Devario in Bangladesh: Species diversity, sibling species, and introgression within danionin cyprinids (Teleostei: Cyprinidae: Danioninae)

Four species of Devario are recorded from Bangladesh: D. aequipinnatus, D. anomalus, D. coxi, new species, and D. devario. Devario aequipinnatus has a wide distribution in northern India and Bangladesh. Devario coxi, from southeastern Bangladesh near Cox’s Bazar, differs from D. aequipinnatus in mtDNA (COI, p-distance 1.8%), colouration, proportional measurements, and meristics. The minor morphological differences and low frequency of overlapping meristics suggest relatively recent separation of D. coxi from other D. aequipinnatus. Devario anomalus occurs only in southeastern Bangladesh and is here reported from localities in addition to the type locality. It differs from the similar D. xyrops in adjacent Myanmar by slender body shape and by 2.3% p-distance in the COI gene. Specimens of D. anomalus from the Sangu River were found to have the mitochondrial genome of D. aequipinnatus from Bangladesh, but agree with other D. anomalus in the nuclear RAG1 gene. Devario devario has a wide distribution on the Indian Peninsula and border regions; in Bangladesh it is restricted in distribution to the Ganga, Brahmaputra, and Meghna drainages. Reports of D. assamensis and D. malabaricus from Bangladesh are misidentifications. Perilampus ostreographus M’Clelland, 1839, is tentatively synonymized with D. aequipinnatus. Phylogenetic analysis of 14 species of striped devarios based on the COI gene results in a polytomy with four unresolved clades. Devario deruptotalea from the Chindwin basin is the sister group of D. aequipinnatus+D. coxi. Devario devario is the sistergroup of D. xyrops+D. anomalus.


Introduction
The cyprinid genus Devario Heckel, 1843, includes about 38 valid species with a combined distribution covering freshwater habitats in most of tropical South and Southeast Asia from the Indus River drainage eastward to the Mekong River drainage [1][2][3]. They are relatively small species from about 50 to 100 mm body length, and distinguished from other danionine cyprinids by the combination of rounded abdominal midline (vs. keeled), neuromast-filled frontal grooves, minute or absent maxillary barbel, a dark spot close to the gill cleft, and in most History Museum, London (BMNH), Kansas University Biodiversity Institute and Natural History Museum, Lawrence (KU) Lawrence, Naturmuseum Senckenberg, Frankfurt/Main (SMF) Florida Museum of Natural History, Gainesville, (UF) or the Zoological Survey of India, Kolkata (ZSI). Measurements were taken with digital calipers to a precision of 0.1 mm. Counts and measurements were made according to Fang [21] and Kullander [22]. Colour pattern terminology follows Fang [23], with modifications for special markings as explained by Kullander [22]. Horizontal stripes and interstripes are identified by alphanumeric annotations. Fin-ray and vertebral counts were recorded from X-radiographs made with a Philips MG-105 low voltage X-ray unit and Kodak X-Omat V plates, or as digital images with a Kevex 130kVP microfocus X-ray source with a Samsung/Rayence 17x17 inch DR panel. Statistics were calculated using SYSTAT v. 13 [24]. The paired supraorbital (actually frontal) groove identified by Fang [4] as a characteristic of Devario and other genera of Devarionini and packed with neuromasts, is here referred to as neuromast grooves; an additional pair of short curved grooves is located anterior to the nostril and are here referred to as rostral neuromast grooves. Juvenile specimens preserved whole for DNA analyses were not measured. 655 basepairs from the 5' end of the mitochondrial cytochrome-oxidase subunit 1 (COI, or COX1) gene was sequenced from 38 morphologically identified specimens of Devario plus two outgroup species (see Table 1 for voucher information and GenBank accession numbers). This fragment was chosen because it is the standard DNA barcoding fragment recommended by the Barcode of Life consortium [25]. To investigate possible hybridization a 1565 bp fragment of the nuclear recombination activating protein 1 (RAG1) gene, exon 3, was sequenced from a subset of 8 selected individuals of Devario plus three outgroup species.
Sequencing of both strands of all fragments was carried out by Macrogen Europe (Amstelveen, The Netherlands). All sequences were assembled and proofread in the software Geneious version 10 [27]. Geneious was used to calculate genetic distances (uncorrected pairwise p-distance, as recommended by [28], and the Geneious plug-in Species Delimitation [29] was used to calculate the probability of reciprocal monophyly under a model of random coalescence. MrBayes version 3.2 [30][31] was used to construct a phylogenetic hypothesis (2 million generations, GTR + Γ + I model), data partitioned by codon position; samples were taken every 1000 generations, and the first 25% of samples were discarded as 'burn-in'. Convergence was checked with Tracer version 1.6 [32]. Leptobarbus sp. (Cyprinidae) was designated outgroup in all phylogenetic analyses.

Ethics statement and collecting permit
Specimens used were available in museums, purchased from fishermen or markets; or collected in the wild using beach seine or hand net and euthanized through immersion in

Results
Devario devario (Hamilton,  Diagnosis. Devario devario is a deep-bodied species (body depth at dorsal-fin origin 33-44% of SL) distinguished from all other species of Devario by the combination of small scales (38-50 in the lateral line, 16 circumpeduncular scales), long dorsal-fin base (iii.14½-17½), infraorbital process absent, danionine notch shallow, rostral barbel absent, maxillary barbel absent or rudimentary; supraorbital process absent; short dark P stripe posteriorly on side, extended on dorsal lobe of caudal fin; light spots or narrow irregular light and dark bars anteriorly on side.

Description.
External aspect is shown in Fig 1. Proportional measurements and some counts are summarized in Tables 2-5. Body laterally compressed, relatively deep, with prominent chest and abdomen, more so with increasing body size; females distinctly more deep-bodied than males (33.1-37.6% SL vs. 38.3-44.3% SL; Table 2, Fig 2).
Predorsal area wide anteriorly, gradually more compressed posteriorly; dorsal-fin base and caudal peduncle compressed, but not acute. Predorsal contour straight or slightly curved except for slight indentation marking transverse lateralis canal. Prepelvic contour steep, straight to level of pectoral fin-base, posteriorly strongly curved with projecting chest, abdominal contour straight horizontal to about pelvic fin insertion, posteriorly horizontal or slightly ascending; chest conspicuously compressed below pectoral fin in larger specimens only; analfin base contour rising; ventral margin of caudal peduncle slightly concave. Snout short, rounded in lateral aspect, shorter than eye diameter. Infraorbital process absent. Supraorbital process absent. Medial margins of dentary pair straight, parallel, except for slight indentation anteriorly, representing shallow danionine notch. Mouth terminal, obliquely directed upwards. Small bony knob at dentary symphysis. Maxilla reaching to below anterior margin of orbit. Jaws equal anteriorly; lower jaw ending anteriorly at horizontal through middle of eye. Lower jaw with narrow band of about 3 rows of densely packed pointed tubercles along lateral margin. Tubercles absent from pectoral fin in all specimens. Rostral barbel absent. Maxillary barbel absent except a very short projection present on one or both sides of head in four specimens (UMMZ 208621). Supraorbital and rostral neuromast grooves present.
Colouration in preservative. Dorsum pale grey, with white or pale yellow ground colour, suffused with grey; abdomen and interstripes white or pale yellow, thin stripe in same colour close along anal-fin base and lower margin of caudal peduncle. Opercle grey or silvery depending on preservation, with several small black spots. Narrow dark brown mid-dorsal stripe from occiput to end of caudal peduncle. Cleithral spot smaller than pupil, brown or black, round or slightly deeper than wide. P stripe black, integer, about 1 scale deep, slightly expanded caudally on caudal peduncle, commencing at vertical from about first to sixth anal-fin ray, extending to end of caudal fin, covering caudal-fin rays D1-5 to their tips and rays V1-3 basally. P stripe bordered dorsally by irregular narrow light interstripe I, which anteriorly is fragmented into a series of light round or vertically extended spots. P+1 stripe dark brown, narrow, irregular, parallel with P stripe. P+2 stripe absent or indistinct, when present bordering interstripe I+1 consisting of small white spots. P-1 and P-2 stripes contained in broad dark field posterior to abdomen. I-1 interstripe short, narrow paralleling P stripe, above anal fin and forward represented by horizontal series of round or vertically extended spots which increasing in size from posterior to anterior. White spots commonly open dorsally and ventrally on middle and anterior side, giving rise to pattern of up to nine or ten narrow alternating light interbars and dark bars.
Fins hyaline, except for caudal fin as described above; dorsal-fin anterior corner with narrow black submarginal stripe and white margin.
Geographical distribution. Devario devario is reported from northern and north-eastern India, Pakistan, Bangladesh, Nepal, and Myanmar [7-8, 17, 34-37]. In Bangladesh it is found in large rivers and beels [7][35] (pers. obs. ; Fig 3), contrasting with other Devario which occur mainly in small streams. Remarks. Barbels are usually considered to be absent in D. devario (e.g., [1]), and that is the case in most specimens, but very short barbels may be present on the maxilla on one or both sides of the head [17], [present paper]. In the field D. devario is readily recognized by the very deep body and the silvery colour in life. ? Danio assamensis Barman, 1984 [18]: 163 (holotype SZI FF1861; type locality Streamlets round about Tangla, Darrang dist., Assam).  Diagnosis. Devario aequipinnatus is distinguished from all other Devario by the particular colour pattern: presence of distinct, straight P+1, P, and P-1 stripes, of which the P stripe is wider than the others, equal in width from anterior to posterior end, or slightly tapering caudad; anterior part of Interstripe I (below P stripe) may be interrupted at one or two points anteriorly. Similar to D. coxi in colour pattern, with distinct P+1, P and P stripes along the side, but P and P-1 integer to close to pectoral girdle, and interstripe I integer except occasional interruption anteriorly (vs. stripes confluent anteriorly and containing light spots). Distinguished from D. coxi also by the combination of less dorsal-fin rays (9½-10½, exceptionally 11½ vs. 11½-12½), circumpeduncular scales (12, rarely 14, vs. 14); slightly more slender body  Tables 3-6.

Material examined
Body laterally compressed, elongate; no sexual dimorphism in body shape. Predorsal area wide and flat anteriorly, gradually more compressed, dorsum acute under dorsal-fin base, postdorsally slightly wider, gradually more compressed to caudal-fin base. Predorsal contour straight ascending or ascending slightly concave anteriorly, raised gradually from transverse lateralis canal; slightly curved to dorsal-fin base; from dorsal-fin base posteriorly to base of caudal fin straight sloping. Prepelvic contour sloping, straight or slightly concave beneath pectoral-fin base, posteriorly nearly horizontal, slightly curved; chest not conspicuously compressed below pectoral fin; anal-fin base contour rising; ventral margin of caudal peduncle about straight, horizontal. Snout short, rounded in dorsal aspect, subtriangular in lateral aspect, about as long as eye diameter. Infraorbital process flat, wider than high or square, width about half pupil diameter, with truncate or slightly rounded distal margin. Supraorbital process absent. Danionine notch caudally margined by well-developed anteromediad projecting blunt laminar dentary process. Skin cover absent from distal part of infraorbital process and anterior margin of supraorbital. Mouth terminal, obliquely directed upwards. Small bony knob at dentary symphysis. Maxilla reaching to below anterior margin of orbit. Jaws equal anteriorly; lower jaw ending anteriorly at horizontal through middle of eye. Upper lip anteriorly much wider than lower lip, which not much wider than barbel. Lower jaw with lateral row of about 25 sharp, short, conical tubercles, followed medially by about six more parallel rows, forming wide band of tubercles extending to the medial margin of the dentary. In males rows of strong, densely arranged sharp-tipped conical tubercles on anterior 5 branched rays of pectoral fin; pectoral-fin tubercles absent in females. Rostral barbels short, length of infraorbital 1 or to base of maxillary barbel or slightly longer; maxillary barbel much shorter, at most half length of rostral barbel. Supraorbital and rostral neuromast grooves prominent. Lateral line complete, along 31-33 trunk scales (Table 3), and 2 scales on caudal-fin base; comprising one tubed scale followed by a canal running steeply caudoventrad under unperforated scales to slightly posterior to pectoral-fin base, where curved caudad and represented by scales with distinct perforation; running in a curve parallel to the ventral body outline and ending low on caudal peduncle and caudal-fin base; vertical section represented by about six scales. Median predorsal scales 14 (15), 15 (5). Lateral scale rows passing between dorsal and pelvic fins ½6+1+2½ (6), ½7+1+2½ (12). Circumpeduncular scale rows 12 (21), 14 (4). A row of scales along anal-fin base. About ¼ of caudal-fin length scaled basally.
Dorsal-and anal-fin ray counts in Tables 4 and 5. Pectoral-fin rays i.10 (5), i.11 (6), i.12 (8), i.13 (1). Pelvic-fin rays i.7 (20). Dorsal fin inserted at highest point of dorsum, little posterior to middle of body; anterior rays gradually longer to last unbranched and 1-2 anterior branched rays; distal margin straight and posterior rays gradually shorter; last ray reaching to middle of caudal peduncle. Anal fin inserted below anterior rays of dorsal fin; anterior rays gradually longer to last unbranched and 1-2 anterior branched rays; distal margin slightly concave; last ray reaching to or very slightly beyond middle of caudal peduncle. Pectoral-fin insertion at about vertical through posterior margin of osseous opercle; tip rounded, extending to pelvic-fin origin. Pectoral-fin axial lobe well developed. Pelvic fin inserted slightly anterior to mid-body, anterior to dorsal-fin origin; not reaching to anal-fin origin. Pelvic axillary scale present. Caudal fin forked, lobes round-tipped, of about equal length, but lower very slightly longer and broader.
Colouration in preservative. Ground colour yellowish white. Sides of head with sparse pigmentation. Dorsum pale grey. Narrow black dorsal midline from occiput to end of caudal peduncle. P stripe about one scale wide, slightly decreasing in width caudally, margins even; dark brown, darker caudally, ending anteriorly 2-3 scales posterior to cleithrum, like other horizontal stripes and interstripes, posteriorly at end of caudal peduncle. Anterior to P stripe diffuse brownish, followed by contrasting yellowish white short vertical bar and wedge shaped black or dark brown vertically oriented cleithral spot. Interstripe I less than one scale wide, paralleling P stripe, anteriorly distinctly narrower than P stripe, posteriorly relatively less distinctly narrower. Interstripe I+1 Paralleling P stripe, anteriorly very narrow, slightly wider caudally, still less than 1 scale wide. P+1 stripe paralleling P stripe, less than one scale wide, pale brown, anteriorly on caudal peduncle fading into pale grey colour of dorsum. Interstripe I +2 short, parallel to, and much narrower than P+1 stripe. P-1stripe brown narrower than P stripe and gradually narrower caudally, ending on caudal-fin base. Interstripe I paralleling P-1 stripe, narrower than P-1 stripe and narrower caudally. P-2 stripe indistinct, formed by scattered melanophores from above posterior anal-fin base anteriorly to anterior level of other stripes and interstripes. In some specimens, one or two light spots anteriorly in P stripe, or I-1 stripe anteriorly partly discontinuous. Pigmentation of P and P-1 stripes continuous anteriorly. Caudal fin hyaline except dark grey indistinct band on lower rays of upper lobe. Dorsal fin dusky basally, corner and distal margin hyaline. Remaining fins hyaline. Specimens in ethanol grey where formalin preserved specimens brown. Cleithral spot variable in shape, wedgeshaped or slightly oval. P stripe of almost equal width in the larger specimens, and also with variable pattern of short darker stripes and light spots at anterior end of dark stripes.
Geographical distribution. In Bangladesh, D. aequipinnatus is found in small numbers in fast flowing small streams within the Karnafuli, Feni, and Meghna drainages (Fig 3). Records from the Tista and Brahmaputra drainages in India suggest that it occurs in similar habitats further to the west or northwest. Devario aequipinnatus was reported from Bangladesh by [11], [35]. It was given a wide range in northern India, Nepal, Bangladesh, Pakistan, Sri Lanka, Thailand and China by Barman [17], and similar wide distribution by other authors (e.g., [36]). In Bangladesh the species appears to be restricted to small forest hill streams.

Description. External aspect is shown in
Body laterally compressed, elongate, females slightly more deep-bodied than males (30.2-33.2% SL vs. 27.8-39.0%), and with greater preanal distance (66.0-69.5% SL vs. 63.6-64.9%). Predorsal area wide and flat anteriorly, gradually more compressed, dorsum acute under dorsal fin base, postdorsally slightly wider, gradually more compressed to caudal fin base. Predorsal contour in small specimens straight ascending; in large adults ascending slightly concave anteriorly, raising gradually from transverse lateralis canal, slightly curved to dorsal-fin base; from dorsal-fin base posteriorly to base of caudal fin straight sloping. Prepelvic contour sloping, straight or slightly concave beneath pectoral-fin base, posteriorly nearly horizontal, slightly curved, more so in females; chest not conspicuously compressed below pectoral fin; anal-fin base contour rising; ventral margin of caudal peduncle about straight, horizontal. Snout short, rounded in dorsal aspect, subtriangular in lateral aspect, about as long as eye diameter. Infraorbital process flat, wider than high, width about half pupil diameter, with truncate or slightly rounded distal margin. Supraorbital process absent. Danionine notch caudally margined by well-developed anteromediad projecting, blunt laminar dentary process. Skin cover absent from distal part of infraorbital process, dentary process, and anterior margin of supraorbital. Mouth terminal, obliquely directed upwards. Small bony knob at dentary symphysis. Maxilla reaching to below anterior margin of orbit. Jaws equal anteriorly; lower jaw ending anteriorly at horizontal through middle of eye. Upper lip anteriorly much wider than lower lip, which not much wider than barbel. Lower jaw with lateral row of about 20-25 short sharp, short, conical tubercles, followed medially by one or more parallel rows, or a wider band of tubercles extending to the medial margin of the dentary; except for lateral row, tuberculation variably developed, from distinct to rudimentary without obvious correlation with sex. In males rows of strong, densely arranged sharp-tipped conical tubercles on anterior six branched rays of pectoral fin; pectoral-fin tubercles absent in females. Rostral barbels short, length of infraorbital 1 or to base of maxillary barbel; maxillary barbel much shorter, at most half length of rostral barbel. Supraorbital and rostral neuromast grooves prominent. Lateral line complete, along 31-33 trunk scales (Table 3), and 2 scales on caudal-fin base; comprising one tubed scale followed by a canal running steeply caudoventrad under unperforated scales to slightly posterior to pectoral-fin base, where curved caudad and represented by scales with distinct perforation; running in a curve parallel to the ventral body outline and ending low on caudal peduncle and caudal-fin base; vertical section represented by about six scales. Median predorsal scales 14 (5), 15 Ã (2). Lateral scale rows passing between dorsal and pelvic fins ½6+1+2½ (6), ½7+1+2½ Ã (2). Circumpeduncular scale rows 14 Ã (7). A row of scales along anal-fin base. About ¼ of caudal-fin length scaled basally.
Dorsal-and anal-fin ray counts in Tables 4 and 5. Pectoral-fin rays i.11 (5), i.12 Ã (2). Pelvicfin rays i.7 Ã (7). Dorsal fin inserted at highest point of dorsum, little posterior to middle of body; anterior rays gradually longer to last unbranched and 1-2 anterior branched rays; distal margin straight and posterior rays gradually shorter; last ray reaching to middle of caudal peduncle. Anal fin inserted below anterior rays of dorsal fin; anterior rays gradually longer to last unbranched and 1-2 anterior branched rays; distal margin slightly concave; last ray reaching to or very slightly beyond middle of caudal peduncle. Pectoral-fin insertion at about vertical through posterior margin of osseous opercle; tip rounded, extending to pelvic-fin origin. Pectoral-fin axial lobe well developed. Pelvic fin inserted slightly anterior to midbody, anterior to dorsal-fin origin; not reaching to anal-fin origin. Pelvic axillary scale present. Caudal fin forked, lobes round-tipped, of about equal length, but lower very slightly longer and broader.
Coluration in preservative. Ground colour yellowish white. Dorsum pale brownish. Narrow black dorsal midline from occiput to end of caudal peduncle. P stripe about one scale wide, of uniform width throughout, margins even; greyish brown anteriorly, darker caudad, dark brown to black posterior to trunk middle; ending at end of caudal peduncle; 1-3 small light spots may be present close to anterior end of stripe. Interstripe I anteriorly replaced with three light spots. Greyish brown pigmentation in P and P-1 stripes continuous anterior to trunk middle. P+1 stripe slightly lighter than and one half as wide as P stripe, ending fading anteriorly on caudal peduncle. P+2 stripe not present or narrower than P+1 stripe, ending below end of dorsal fin base. P-2 stripe along abdominal side, faint, barely visible. P-1 stripe as wide as P stripe and light greyish brown anteriorly; indistinct lower border; gradually narrower posteriorly reduced to thin line on caudal peduncle. Interstripes I and I+1 each about half as wide as P stripe, straight, uniformly light, Interstripe I+2 slightly wider han P+1, ending diffusely anteriorly on caudal peduncle. Black, vertically oriented cleithral spot covering part of first lateral-line scale; posterior to it two greyish brown bars, slightly curved (concave rostrally), from anterior end of P+1 stripe to anterior end of P-1 stripe. Dark brown dot on caudal-fin base at end of P and P-1 stripes. Caudal fin hyaline except dark grey indistinct band on middle rays. Dorsal fin with scattered melanophores along middle; and white anterior corner and distal margin. Anal fin with scattered melanophores and white anterior corner. Pectoral and pelvic fins hyaline.
Live colouration. Adults observed in the field with dorsum steel blue, P+1 stripe green, P and P-1 stripes pale blue or blue-green, abdomen and chest white. Light interstripes yellow. Middle rays of caudal fin grey-green, lobes red.

Geographical distribution.
Known only from a small stream near Cox's Bazar, and the lower Matamuhuri River in Bangladesh (Fig 3).
Etymology. The specific name is a noun in the genitive case, referring to the type locality area (Cox's Bazar), but could also be understood as referring to the gene fragment used to identify the species (cytochrome c oxidase subunit 1, often shortened to COXI); or to Hiram Cox (1760-1799), after whom Cox's Bazar was named.
Remarks. Only two specimens were available for sequencing; remaining specimens referred to this population were identified by the unique combination of 14 circumpeduncular scales and 11½-12½ unbranched dorsal-fin rays. Diagnosis. Devario anomalus shares uniquely with D. xyrops an interrupted P stripe; the anterior portion is represented by several short vertical bars, the posterior portion by a dark stripe tapering caudally, and not continued onto the caudal fin. Narrow P+1 and P-1 stripes may be present posteriorly on the side. Distinguished from D. xyrops by more slender body, 26.3-30.2% SL vs. 31.4-35.8% SL in females (Fig 6) Description. External aspect is shown in Fig 5. Some meristics is presented in Tables 3-5; proportional measurements are summarized in Table 8.

Devario anomalus Conway, Mayden & Tang, 2009
Body laterally compressed, elongate. Predorsal contour straight or slightly curved, ascending, with or without slight indentation marking position of transverse lateralis canal; sloping posteriorly from dorsal-fin insertion. Prepelvic contour curved, more so in females; chest more compressed below pectoral fin, but not keeled. Snout short, rounded in dorsal aspect, subtriangular in lateral aspect, about as long as eye diameter in lateral aspect. Infraorbital process laminar, pointed or broader than high with truncate distal margin. Danionine notch caudally margined by blunt anteromediad projecting, laminar dentary process. Skin cover absent from distal part of infraorbital process, dentary process, and anterior margin of supraorbital. Mouth terminal, obliquely directed upwards. Small bony knob at dentary symphysis. Maxilla reaching to below anterior margin of orbit. Jaws equal anteriorly; lower jaw ending anteriorly at horizontal through middle of eye, posteriorly at vertical through middle of eye. Lower jaw with large conical tubercles in several rows in juveniles (including smallest specimen, 19.4 mm SL); tubercles gradually reduced in relative height with increasing body size; in adults lower jaw with 2-3 rows of minute conical tubercles concentrated on lateral surface, occasionally absent, in large specimens a few tubercles along margins of danionine notch; tuberculation in adults variably developed from hardly visible to well-developed without correlation with sex. Seven specimens, 50.8-79.2 mm SL, with two rows of strong, densely arranged sharp-tipped conical tubercles anterior 5-6 branched rays of pectoral fin. Rostral barbel short, length of infraorbital 1 or to base of maxillary barbel; maxillary barbel much shorter, at most half length of rostral barbel.
Lateral line complete, along 30-33 body scales (Table 3) and 2 scales on caudal-fin base; comprising one tubed scale followed by a canal running steeply caudoventrad under unperforated scales to slightly posterior to pectoral-fin base, where curved caudad and represented by perforated scales running in a curve parallel to the ventral body outline and ending low on caudal peduncle and caudal-fin base; vertical section represented by about six scales; continued by two scales on caudal-fin base. Median predorsal scales 14 (2), 15 (10). Lateral scale rows passing between dorsal and pelvic fins ½6+1+2½ (7) ½7+1+2½ (5). Circumpeduncular scale rows 12 (12). A row of scales along anal-fin base. About ¼-1/3 of caudal-fin length scaled basally.
Colouration in preservative. Dorsum pale grey, sides pale yellowish white, abdomen white. Opercle brownish or silvery depending on preservation. Sides of head sparsely pigmented. Narrow dark brown middorsal stripe from occiput to end of caudal peduncle. Dark brown or black, round or irregularly triangular cleithral spot covering part of first lateral-line scale and scale above. Along middle of side anterior to about vertical from anal-fin insertion a row of about 6-10 short dark vertical bars occupying a space equivalent to stripes P +1, P, and P-1. Anterior two bars greyish brown and narrow, followed by 4-5 wide, deep and contrasting black bars, preceding grey posterior bars. P stripe commencing posterior to bars, above anterior anal fin rays, extending to end of caudal peduncle; wide, black, with irregular borders, P stripe bordered dorsally by irregular narrow interstripe I. P+1 stripe dark brown, narrow, irregular, parallel with P stripe. P-1 stripe represented by indistinct spots detached from vertical bars may be present anteriorly on side, and present as 2-4 indistinct spots below P stripe. Dark markings predominantly brown in specimens preserved for a longer time, black in fresh specimens.
Dorsal fin pale grey basally, distal margin white. Anal fin pale grey basally, distal margin white. Pectoral and pelvic fins hyaline. Caudal fin semihyaline or pale grey; grey stripe along middle rays. Juveniles (Fig 3) with dark brown cleithral spot, about 5 pale vertical bars from about 20 mm SL; broad P stripe, indistinct P-1 stripe and contrasting interstripes I and I-1 posteriorly on side.
Geographical distribution. Known only from Bangladesh, recorded from coastal streams between Himchori and Teknaf, and the middle Sangu River upstream of Bandarban (Fig 3).
Comments. Devario anomalus was described from six large specimens, 57.0-81.0 mm SL, from south of Cox's Bazar [9], probably at Himchori based on the coordinates in the original description. Data in the original description are confirmed by the present observations, except for the number of lateral line scales (33-35 according to [9]; 30-33 in our material), and different ranges of otherwise overlapping proportional measurements. Devario anomalus was already reported from Barachora and Kudung Cave [8]). The Donumia and Sangu localities are new records for the species, which is probably present in small streams in a wider area of the Cox's Bazar district and Sangu River. Two specimens (CAS 94595) from the Sangu River that were reported [20] as not clearly identifiable as belonging to either D. xyrops or D. anomalus, fall within the variation of D. anomalus reported here. The sequenced Sangu specimens (NRM 67936-67937) do not depart morphologically from other D. anomalus, but the COI sequence corresponds to D. aequipinnatus. Devario anomalus is most similar to D. xyrops from western Rakhine Yoma in Myanmar. They share a unique colour pattern in the genus with contrasting black bars in a group anteriorly, and a wide black P stripe posteriorly, separated by a few paler vertical bars. In D. xyrops the pale area is more obvious than in D. anomalus, however. Females of D. anomalus are more elongate than females of D. xyrops (body depth 26.3-30.2% SL vs. 31.4-35.8%; Fig 6). Devario anomalus also has a slightly narrower interorbital space (9.6-10.5% SL vs. 10.6-11.8%) than D. xyrops, a character which does not show sexual dimorphism. In the material available all male D. xyrops are larger than the male D. anomalus, and regression on standard length cannot be compared, but proportional measurements overlap (body depth 29.1-35.8% SL in D. xyrops, 25.9-30.2% in D. anomalus). Adult females of both species tend to be more deep-bodied than males but this is not expressed in proportional measurements, perhaps due to a sample bias because males are few and large males matching large females are available only for D. xyrops. The two largest specimens of D. anomalus, 96.7 and 79.2 mm SL (NRM 68131), are female and male, respectively, but not in good condition and not included in the measurements.  Fig 7), separates D. devario from the rest based on a prominent PC2 in which the long dorsal-fin base is dominant. Removing D. devario results in more scatter (Table 10, Fig 8), probably marked by individual variability from different preservation and size ranges among species samples. Nonetheless, the PCA indicates different clusters for D. anomalus and D. xyrops, as well as for D. aequipinnatus and D. coxi.

Phylogenetic analysis and species delimitation
The phylogram based on COI sequences from a large number of striped species of Devario ( Fig  9) is unresolved, with four distinct clades, (a) consisting of species from Thailand and Laos (Mekong River and Peninsular Thailand), (b) species from Myanmar (Irrawaddy and Salween Two specimens of Devario anomalus from the Sangu River cluster with D. aequipinnatus in COI sequences (Fig 9, Table 11). In the nuclear RAG 1 sequences (Table 1, Fig 10), those specimens are 99.5% similar to D. anomalus from near Cox's Bazar, but only 97.3% similar to D. aequipinnatus.

Other nominal species of Devario reported from Bangladesh
Ahmed et al. [8], reported D. assamensis from Bangladesh. Their photo, Fig 2, however, shows D. aequipinnatus. Ahmed et al. [8] reported D. malabaricus from Bangladesh. Their photo, Fig  2, however, shows D. anomalus. Conway et al. [9] discussed the status of Perilampus ostreographus M'Clelland, 1839 [13] in connection with Bangladeshi Devario, but did not have any specimens to demonstrate the existence of the species in Bangladesh. Perilampus ostreographus was synonymized with Devario devario by Day [39] and later authors.

Taxonomy of Devario aequipinnatus
Devario aequipinnatus is one of the most commonly reported species of Devario, with numerous records from India, e.g., [17,[39][40], some records from Nepal [9,37], and few, imprecise  records from Bangladesh [11] [35]. It is even reported from China, Myanmar, Pakistan, Sri Lanka, and Thailand [17]. Specimens reported from "rice paddy and creek east of Cox's Bazar" [9] may represent D. coxi. Devario aequipinnatus was described by M'Clelland [13] as Perilampus aequipinnatus based apparently on a single preserved specimen that meanwhile probably has been lost. M'Clelland indicated that the specimen had been collected in Assam. At the time Assam covered a large part of northeastern India and northwestern Bangladesh. Since then, numerous species of Devario have been reported from South and Southeast Asia, many of them very similar to each other in appearance, and frequently synonymized with D. aequipinnatus, which name has served as a convenient container particularly for all species of Devario with a predominantly striped colour pattern. M'Clelland's 1839 monograph [13] on the Indian Cyprinidae was presented on 5 September 1838, and published in 1839. In the description of Perilampus aequipinnatus he makes special mention that his material was found among plants in spirits collected by William Griffith in Assam. He also mentions that "It has already been three years in spirits. . ." By twice referring to the studied material as "it" it seems likely that only a single specimen was available to M'Clelland. From the reference to Assam and that the specimen had been preserved since three years, it is obvious that it came from the expedition by M'Clelland, Wallich, and Griffith  [41]. The specimen of D. aequipinnatus could have been obtained in present-day Bangladesh as well as in any of the Indian localities visited by the expedition, although it seems more likely that it was obtained in the focus area of the expedition, in present-day Assam or Arunachal Pradesh.
The original description of D. aequipinnatus [13] is relatively brief, but compatible with an elongate species of Devario, except for the count of nine pelvic-fin rays. Remaining counts are recorded as D. 13, P. 13, A. 13, C. 19; lateral line scales 32; scale rows between dorsal and pelvic fins 8. There is no colour description, which is remarkable, as D. aequipinnatus has been identified by later authors as a species with distinctive striped colour pattern, and stripes are mentioned in the colour description and shown in the drawing of D. ostreographus in the same paper [13], p. 392, pl. 46 [= 45] , Fig 3). The tongue is described as thick and corrugated, and whereas a tongue is absent in Devario, the skin covering the floor of the gape is thick with transverse grooves. M'Clelland also gives the length of the intestine and stomach, showing that the specimen was dissected. M'Clelland's illustration [13], pl. 60, Fig 1, shows small dark spots in two or three rows on the side close to the back, but no other colour pattern.
There is no indication that the specimen described by M'Clelland is in existence. The identification of the fish as a species of Devario is not fully supported by an unambiguous autapomorphy, and the taxon should properly have been treated as a species inquirenda. Nevertheless, the widespread application of the name on a species of Devario merits its recognition.
The type locality area for D. aequipinnatus, as deduced from the itinerary of Griffith's 1835 expedition includes present-day Bangladesh, and the Indian states of Meghalaya and Assam. We have not had access to specimens coming from specific places mentioned by Griffith [41] and listed above, but we base our concept of D. aequipinnatus on specimens from within the area covered by the general area of the expedition and which correspond to M'Clelland's description except for probable differences in counting fin-rays, and the possession of a welldeveloped striped colour pattern. M'Clelland based D. aequipinnatus on the equal size of the dorsal and anal fins, probably referring to both having 13 fin rays. Although it is uncertain whether M'Clelland included or excluded the two shorter unbranched dorsal-and anal-fin rays in his count of 13, there are no alternative identifications than a species of Devario, as no other cyprinid genus in the area has partly opposed dorsal and anal fins with more than ten unbranched rays. In Salmophasia, Barilius, and Opsarius the dorsal fin, with usually seven branched rays, has a much shorter base than the anal fin, with more variation in number of fin rays, and the dorsal fin inserts well anterior to a vertical from the insertion of the anal fin; in Chela, Laubuka, and Esomus, with opposed dorsal and anal fins, the dorsal fin, with only 6-10 unbranched rays, has a much shorter base than the anal fin (data from [36]). In D. aequipinnatus, as diagnosed here, there are 1-2 more rays in the anal fin than in the dorsal fin, but there is no known species within the type locality area that comes closer to M'Clelland's description.
Aside from D. assamensis and D. ostreographus, the only potential synonym of D. aequipinnatus is Leuciscus lineolatus Blyth, 1858, already synonymized by [39], followed by [17] and [1]. Blyth's description [38], based on a single specimen from Darjeeling, agrees in scale and fin counts (D. 12, A. 14, P. 11, V. 8; 10 transverse scales, about 32 lateral line scales), and the colour markings mentioned correspond to the cleithral spot and P, P+1, and P-1 stripes, with interstripes: "A dusky spot behind the gill-covers, placed in a whitish space; beyond which a broad darkish band extends to the middle of the tail, bordered by a narrow pale line above and below, the lower not reaching so far forward as the upper: below this again another dark band and then white; and above a second and trace of a third pale line." No type specimens are known to exist.
Jayaram [40] presented a review of Devario malabaricus and D. aequipinnatus from southern India, focusing on the diagnosis of D. malabaricus. The specimens he listed are from the Krishna, Kaveri (Coromandel Coast), and Bharathappuzha (Malabar Coast) drainages. His illustration of D. aequipinnatus shows a specimen from the upper Krishna River drainage, not far from Pune, agreeing with his view of D. aequipinnatus as a slender species with 31-34 lateral line scales. Two specimens, CMK 7250 with the same locality data represent a very slender species of Devario, with about 34 lateral-line scales, extremely short barbels, 12 circumpeduncular scales, infraorbital process absent; and a colour pattern including several short vertical bars anteriorly on the side, followed by horizontal stripes P, P+1 and P-1. It is consequently different from D. aequipinnatus in barbel development, absence of infraorbital process, and colour pattern. Several nominal species of Devario have been described from southern and western peninsular India, including D. aurolineatus (Day, 1865), D. canarensis (Jerdon, 1849), D. fraseri (Hora & Mukerji, 1935), D. malabaricus, D. mysoricus (Jerdon, 1849) and D. neilgherriensis (Day, 1867), all requiring taxonomic revision.
Numerous COI sequences identified as Devario aequipinnatus were downloaded from Gen-Bank in August 2016. They reflect extensive barcoding projects currently underway in India. About half of those sequences do not concern the species here identified as D. aequipinnatus, but are from D. malabaricus and similar species in southern India. "Barcode sequences" from the Brahmaputra in Assam/Meghalaya (Genbank accession numbers FJ459485, KF511497-511499, KJ936744-48), the Tuicheng River, a tributary of the Kaladan River (Genbank accession numbers KJ936784-KJ936788) in Mizoram, and from Kamalganj, on the border with Tripura, apparently in the Meghna drainage (Genbank accession number KT364769) agree with our sequences of D. aequipinnatus. Based on preserved specimens and published barcode sequences, it seems that D. aequipinnatus is restricted to the Brahmaputra, Feni, Meghna, and Karnafuli basins, including the upper Tista River, and potentially present in the upper Kaladan drainage.  Table 4). The colour pattern is slightly different as in D. aequipinnatus the P, P+1 and P-1 stripes are integer, with the exception that the I-1 stripe may be interrupted anteriorly (Fig 4A-4C); whereas in D. coxi the P+1 stripe is irregular anteriorly, and the P and P-1 stripes are joined anteriorly on the side in a wide dark area incorporating light spots (Fig 4D and 4E). Devario coxi is represented by large adults, and D. aequipinnatus by many small specimens, rendering a morphometric comparison difficult, as size allometry may influence proportional measurements. Nevertheless, the PCA (Fig 8) supports distinct morphologies. Although very similar to D. aequipinnatus, genetic distinctness, fin counts, scale counts, colouration and morphometrics combined provide evidence for recognition of D. coxi as a distinct species.

Devario coxi
Pairwise p-distance is the proportion of differences, or divergence, between two DNA sequences, expressed in percent. As a coarse empirical rule-of-thumb, for the mitochondrial COI gene, the pairwise p-distance between any two members of the same species is typically <1%, while the pairwise p-distance between members of different species is typically !2% [43], and can be used as an indicator of species affinity.
The p-distance among striped Devario range from 1.8 to 12.8 (Table 11). The distance between D. aequipinnatus and D. coxi is 1.8%, which is low for the genus but similar to other closely related similar species like D. kakhienensis and D. fangae which occur in separate headwaters of the Irrawaddy River in Myanmar and China. At the other end, D. deruptotalea is more similar (98%) to D. coxi, than to D. aequipinnatus (94%), conforming to the close relationship in the phylogenetic analysis. Rather than insisting on a rule of thumb of 2% p-distance as species criterion, we accept the morphologically differentiated D. aequipinnatus+D. derupto-talea+D.coxi and D. browni+D. fangae+D.kakhienensis as two groups of closely related species.
Masters et al. [29] proposed P ID(Liberal), a statistical comparison of the distribution of intraspecific variation within a putative species to the interspecific distance and intraspecific variation within its closest relative, which estimates the mean (95% confidence interval) probability of correctly identifying an unknown member of the putative species, with the criterion that it should fall within or sister to the putative species. P ID(Liberal) supports the distinctiveness of the D. coxi (0.97).
Monophyletic groups in gene trees can be artefacts caused by stochastic gene coalescence within a larger panmictic group. Rosenberg's P(AB) statistic [44] is an estimate of the probability that a putative species is such a randomly monophyletic clade relative to its closest sister clade under a null model of random coalescence. Rosenberg's P(AB) supports D. coxi as a monophyletic group separate from D. aequipinnatus.
The RAG1 sequence of D. coxi is identical to that of D. aequipinnatus (Fig 10), and seems likely that the two represent relatively recently isolated species.
Other species of Devario reported from Bangladesh Devario assamensis was described by Barman [18] on the basis of three specimens from Tangla, Darrang district, Assam, characterized by a very deep body, (2.87-3.1 times in SL), 40-41 lateral-line scales, dorsal fin rays ii.12, anal-fin rays ii. [16][17]. It was synonymized with D. regina by Talwar & Jhingran [36] without explanation. Tilak & Jain [19] attempted to correct Barman's description of the type specimens of D. assamensis. The only significant difference, however, seems to be their count of 36 instead of 40-41 scales in the lateral line, and presence instead of absence of a supraorbital process ("pointing forward above the anterior superior margin of the orbit"). Unpublished notes on the holotype made by the late Fang Fang mention an infraorbital process, but no supraorbital process and it is not showing on her photograph of the holotype. It may be that the supraorbital process observed by Tilak & Jain is the anterior projection of the orbital shelf formed by the supraorbital, or a sharp curved process of the lateral ethmoid supporting the nasal cavity. The latter is frequently exposed in poorly preserved or desiccated specimens of Devario. The supraorbital process in Devario spinosus, referred to by Tilak & Jain, is a flat, blunt, erect projection from the supraorbital (pers. obs.). We count 34/35±1 scales in the lateral line and 34 scales along the middle of the side on the photograph of the holotype of D. assamensis, and at least 12½ or 13½ branched anal-fin rays. The holotype has lost most of the caudal fin and also the other fins are damaged. It shows only faint traces of the colour pattern, but it is consistent with the description and photograph by Barman [18], pl. 84, including P+1, P, and P-1 stripes. The colour pattern is thus similar to that of D. aequipinnatus. With the reservation that both Barman's and Tilak & Jain's counts may be incorrect (e.g., 2 instead 3 of unbranched rays in the dorsal and anal fins), their count of 16-17 (vs. 11.5-13½ branched anal-fin rays, and 36 (vs. 30-33) lateral line scales suggest that D. assamensis may be a distinct species. Devario assamensis is so far known only from the type series.
Conway et al. [9] discussed the status of Perilampus ostreographus M'Clelland [13] in connection with Bangladeshi Devario, removing it from the synonymy of D. devario. M'Clelland gave P. ostreographus a wide distribution which potentially includes present-day Bangladesh: "throughout Assam in small indolent streams, as well as in the larger rivers, . . . also abundant in some parts of Bengal." The species was referred to Devario by Heckel [45], p. 1015, as D. osteographus. Devario cyanotaenia Bleeker, 1860, is a junior objective synonym, based on M'Clelland's [13] drawing of D. ostreographus. No type specimens are known to exist. According to the data in the original description and the accompanying figure, reproduced by Conway et al. [9], Fig 6, it is a deep-bodied, striped danionin, characterized by M'Clelland by body depth almost equal to half of the total length (to be compared with body depth one-third to one-fourth of the total length in his description of D. aequipinnatus), mouth and head directed upwards; 35 scales along the sides, 15 scales in a transverse row; fin counts D. 12, A. 16, P. 15, V. 8, C. 19; colour pattern with purple stripes along the side "to the extremity of the caudal"; prominent knob at the tip of the lower jaw; tongue thick and corrugated; and "caudal almost entire, the middle rays being very little shorter than the outer". The pectoral and pelvic fin counts are probably incorrect, as there are rarely more than 12 branched pectoral-fin rays in Devario. The description of the colour pattern is compatible with striped species of Devario, except that in Devario the dark stripes along the side do not continue on the caudal fin, except for the continuation of the P stripe. The image, however does not show any stripes on the caudal fin. In Danio the stripes on the side of the body may be continued on the caudal fin, as in D. dangila and D. rerio. With a size of "usually about three inches" (76 mm), P. ostreographus is too large for D. rerio. Adult Devario always possess a distinctly emarginate caudal fin, whereas the caudal fin becomes almost subtruncate in large specimens of Danio. The trunk colour pattern of D. dangila and other members of the chain danio group is different from that of D. ostreographus in being composed of a combination of narrow stripes and contiguous rings. M'Clelland does not mention presence of barbels, suggesting that the species does not have barbels, or only very short barbels, whereas most species of Danio have very long barbels. It is clearly different from the other deep-bodied Devario in Bangladesh and northern India, D. devario, in presence of multiple flank stripes and shorter dorsal fin (12 vs. 14½-17½ branched rays). Based on the colour pattern and the fin counts, and the option that P. ostreographus represents large deep-bodied females, it is potentially the same species as D. aequipinnatus. Both species were described in the same paper, but have never before been considered to be the same species. As First Revisers we give precedence of priority to Perilampus aequipinnatus, under the International Code of Zoological Nomenclature, Article 24.2. [46]. We do not think it is necessary to fix a neotype for P. ostreographus, as the synonymy is tentative and the fish fauna of its area of occurrence is still incompletely known. There is at least no indication of occurrence in Bangladesh of a deep-bodied species of Devario conforming to the description by M'Clelland [13].
Because D. ostreographus and D. assamensis are the only known deep-bodied Devario known from the Ganga-Brahmaputra basin besides D. devario it seems possible, potentially supported by the similarity in reported shape and colour pattern, and anal-fin count, that D. assamensis is a junior synonym of D. ostreographus, and in extension, a synonym of D. aequipinnatus. The 1835 expedition to Assam, in which McClelland participated [41], passed Darrang on the way, and it is possible that P. ostreographus was collected in that part of the Brahmaputra.
No specimens identified by McClelland as P. ostreographus are known to have been preserved. Four specimens of D. devario in SMF 8835, donated by McClelland in 1844, are in a very bad state of preservation but are definitely referable to D. devario. Associated collection information does not indicate that they represent D. ostreographus (SMF, pers. comm.) The species is not listed in the catalogue of type specimens in the Zoological Survey of India, Calcutta [47].
Day [48] synonymized species currently in Devario (Danio micronema, Devario cyanotaenia, Perilampus canarensis, P. malabaricus, P. mysoricus) under the name of Danio osteographus without explanation, and later [39], he synonymized D. ostreographus with Devario devario, and synonymized the other species under D. malabaricus. Later ichthyologists followed Day's authority, and all treated D. ostreographus as a junior synonym of D. devario [1], [17], [36], [49], until Conway et al. [9] re-validated D. ostreographus as a species without known specimens. remaining D. anomalus. These two specimens differ from D. aequipinnatus by 0.3% or less, strongly indicating that they are conspecific. However, their nuclear RAG1 sequences place them with other D. anomalus (Fig 10), 0.5% dissimilar from the other D. anomalus, and 2.7% dissimilar from D. aequipinnatus. These two specimens consequently represent introgression of D. aequipinnatus. This seems to be the first case of hybridization in the wild between two species of danionine cyprinids. Hybridization is reported from many other cyprinids, but then mainly from Europe and the United States and usually in anthropogenic conditions [53], and no documented case of hybridization in nature in Asia. The present case is noteworthy because there is no record of D. aequipinnatus from the Sangu River. The closest known populations of D. aequipinnatus are in the Karnafuli drainage. It is not impossible that D. aequipinnatus has been overlooked in the Sangu drainage, or the introgression is due to an ancient river connection and faunal exchange.

Biogeography
Bangladeshi species of Devario show a distinct pattern by presence and absence data. Most noteworthy is that Devario devario is absent from southeastern Bangladesh, although there are suitable habitats in the major rivers, the Karnafuli and Sangu. There are also no collecting records from the Sundarbans or elsewhere in the lowland southwestern portion of Bangladesh. Outside Bangladesh the geographical range of D. devario stretches along the Brahmaputra up to Dibrugarh, and in the Ganga River basin in Nepal and India west to the Indus River drainage [17] [54].
Devario aequipinnatus turns out to be a widespread species in the Brahmaputra drainage in northeastern India, eastern Nepal, and in Bangladeshi rivers Meghna, Feni and Karnafuli; but absent from the Sangu River and coastal streams, except for the presence of D. aequipinnatus mtDNA in D. anomalus in the Sangu River. This is still a much more restricted distribution than claimed by, e.g., Barman [17].
The sister-group relationship of D. anomalus and D. xyrops, the latter endemic to the western slope of the Rakhine Yoma-and the only species of Devario present there-suggests allopatric speciation, although there are no obvious geographical barriers between the Cox's Bazar and Teknaf region and the Rakhine Yoma, except possibly the large Kaladan River, which forms the border between India and Myanmar, emptying in the Bay of Bengal at Sittwe. The collecting record of D. xyrops and D. anomalus is still meagre, and should be expanded in order to provide material for a biogeographical analysis.
Devario aequipinnatus is a widespread species in northeastern India and Bangladesh, including the Feni and Karnafuli Rivers, but absent from the Sangu River and coastal streams, except for the distinctive population near Cox's Bazar, and the presence of D. aequipinnatus mtDNA in D. anomalus in the Sangu River.
Whereas the collecting records of Devario definitely indicate southeastern Bangladeshwith two endemic species-as distinct from the northern and western parts of the country, and possibly pertaining to the distinctive fauna of the Rakhine Yoma, the geographical coverage of available collections is still limited. The presence of D. devario and D. aequipinnatus in the rest of the country represents a portion of two widely distributed species in the Brahmaputra basin. Published inventories (e.g., [8]) nevertheless suggest that the four species of Devario reported here represents the total presence of the genus in Bangladesh.
Bangladesh using DNA barcodes (Swedish Research Council, contract D0674001 to SK and ARM). For access to other material and information we thank past and present curators and collection staff at the California Academy of Sciences, Kansas University Biodiversity Institute and Natural History Museum, University of Michigan Museum of Zoology, The Natural History Museum and the Naturmuseum Senckenberg in Frankfurt. We thank reviewers for valuable input.