Genetic evidence on the occurrence of Channa harcourtbutleri (Annandale, 1918) in Eastern Ghats, India: first report from mainland India

: Channa harcourtbutleri (Annandale) was described from Inle Lake (Southern Shan State) in Myanmar, and is currently considered as a valid species in the Channa gachua species-group. Notwithstanding several detailed studies on Channa from India in the recent, none has mentioned the occurrence of C . harcourtbutleri in the Indian mainland. In continuation to the faunal diversity exploration in Eastern Ghats, India, a few specimens in the C. gachua species-group were collected from the river Sabri sub-basin of the river Godavari basin in the East Godavari District of Andhra Pradesh which was identified as C. harcourtbutleri through DNA barcoding. This is a first report on occurrence of the species in the wild in the Eastern Ghats, India.


INTRODUCTION
Snakehead fishes belong to the family Channidae, are most popular in ornamental fish trade. The species in the genus Channa Scopoli, 1777 are currently distributed in southern, eastern, and southeastern Asia while their allied species in the genus Parachanna Teugels & Daget, 1984 are endemic in Africa. Snakehead fishes are broadly categorized into two major groups based on the presence of gular scales. The first group having gular scales comprises of all the African species of the genus Parachanna and four Asian species of the genus Channa while the second group lacking gular scales comprises of all the rest of the channid species (Musikasinthorn & Taki 2001;Zhang et al. 2002). Further, the Channa gachua species group (sensu Britz 2008) was characterized by a varying number of dark and light semi-circular bands on the pectoral-fin, which has been subsequently followed by other ichthyologists Praveenraj et al. 2019). The presence of gular scales is regarded as a plesiomorphic state within the Channidae (Li et al. 2006). As mentioned in Li et al. (2006), the African taxa of Parachanna differ from the Asian taxa of Channa by the absence of supporting 'lamellas' or process of the first epibranchial and hyomandibular in the suprabranchial organs (Senna 1924;Bonou & Teugels 1985). Further, the study has shown that the species lacking pelvic fins clade with the species having the pelvic fins, and has explained that the loss of the pelvic fins occurred at least three times independently during the evolution of the taxa in Channidae (Bonou & Teugels 1985). The gular region with scales or without scale is one of the key characters in channid taxonomy (Talwar & Jhingran 1991;Li et al. 2006), and the shape of isthmus was also shown to be a key feature in differentiating marulius and gachua speciesgroups (Vishwanath & Geetakumari 2009). Among the currently reported 24 channid species in India, 18 species (15 with pelvic fins and 3 without pelvic fins) are included in the C. gachua species-group (sensu Britz 2008).  mentioned that snakehead fishes have a centre of diversity in the eastern part of the Himalaya Biodiversity Hotspot (Conte-Grand et al. 2017;Rüber et al. 2019). In the recent two decades, quite a good number of new species, mostly in the C. gachua species-group, have been described. However, a few of the recent descriptions have been retained in synonymy by Britz et al. (2019).
Channa harcourtbutleri (Annandale) was described from Inle Lake (southern Shan State) in Myanmar (Annandale, 1918), but it was placed in synonymy with C. gachua (Hamilton) by Hora & Mukerji (1934). However, since the latter species is younger than the former, this synonymy at its first instance appears incorrect. Ng et al. (1999) resurrected the species and discussed the differences between them. The taxonomy of C. gachua has been a complex problem (Ng et al. 1999), but the recent phylogenetic study suggested two distinct lineages within the C. gachua species-complex (Conte-Grand et al. 2017). The true C. gachua as referred in Conte-Grand et al. (2017) is restricted to the area west of the Indo-Burman ranges (i.e., Rakhine Yoma and Chin Hills) and covers Sri Lanka, India, Nepal, Bangladesh, and the Rakhine area of Myanmar. The taxon previously recorded as C. gachua from Sri Lanka has been revalidated as C. kelaartii (Gunther), and has its population also distributed in southern peninsular India (Sudasinghe et al. 2020). The eastern lineage of the C. gachua species-complex, nominally referred to as C. limbata, is distributed to the east of the Indo-Burman ranges from Myanmar reaching east to Vietnam and southern China and south to Indonesia and Malaysia (Conte-Grand et al. 2017).
Notwithstanding several detailed studies on channid taxa from India recently, none has mentioned the occurrence of C. harcourtbutleri in mainland India (Conte-grand et al. 2017;Britz et al. 2019;Sudasinghe et al. 2020).
In continuation to the faunal diversity exploration in Eastern Ghats, India, several specimens of the genus Channa were collected from various localities. Among the examined specimens, a few specimens collected from the river Sabri sub-basin of the river Godavari basin in the East Godavari District of Andhra Pradesh, were morphologically identified as similar to C. harcourtbutleri and were confirmed through DNA barcoding. C. harcourtbutleri is a Burmese species, hitherto not recorded from mainland India.

MATERIALS AND METHODS
The study incorporates several specimens of the genus Channa from various localities within India. However, this study is specifically aimed to resolve the identity of the specimens in the Channa gachua species-group collected from the northern Eastern Ghats, within a range of around 30 km to the north-east of Papikonda National Park.

DNA isolation, PCR and DNA sequence
DNA isolation followed basic methods after partial modification (Sambrook & Russell 2001;Laskar et al. 2018). Partial segment of the mitochondrial cytochrome oxidase C subunit I (COI) gene was amplified using the primer pairs FishF1-FishR1 (Ward et al. 2005). A total of 28 COI sequences for nine channid species from India

J TT
were generated in this study. All the examined specimens were registered in the National Zoological Collections of Zoological Survey of India (Freshwater Biology Regional Centre), Hyderabad, and the COI sequences generated in this study were submitted to NCBI GenBank and BOLD. The accession numbers are given in the material examined section as well as in the phylogenetic tree and are marked by orange stars. We also retrieved COI sequences from NCBI, and BOLD. Representative sequences for 22 clearly defined channid taxa from India as referred in Conte-Grand et al. (2017) were retrieved from databases. Further, the sequences of a few recently described species whose accessions are referred by their original authors were retrieved from NCBI. The dataset of 157 COI sequences included an outgroup Parachanna obscura (MK074551). Genetic divergence analyses and the neighbor-joining phylogenetic tree visualization were performed in MEGA7.0 (Kumar et al. 2016). Bayesian inferences were drawn in Mr. Bayes (Ronquist & Huelsenbeck 2003) and the tree topology was developed in iTOL (Letunic & Bork 2007). We used the Kimura 2-parameter model, mostly applied in DNA barcoding studies (www.bold.org), to calculate the mean genetic distance between the groups. The study is limited by the lack of COI sequence of a few of the recent species, like, C. brahmacharyi, C. pomanensis. Further, we limit to discuss only the conspecific status of the COI sequences generated in the study, C. harcourtbutleri in particular.

RESULTS
The phylogenetic tree shows distinctive cladding of the Asian channid taxa with reference to the African Parachanna used herein to root as out-group. The generated sequences of the C. gachua species-group from the Yelagiri Hills in Tamil Nadu form a distinct clade that comprises of a few database sequences that were referred in a recent study as C. kelaartii. Thus, this study contributes further specimens of C. kelaartii from the southern Eastern Ghats. The sequences of the C. gachua like specimens collected from near to the type locality of Ophicephalus marginatus and various other localities in the Godavari River basin are nested as a subclade to the topotypic C. gachua. Our generated sequences of other congeners like C. marulius, C. punctata, C. stewartii, C. bleheri, C. aurantimaculata, and C. striata show distinct cladding and each comprises of conspecific database sequences wherever available. The sequence of C. pomanensis (referred in Praveenraj et al. 2019) was found to be cohesively claded with a sequence of C. melanostigma. Hence, in the absence of further specimens for both the taxa, the identity of the two sequences is not confirmed and remained beyond the limit of this study. Our sequences of C.  The present dataset is comprised of representative sequences from the clearly defined taxa. However, in order to estimate the range of intra-species genetic divergence, the sequences are grouped based on the clustering inferred from neighbor-joining phylogeny (Figure 1). Sequences in the cluster with topotypic C. gachua (referred in Britz et al. 2019) are named as C. gachua Topotypic. The sequences of C. gachua like specimens from near to the type locality of O. marginatus are named as C. gachua Godavari. Similarly, the sequences of C. royi from Andaman Islands (Praveenraj et al. 2019) are tentatively named as C. royi, sequences in the cluster with C. harcourtbutleri BIN:AAC3926 from Myanmar (Conte-Grand et al. 2017) are named as C. harcourtbutleri Myanmar, and the sequences of C. gachua species-group from near the Papikonda National Park, East Godavari District, along with other database sequences in the same cluster are named as C. harcourtbutleri India. The overall genetic divergence between groups in the dataset is lying in the range from 1.4-25.8 % ( Table 1). The divergence matrix revealed that the sequences of C. gachua Godavari are genetically diverged by 3.2% K2P distance from C. gachua Topotypic, and maintained 9.4-22.8 % K2P distance from all the congeners. C. pseudomarulius is diverged by 4.3% K2P from C. marulius, C. pardalis is diverged by 5.8% from C. bipuli, C. melanostigma by 6.3% from C. aurantimaculata, and so on. Similarly, the divergence matrix revealed that the sequences of C. harcourtbutleri India are genetically diverged by 1.4% K2P distance from C. harcourtbutleri Myanmar (BIN: AAC3926), 2.3% K2P distance from C. royi Andaman Islands, and maintained 13.4-24.3 % K2P distance from all the congeners. -Grand et al. (2017) recovered a total number of 90 BINs in their dataset having a total number of 38 valid species at time, and inferred higher species diversity in snakeheads. However, they neither included any COI sequence data of C. limbata in the phylogenetic analysis nor assigned any BIN for the species. As of now, a search for BINs with the name 'Channa' in BOLD yielded a record of 93 BINs. Conte-Grand et al. (2017) had an extensive dataset covering various geographical areas and populations. In fact, the representative sequences for almost all the species described or validated after 2017 were present either as an unnamed clade (potential new BIN in Conte-Grand et al. 2017) or has been assigned with BOLD BIN.

Conte
The taxonomic history of the snakehead fishes finds two descriptions, O. marginatus and O. coramota, with their type locality in Vizagapattam (=Visakhapatnam, Andhra Pradesh), part of the Eastern Ghats region. However, both these species have been synonymized with C. gachua (Roberts 1993;Ng et al. 1999;Courtenay & Williams 2004;Kottelat 2013). In fact, Britz et al. (2019) examined specimens of topotypic O. marginatus, and found a very similar colour pattern as well as a very little genetic difference with topotypic C. gachua (2.4% uncorrected p-distance). Britz et al. (2019) pointed out several flaws in the description of C. shingon by Endruweit (2017) and provided various valid reasons to consider C. shingon as a junior synonym of C. harcourtbutleri. Britz et al. (2019) was also not convinced enough by the morphological descriptions to consider C. royi as a distinct species from C. harcourtbutleri, and a very low genetic distance of 2.4-2.8 % uncorrected p-distance was stated to be in the range of intra-species variation, and therefore considered Andaman C. royi as a junior synonym of C. harcourtbutleri. However, the distribution limit of C. Following the previous studies, it may be figured out that the nominal species in the genus Channa with fewer or no morphological differences can have intra-species genetic divergence as high as 2.2-2.4 %, and the nominal taxa falling within such range of genetic divergence could be considered as a single species. Therefore, a clear understanding of the range of intra-species genetic divergence would be helpful in taxonomic assignment of the channid taxa. Based on the analysis of COI barcode sequences, we confirm that the specimens of C. gachua species-group from East Godavari District, Eastern Ghats, India, along with the sequences from southern India, are actually a single species which may be named as C. harcourtbutleri because of low genetic divergence with the conspecific sequences from Myanmar. Thus, this study claims the presence of C. harcourtbutleri in the wild in Eastern Ghats region, in mainland India. Nonetheless, C. harcourtbutleri has already been recorded to be distributed