Integrative Taxonomic Approach for Describing a New Cryptic Species of Bush Frog (Raorchestes: Anura: Rhacophoridae) from the Western Ghats, India

A new cryptic species of bush frog Raorchestes honnametti sp. nov. is described from the south-eastern part of the Western Ghats, India. This newly described species belongs to the Charius clade and is morphologically similar to other clade members—R. charius and R. griet. Therefore, an integrative taxonomic approach based on molecular and bioacoustic analysis along with morphology was used to delimit the new species. Raorchestes honnametti sp. nov., is currently known only from Biligiri Rangaswamy Temple Tiger Reserve, a part of Biligiri Rangaswamy horst mountain range (a mountain formed due movement of two faults) formed during the Late Quaternary period (1.8–2.58 Ma). Discovery of cryptic species from a highly speciose and well-studied genus Raorchestes hints at the possible existence of several more cryptic species in this genus. We discuss the possible reasons for crypsis and emphasize the need for continued systematic surveys of amphibians across the Western Ghats.


Introduction
Cryptic species are morphologically similar and difficult to distinguish from their closest congeners [1]. Studies have shown that cryptic species are common and many of them go undetected when studied solely on the basis of morphological evidence. In recent years, molecular genetic techniques and bioacoustics have helped in the discovery of cryptic anurans [2,3]. Identifying cryptic species is of paramount importance not only for biodiversity conservation but also to understand processes of evolution and biogeography [1,4].
The Western Ghats of India is known for high diversity of amphibians with 220 species recorded. Along with high diversity, many studies have also suggested presence of high number of cryptic species in the Western Ghats. Using molecular techniques, Nair et al [5] carried out a study on endemic genus Indirana, and demonstrated high cryptic diversity among them. Similarly, two new genera in the family Rhacophoridae were discovered, in a cryptic genus Polypedates [6]. Using bioacoustics and molecular techniques, cryptic species were discovered from widely distributed genera like Euphlyctis (2 species, [7]) and Fejervarya (4 species, [8]).
The genus Raorchestes belongs to family Rhacophoridae. It includes bush frogs with adult size ranging from 10 mm to 50.5 mm. They are distinguished by the presence of transparent/ translucent vocal sac and the absence of vomerine teeth. Adults are nocturnal and lack free swimming larvae [9]. There are 59 valid species in the genus Raorchestes [10], of which 50 species are from the Western Ghats, 4 from Eastern Himalaya, 1 from Eastern Ghats and 4 species from China, Myanmar and Vietnam. Bee et al [11] indicated the possible existence of several undescribed cryptic species in the genus Raorchestes. However, till date, there is no published literature about cryptic species from this genus. Here we describe a new cryptic bush frog species from south eastern region of the Western Ghats. The description is based on an integrative taxonomic approach involving morphology, molecular techniques and bioacoustics. We discuss the possible reason for cryptic speciation and the need for more detailed studies on cryptic species of Western Ghats.

Ethics statement
Fieldwork and sampling was carried out in the Biligiri Rangaswamy Temple Tiger Reserve (BRTTR) of Karnataka, with due permission from the Director, BRTTR dated 10 October 2012. Tissue sampling was carried out under the strict supervision of local foresters and used solely for scientific research. Minimum samples were collected (six individuals) as required for the study and the sampling do not in any way affect effect the population of this species. Field Permit issued by Principal Chief Conservator of Forests and Chief Wildlife Warden, Karnataka State Forest Department.

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:pub:93347FEF-6BF8-40A2-9EEB-2F2C743E70FC. 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.

Study area
The study was carried out in BRTTR (Biligiri Rangaswamy hill range; latitude: 11°-12.15°N, longitude: 77°-77.2667°E, altitude: 600-1800 m amsl). Specimens were collected from shola forests (shola forests are a type of forest in the valleys of a mountain separated by montane grasslands) in Honnametti (11.8987°N, 77.1741°E, 1659 m amsl) and Dodda Sampige (11.9473°N, 77.1836°E, 1142 m amsl) within BRTTR (Figs 1 and 2). The reserve got its name "Biligiri" from the white cliff on which the temple of Lord Rangaswamy is situated. This tiger reserve is located to the east of the Western Ghats in Karnataka State. The total area of the tiger reserve is 540 km 2 . The landscape is very heterogeneous and undulating. The annual rainfall, received mainly during the southwest monsoon, ranges from 600 mm at the base to 3000 mm at the hilltop. The variation in temperature ranges from 8°C to 25°C. The wide range of climatic conditions as well as altitudinal variations within the small area has resulted in varied forest types like scrub forest, dry deciduous, moist deciduous, riparian, evergreen, sholas and  grasslands [12]. The extent of the major forest types represented as percent of total area is as follows: deciduous (moist and dry) = 61.1%, scrub = 28.2%, grassland = 3.4%, evergreen = 6.5% and high altitude sholas = 0.8% [13].

Specimen collection
Six adult calling males were collected in October 2012. Specimens collected were euthanized and fixed in 5% formalin for 24h and preserved in 70% alcohol. Thigh muscle tissue samples for genetic analysis were collected before fixing the specimen. Color of live specimens and natural history notes were recorded at the type locality during multiple visits. Specimens were deposited in the Bombay Natural History Society (BNHS) museum, Mumbai.

Genetic analysis
For DNA extraction, we followed the method described in Vences et al [14]. A small amount of thigh muscle tissue was excised from the holotype for extraction. The PCR amplification and sequencing of 16S rRNA and ND1 genes (subunit of NADH dehydrogenase) were done following Palumbi et al [15] and Bossuyt et al [16] respectively. The amplified products were sent to Amnion Sequencing services, Bangalore, India. The sequences were checked manually using program Chromas lite 2.01 (http://www.technelysium.com.au/chromas_lite.html). The sequences were aligned using MAFFT algorithm [17] and manually corrected in MEGA 5.10 software [18] and deposited in GenBank (Accession numbers, KT151650-KT151653). The final dataset consisted of 1032 base pairs. 16S and ND1 genes of 49 valid species of the genus Raorchestes endemic to the Western Ghats [19,20,21] were used in phylogenetic analysis.
GenBank accession numbers for 16S rRNA and ND1 genes of 49 valid species used in the molecular analysis are given in S1 and S2 Tables, respectively. The phylogenetic analyses were performed using Maximum likelihood (ML) algorithm and Bayesian inference methods. The ML analysis was executed in raxmlGUI v1.3 [22] with GTR+G model selected as the best-fit nucleotide substitution model in jModel test [23] for 1000 bootstrap replicates. The Bayesian analysis was performed in MrBayes 3.2.4 [24]. The combined data set of 16S and ND1 partitioned as separate gene fragments were used for the analysis with TIM2+I+G and TPM3uf+I +G selected as best fit models for 16S and ND1 genes respectively in jModel test. The Markov chain Monte Carlo analysis for the dataset was run for 50 million generations and trees were sampled every 500 cycles. The convergence of the runs was analyzed by assessing the split frequency standard deviations (<0.001) and potential scale reduction factor (PSRF~1.0). The first 10% of the sampled trees were discarded as burn-in and remaining samples were used to generate majority rule consensus tree. For estimating the genetic divergence, uncorrected pairwise genetic distance between the species was calculated in MEGA 5.10.

Morphology
The new species was compared with all valid species of Raorchestes both from published literature and through examination of type specimens deposited in India (S3 Table). After the phylogenetic analysis, initial morphological comparisons were made with the members of the Charius clade. Raorchestes thodai [25] from Western Ghats lacked genetic data and was morphologically compared with the new species. Similarly, other non-Western Ghats species of Raorchestes, namely, R. terebrans from Eastern Ghats [26], R. annandalii [27], R. manipurensis [28], R. sahai [29] and R. shillongensis [30] from Eastern Himalayas; R. gryllus [31], R. longchuanensis [32,33,34], R. menglaensis [32,35] and R. parvulus [36] from Myanmar, Vietnam and South-China were also compared morphologically with the new species. Morphological comparisons were made with synonymised species based on their original description, namely, R. emeraldi [21], R. montanus [37,38,39] and R. neelanethrus [40] to verify the availability of those names to the new species. Morphological measurements were taken using a Mitutoyo 1 digital slide calliper to the nearest 0.1 mm. Measurement and terminology follows Biju and Bossuyt [20] and Gururaja et al [41]. Abbreviations used are as follows: snout-vent length (SVL); head width, at the angle of the jaws (HW); head length, from the rear of the mandible to the tip of the snout (HL); inter upper eyelid width, i.e. the shortest distance between the upper eyelids (IUE); maximum upper eyelid width (UEW); snout length, measured from the tip of the snout to the anterior orbital border of the eye (SL); eye length, i.e. the horizontal distance between the bony orbital borders of the eye (EL); internarial distance, i.e. least distance between the inner margins of nares (IN); nostril-snout distance, i.e. distance between middle of nostril and tip of snout (NS); eye to nostril distance, i.e. distance between anterior-most point of eye and middle of nostril (EN); tympanum-eye distance, i.e. anterior rim of tympanum to posterior of eye (TYE); distance from the rear of the mandible to the nostril (MN); distance from the rear of the mandible to the anterior orbital border of the eye (MFE); distance from the rear of the mandible to the posterior orbital border of the eye (MBE); distance between anterior corner of eyes, i.e. the shortest distance between the anterior orbital borders of the eyes (IFE); distance between posterior corner of eyes, i.e. the shortest distance between the posterior orbital borders of the eyes (IBE); largest tympanum diameter (TYD); forelimb length, measured from the elbow to the base of the outer palmar tubercle (FLL); hand length, measured from the base of the outer palmar tubercle to the tip of the third finger (HAL); thigh length (TL); shank length (ShL); foot length, measured from the base of the inner metatarsal tubercle to the tip of the fourth toe (FOL); distance from the heel to the tip of the fourth toe (TFOL); disc width on  fingers I, II, III and IV (FD I, II, III and IV); width of finger I, II, III and IV measured at the  base of the disc (FW I, II, III and IV); lengths of fingers I, II, III and IV measured from base of  proximal subarticular tubercle to fingertip (FL I, II, III and IV); tibia width, i.e. width of tibia at its widest region (TW); disc width on toes I, II, III, IV and V (TD I, II, III, IV and V); width of  toes I, II, III, IV and V (ToW I, II, III, IV and V) measured at the base of disc; length of toes I, II, III, IV and V measured from base of proximal subarticular tubercle to tip of toe (ToL I, II, III, IV and V); length of inner metatarsal tubercle (IMT); distance from distal edge of metatarsal tubercle to maximum incurvature of web between fourth and fifth toe (MTFF); distance from distal edge of metatarsal tubercle to maximum incurvature of web between third and fourth toe (MTTF); distance from maximum incurvature of web between fourth and fifth toe to tip of fourth toe (FFTF); distance from maximum incurvature of web between third and fourth toe to tip of fourth toe (TFTF). Intercalary ossification, which is the cartilaginous structure between distal and penultimate phalanges in fingers and toes was observed. The presence of an intercalary ossification can be noticed without anatomical sectioning as a glandular projection between phalanges [42]. We carried out linear discriminant analysis (LDA) using R v. 3.1.3 [43] and R Studio v. 0.98.1102 (http://www.rstudio.com/) on morphological data. The data was converted as ratio of SVL and tested for normality. Eight morphological characters viz., HW, HL, UEW, SL, HAL, TL, ShL and FOL were subjected to LDA to get a linear combination of these characters to discriminate R. charius, R. griet and R. honnametti. To identify multivariate differences among species MANOVA was performed and Tukey HSD post-hoc test was used for pairwise comparison of characters.

Advertisement call recording and analysis
Call recordings were made using Olympus stereo digital voice recorder (LS-11) for R. charius and R. honnametti. Calls of R. griet were recorded using Canon EOS 600D. Calls with less background noise were manually selected from different individual call records and were analysed using Audacity Ver.1.3 (Beta) and Raven Pro 1.5. Nineteen calls from 3 individuals of R. honnametti; 20 calls from 4 individuals of R. charius and 8 calls from 1 individual of R. griet were selected for further analysis. Duration, inter-call interval duration, dominant frequency and number of pulses of each call was recorded. Call terminology was based on Kok and Kalamandeen [44]. Air temperature and relative humidity were recorded using TFA digital Thermo-Hygrometer. We used Welch t-test in R v.3.1.3 to compare the means of call characteristics between two species, assuming normal distribution of call samples with unequal variance and sample size.

Genetic analysis
The molecular data analysis suggested that Raorchestes honnametti sp. nov. belonged to the 'Charius clade' [21] comprising of R. charius and R. griet The uncorrected pairwise genetic distance (Mean ± SD) between R. honnametti and R. charius is 4.14 ± 0.14% and between R. honnametti and R. griet is 6.47 ± 0.17%. Fig 3 shows the result of ML tree with a strong bootstrap support (100%) for the R. honnametti and R. charius. Hence, the morphological comparisons are made only with the closest relatives of the species, i.e. R. charius and R. griet. The overall genetic divergence of 50 species, including R. honnametti for the combined dataset of 16S rRNA and ND1 genes ranges from 2 to 32% (S4 Table).  Diagnosis: Raorchestes honnametti belongs to the genus Raorchestes as they are relatively small sized frogs (15-45 mm), active in night, vomerine teeth absent, transparent/translucent vocal sac while calling and direct development without free swimming tadpoles. It is a small sized adult (male: 21.7-24.8 mm, n = 6); snout longer than the horizontal diameter of eye; groin uniform light brown with 3-4 yellow blotches; both anterior and posterior part of thigh uniform light brown with small round to oval shaped yellow blotches and relatively short hind limbs ShL/SVL ratio <0.5. It belongs to the Charius clade and morphologically similar to R. charius and R. griet.
Description of holotype: A small sized frog (SVL = 24, Fig 4A-4H, Table 1 New Cryptic Bush Frog from the Western Ghats chest, belly, lower flanks and ventral sides of forelimbs and thighs granular; chest with larger granules; no dorso lateral fold, Colour in life: Dorsum light grey with two dark grey concave stripes behind eyes to groin; light brown stripe between eyes demarcates triangular light grey snout, limbs barred, brown band from snout passing through loreal, mid iris and tympanic region till end of supratympanic fold. Iris golden above and light golden below, horizontally separated by brown band; dorsal side of forelimbs and hind limbs with dark grey cross bands. A triangular white spot on the tip of the snout. Venter uniform cream white, vocal sac yellow translucent with granular grey spots. Webbing cream white. Groin uniform light brown with 3-4 yellow blotches, both anterior and posterior part of thigh uniform light brown with small round to oval shaped yellow blotches.
Colour in preservative: Dorsum light grey, dorsal side of forelimbs and hind limbs with faint cross bands. Venter uniform cream white, vocal sac with granular grey spots. Webbing cream white. Groin uniform pale brown with 3-4 cream white blotches, both anterior and posterior part of thigh uniform pale brown with small round to oval shaped cream white blotches.
Etymology: Named after the locality of holotype-Honnametti. Honnametti is treated as an invariable noun in apposition to the generic name.
Variation: Details of the morphometric variations observed in 6 individuals is provided in Table 1. The dorsum colouration varied from light grey to dark brown. A triangular white spot on the tip of the snout are seen in couple of individuals. Flanks light grey to dark brown, yellow blotches in the groin varies both in numbers (2 to 5) as well as in size. Variations included distinct cross bands on both forelimbs and hind limbs.
Comparison: Molecular data suggested that Raorchestes honnametti belonged to the Charius clade within the Western Ghat endemic species of Raorchestes and hence initial morphological comparisons were made only with R. charius and R. griet. Raorchestes charius differs from R.   There is an overlap between R. honnametti and R. charius. Raorchestes griet, however, forms a distinct cluster. Factor loading for axis 1 and 2 are given in S5 Table. MANOVA did not result in significant differences between the three species (F 2,16 = 1.61, p = 0.17). Tukey's HSD post-hoc test also did not exhibit significant differences in pairwise comparison of characters except for SL in R. griet and R. honnametti (Q = 4.73, p = 0.012); TL in R. griet and R. charius (Q = 3.83, p = 0.041) and TL in R. honnametti and R. griet (Q = 5.03, p = 0.008). Morphological comparisons of R. honnametti with R. thodai and with non-Western Ghats species of Raorchestes are given in S6  Tables 2 and 3). Phonetically, the advertisement call sounds like 'treenk. . ...treenk. . ...treenk. . .', however they vary in number of pulses, call duration, inter-call interval duration and dominant frequency. Table 2 provides detail on call characteristics of R. honnametti, R. charius and R. griet. Calls of R. honnametti had 6-9 pulses (Mean ± SE, 6.4 ± 0.18) in each call. Average dominant frequency was 2635.9 ± 11.75 Hz, Call duration was 0.1 ± 0.003 s, inter-call interval duration of a lone individual was 6.3 ± 0.31 s and in the presence of multiple calling individuals, inter-call interval duration was 2.45 ± 0.02 s. Calls of R. charius had 13-18 pulses in each call. Average dominant frequency was 2478.57 ± 12.56 Hz; Call duration was 0.09 ± 0.004 s; inter call duration was 2.51 ± 0.09 s in the presence of multiple calling individuals. Calls of R. griet had 13-15 pulses in each call. Average dominant frequency was 3699.88 ± 9.51 Hz, Call duration was 0.09 ± 0.0007 s, and inter-call interval duration was 5.22 ± 0.2 s in the presence of multiple calling individuals. Welch t-test showed significant differences in dominant frequency, call duration and number of pulses of all the three species (Table 3).
Natural history: Raorchestes honnametti sp. nov. is known only from Biligiri Rangaswamy hills and is one of the very common frogs in that landscape. It is found in shola forests, evergreen forests, semi-evergreen forests and around human habitations. Individuals were found calling in an open area within Ageratina adenophora (Asteraceae) and Strobilanthus bushes. Some individuals were also found on tree saplings in the understory. Individuals call at a perched height between 0.48-1.00 m from ground. Call starts at around 6 pm and goes till early morning. During monsoon (June to September), individuals call almost throughout the day except on days with heavy rains or dry days. Other anuran species like Pseudophilautus sp., Hylarana sp., Fejervarya sp., Duttaphrynus melanostictus, Microhyla sholigari, M. ornata, M. rubra and Euphlyctis cyanophlyctis co-occur with R. honnametti in Biligiri Rangaswamy hills.

Conservation
There are no immediate threats from human activities to this newly described species as Biligiri Rangaswamy hills is a tiger reserve and enjoys high level of protection. However, in the last one decade, a significant area of the Reserve has been taken over by highly invasive species like Lantana camara (Verbenaceae) and Ageratina adenophora (Asteraceae). On subsequent visits, we have seen several calling males of R. honnametti on Lantana and Ageratina bushes, indicating that this species might have adapted to the presence of these invasive species. However, a systematic research needs to be undertaken to assess the impact of invasive species on R. honnametti.

Discussion
The bush frogs of the genus Raorchestes are distributed across India, southern China, Myanmar and Vietnam. India harbours 55 of these species, with 50 species distributed in the Western Ghats. With the addition of R. honnametti the total number of species in genus Raorchestes stands at 60. Discovery pattern shows that 42 species of Raorchestes were described in last 15 years (Fig 9), indicating that there may be many more species to be described from the Western Ghats and needs a continued systematic surveys of amphibians across the region. Such patterns  in discovery further strengthen our expectations that not only morphologically distinct species, but also many cryptic species are waiting to be discovered from the Western Ghats. Multivariate analysis of morphological data using LDA showed an overlap between R. charius and R. honnametti individuals while R. griet formed a distinct cluster. MANOVA and Tukey's HSD post-hoc tests on morphological data did not show significant differences between R. charius, R. griet and R. honnametti, except for SL and TL. This validates our observations on morphological similarities between the three species and hence morphological characters alone cannot distinguish cryptic species. Tools like bioacoustics and molecular genetic technique can be a better aid in describing a cryptic species through integrated taxonomic approach. Studies have shown that a bioacoustic character like call dominant frequency has a highly significant negative relationship with body size in anurans [45]. A similar significant negative relationship was exhibited among the Charius clade members between dominant frequency (DF) and body size (SVL) (Pearson's correlation coefficient, r = -0.74, p = 0.0004). External factors like temperature, relative humidity and altitude influence advertisement call characteristics [46]. Due to lack of data, we did not analyze the effect of external factors on advertisement call characteristics of Charius clade members. Biju and Bossuyt [20] through their study on genus Raorchestes showed that R. charius, R. griet and R. coonoorensis formed a distinct clade. Recently, Vijayakumar et al [21] revised the Charius clade which involved only R. charius and R. griet, with R. coonoorensis forming a separate clade basal to the Charius clade. From our study, the Charius clade now includes R. charius, R. griet and R. honnametti with R. charius and R. honnametti forming sister taxa, while R. griet remains basal to the other two. The distribution range of these three species is restricted to separate regions of Western Ghats. According to the present distribution (Fig 1), R. charius is found in Bababudangiri to Brahmagiri hill ranges of central Western Ghats. It is 169 km north-west from R. honnametti and 251 km north from R. griet distribution. R. griet is found in Munnar and Valparai regions of southern Western Ghats below Palghat Gap. It is 169 km south west from R. honnametti, whereas R. honnametti (from this study) is found only in Biligiri Rangaswamy hills, south east of the Western Ghats. Absence of R. charius and R. griet in Biligiri Rangaswamy hills indicates that each of the three species has a smaller non-overlapping geographical range.
Studies understanding cryptic speciation signify the need to understand role of historical and biogeographical processes. The phenomenon of morphological stasis and genetic divergence in cryptic species could be because of recent speciation events owed to climatic oscillations and geographic barriers [47]. Biju and Bossuyt [20], stated that species of genus Raorchestes showed a high level of local endemism and emphasized the role of isolated hilly regions as a major reason for restricted species distribution. Studies understanding the divergence in frog lineages indicate that not just mountains but ecology, climatic conditions, forest types and rivers can act as barriers [48,49,50]. Raorchestes honnametti is a nocturnal, arboreal and a direct developing frog. Though detailed ecological study on R. honnametti is warranted, above mentioned ecological factors could have acted as barriers resulting in speciation of R. honnametti. The geological studies on Biligiri Rangaswamy hills suggest that it is a horst mountain formed due to the movement of Kollegal fault running N-NE to S-SW direction (Fig 10) and might have formed recently during late Quaternary period (1.80 to 2.58 Ma) along the eastern borders of Mysore Plateau [51]. The underlying rocks date back to 3.6 Ga of Dharwad Craton. This uplift of Biligiri Rangaswamy hills-Male Mahadeshwara hills and subsequent isolation of Biligiri Rangaswamy hills from Niligiris due to Moyar Gorge could be one of the reasons for diversification of R. honnametti from R. charius and R. griet.
The restricted distribution of R. honnametti could also be due to the present day disjunct distribution of evergreen forests in Biligiri Rangaswamy hills. Studies have shown that such forests of India got restricted due to variation in climatic conditions (Miocene and Quaternary period) and anthropogenic factors resulting in a disjunct distribution of species [52]. Molecular dating can further help in understanding the diversification in R. honnametti as well its other clade members. Nevertheless, the discovery of R. honnametti opens up interesting evolutionary questions on cryptic species and studies understanding ecological and evolutionary mechanisms can help in appreciating the conserved morphology of such cryptic species.
Supporting Information S1