between nasal scale and granular scales of anterior border of orbit; postmental scales (Postm), number of throat scales (Throat), number of scales between end of jaws; third finger lamellae (3ForefLm), number of subdigital lamellae beneath third finger from first scale/lamella at digits' insertion point to claw; fourth toe lamellae (4HindfLm), number of subdigital lamellae beneath fourth toe from first scale/lamella at digits' insertion point to claw; midbody scales (Midb), number of scales around midbody; dorsal scales 1 (Dorsal1) number of dorsal scale rows with keels oriented parallel to the keels of the dorsal crest scales); dorsal scales 2 (Dorsal2), number of dorsal scale rows with keels oriented diagonally downward; nuchal spines (NuchalS), number of enlarged nuchal crest scales at least twice as large as enlarged vertebral scales; dorsal crest (DorsCrst), presence or absence; nuchal crest scales' shape (NucCrstSS), lanceolate, cresent-shaped , or triangular.
Aspects of body shape were examined by converting paired traits to proportions, such as TailL/SVL, ForefL/ForelL, ForelL/HindlL, HeadW/HeadL. Measurements, proportions, and scale counts were analyzed by basic parametric statistics, Systat 12. We used a simple Student's t test to examine dimorphism between adult females and males, significance α ≤0.05. Statistical analysis used SYSTAT 12.
Specimen number abbreviations are as follows: CAS (California Academy of Sciences), EHT (Edward H. Taylor), LSUHC (La Sierra University Herpetology Collection), USNM (United States National Museum, National Museum of Natural History), TNHC (Texas Natural History Collection, Univ. of Texas, Austin), ZMKU (Zoological Museum of Kasetsrat University, Bangkok, Thailand) and ZSI (Zoological Society of India).

Results
We obtained ND2 sequence data from eight individuals ranging in size from 1,260-1,369 bp that were 98.5-99.8% identical to one another (un-corrected sequence divergence). Our samples have a truncated origin of light strand replication (21 bp) between tRNA ASN and tRNA CYS that appears functional (folds using the DNA-Matthews 1999 Energy Model in Geneious) and tRNA CYS shows a d-arm replacement loop (Macey et al. 1997) consistent with 45% of Draconinae agamids (Seligmann and Labra 2014). Our samples formed a well-supported clade (100%) with each other and were placed sister to Bronchocela rayaensis with 100% bootstrap support (Fig. 2). Our samples ranged from 93.4-95.0% identical (un-corrected sequence divergence) to the B. rayaensis specimens in GenBank (KR053115-KR053116, KY498356). Our analyses recovered the Phuket Island B. rayaensis sister to the Pulau Langkawi samples with 94% bootstrap support (Fig. 2). The rest of our tree was nearly identical to that recovered by Grismer et al. (2015) with the exception of the inclusion of B. cristatella (AF128495) and Aphaniotis fusca (AF128497); the latter was placed sister to Aphaniotis fusca (AF288228).
We attempted to code a few coloration traits but were unsuccessful because preservation had altered life-colors and pattern. The manner of field work and genomic tissue sampling did not permit the recording of coloration in living specimens. The general impression of the Tanintharyi Bronchocela is green with transverse bands of white spots on trunk and alternating bands of green and white on the tail (Fig. 3). Preserved specimens have a broadly mottled pattern of dark rufous brown and dusky turquoise dorsally and laterally. These colors form broad swatches with no consistence of position on the body although the majority (eight of nine) of the specimens have the snout and interorbital (not the superciliary area) dark. The middle third of the tail (dorsally) shows a vague banding, broad bands of medium brown and dusky tan. The venter from the tip of the chin onto the tail base is light turquoise with the chin and throat lighter that chest and abdomen. The temporal area (between orbit and tympanum) is bright rufous in eight of nine specimens although the size of this rufous patch varies from a small spot to the entire area. The eye sheath is dark, the tympanum light. The loreal area also is rufous and size of rufous patch varies from a small patch to the entire area. The ventrolateral neck folds range from uniform dark rufous to uniform turquoise. In two instances of the latter, the posterior edge of the fold bears small, illdefined rufous spots.

Discussion
The molecular (ND2) results support our Tanintharyi specimens as a discrete clade, 5-6.6% different from the closest species Bronchocela rayaensis available. Our samples were placed sister to B. rayaensis, a species known from Pulau Langkawi (island) off the west coast of Peninsular Malaysia and Phuket Island, Thailand (Fig. 2). Our phylogeny recovered the Phuket Island B. rayaensis sister to the Pulau Langkwai samples with 94%  Grismer et al. 2015) from Polillo Island, Philippines as B. marmorata, but did not update the taxonomy of those sequences in GenBank. Here, we included all samples from the Philippines from Grismer et al. (2015), the two from Polillo Island and two from Luzon Island, Philippines [KR053111-KR053112, CDS2105 (KU305472) and RMB9878 (KU315805), respectively]. These samples form a well-supported clade and are 94.4-100% identical to one another for the ND2 sequences. Because there may be more than one lineage of Bronchocela in the Philippines (R. Brown, pers. com.), we refer to these specimens as B. cf. marmorata and they are sister to a clade containing B. shenlong from Peninsular Malaysia and B. cristatella from Malaysia, Borneo, and Indonesia (Pulau Natuna Besar). Though the sister relationship between B. shenlong and B. cristatella received 100% support (Fig. 2), the relationship between them and B. cf. marmorata received much less support, both in our study (51%, Fig. 2) and in that of Grismer et al. (2015, 72% ML bootstrap support); though with fewer individuals analyzed, this relationship received 100% support (Grismer et al. 2016). We included all specimen in GenBank from Grismer et al. (2015), including outgroups, the B. rayaensis from Phuket Island (Grismer et al. 2016), and an additional Bronchocela cristatella sequence in GenBank (AF128495) and Aphaniotis fusca (AF128497). These specimens were reported as being switched (Zug et al. 2006), which our study confirms. The sequence AF128495 is 99% identical to B. cristatella from Selangor, Malaysia (KR053100-KR053102) and was placed in the B. cristatella clade sister to these specimens from the same state (Selangor) as TNHC57943, the voucher of AF128495 (Fig. 2). The identifications of these sequences were recently corrected in GenBank. The Fig. 5 of Macey et al. (2000b) appears to display the correct tRNA CYS for these genera; however, our ARWEN model depicts the lower three base-pairs non-binding, extending the d-arm replacement loop to "AAAGTG." The tRNAs of Grismer et al. (2015) appear to be improperly annotated because virtually all tRNAs are annotated as transcribing in the light strand direction, whereas tRNA ALA , tRNA ASN , tRNA CYS , and tRNA TYR transcribe on the heavy strand in all other agamids examined (e.g. Macey et al. 1997), their sequences have "tRNA ASX " (the transcribed product) in place of tRNA ASN , and none of the origin of light strand (OL) sequences are identified. Secondary structure inspection of one of their specimens (KR053067) in ARWEN reveals the tRNA ASN does exists, it is transcribed in the complimentary direction, and their tRNA ALA as annotated does not form a tRNA, but the bases between tRNA TRP and tRNA ASN do form tRNA ALA . Agamids, in particular, show interesting variation in the relationship between the condition of the OL and the neighboring tRNA's d-arm replacement loop, which can be useful for the study of evolution of mtDNA replication (Seligmann and Labra 2014) and can be useful as phylogenetic information content (Macey et al. 2000b) if properly annotated.
The morphology of Bronchocela burmana and B. rayaensis are similar; however, the differences indicate adaptive divergence between the southern Burmese and Langkawi Malaysian populations. The Burmese lizards show sexual dimorphism with males larger than females with only a slight overlap. The Langkawi population seems to reverse this dimorphism with females larger than males (Table 2) although that sample includes only one adult of each sex.
The newly discovered Phuket Bronchocela rayaensis is a gravid female yet smaller than the paratype adult female ( Table 2). As would be expected, its absolute dimensions are smaller than the typic female and surprisingly also smaller than the typic male. Also several of its body proportions (HeadH/HeadL, TympD/OrbD, ForelL/ SVL, HindlL/SVL) are smaller than those of the types. Because the Phuket sample is a single individual, we note these differences, but also note that in B. burmana adult size remains larger, midbody and nuchal crest scales remain fewer, thereby supporting the Phuket individual as B. rayaensis.
Bronchocela rayaensis possess more nuchal crest scales than B. burmana-with a slight overlap 6-9 vs. 9-13. Third and fourth fingers are nearly equal in B. rayaensis, third finger is about half a claw length longer than fourth finger, and midbody scales 71 in holotype of B. rayaensis vs. 59 (55-67) scales around midbody in our Tanintharyi sample At this time, we cannot define the distribution of B. burmana in Thailand. The distribution map in Chan-ard et al.'s field guide (2015: p. 94) shows a Thai occurrence of B. cristatella in the Tenasserim range from Khanchanburi Province southward into southern Peninsular Thailand to the Malaysia border. Their morphological description appears to mix characteristics of B. burmana and B. cristatella. We assume that this characterization demonstrates that the distribution of their Thai Bronchocela specimens was a mix of B. burmana and B. cristatella. The Phuket and Khura Bun specimens (Grismer et al. 2016) confirm the presence of B. rayaensis in Thailand, at least south of the Isthmus of Kra. In an earlier review of Thai lizards, Taylor (1963) considered cristatella as a Calotes. His description was based on a specimen from Yala Province and his illustration on an individual from Sangkhla. Both of these specimens are clearly B. cristatella, demonstrating that B. cristatella occurs in southern-most peninsular Thailand. Chan-ard et al.'s (1999)  Bronchocela rayaensis has been recently characterized by Grismer et al. 2015 and 2016 and is not repeated here. The following is a redescription of B. burmana based exclusively on our recently vouchered specimens from southern Tanintharyi.
Holotype. Adult ZSI 5337 collected from '...near Tavoy'. We have not examined this specimen as it was not available to us. Definition. A Bronchocela lizard with a short nuchal crest of six to nine erect triangular crest scales; no middorsal crest of raised scales on trunk. Snout-vent length of adults range from 80 to 94 mm with tail length 240 to 360% of snout-vent length; limbs slender, forelimbs 42-52% of SVL, hindlimbs 86-97% of SVL; digits long and slender with third finger slightly longer than fourth finger, fourth toe distinctly longer that third toe; head medium sized (25-27% of SVL); head with distinct canthal ridge, narrow triangular shaped from dorsal view, length > width ≈ height and approximately 26 % of SVL; moderately large eye (OrbD/HeadL ~26-28%) and about twice diameter of tympanum (continuous with temporal surface).
General description. Detailed metric and scalation features are presented above in the Results section, also Table 2. Bronchocela burmana is a slender green lizard with long tail, usually 2.5-3.5X snout-vent length. In spite of its 80 to 94 mm body length, its slenderness and thin legs give it a delicate appearance and make it immediately recognizable among the other lizards of southern Tanintharyi.
In life, Bronchocela burmana appears uniformly green (Fig. 2). Preservation changed the overall coloration to light olive but highlights a light rufous vertical bar in the temporal area.
Distribution. Bronchocela burmana is presently confirmed for only southern Tanintharyi Division, Myanmar (Fig. 1). The type-locality is given as Tavoy, now Dawei; however, two years of surveys in both the dry and wet seasons at the Tanintharyi Nature Reserve, just north of Dawei, did not confirm the presence of this species in the former Tavoy area (approximately 350 km N of Lenya). The morphological description of B. cristatella in Chan-ard et al.'s field guide (2015: p. 94) appears to be a composite of B. burmana and B. cristatella. As noted in the preceding section, we interpret their distribution map also to be a composite of the two species' occurrence in western Thailand. We suggest that the two species are allopatric with B. burmana in the north, likely north of the Isthmus of Kra and B. cristatella in the south. Taylor's description and photograph of B. cristatella (1963) are definitely of that species and are plotted ( Fig. 1) to confirm its occurrence in southern-most Thailand.
Etymology. Blanford did not explain his choice of burmana. His selection seems obvious as it was the first Bronchocela from British Burma and distinctly different from B. cristatella and the other species recognized at that time.