Revision of Gangesia (Cestoda: Proteocephalidea) in the Indomalayan Region: Morphology, Molecules and Surface Ultrastructure

Tapeworms of Gangesia Woodland, 1924 (Cestoda: Proteocephalidea) parasitic in freshwater fishes in the Indomalayan Region were critically reviewed. Evaluation of type specimens and newly collected materials from Bangladesh, Cambodia and India, as well as critical examination of extensive literature have shown that only the following four species, instead of 48 nominal species of Gangesia and Silurotaenia Nybelin, 1942 reported from this region (36 new synonymies proposed), are valid: Gangesia bengalensis (Southwell, 1913), type-species of the genus and most common parasite of Wallago attu (Siluridae), G. macrones Woodland, 1924 typical of Sperata seenghala (Bagridae), both species characterized by the possession of two circles of hooks on the rostellum-like organ and several rows of hooklets on the anterior margins of suckers; G. agraensis Verma, 1928 from W. attu (typical host), which has the scolex with only one circle of hooks and 1–3 incomplete rows of tiny hooklets on the suckers; and G. vachai (Gupta and Parmar, 1988) n. comb. from several catfishes, which possesses 4–6 circles of hooks and 5–11 rows of hooklets on the anterior half of suckers. Scolex morphology, including surface ultrastructure (microtriches), of all but one species (G. vachai) is described for the first time using scanning electron microscopy. A phylogenetic analysis based on the partial sequences encoding the large nuclear ribosomal subunit RNA gene has shown that three Indomalayan species, namely G. bengalensis, G. macrones and G. vachai, form a monophyletic group within Gangesia, whereas G. agraensis tends to form a clade with the Palaearctic species of the genus. A table with differential characters of all species from the Indomalayan Region is also provided together with a key to identification of genera of the subfamily Gangesiinae. The present study demonstrates that species of Silurotaenia do not occur in the Indomalayan region.

Subsequently, a number of species have been described from the same fish hosts, especially from W. attu and S. seenghala (see Tables 1-3). However, descriptions of these species were inadequate and were based on apparently decomposed specimens, often with missing hooks due to post mortem disintegration of tissues, including detachment of the tegument and surface structures (hooks, hooklets and microtriches) (see, e.g., [7]). In addition, type specimens or vouchers of all but one (G. sindensis) taxa are not known to exist and all requests for their loan have remained unanswered, which casts serious doubts upon reliability of these species descriptions.
Recently, extensive materials of Gangesia tapeworms were collected from W. attu, S. seenghala and Mystus spp. from Bangladesh, Cambodia and India. This new, well fixed and Table 1. Selected morphological characteristics of Gangesia bengalensis (Southwell, 1913) and its synonyms.   properly processed material made it possible to critically assess the taxonomic status of tapeworms from these hosts and to redescribe the species that are considered to be valid. New morphological data, including those on scolex ultrastructure as revealed by scanning electron microscopy (SEM), are provided (see Table 4). For the first time molecular data of newly collected material of these species were used for confirmation of conspecificity of individual taxa and to assess their phylogenetic relationships. Fish examined were sold in fish markets and/or from fishermen. This implies they were aimed at human consumption and thus no permission was needed to get these fish that were caught just for commercial purposes. In some cases (big Wallago attu catfish), only guts were purchased from fish sellers. Fish that were still alive when examined were killed by dorsal pithing (spinal cord and blood vessels cutting immediately behind the head), which is a method allowed by the law of the Czech Republic, European Directive and local animal care committees.

Materials and Methods
Tapeworms, obtained from live or fresh fish, were gently rinsed in 0.9% NaCl solution. A small piece, usually a few posteriormost proglottides, was cut off and placed in molecular-grade ethanol for DNA sequencing (see below). The worms were then placed into a small amount of saline in a beaker or big vial, and hot (boiling) 4% formaldehyde solution ( = formalin) was added, which kept the worms relaxed, not contracted or deformed (see Oros et al., 2010 [8] for more data on this fixation procedure). After 2-3 weeks, formalin was replaced by 70% ethanol for storage before further processing of specimens, i.e. staining, sectioning or preparation for scanning electron microscopical (SEM) observations. For light microscopy, specimens were stained with Mayer's hydrochlorid carmine, destained in 70% acid ethanol (i.e. ethanol with several drops of HCl), dehydrated through a graded ethanol series, cleared in clove oil (eugenol), and mounted in Canada balsam as permanent preparations [9]. Pieces of strobila and scoleces were embedded in paraffin wax, sectioned at 12-15 mm (cross sections of strobila and longitudinal sections of scoleces), stained with Weigert's haematoxylin and counterstained with 1% acidic eosin B solution [10]. Illustrations were made using a drawing attachment of an Olympus BX51 microscope with the use of Nomarski interference contrast. Measurements were taken using Olympus Image-Pro programme. Eggs from the uterus of unstained, unmounted worms were measured and illustrated in the water. For SEM observations, specimens were dehydrated through a graded ethanol series, transferred to hexamethyldisilazane (HMDS -see [11]), dried in air, sputtered with gold (approximately 10 nm thick) and examined with a Jeol JSEM 7401F microscope.
In morphological descriptions, measurements are usually those taken from mature segments (only if the number of mature proglottides was too low, measurements were taken from the first pregravid proglottides) and are in micrometres (mm) unless otherwise stated; classification of microtriches follows Chervy [42]. The scientific and common names of fish hosts follow the FishBase online database [43]. Abbreviations of the terms used in redescriptions are as follows: n = number of measurements; L/ W = length/width ratio; O/P = ratio of the width of the ovary and width of the proglottis; C/P = ratio of the length of the cirrus-sac and width of the proglottis; R/S = ratio of the diameter of a rostellum-like organ and diameter of the suckers.
Twenty-eight new sequences of the large subunit nuclear ribosomal RNA gene (lsrDNA or 28S rDNA) of 9 nominal taxa were characterized with focus on the Indomalayan Gangesia species but also included other genera of the Gangesiinae (see Table 5). Genomic DNA was extracted using a standard phenol-chloroform protocol [44] from 96% ethanol preserved samples. The D1-D3 lsrDNA region was amplified by PCR using the primers and conditions described in Brabec et al. [45]. All products were verified on a 1% agarose gel and purified using exonuclease I and shrimp alkaline phosphatase enzymes [46]. BigDyeH Terminator v3.1 cycle sequencing kit and PRISM 3130xl automatic sequencer (Applied Biosystems) were used for bidirectional sequencing of the PCR products using the set of PCR and internal sequencing primers (see [45]). Newly characterized sequences were deposited in the GenBank under the Accession Nos. JX477427-JX477454. Sequences were assembled and inspected for errors in Geneious Pro 5.3.6 [47], aligned using the E-INS-i algorithm of the program MAFFT [48] and the ambiguously aligned positions were manually excluded from resulting alignments in MacClade 4.08 [49]. The phylogenetic relationships were evaluated under the maximum likelihood (ML) criteria in the program RAxML 7.2.8 [50], [51], employing the GTR+C substitution model. All model parameters and bootstrap nodal support values (1000 repetitions) were estimated using RAxML. An alternative hypothesis than the one represented by the topology of the best ML tree was tested using the approximately unbiased (AU) test implemented in the program Consel 0.20 ([52], [53]). ML branch lengths of the constrained topology as also the per-site log-likelihood values of both unconstrained and constrained trees were computed in RAxML using the ML settings described above. Subsequently, p values of different likelihood-based tests were calculated with Consel.

Species Composition and Differential Criteria
A critical review of literature (papers from years 1948-2012) has shown that: (1) most specimens used for descriptions of new taxa were decomposed; (2) descriptions were incomplete; e.g., cross sections were always missing, data on the number of uterine branches were often omitted, etc.; (3) many data were undoubtedly erroneous and individual structures were misinterpreted, e.g., vitelline follicles were counted as testes and muscle cells were considered to be hooks, which were in fact lost due to detachment of the tegument; (4) data in the text did not correspond to those inferred from illustrations, which were always very schematic and incomplete; (5) species were differentiated on the basis of questionable taxonomic characters, such as negligible differences in the number of testes (often erroneous) and unspecified shape of the cirrus-sac; (6) types are not known to exist and never were available upon request for all but three species of Gangesia and Silurotaenia described from the Indian subcontinent; and (7) identification of some fish hosts is questionable.
To clarify this confused situation, the following approach was applied (see also [54]): (1) new, well fixed material was compared with the oldest available descriptions to avoid other inflation of new names in the literature; and (2) the following key taxonomic (differential) characteristics were selected to assess the validity of inadequately described species (see Table 4 and Remarks on individual species): (i) number of rows of rostellar hooks and their size; (ii) number of rows of hooklets on the suckers; (iii) ratio of the width of the rostellum-like organ and diameter of the suckers; (iv) relative length of the ovary, i.e. ratio of its length to the length of proglottides; (v) width of the scolex; (vi) diameter of the suckers; and (vii) relative width of ventral osmoregulatory canals, i.e. ratio of their width to the width of the proglottis.
Examination of newly collected material has shown that only four species, instead of 48 nominal taxa, of Gangesia and Silurotaenia, namely Gangesia bengalensis (Southwell, 1913) (type-species of the genus), G. agraensis Verma, 1928, both parasitizing most commonly Wallago attu; G. macrones Woodland, 1924, a typical parasite of Sperata seenghala; and G. vachai (Gupta and Parmar, 1988) n. comb., from catfishes of several families, are valid. Their redescriptions based on type materials, if available, and newly collected specimens are provided below. The taxonomic status of some species, the descriptions of which were not available, could not be clarified (see ''Taxa of unclear status'').
Vitelline follicles medullary with some follicles paramuscular (penetrating between muscle fibres of inner longitudinal musculature), in two longitudinal bands on both sides of proglottis, occupying almost its total length (95-99% on poral side, 94-99% on aporal side in mature proglottides; n = 10); bands interrupted at level of terminal genitalia on ventral side ( Fig. 1D-F), with few follicles on dorsal side. Uterus medullary, with development of type 1 according to de Chambrier et al. [55], defined as follows: in immature proglottides, uterine stem present as longitudinal concentration of chromophilic cells along median line. Lumen of uterus appears in last premature proglottides, gradually extending to form thick-walled tubular structure. Eggs appear simultaneously with formation of lateral, thin-walled diverticula. In pregravid proglottides, lateral diverticula remain thin-walled, 18-25 in number on each side (n = 20), occupy up to 62% of proglottis width (Fig. 1F), may partially overlap ovary (Fig. 1F). Uterus with slit-like openings (Fig. 1F).
Woodland [1] in fact studied two separate species, one corresponding to O. bengalensis, whereas the other representing undescribed species, for which Verma [2] proposed the name Gangesia agraensis. Examination of type specimens of G. wallago by the present authors confirmed that Woodland [1] actually described two different species under the name G. wallago (see Fig. 3A-C). Southwell [5] listed Gangesia bengalensis as the type species of the genus, but synonymized G. agraensis with G. bengalensis, which was then accepted by some authors (e.g. Yamaguti [57], Schmidt [6]).
Newly collected material from India (Assam, Maharashtra and West Bengal) enabled us to redescribe G. bengalensis, which is typified by the following morphological characteristics: (i) the rostellum-like organ is armed with two rows of hooks of a similar length (36-38 mm long), 47-54 in number (upper row with 24-28 and lower row with 23-26 hooks), with broader hooks in the anterior (upper) circle; (ii) the outer rim of the anterior half of the suckers is covered with minute (length 3-4 mm) hooklets arranged in 5-8 rows (Figs. 1A; 2A,D); (iii) base of the hooks of the upper row consists of a massive base plate, much wider in the anterior part, with deep constriction in the middle, whereas the hooks of the lower row possess a narrow base plate widened posteriorly; (iv) vitelline follicles are medullary, with a few follicles paramuscular, i.e. penetrating between the inner longitudinal musculature (Fig. 1G,H); (v) cavity of suckers covered with acicular filitriches, with few tufts of capilliform filitriches; (vi) ventral osmoregulatory canals occupy 4-11% of the width of mature proglottides; (vii) the uterus possesses slit-like openings in pregravid proglottides.
The type host of G. bengalensis was not designated by Southwell [56], who listed snakehead murrel Channa striata (Channidae) and rahu Labeo rohita (Cyprinidae) as the definitive hosts of O. bengalensis. However, the former fish host was heavily infected, whereas tapeworms in L. rohita were few in numbers [56]. Southwell [58] then reported Wallago attu as an additional fish host. The absence of reliable records of G. bengalensis from C. striata and L. rohita indicates that these fishes were most probably only atypical or accidental hosts. The present authors dissected 32 specimens of C. striata from Bangladesh, India and Indonesia, and 12 specimens of L. rohita from Bangladesh and India but have never found any proteocephalidean cestode.
Except for G. wallago and G. sindensis, type material of any species of Gangesia described from W. attu and other fish hosts from India and Pakistan has not been available and it is not known to exist. Consequently, the validity of all species had to be assessed solely by comparison of the data from the original descriptions, which, however, suffered from many flaws and deficiencies (see below). As a result, most synonymies proposed in this paper rely on the characteristics considered to be diagnostic (species-specific) as listed above. In contrast, differential criteria used to distinguish allegedly new species of Gangesia and/or Silurotaenia were found to be doubtful and unreliable, which justifies proposed invalidation of all the species discussed below. The present authors are aware that such arbitrarily taken decision does not exclude the possibility that some of the taxa synonymized with G. bengalensis are valid, but their validity must be justified by reliable and convincing arguments based on adequate morphological descriptions of well-fixed specimens that are deposited in institutional collections and available upon request.
Singh [59] described Gangesia lucknowia from the schilbeid catfish Eutropiichthys vacha from Lucknow, Uttar Pradesh, India. The species was characterized by the possession of two rows of rostellar hooks of the same size and shape, which is also a typical feature of G. bengalensis. In addition, most differential characteristics overlap or are identical in both species (see Table 1 and the redescription of G. bengalensis above). Consequently, we accept Malhotra et al.'s [16] conclusion that G. lucknowia is a synonym of G. bengalensis.
Rehana and Bilqees [60] proposed a new species, G. sindensis, on the basis of two partially digested, strongly flattened and thus deformed specimens found in W. attu from Pakistan (see Fig. 1A-E in Rehana and Bilqees [60]). The main distinguishing features of the species were the alleged absence of hooklets on the suckers, number and size of the rostellar hooks (single row) and the number of uterine diverticula. However, examination of the existing syntypes (though designated as vouchers) of G. sindensis (BMNH 1982.5.13.27) has shown that suckers are in fact covered with hooklets (see Fig. 3D) and that there are two rows of rostellar hooks (see Fig. 3D,E and Table 1). Hence, G. sindensis is synonymized with G. bengalensis.
The same authors [61] described another new species, G. spinocirrosa, from the same host (W. attu) and locality from Pakistan, but this original description was not available to the present authors. However, R. Rehana provided a detailed description of G. spinocirrosa, including illustrations, in her PhD thesis [62]. She distinguished this species from congeners by the possession of a spiny cirrus. However, it is obvious from Figs. 106 and 107 in Rehana [62] that the worms were decomposed and muscles fibres of the cirrus were misinterpreted as spines. Otherwise, all other characters including the arrangement and number of hooks correspond to those of G. bengalensis (see Table 1). Therefore, G. spinocirrosa becomes a junior synonym of G. bengalensis, too.
Dhar and Fotedar [7] described G. kashmirensis from a catfish identified as Glyptosternum sp. from Kashmir, India and distinguished the new species from G. bengalensis (based on a very brief species diagnosis provided by Southwell [56]) by several characters, such as shape of the scolex, number of rows of rostellar hooks, number of testes, etc. (see Table I of Dhar and Fotedar [7]). In fact, G. kashmirensis does not differ from G. bengalensis as redescribed herein, which is obvious from comparison of their morphology and measurements (see Table 1), and it is also synonymized with this species. G. mehamdabadensis was described by Malhotra et al. [15] from Mystus tengara (misspelled as M. tengra in the original paper) from Gujrat, India. Morphological description seems to be comprehensive, but cross sections were not provided (even though they were mentioned in the Material & Methods section). The taxon is in fact identical with G. bengalensis in all but one characteristic, the only alleged difference being the arrangement of the hooks on the rostellum-like organ in one, instead of two rows. However, the specimens decribed by Malhotra et al. [15] were strongly flattened, as obvious from their Fig. 1. It has been observed by the present authors that double rows of hooks may appear as a single one in strongly flattened worms. Malhotra et al. [15] reported the holotype of G. mehamdabadensis to have been deposited in the Commonwealth Agricultural Bureaux Institute of Parasitology in St. Albans, UK, but no specimens are now in The Natural History Museum in London, where this collection is currently placed (E. Harris, Curator -pers. comm.). It is thus not possible to reliably count the number of rows of hooks in the species, which is invalidated as a junior synonym of G. bengalensis. In the same paper [15], a new species, Gangesia etawaensis, was mentioned, but it was apparently an error and G. mehamdabadensis ( = G. bengalensis) was incorrectly named. Since no description of the former species appeared in that article, G. etawaensis becomes a nomen nudum.
Gupta and Parmar [17] described G. indica from Wallagonia attu ( = Wallago attu) from Lucknow, India. Their new species is mainly characterised by the presence of 24-26 rostellar hooks arranged in two rows and suckers without hooklets. However, this does not correspond to figures: 12+13 hooks, i.e. 25 in total, are illustrated, but just on one side, which means that the total number of hooks is about 50. The hooks are of the same type, not of two types as written in the text. The hooks illustrated in Fig. 1 of Gupta and Parmar [17] are larger (length 25-28 mm) than mentioned in the text (length 10-25 mm only). Thus the species is in fact indistinguishable from G. bengalensis with which is synonymized.
Gangesia fotedari was described by Dhar and Majdah [18] from a catfish tentatively identified as Glyptothorax sp. from Kashmir, India. This cestode was differentiated from G. kashmirensis Dhar and Fotedar, 1979 ( = G. bengalensis; see above) by very few characters, namely the number of hooks and their size, and the size of the scolex (see Table 1). However, these allegedly differential characters should be considered with caution, because the characteristics reported in the morphological description by Dhar and Majdah [18] do not correspond to those illustrated in figures (see Table 1). Based on general resemblance of most morphological characteristics, G. fotedari is synonymized with G. bengalensis until adequate redescription of the former species, based on good quality material, confirms validity of this poorly described taxon.
Jadhav et al. [19] described G. paithanensis from Barbus ticto ( = Puntius ticto) from the Godavari River at Paithan, India. The description was very brief and incomplete (e.g., no data on eggs were provided) but it is evident that the specimens were digested and hooks detached due to decomposition (see Figs. A and B of Jadhav et al. [19]). Low number of the hooks (only 11!) reported in the paper is apparently incorrect. The authors observed two types of hooks, which is a characteristic typical of G. bengalensis, and their length was similar (16-18 mm). Other morphological features of G. paithanensis also correspond to those of G. bengalensis (see Table 1) and thus G. paithanensis is synonymized with the former species.
The same authors [21] described Silurotaenia tictoi from the same fish host (Puntius ticto) and locality (Paithan), but this description was based on completely digested material with all hooks lost (see Plate 4A in Shinde et al. [21]). It is evident from Plates 4B and 4C that the authors misinterpreted the muscles in the rim of the rostellum-like organ as hooks and vitelline follicles as testes. Morphological description was incomplete and illustrations erroneous (data on the uterus were missing and the uterus was not illustrated in Plate 4C, etc.). Type specimens of this species never existed (B. Jadhav -pers. comm.), but a large size of the scolex and relatively narrow ventral osmoregulatory canals indicate that G. paithanensis is conspecific with G. bengalensis, with which it is synonymized.
Hiware [29] described G. seenghali from Mystus seenghala ( = Sperata seenghala) from Satara, Maharashtra, India. Description of this species is very poor (e.g., vitelline follicles are missing in Fig.  C), but the hooks are identical with those of G. bengalensis (Fig. A) and also other morphological characteristics correspond to those of the latter species. Consequently, G. seenghali becomes a new synonym of G. bengalensis.
Hiware and Jadhav [27] described G. maharashtrii from W. attu from Satara, Maharashtra, India. Later, researchers from the same laboratory in Aurangabad proposed additional three species found in the same host (W. attu) and locality, namely G. dharurensis Jadhav and Tat, 1997; G. ramkaei Pawar and Hiware, 2008; and G. bendsurensis Reddy, Wankhede, Dhole and Anand, 2011. The validity of these taxa is questionable because the authors studied contracted and digested specimens (see Figs. A-D of Hiware and Jadhav [27]; Figs. A and B in Jadhav and Tat [28]; very lowquality photomicrographs of Reddy et al. [39]). Moreover, illustrations are very schematic and undoubtedly do not document the actual morphology of the worms, as obvious from various incorrect structures (see figures of Hiware and Jadhav [27]; Jadhav and Tat [28]; Pawar and Hiware [37]; Reddy et al. [39]), such as the shape of hooks, erroneous morphology of the cirrus-sac and genital atrium, extent and position of vitelline follicles, and shape of the eggs (Pawar and Hiware [37] reported the eggs to be elongate, resembling those of caryophyllidean or bothriocephalidean cestodes, whereas all species of Gangesia have spherical eggs - [3], [63]).
The morphological descriptions mentioned above also contain apparent mistakes, such as the absence of the neck in Hiware and Jadhav [27] and the diameter of the suckers larger than the width of the scolex [sic!] in Reddy et al. [39]. Discrepancies also exist between the text and figures: the number of hooks was reported in the text to be 40-45, but only 35 hooks were illustrated in Fig. A of Hiware and Jadhav [27], the number of the testes mentioned in the text (60-70) markedly differs from that illustrated by Jadhav and Tat [28] in Fig. B (32 testes only). As a rule, type specimens do not exist, as confirmed on site (Dr. Babasahed Ambedkar Marathwada University, Aurangabad, Maharashtra) by one of the present authors (M.O.). Since all four species mentioned above, i.e. G. bendsurensis, G. dharurensis, G. maharashtrii and G. ramkaei, are almost indistinguishable from G. bengalensis (see Table 1), they are synonymized with it unless new data based on well-fixed material provide unequivocal evidence of their validity.
Gangesia cirrhinae was described by Patel et al. [30] from the cyprinid fish Cirrhina mrigala ( = Cirrhinus cirrhosus) from Maharashtra, India. The original description was based on decomposed specimens (see Fig. 1A, B of Patel et al. [30]) and morphological description was very poor and erroneous, e.g., hooks on the rostellum-like organ were described to form as many as 5 rows (but only 27 hooks were reported to be present, which would imply that each row contains only 5-6 hooks, which is non-sense), hooklets on the suckers were not reported (they are in fact present in all species of Gangesia -[1], [2], [56], [57], [63]), no cross sections were provided and the number of uterine diverticula was not reported. However, it is obvious from Fig. 1B of Patel et al. [30] that the hooks are arranged in two rows, which is a typical feature of G. bengalensis. Even though conspecificity of both species cannot be confirmed (type specimens were never deposited -B. Jadhav, pers. comm.), G. cirrhinae is synonymized with G. bengalensis.
Shinde et al. [31] very briefly described G. rohitae from Labeo rohita from Maharashtra, India. It seems that the description was based on immature specimens (the uterus was not yet developedsee Fig. B in Shinde et al. [31]) and it is apparently erroneous in several characteristics (e.g., the number of hooks was mentioned to be 60 in the text whereas only 44 hooks, i.e. 25 in the upper row and 19 in the lower row, were illustrated - Fig. A). Based on the presence of two rows of hooks of similar length and other morphological characters (see Table 1), G. rohitae is considered to be conspecific with G. bengalensis.
In 2004, G.B. Shinde with co-authors [34] described another new species from the same host (L. rohita) and locality in Maharashtra and proposed the identical name [sic!], i.e. Gangesia rohitae. This implies that Gangesia rohitae Pawar, Lakhe, Shinde and Patil, 2004 becomes a homonym of G. rohitae Shinde, Mahajan and Begum, 1999, the latter taxon being a synonym of G. bengalensis.
Bhure et al. [40] reviewed the species of Gangesia from freshwater fishes from Maharashtra, India, and described another new species, G. marathwadaensis from W. attu. The description is very short (no measurements of any structure are given), erroneous (description of G. mastacembali Wankhede, 2005 is provided instead of that of G. marathwadaensis) and supplemented with very schematic and uninformative line drawings (see Figs. 2Q and 3Q in [40]). The most characteristic feature of this species, i.e. the number of hooks, corresponds to that of G. bengalensis, with which it is now synonymized.
Bhavare and Shukla [41] described a new species G. jayakwadensis from walking catfish Clarias batrachus from Aurangabad, India, but the description is brief, with several errors and illustrations are very schematic. However, the diagnostic characteristics such as a double circle of rostellar hooks and their number (see Table 1) correspond to those of G. bengalensis, with which it is synonymized.
Chavan [64] proposed two new species, G. attui and G. parbhaniensis, from W. attu, both being indistinguishable from G. bengalensis (see Plates iv and ii of Chavan [64]). However, descriptions of both species were presented in an unpublished thesis and thus G. attui and G. parbhaniensis become nomina nuda.
Fernando and Furtado [65] reported G. bengalensis from W. attu from northern Ceylon (Sri Lanka), but they in fact found two species, G. bengalensis and G. agraensis, as obvious from their morphological descriptions and illustrations (Figs. 1g and 1h in Fernando and Furtado [65]). Gupta and Arora [14] misidentified specimens of G. bengalensis from W. attu as G. lucknowia and G. macrones, as obvious from comparison of their morphology, including the armature of the scolex (see Figs. 4, 5, 15 and 16 of Gupta and Arora [14]).
Vitelline follicles medullary, in two longitudinal bands on both sides of proglottis, occupying almost its total length in mature proglottides (96-99% on poral side, 95-99% on aporal side; n = 8); bands interrupted at level of terminal genitalia on ventral side, with few follicles on dorsal side (Fig. 4D).
Uterus medullary, with development of type 1 according to de Chambrier et al. [55], defined as follows: in immature proglottides, uterine stem present as longitudinal concentration of chromophilic cells along median line. Lumen of uterus appears in last premature proglottides, gradually extending to form thick-walled tubular structure. Eggs appear simultaneously with formation of lateral, thin-walled diverticula. In pregravid proglottides, lateral diverticula remain thin-walled, encircled by thick chromophilic cell layer, 14-20 in number on each side (n = 20), occupy up to 63% of proglottis width (Fig. 4I), may completely overlap ovary (Fig. 4I). Uterus opened with 5-8 pore-like openings (Fig. 4I).
Remarks: The species was described by Verma [2], who first realized that two species of Gangesia, differing from one another by the number of rows (circles) of hooks on the rostellum-like organ and other morphological characteristics, may occur simultaneously in Wallago attu. Verma [2] provided a reasonably good morphological description of G. agraensis, but his specimens were contracted and strongly flattened (see his Figs. 38-40 and 43-48). However, Southwell [5] synonymised G. agraensis with G. bengalensis, which was followed by most subsequent authors (see below and remarks on G. bengalensis).
The present study provides unequivocal support of the validity of G. agraensis based on morphological and molecular data and thus fully confirms Verma's [2] conclusions. The present study also provides new information on the morphology of G. agraensis: (i) the anterior half of the outer rim of the sucker is covered by 2-3 rows of small hamulate shaped hooklets, whereas Verma [2] reported only one, occasionally two rows of minute spines; (ii) ventral osmoregulatory canals are markedly wide, occupying 16-22% of the width of mature proglottides (wide canals were observed in all specimens, which were fixed in the same way as those of G. bengalensis, in which ventral canals are always much narrower; this implies that the canals in G. agraensis were not inflated during fixation); (iii) base of hooks on the rostellum-like organ is almost rectangular in shape with shallow constriction in the centre; (iv) vitelline follicles are exclusively in the medulla, unlike those of G. bengalensis, G. macrones and G. vachai, in which some follicles are paramuscular; (v) suckers are uniformly covered with papilliform filitriches.
Since Verma's [2] comprehensive account, a number of new species have been described from the Indian subcontinent, especially from Maharashtra. Unfortunately, all these descriptions were incomplete and/or erroneous and were always based on deformed (strongly compressed) or macerated (with detached hooks and disintegrated tissues) specimens. Type material of none of these species, which seem to be indistinguishable from the Verma's taxon, is available and it is not known whether it actually exists (any request for specimens to the authors of individual taxa has never been replied).
G. haryanae was described by Gupta and Arora [14] from W. attu from Ambala, India. Diagnostic features of G. haryanae are irregularly arranged 2-20 [sic!] hooks on the rostellum-like organ and suckers without hooklets, but it is evident that the specimens were partially digested and hooks and hooklets on the suckers were lost due to tissue decomposition (only six hooks irregularly scattered on the scolex are present in Fig. 1 of Gupta and Arora [14]). The description was very brief and incomplete (e.g., the information on uterine branches is missing). Except for the absence of hooklets on the suckers, which is undoubtedly a result of disintegration of the tegument, G. haryanae possesses all characteristics that are typical of G. agraensis. Therefore, both taxa are considered to be conspecific and G. haryanae becomes a junior synonym of G. agraensis.
Malhotra et al. [16] described G. sanehensis from the cyprinid Cirrhina mrigala ( = Cirrhinus cirrhosus) and from W. attu from Garhwal ( = Uttarakhand), India. Morphological description appears to be detailed but illustrations are very schematic and do not provide sufficient evidence about reliability of morphological data. For example, hooks on the rostellum-like organ were not illustrated in detail to confirm their shape and unusually high size variability, which was not observed in any other species. The illustration of a mature proglottis (Fig. 3 of Malhotra et al. [16]) was insufficient because it did not provide accurate information on the distribution and number of testes, internal structure of the cirrus-sac, position (extent) of vas deferens and other features. Vitelline follicles in a gravid proglottis (Fig. 4 of Malhotra et al. [16]) were not correctly illustrated and the ovary was missing. Despite these obvious mistakes and deficiencies in the morphological description, it is unquestionable that G. sanehensis shares key diagnostic characteristics with G. agraensis, i.e. a single row of hooks of identical shape, suckers armed with a few rows of hooklets, a large, massive strobila, and a similar number of testes (see Table 2). Consequently, G. sanehensis is synonymized with G. agraensis.
Gangesia shindei was described by Deshmukh and Shinde [23] from the barb Barbus ticto ( = Puntius ticto) from Aurangabad, India and the species was named after one of the authors themselves (!). Species description was very brief and incomplete (e.g., there are no data on uterine branches and eggs). It is also evident from the very schematic figures that the figured specimen was partially decomposed and extremely flattened (see Fig. 1 of Deshmukh and Shinde [23]). However, the characteristics considered to be species-specific (see above -Species composition and differential criteria), such as a single row of hooks on the rostellum-like organ, their number and size, and the number of testes, are fully consistent with those of G. agraensis (see Table 2). Thus G. shindei is synonymized with G. agraensis.
Jadhav et al. [32] described G. clariusae from the walking catfish, Clarias batrachus, from Karnataka, India, and Begum [36], who worked in the same laboratory (Aurangabad), described G. batrachusi from the same host and locality. The descriptions of both taxa were based on decomposed specimens and were very brief and erroneous (e.g., subtegumental cells were misinterpreted as vitelline follicles, see Fig. C of Jadhav et al. [32] and Fig. 1C of Begum [36]). Figs. 1A and 1B in Begum [36] and Fig. A in Jadhav et al. [32] clearly show that only one row of hooks is present in both species and that their base is rectangular (see Fig. B of Jadhav et al. [32] and Fig. 1B of Begum [36]), which fully corresponds to the morphology of G. agraensis (see Table 2). Based on this morphological similarity, both the taxa are synonymised with G. agraensis.
Hiware et al. [33] described G. ambikaei from W. attu from Maharashtra, India. Description was based on material of very poor quality as obvious from the illustrations provided (see Figs. A-C of Hiware et al. [33]). Hooks were illustrated very schematically (Fig. B of Hiware et al. [33]) and the uterine stem was apparently incomplete; vitelline follicles and vas deferens were almost completely missing (Fig. C of Hiware et al. [33]). Due to poor quality of the material and incomplete morphological description, it is difficult to rely on the data provided (type specimens do not exist -B. Jadhav, pers. comm.). In fact, some measurements were apparently erroneous, as obvious from discrepancies between the text and figures (some testes are illustrated as postovarian in Fig. C of Hiware et al. [33], which is apparently error because no species of Gangesia has testes posterior to the ovary). The position of the hooks indicates that only one row was present, but their number was reported to be slightly higher than that of G. agraensis. Despite this negligible difference, both species are considered conspecific and G. ambikaei becomes a new synonym of G. agraensis.
Fotedar and Dhar [12] described G. jammuensis from W. attu from Jammu and Kashmir, India. This species is in fact G. agraensis, and was already invalidated by Ash et al. [66]. Chavan [64] proposed a new species, G. wallagoi, which is indistinguishable from G. agraensis (see Plate III of Chavan [64]). However, description of the former species has never been published and G. wallagoi should be considered as nomen nudum.
Shinde et al. [13] described Silurotaenia nybelini from Pseudeutropius taakree ( = Proeutropiichthys taakree) from Maharashtra, India, but there was conspicuous discrepancy between the text and figures: the text claims the presence of minute hooks on the rostellar-like organ, but they were not illustrated -see Fig. A in Shinde et al. [13]. There were also no traces of hooklets on the suckers. It seems that the authors observed Gangesia specimens devoid of hooks and hooklets due to their post mortem decomposition. They also provided erroneous data, e.g., measurements of the rostellum-like organ (470 mm6720 mm) are greater than the diameter of the whole scolex (140 mm6100 mm) and the suckers are much bigger in Fig. A than those described in the text. Silurotaenia nybelini possesses significantly wide ventral osmoregulatory canals (see Figs. B, C of Shinde et al. [13]) and also other characteristics correspond to those typical of G. agraensis (see Table 2). Based on this similarity, S. nybelini is considered to be a synonym of G. Redescription (based on ten specimens collected from S. seenghala from Bangladesh and India, including three scoleces observed in SEM): Strobila with acraspedote proglottides, more than 65 mm long (28-56 mm according to Woodland, 1924 [1]) and up to 1.15 mm wide (a little over 1 mm according to Woodland, 1924 [1]). Proliferative zone 3.6-4.3 mm long and 165-195 wide (n = 4). Strobila consists of more than 110 proglottides (n = 3): 80-86 immature, only 3-4 mature, 10-20 pregravid and gravid (actual number is higher because posteriormost proglottides were cut for molecular analyses). Proglottides variable in shape, mature proglottides usually wider (0.96-1.16 mm) than long (0.63-1.10 mm; n = 8), but gravid proglottides longer (1.4-3.1 mm) than wide (0.73-1.2 mm; n = 15).
Vitelline follicles medullary, with some follicles paramuscular (penetrating between muscle fibres of inner longitudinal musculature), in two longitudinal bands on both sides of proglottis, occupying almost its total length (94-95% on poral side, 93-95% on aporal side in mature proglottides; n = 8); bands interrupted at level of terminal genitalia on ventral side, with few follicles on dorsal side (Fig. 6E).
Uterus medullary, with development of type 1 according to de Chambrier et al. [55], defined as follows: in immature proglottides, uterine stem present as longitudinal concentration of chromophilic cells alongside median line. Lumen of uterus appears in last premature proglottides, gradually extending to form tubular structure. Eggs appear simultaneously with formation of lateral, thin-walled diverticula. In pregravid proglottides, lateral diverticula remain thin-walled, 25-35 in number on each side (n = 25), occupy up to 83% of proglottis width (Fig. 6F), may partially overlap ovary (Fig. 6F). Uterus opened with 7-10 pore-like openings (Fig. 6F).
Type-host: Sperata seenghala (Sykes, 1839) (Siluriformes: Bagridae). Distribution: Bangladesh and India (Maharashtra, Uttar Pradesh). Remarks: Woodland [1] described G. macrones from seven or eight flattened specimens (but only with two scoleces) from the giant river catfish Sperata seenghala. The original description was not complete (e.g., no data on eggs were provided) and contained some errors (e.g., the number of hooks reported in the text was 33 but as many as 47 hooks were illustrated in Fig. 26 of Woodland  [1]). Material of G. macrones recently collected in Maharashtra, India enabled us to provide new morphological data on this cestode, which is typified by a number of morphological characteristics: (i) rostellum bears a double circle of hooks: the upper row with 38-40 larger hooks (10-13 mm long) and the lower row with 42-45 smaller hooks (5-7 mm long); (ii) the outer rim of the anterior half of the suckers is covered by minute (length 1-2 mm) hooklets arranged roughly in 9-12 rows (Figs. 6A; 7C); (iii) rostellum-like organ is markedly larger (1.3-2.0x) than the suckers; (iv) scolex is comparatively small (160-220 mm long by 215-275 mm wide), similarly as suckers (80-90 mm in diameter); (v) testes (90-140 in number) form a single layer; (vi) ovarian region is comparatively short, occupying at maximum 1/4 of the length of mature proglottides; (vii) cirrus-sac is long and narrow, the lengthwidth ratio being 3.1-3.7; (viii) suckers are covered with capilliform filitriches; (ix) vitelline follicles are medullary, with some follicles paramuscular, i.e. penetrating between the inner longitudinal musculature (Fig. 6G,H).
As many as five Silurotaenia spp., namely S. paithanensis Shinde, Majid and Solunke, 1983; S. macroni Shinde, Kadam and Jadhav, 1984; S. singhali Shinde, Kadam and Jadhav, 1984; S. shastri Gavhane and Jadhav, 1991; and S. raoii Bhure, Pathan, Nanware and Dhondge, 2010, have been described from S. seenghala from India, all from the same state (Maharashtra). Descriptions of these species were very brief, incomplete (e.g., cross sections necessary for confirmation of the position of internal organs were never  , and the number of uterine branches was not reported in all but one species; see also Table 3) and often erroneous (e.g., all species were reported to possess the suckers devoid of hooklets).
It is obvious from the figures in the original descriptions that all the above-mentioned Silurotaenia spp. were described from decomposed specimens that were also strongly flattened and thus deformed. Some of the authors of these taxa also misinterpreted decomposed muscle fibres (see Figs. 1A and B of Shinde et al. [20]; Plates 1A and 2A of Shinde et al. [21]) or gland cells (see Fig. A of Gavhane and Jadhav [26] and Fig. 2A of Bhure et al. [38]) as hooks on the rostellum-like organ. One of the present authors (S.P.C.) fixed originally a few decomposed specimens of G. macrones from the same host (S. seenghala) and same state (Maharashtra) in cold formalin. These worms, including their scoleces, showed a similar morphology (Fig. 6C) to that of Silurotaenia spp. listed above. In fact, all the above-mentioned Silurotaenia spp. possess all characteristic features of G. macrones, such as the shape and size of the scolex, which is small and bears conspicuous suckers, a large rostellum-like organ (its diameter is markedly larger, up to twice, than the diameter of the suckers), and a relatively short ovary (ratio of its length to the length of proglottides is about 1/4 only; see Table 3). Based on this morphological similarity, Silurotaenia macroni, S. paithanensis, S. raoii, S. shastri and S. singhali are synonymized with G. macrones.
Two more species of Silurotaenia were described from Maharashtra, both from fish hosts not typical of proteocephalidean cestodes, namely S. barbusi from the barb Barbus ticto ( = Puntius ticto) (Cyprinidae) by Shinde et al. [21], and S. behairvnathi from the zigzag eel Mastacembelus armatus (Mastacembelidae) by Deshmukh and Shinde [24]. Similarly as the previously described taxa, these species were described very briefly and from decomposed material (see Plates 3, 4 of Shinde et al. [21] and Figs. A-C of Deshmukh and Shinde [24]). Both descriptions suffered from many flaws and deficiencies, e.g., muscle fibres or gland cells were considered to represent hooks on the rostellum-like organ (see Plate 3A and 4B of Shinde et al. [21] and Fig. A of Deshmukh and Shinde [24]), subtegumental cells as vitelline follicles (see Plate 3B of Shinde et al. [21]) and vitelline follicles were completely missing in illustrations (Figs. B, C of Deshmukh and Shinde [24]). Moreover, measurements provided by Deshmukh and Shinde [24] were obviously erroneous, e.g., the diameter of suckers (23 mm) was lower than the diameter of testes (27-53 mm) and the hooks were longer (71 mm) than the diameter of the rostellum-like organ (45 mm642 mm). If doubtful characteristics and measurements are not considered, S. barbusi and S. behairvnathi possess all features typical of G. macrones, such as a small scolex, large rostellum-like organ, and similar strobilar morphology (see Table 3). For this reason, S. barbusi and S. behairvnathi are synonymized with G. macrones.
Wankhede [35] described G. mastacembali from Mastacembelus armatus (misspelled as Mastacembalus armatus in the original paper) from Aurangabad, but no illustrations were provided. Bhure et al. [40] provided two figures of this species, which clearly show that the scolex is small and possesses a large apical organ (muscle fibres of the decomposed scolex were apparently considered to be hooks). Scolex morphology and low number of testes correspond to those of G. macrones and thus G. mastacembali is synonymized with the former species. Redescription (based on five specimens from W. attu and Mystus spp. from Bangladesh and India; measurements from the original description [22] in parentheses): Strobila with acraspedote proglottides, more than 25 mm long and up to 660 (600-680) wide. Proliferative zone about 0.68-1.10 mm long and 130-190 wide (n = 2). Strobila consists of more than 55 proglottides: more than 41 immature, only 3-4 mature, 5-10 pregravid and gravid (actual number is higher because posteriormost proglottides were cut for molecular analyses). Proglottides variable in shape, mature proglottides 505-630 long and 555-660 wide (n = 4).
Retractor muscle bundles well developed, surround rostellumlike organ, joined each other at level of suckers, form wide band at neck region (Figs. 8A,D; 9D).
Vitelline follicles medullary, with some follicles paramuscular (penetrating between muscle fibres of inner longitudinal musculature), in two longitudinal bands on both sides of proglottides, occupying almost their total length (96-97% on poral side, 94-96% on aporal side in mature proglottides); bands interrupted at level of terminal genitalia on ventral side, with few follicles on dorsal side.
Eggs with hyaline, spherical outer envelope, 65-95 in diameter (measured in eggs from uterus in distilled water; n = 30); embryophore thick (4-7 wide), spherical, 30-35 in diameter, consisting of two layers; outer layer thinner than nuclei-containing inner envelope; oncosphere spherical, 20-25 in diameter, with three pairs of embryonic hooks, 7-10 long (Fig. 8C).   Remarks: Specimens recently found by the present authors are considered to be conspecific with Silurotaenia vachai based on their morphological similarity. This species was described by Gupta and Parmar [22] from the schilbeid catfish Eutropiichthys vacha from Lucknow, India.

Taxonomic Summary
It actually somewhat resembles Silurotaenia siluri, a parasite of European wels (Silurus glanis L.), in its morphology (see [70] and [71]). However, it is transferred to Gangesia as G. vachai n. comb. because it shares the following morphological characteristics with the remaining species of Gangesia: (1) the anterior rim of suckers is covered with several rows of small hooklets, which are markedly different from spinitriches that cover the surface of the suckers (in S. siluri, the suckers are covered with spinitriches only -see Figs. 1, 2 in Scholz et al. [71]); (2) the rostellum-like organ is connected with well-developed retractor muscles that form a wide band of muscles (only lateral retractor muscles are present in the scolex and neck region in S. siluri -see Figs. 1A-C in Scholz et al. [71] and Fig. 9E; (3) ventral osmoregulatory canals are median (internal) to vitelline follicles (versus lateral, i.e. external, to vitelline follicles in S. siluri); (4) rostellum-like organ is as large as, or slightly larger than, the suckers (versus always markedly smaller in S. siluri) [71].
In addition, S. vachai forms a strongly supported clade with Gangesia macrones and G. bengalensis in molecular analyses, whereas S. siluri is a sister taxon of the clade formed by all species of Gangesia, including S. vachai (Figs. 10, 11). Therefore, molecular data support transfer of S. vachai to Gangesia, which well corresponds to its distribution, i.e. the Indomalayan Region (S. siluri occurs in the European part of the Palaearctic region).
Gangesia vachai is morphologically well differentiated from other Gangesia spp. by the presence of 4-6 irregular rows of rostellar hooks. Otherwise, it possesses similar proglottides and a small scolex like G. macrones, but can be easily differentiated by having a comparatively longer ovary (occupying about 1/3 of mature proglottid length) than the latter species (see Table 4).
The present authors found only two intact worms, but from two different hosts and localities, one from Durgapur, Bangladesh (field no. BAN 186) and another from Phulbari dam, Siliguri, India (field no. IND 906). They slightly differ in some morphological details (shape of the scolex, arrangement of hooks and distribution of hooklets of suckers; see Fig. 8A,D) and their lsrDNA sequences differed in 3 bp (0.2%), which is considered to represent intraspecific variability.

Taxa of Uncertain Status
The systematic position and validity of three species of Gangesia and Silurotaenia described from India could not be confirmed. Their original descriptions were not available despite numerous requests to the editorial offices of relevant journals, the authors of the papers, and exhaustive search in internet databases and major libraries. Until these descriptions are available (type material is not known to have been deposited), the validity of each of the following species remains doubtful: (1) Gangesia hanumanthai  Chavan [64] proposed two Silurotaenia species, namely S. makniensis and S. gangakhedensis from W. attu, in his unpublished PhD thesis and, thus, both of them are invalid according to the International Code of Zoological Nomenclature [72].
Wankhede [35] listed the species Gangesia godavari Kadam et al. 1983, but did not provide citation of the original description, which was unavailable to the present authors. In contrast, Bhure et al. [40] listed Jadhav, Shinde and Kadam (1983) as authorities of this species, but this paper [19] does not include any mention of G. godavari. However, the species is indistinguishable from G. agraensis as obvious from its brief description and figures (Figs. 2H and 3H) provided by Bhure et al. [40]. Therefore, G. godavari is considered to be a synonym of G. agraensis.

Phylogenetic Relationships
Twenty-three 1556-1644 bp long lsrDNA sequences of Gangesia have been characterized within the scope of this study, including five samples of Gangesia spp. from other than the Indomalayan zoogeographical region. In addition, species of all remaining genera of the subfamily Gangesiinae (Electrotaenia Nybelin, 1942;Postgangesia Akhmerov, 1969; Ritacestus de Chambrier, Scholz, Ash and Kar, 2011; Vermaia Nybelin, 1942) have been sequenced and their sequences combined with those available in GenBank. Whereas no intraspecific variability was observed within G. bengalensis (n = 4; all samples from India) and G. macrones (n = 4; Bangladesh and India), sampled specimens of G. agraensis (n = 8; Cambodia and India) and G. vachai (n = 2; Bangladesh and India) differed in 1-5 and 3 bp, yielding 3 and 2 unique lsrDNA sequences per given species, respectively (see inset table in Fig. 10). Specimens of G. oligonchis from Russia (n = 5) were identical.
ML analyses of the Gangesia specimens studied (Fig. 10) confirmed the monophyletic status of each of the four redescribed Gangesia species from Indomalayan region. However, the observed genetic distances within individual species were not consistent (see inset table in Fig. 10). The morphologically noticeably distinct species G. bengalensis, G. macrones and G. vachai differed in a relatively low (4-6) number of nucleotides, but G. agraensis differed in 51-64 bp. The Palaearctic species (G. oligonchis and G. parasiluri) differed in 17 bp, whereas differences of these taxa from those of the Indomalayan region were markedly higher (36-51 bp from G. agraensis and 56-81 from the three remaining species).
The phylogenetic analysis of the representatives, usually typespecies, of all genera of the subfamily Gangesiinae (Fig. 11) confirmed the monophyly of the genus Gangesia, represented here by six of nine currently valid species (see Discussion). Palaearctic Silurotaenia siluri appeared as a sister lineage to the Gangesia clade, Electrotaenia malopteruri then being a well-supported sister lineage of the two. Monophyly of the Indomalayan lineages of Gangesia, as detected on Fig. 10, was not revealed by the phylogenetic analysis involving all Gangesiinae. Here (Fig. 11), G. agraensis rather grouped with the Palaearctic Gangesia species, although without statistical support. Subsequent statistical topology test for a hypothesis constraining G. agraensis as a sister lineage to the rest of Indomalayan Gangesia species did not reject this scenario since the p value of the AU test reached a non-significant value of 0.423 (rejection at p,0.05).

Discussion
The present study has revealed that the number of nominal species of proteocephalidean cestodes described as members of Gangesia and Silurotaenia from freshwater fishes in the Indomalayan region conspicuously exceeds that of actually valid taxa, similarly as in cestodes of the order Caryophyllidea (see [54], [73]). Instead of 48 spp. in two genera, only four species of Gangesia are considered to be valid. As many as 41 species are either invalidated as synonyms of one of the three species described between 1913 and 1928 or represent nomina nuda. It is also concluded, and molecular data strongly support this assumption, that no species of Silurotaenia occurs in India and neighbouring countries. In fact, all nominal species of Silurotaenia from the Indian subcontinent are synonyms of some of the four species of Gangesia. The genus Silurotaenia thus seems to be limited in its distribution to the Palaearctic region, with one species occurring in Europe and another one in China ( [3], [70]). In the silurid catfish Wallago attu, which is one of the most valuable fishes sold on markets in India [74], only two species of Gangesia occur commonly, which was first noticed by Verma [2]. All the remaining species described from this catfish appear to be merely synonyms of G. bengalensis or G. agraensis. Another economically important catfish, Sperata seenghala, has also been reported to host a number of proteocephalidean species, all but one (G. macrones) placed in Silurotaenia. However, it seems that the only valid species specific of S. seenghala is the species described by Woodland [1] and all seven species of Silurotaenia are its synonyms. The fourth species, Gangesia vachai, has been found in catfishes of as many as three families, but more data are necessary to clarify the host specificity and distribution of this apparently uncommon species. As a result of the present revision, a comparative table with differential characters of the four valid species parasitizing catfishes in the Indomalayan region is provided to facilitate their identification (Table 4).
It is unfortunate that all descriptions published after 1928 were inadequate, with many errors and were based on decomposed specimens with detached hooks and hooklets, and/or on deformed (contracted) tapeworms, the morphology of which did not reflect natural state. In addition, authors of these descriptions almost completely ignored previously published data, in particular a key paper by Verma [2], who first clarified the confused taxonomic situation of Gangesia tapeworms parasitic in W. attu.
Similarly as in other groups of cestodes parasitic in freshwater fishes in the Indian subcontinent [54] [73], there is indirect evidence that no type specimens of any species of Gangesia described during the last seven decades have been deposited in any collection. This is apparent violation of the basic rules of the International Code of Zoological Nomenclature [70] and editors of all journals declared to be scientific should avoid this deplorable practice.
Gangesia tapeworms have also been recorded from non-siluriform fishes, such as cyprinids (Cirrhinus, Labeo, Puntius) and the mastacembelid Mastacembelus armatus. However, no vouchers of any of these remarkable findings have been deposited, which makes these records doubtful. For this reason, it is impossible to assess reliably the actual role of these fish hosts in the transmission of Gangesia tapeworms, which have otherwise been found exclusively in catfishes ( [3], [6], [75]). Misidentification of fish hosts (cyprinids and zig-zag eel versus siluriforms) is impossible but mislabelling of samples cannot be excluded.
Records of three Gangesia tapeworms, G. batrachusi, G. clariusae and G. jayakwadensis, which are newly synonymized with G. agraensis (first two) and G. bengalensis (last one), from the walking catfish, Clarias batrachus, are unreliable (no reference specimens exist). Mislabelling of the samples from W. attu seems to be the most probable explanation. It is worth mentioning that these three doubtful records were published by researchers from the laboratory, from which several tens [sic!] new species and several new genera of caryophyllidean tapeworms have been described from C. batrachus, all of them being in fact invalid (see [54]). The present authors dissected 235 C. batrachus and 105 specimens of other Clarias species from Bangladesh, Cambodia, India, Indonesia and Vietnam, but proteocephalidean tapeworms have never been found.
Future surveys, especially those in the Indian subcontinent, should focus on these atypical hosts, but adequate methods should be applied, including correct labelling of samples and their deposition in internationally accessible collections. The present authors have examined 363 specimens of six species of Puntius, 112 specimens of Mastacembelus armatus and 18 specimens of Labeo spp. from Bangladesh, India (Assam, Maharashtra and West Bengal) and Vietnam, but did not find any proteocephalideans.
Molecular lsrDNA data support the conclusions on the species composition of Gangesia tapeworms in the Indomalayan region based on morphological revision of type specimens and new material. Each of the four Indomalayan Gangesia species represents a monophyletic lineage with G. bengalensis, G. macrones and G. vachai forming a strongly supported clade (with low interspecific differences) apart from G. agraensis that most probably forms their sister lineage. The other strongly supported clade consists of two Palaearctic species, G. oligonchis from Russian Far East and G. parasiluri from Japan.
Gangesia, here represented by a majority of its species (six of nine), is a monophyletic assemblage, with European Silurotaenia siluri representing its sister taxon. Another strongly supported node found on the phylogenetic tree of the Gangesiinae represents Electrotaenia as a sister lineage to the clade composed of Silurotaenia and Gangesia. The phylogenetic relations of the remaining Gangesiinae, i.e. Postgangesia, Ritacestus and Vermaia, should remain considered as unclearly resolved.
Gangesia was one of the most species-rich genera of proteocephalidean cestodes with as many as 43 nominal taxa. However, the present study has demonstrated that the number of valid species is in fact much lower because as many as 32 species are invalidated. Based on the present revision, the genus consists only of G. agraensis Verma, 1928; G. bengalensis (Southwell, 1918); G. macrones Woodland, 1924; G. margolisi Shimazu, 1994; G. oligonchis Roitman and Freze, 1964; G. parasiluri Yamaguti, 1934; G. polyonchis Roitman and Freze, 1964; G. pseudobagrae Chen, 1962; and G. vachai (Gupta and Parmar, 1988). Validity of G. spasskajae Demshin, 1987, the type specimens of which are deposited in the Institute of Biology and Soil Science in Vladivostok, Russia, is questionable and requires confirmation.
The present study has also revealed that all Indian species of Silurotaenia are invalid, having been synonymized with some of the four species of Gangesia. The former genus thus seems to be restricted in its distribution to the Palaearctic region. Gangesia vachai, originally assigned to Silurotaenia, actually resembles in its morphology S. siluri, the type species of the genus, but this resemblance is apparently a result of convergence as evidenced by molecular data (Fig. 11). To better circumscribe individual genera of the Gangesiinae, a simple key based on morphological characteristics is provided.
Key to the Genera of the Subfamily Gangesiinae Mola, 1929