Molecular identification, description, and phylogenetic implications of the tadpoles of 11 species of Malagasy treefrogs, genus Boophis

Based on specimens identified by DNA barcoding, we describe the tadpoles of 11 species of treefrogs (Boophis) in the Malagasy family Mantellidae. All tadpoles belong to species of the stream‐breeding clade within Boophis. Based on these and other published descriptions of Boophis tadpoles which develop in running water bodies, we tentatively distinguish three ecomorphological guilds for these larvae. Guild A, in which we describe the larvae of B. boehmei, B. reticulatus, B. pyrrhus, B. tasymena, and B. viridis which have few lotic adaptations, their oral disc width being 31–43% of body width, with a single row of 48–81 marginal papillae, and the first upper keratodont row having 58–144 keratodonts. Guild B, in which we describe the tadpoles of B. albilabris, B. madagascariensis, B. luteus, and of an undescribed species here named B. sp. aff. elenae, is intermediate, with an enlarged oral disc, an increasing number of keratodont rows and a lower height of the caudal fin. In these tadpoles, oral disc width is 43–63% of body width, they have one or two rows of 69–164 marginal papillae, and the first upper keratodont row has 164–238 keratodonts. Guild C contains tadpoles with a very large oral disc, living on submerged rocks and stones in stream sections of strong current. In this guild we describe the tadpoles of B. marojezensis and B. sibilans. Their oral disc width is 63–89% of body width, there are multiple rows of many marginal papillae, and the first upper keratodont row has many small keratodonts which are difficult to count, but consistently amount to over 200. In B. marojezensis, the dorsal gap in the marginal papillae rows, apparent in all other species, is closed. These larval morphologies show a rather good fit with recently published molecular phylogenetic data: species groups that were confirmed to be monophyletic in most cases have similar larval morphologies, and, in contrast, where species of the same group have disparate larval morphologies the monophyly of the group is questionable (e.g. the B. majori group). Nevertheless, some cases of convergent evolution are apparent, such as the highly specialized Guild C morphology, which may have evolved separately in the B. albipunctatus group, B. mandraka group, and in some species of the B. majori group.


Introduction
Anuran amphibians are characterized by a unique type of larva, commonly named tadpole (Altig and McDiarmid 1999). Whereas adult frogs are carnivorous, most tadpoles are omnivorous and have a highly modified bauplan. Frogs in general are less variable in terms of morphology than in terms of reproductive biology. Many instances of partly or fully terrestrial reproduction are known, including direct development without a free-swimming larval stage, a mode that has evolved multiple times and mostly in tropical environments (Duellman and Trueb 1986). Aquatic tadpoles occur in a great variety of morphologies, and have adapted to a multitude of environments. Knowledge of the morphology and ecology of these larval stages is crucial to understand the ecological requirements of any particular frog species, an issue that becomes especially important in light of the delicate conservational status of amphibians as a whole (Stuart et al. 2004). Unfortunately, the identification of frog larvae is a complex task that involves either rearing of eggs obtained from a well-identified mating pair of frogs, or rearing of wild-collected tadpoles and tentative assignation of the metamorphosing juveniles to a species known to occur at the collecting locality. This latter method especially has caused a large number of misidentifications in the past and encounters major problems in megadiverse tropical anuran communities, because these often contain numerous sibling species which are hard to distinguish morphologically in their adult stage, and impossible to identify morphologically as juveniles.
The rainforests of Madagascar are known to harbour very species-rich amphibian communities (Blommers-Schlö sser 1979a, 1979bBlommers-Schlö sser and Blanc 1991) which are characterized by numerous instances of sibling species . A group where morphologically similar but bioacoustically disparate sibling species are particularly common is the genus Boophis. These are treefrogs (Cadle 2003), part of the endemic mantellid radiation of Madagascar and the Comoros , and currently comprise more than 50 species Vences and Glaw 2005). Many more species have been identified and are in the process of taxonomic description. Except for a few highland species, Boophis are strictly arboreal, and can be divided into two major clades which are characterized by either breeding in stagnant or running water . The stream-breeding clade is more speciose and represents a radiation mainly of the humid east of Madagascar . Tadpoles of 15 species of Boophis have been described by Blommers-Schlö sser (1979b): B. erythrodactylus, B. goudoti, B. granulosus, B. hillenii, B. idae, B. luteus, B. madagascariensis, B. majori, B. mandraka, B. microtympanum, B. rappiodes, B. tephraeomystax, B. williamsi, and two species of uncertain identification, one assigned to the B. luteus group, a second one reported under the name B. untersteini which is presently a junior synonym of B. goudoti. A few additional larval morphologies have been reported since: Boophis jaegeri and B. ankaratra Vences 1992, 1994); B. occidentalis ; B. laurenti ; B. rufioculis (Grosjean et al. 2006). Most of the early tadpole identifications were made by assigning reared juveniles to species, and are therefore rather reliable at genus and speciesgroup level, but require confirmation at species level. Although all Boophis tadpoles known so far have relatively generalized, exotrophic and benthic morphologies, conforming to the ranoid type (type IV of Orton 1953), some stream-breeding species have evolved obvious adaptations to the lotic environment, such as various degrees of enlargement of the oral disc, and an increase in the number of keratodont rows, of keratodonts and papillae (Blommers-Schlö sser 1979b). Such adaptations are known from various phenetic species groups of Boophis, such as the B. rappiodes group, B. luteus group, B. albilabris group, B. mandraka group, B. majori group, and B. microtympanum group, but since the monophyly of these species assemblages was uncertain until recently, the phylogenetic significance of these larval adaptations has not been analysed so far.
The advent of molecular techniques has led to an impressive acceleration of the understanding of phylogenetic relationships among organisms, among them mantellid frogs and the genus Boophis (e.g., Richards et al. 2000;Vences et al. 2002Vences et al. , 2003Lehtinen and Nussbaum 2003;Lehtinen et al. 2004;Vences and Glaw 2005;Glaw and Vences 2006). In addition, DNA barcoding techniques allow a largely unequivocal and fast identification of larval stages, once a proper database for adults has been assembled (Hebert et al. 2003;Blaxter 2004), a method that has already been applied to some amphibians . We here use this DNA barcoding approach to identify and describe the tadpoles of 11 species of Boophis, representing six species groups, and discuss their patterns of morphological evolution in relation to molecular phylogenies that have recently been published.

Material and methods
We collected tadpoles in the field from several streams in or near rainforest during the rainy season (December 2001to January 2002, using a variety of large and small nets, adjusted to the specific conditions of each water body. General abiotic characteristics of the streams were recorded in as much detail as possible. Physico-chemical parameters of water bodies were assessed using portable devices in the field (temperature, depth, oxygen content, pH) or in the laboratory using standardized procedures. Tadpoles were euthanized by immersion in chlorobutanol solution, and, using a stereomicroscope, immediately in the field sorted into homogeneous series based on morphological characters. From each series one specimen was selected, and from this individual, a tissue sample from its tail musculature or fin was taken and preserved in 99% ethanol. This specimen is here named ''DNA voucher''. The drawings and morphological descriptions are based on this unequivocally identified specimen, whereas the additional tadpoles of a series are discussed under ''Variation''. In the tadpole drawings (Figures 1-11) the missing tissue pieces of the tail were completed by comparison with additional specimens of the series. After tissue collection, all specimens were preserved in 4% formalin. DNA vouchers have been deposited in the Zoologische Staatssammlung, Munich, Germany (ZSM). Further specimens of the series are so far uncatalogued and will be integrated in the collections of ZSM and of the Département de Biologie Animale, Université d'Antananarivo at a later stage; they are here reported with their field numbers of the first author (L.R.). Comparative specimens were examined from the herpetological collection of the Zoological Museum, Amsterdam, Netherlands (ZMA).
Tadpoles were identified using a DNA barcoding approach based on a fragment of the mitochondrial 16S rRNA gene, which is known to be sufficiently variable among species of Malagasy frogs . The 550 bp fragment was amplified using the primers 16Sa-L and 16Sb-H from Palumbi et al. (1991) and applying standard protocols, resolved on automated sequencers, and compared to a near-complete library of sequences of adult Malagasy frog species. Identification was considered to be unequivocal when the tadpole sequence was 99-100% identical to an adult specimen from the same geographical region (except for B. sp. aff. elenae; see Discussion below), and not more similar to any sequence from another species. DNA sequences were deposited in GenBank (accession numbers DQ792462-DQ792499; accession numbers of comparative adult specimens are included in the sequence sets AY847959-AY848683 and AJ315909-AJ315913).
All detailed tadpole descriptions in the following are based on one DNA voucher specimen, whereas variation is described based on further specimens of the same series, and DNA vouchers and specimens of additional series. Morphological terminology follows Altig and McDiarmid (1999); keratodont row formula (KRF) follows Dubois (1995) and developmental stages were determined according to Gosner (1960). Measurements were taken with a graduated ocular attached to a stereomicroscope. The landmarks are largely those shown in Altig and McDiarmid (1999, p 26, Figure 3.1); for others see Grosjean (2001). Drawings were made with the aid of a camera lucida.
The abbreviations used in the descriptions are the following: BH, maximum height of body; BL, body length; BW, maximum width of body; DG, maximum size of dorsal papilla gap; ED, maximum diameter of eye; MTH, maximum tail height; NN, internarial distance; NP, distance from centre of nares to anterior margin of eye; ODW, oral disc width; PP, interpupilar distance; RN, rostro-narial distance; SS, distance from tip of snout to opening of spiracle; TMH, tail muscle height; TMW, tail muscle width; TL, total length.

Characteristics of study area and water bodies
Specimens analysed for this study came from four localities in central-eastern Madagascar, all close to Andasibe village: (1) a relatively fast-flowing stretch of Ranomena river next to Vohidrazana, within remnants of primary rainforest; (2) a small stream and dam next to the village Andasibe, within Analamazaotra Special Reserve; (3) a relatively slow-moving stretch of the Analamazaotra river around the bridge of the road leading to the village of Andasibe, at the border of Analamazaotra Special Reserve, surrounded by a mix of degraded vegetation, secondary forest and primary rainforest; (4) a stretch of Andranomanaponga river within Mantadia National Park, flowing through primary rainforest. A few additional specimens were collected from (5) a small stream at a place called Andranomena, located south-west of Moramanga. Water was clear and transparent at all sites except no. 3. The bottom was composed mainly of large granitic boulders and rocks at site no. 1, and of sand and debris at site nos 2-5. Geographic location and physicochemical water parameters are summarized in Table I. In the following accounts, collecting localities 1-5 refer to the sites listed above and in Table I.

Species accounts
In the following accounts, we have grouped species according to phenetic species groups as recognized by Vences (1994, 2006), in alphabetical order within groups and order of species groups according to their similarity and putative phylogenetic relationships.
In dorsal view (Figure 1a), body ovoid elongate, snout rounded. In lateral view (Figure 1b), body depressed, BW 111% of BH, snout rounded. Eyes of moderate size, ED 11.4% of BL, slightly bulging, not visible in ventral view, positioned and directed dorsolaterally, situated at about the anterior third of the body. Pineal ocellus present between eyes. Nares elliptical, moderately sized, rimmed, positioned dorsolaterally, directed anterodorsolaterally and equidistant from tip of snout to anterior edge of eyes, RN 100% of NP; NN 62% of PP. Spiracle sinistral, a short tube free at opening, well visible from dorsal view; spiracular opening oval, directed posteriorly, situated at the height of the longitudinal axis of caudal musculature and closer to end of body than to snout, SS 68% of BL. Vent tube short, dextral and opening at ventral edge of fin, its right wall displaced anterodorsally. Caudal musculature moderately developed, TMH 45% of BH and 68% of MTH, TMW 47% of BW; its height three-fifths of total height at mid-length of tail. Caudal fins shallow anteriorly and deepest at about midway of tail, height of caudal fins decreasing progressively towards the nearly pointed tail tip, MTH 66% BH; dorsal fin beginning next to the dorsal tail-body junction. Oral disc (Figure 1c) of moderate size, ODW 24% of BL and 40% of BW, positioned and directed ventrally, emarginate. A single row of marginal papillae interrupted by a large medial gap on the upper labium (DG 94% of ODW) and by a small medial gap on the lower labium; total number of marginal papillae about 64 (33 on the left side, 31 on the right side). A few submarginal papillae positioned on the lateral parts of the anterior and posterior labia. Papillae of moderate size, round or conical with rounded tip. No denticulate papillae. Keratodont row formula 1:3+3/1+1:2. The density of keratodonts on A1 is about 50 per mm (a total of ca 110). The length of interrupted anterior keratodont rows (A2, A3, and A4) decreases gradually towards the centre of the disc, length of P3 about one-third of P2. Both jaw sheaths coarsely serrated; upper jaw completely black with a large medial serration surrounded by smaller serrations on each side forming a slight convexity; lower jaw partially pigmented, V-shaped.
Coloration in preservative. Dorsally: body and tail musculature brownish, covered by scattered black spots. Some concentrated pigmentation forming dark patch between eyes. Laterally: intestinal coils well visible; caudal musculature mottled, dorsal fin slightly pigmented; ventral fin clear. Ventrally: the whole surface of ventral body transparent, branchial and cardial region translucent, easily visible through ventral surface.
Variation. TL and BL of seven tadpoles at stages 25-31, all from locality 2 (ZSM 533/ 2004-537/2004 andLR 147, LR 165) (Figure 2b), body depressed, BW 131% of BH, snout rounded. Eyes moderately large, ED 14.0% of BL, bulging, not visible in ventral view, positioned dorsolaterally but directed slightly more laterally than anterolaterally or dorsolaterally, situated at about the anterior quarter of the body. Nares elliptical, rimmed with a mediodorsal projection giving them a bean shape, moderately small, positioned dorsolaterally, directed anterodorsally and about equidistant between tip of snout and anterior edge of eyes, RN 106% of NP; NN 53% of PP. Spiracle sinistral, moderately sized, tubular, well visible from dorsal view, distal half free from body, spiracular opening orientated posterodorsally, closer to end of body than to snout, SS 65% of BL, situated at the height of the lower part of tail musculature. Vent tube moderately sized, short, dextral, opening dextral at ventral edge of fin, its right wall displaced anterodorsally, directed posterolaterally, linked to ventral tail fin except its tip. Tail long and tapered. Caudal musculature strong, TMH 64% of BH and 87% of MTH, TMW 46% of BW, at mid-length of tail its height is equal to half of total tail height, reaching tail tip. Caudal fins regular with straight edges, moderately weakly developed, their general appearance almost similar throughout tail musculature, point of maximum height attained in the second half of the tail, MTH 74% of BH; dorsal fin originating at the dorsal tail-body junction and slightly taller than ventral fin at mid-length of tail; tail tip bluntly pointed, its state as reproduced in Figure 2 (caudal fins not reaching tail tip) probably being an artefact of preservation or damage during transport. Oral disc (Figure 2c) of moderate size, ODW 21% of BL and 38% of BW, positioned and directed ventrally, emarginated. A single row of marginal papillae interrupted by a large median gap on the upper labium (DG 73% of ODW) and by a small medial gap on the lower labium; total number of marginal papillae about 81 (43 on the left side and 38 on the right side). A few submarginal papillae positioned in the lateral parts of anterior and posterior labia. Papillae moderately large, cylindrical with rounded or pointed tips. No denticulate papillae. Keratodont row formula 1:3+3/1+1:2. The density of keratodonts on A1 is about 55 per mm (a total of ca 144). The length of interrupted anterior keratodont rows (A2, A3, and A4) decreases gradually towards the centre of the disc, P1 and P2 are of about similar size, slightly longer than P3. Both jaw sheaths coarsely serrated; upper jaw completely black with a large medial serration surrounded by smaller serrations on each side forming a slight convexity; lower jaw partially pigmented, strong and V-shaped.
Coloration in preservative. Dorsally: whole body and caudal musculature dark brown and covered by scattered black dots; dark pigments concentrated dorsally forming dark patches slightly visible on the interorbital area and beside the nares; a dark line slightly visible middorsally on body. Laterally: intestinal coils well visible; body and caudal musculature with scattered dark pigmentation; dorsal fin transparent with a few sparse black dots; ventral fin clear. Ventrally: branchial and cardial region opaque through ventral body wall; intestinal coils well visible.

Boophis madagascariensis (Peters, 1874) (Figure 3)
The following description is based on one tadpole in developmental stage 35 (ZSM 519/ 2004, field number LR 5, TL 34.7 mm, BL 13.7 mm), from locality 5. In dorsal view (Figure 3a), body ovoid elongate, snout rounded. In lateral view (Figure 3b), body depressed, BW 117% of BH, snout rounded. Eyes of moderate size, ED 11.0% of BL, very slightly bulging, not visible in ventral view, positioned dorsolaterally but directed almost laterally, situated at about the anterior third of the body. Pineal ocellus in line with the anterior margin of eyes. Nares moderately sized, elliptical, rimmed with a flat mediodorsal projection giving them a bean shape, positioned dorsolaterally, directed anterolaterally and slightly dorsally, nearer to anterior edge of eyes than to snout, RN 109% of NP; NN 55% of PP. Spiracle sinistral, moderately sized, well visible from dorsal view, oval, its tip free from body; spiracular opening orientated posteriorly, closer to end of body than to snout, SS 66% of BL, situated at the height of the lower part of tail musculature. Vent tube moderately sized, short, dextral, opening dextral at ventral edge of fin, its right wall displaced anterodorsally, directed posteriorly. Caudal musculature moderately strong, TMH 54% of BH and 58% of MTH, TMW 49% of BW, at mid-length of tail its height is two-fifths of the total tail height, reaching tail tip. Caudal fins regular with straight edges, MTH 93% of BH; dorsal fin low, originating on tail muscle and wider than ventral fin; tail tip pointed.
Oral disc (Figure 3c) of moderate size, ODW 26% of BL and 43% of BW, positioned and directed anteroventrally, emarginated. A single row of marginal papillae interrupted by a large median gap on the upper labium (DG 81% of ODW) and without visible gap on the lower labium; total number of marginal papillae about 95. Submarginal papillae present, some of them clustered at the corners of the disc but most localized on posterior labium and distributed in two rows. Papillae of moderate size, elongate with rounded tip. No denticulate papillae. Keratodont row formula 1:5+5/1+1:2. The density of keratodonts on A1 is about 60 per mm (a total of 164). The length of interrupted anterior keratodont rows (A2-A6) decreases gradually towards the centre of disc, P1 interrupted by a gap of less than 0.1 mm, keratodont rows of lower labium subequal in size. Both jaw sheaths completely dark pigmented; upper jaw a wide flattened, medially convex arch, finely serrated; lower jaw V-shaped, moderately finely serrated.
Coloration in preservative. Dorsally: body grey with sparse dark markings, caudal musculature with scattered dark pigmentation. Laterally: intestinal coils visible; fins translucent with sparse black dots. Ventrally: branchial and cardial region not visible through ventral body wall; intestinal coils obscure.
Variation. TL and BL of six tadpoles at stages 28-37, all from locality 5 (ZSM 517-520, and LR 4, LR 17) are 24.5-35.6 and 10.8-15.6 mm, respectively. The ratios vary in the following proportions: BW 117-157% of BH; ED 10. In dorsal view (Figure 4a), body elliptical, snout rounded. In lateral view (Figure 4b), body depressed, BW 116% of BH, snout rounded. Eyes moderately large, ED 13.0% of BL, bulging, not visible in ventral view, positioned dorsolaterally but directed anterolaterally and dorsolaterally, situated at about the anterior quarter of the body. Pineal ocellus placed anteriorly to margin of eyes. Nares round, moderately sized, rimmed, positioned dorsolaterally, directed anterolaterally and opening dorsolaterally, positioned nearer to tip of snout than to anterior edge of eyes, RN 86% of NP; NN 44% of PP. Spiracle sinistral, moderately small, slightly conical, slightly visible from dorsal view, free at tip, orientated posterodorsally; spiracular opening oval, situated closer to end of body than to tip of snout, SS 58% of BL, situated below the longitudinal axis of tail musculature. Vent tube short, dextral, opening dextral, its right wall displaced anterodorsally, directed posterolaterally, entirely linked to ventral tail fin. Caudal musculature moderately strong, TMH 55% of BH and 62% of MTH, TMW 51% of BW, at mid-length of tail its height about two-fifths of the total tail height, reaching tail tip. Caudal fins moderately developed, point of maximum height attained at the proximal third of the tail, MTH 89% of BH; dorsal fin convex, originating at the dorsal tail-body junction and taller than ventral fin at mid-length of tail, ventral fin originating immediately behind body; tail tip pointed.
Oral disc (Figure 4c) moderately large, ODW 25% of BL and 47% of BW, positioned and directed ventrally, emarginated. Oral disc surrounded by a single row of 69 (34 on the left side and 35 on the right side) marginal papillae interrupted by a large median gap on the upper labium (DG 71% of ODW), and by a small medial gap on the lower labium. A few submarginal papillae positioned on the lateral parts of the upper labium, and lateral and posterolateral parts of the lower labium. Papillae of moderate size, cylindrical with a rounded tip. No denticulate papillae. Keratodont row formula 1:5+5/1+1:2. Gap on P1 shorter than 0.1 mm; about 188 keratodonts on A1, the longest keratodont row of the oral disc (ca 61 per mm); P1 and P2 of about similar size, slightly longer than P3. Upper jaw completely black, moderately serrated, a large arch with a large median convexity; lower jaw V-shaped, coarsely serrated, mostly coloured in black but its base less pigmented.
Coloration in preservative. Dorsally: body and tail musculature mottled with black stippling; dark pigmentation concentrated dorsally forming dark patches between eyes and next to nares. Laterally: intestinal coils well distinct; dorsal fin with scattered black dots; ventral fin generally clear except near tail tip; at about mid-tail many small dark patches visible along the upper tail musculature. Ventrally: whole surface of ventral body transparent; branchial and cardial organs translucent but slightly visible through ventral body wall.
Variation. TL and BL of two tadpoles at stages 33 and 41, from locality 3 (ZSM 588/2004(ZSM 588/ -589/2004  Boophis majori group Boophis pyrrhus Glaw, Vences, Andreone and Vallan, 2001 ( Figure 5) The following description refers to one specimen in developmental stage 31 (ZSM 580/ 2004, field number LR 231a, TL 41.3 mm, BL 15.0 mm), from locality 3. In dorsal view (Figure 5a), body ovoid elongate, snout rounded. In lateral view (Figure 5b), body depressed, BW 117% of BH, snout rounded. Eyes of moderate size, ED 11.3% of BL, bulging, not visible in ventral view, positioned anterolaterally but directed anterolaterally and dorsolaterally. Nares round, moderately small, rimmed, positioned dorsally, directed anterolaterally, nearer to anterior edge of eyes than to tip of snout, RN 123% of NP; NN 40% of PP. Spiracle sinistral, moderately small, square, well visible from dorsal view, free at tip, orientated almost posteriorly; spiracular opening oval, situated closer to end of body than to tip of snout, SS 64% of BL, situated at the height of the longitudinal axis of tail musculature. Vent tube short, dextral, opening dextral, its right wall displaced anterodorsally, directed posterolateroventrally. Caudal musculature moderately developed, TMH 50% of BH and 60% of MTH, TMW 40% of BW, at mid-length of tail its height about two-fifths of the total tail height. Caudal fins moderately weakly developed, MTH 83% of BH; dorsal fin originating at the dorsal tail-body junction, ventral fin originating immediately behind body; tail tip nearly rounded.
Oral disc (Figure 5c) of moderate size, ODW 20% of BL and 37% of BW, positioned and directed ventrally, emarginated. A single row of marginal papillae interrupted by a large median gap on the upper labium (DG 53% of ODW), and by a very small medial gap on the lower labium. A few submarginal papillae positioned in the lateral parts of the upper and lower labia. Papillae stout, of moderate size, conical with a rounded tip. No denticulate papillae. Keratodont row formula 1:3+3/1+1:2. About 92 keratodonts on A1 (ca 54 per mm). P1 interrupted by a gap of less than 0.1 mm and slightly longer than P2; P3 about two-thirds of P2. Upper jaw completely black, a large arch with a slight median convexity, moderately coarsely serrated; lower jaw strong, V-shaped, coarsely serrated.
Coloration in preservative. Dorsally: the whole dorsum body and the caudal musculature with dark scattered pigmentation. Laterally: intestinal coils completely visible; lateral body and caudal musculature covered with sparse black dots; caudal fins moderately pigmented. Ventrally: the whole surface of ventral body transparent; branchial and cardial region visible through ventral body wall. Variation. TL and BL of 11 tadpoles at stages 31-41 (ZSM 580/2004-587/2004, andLR 231c, LR 231d, LR 231i) Boophis marojezensis Glaw and Vences, 1994 ( Figure 6) The following description is based on one tadpole in developmental stage 41 (ZSM 523/2004, field number LR 66, TL 26.5 mm, BL 10.0 mm), from locality 1. Hindlimbs of this specimen were largely used for tissue sampling and are therefore not shown in Figure 6. In dorsal view (Figure 6a), body ovoid elongate, snout rounded. In lateral view (Figure 6b), body greatly depressed, BW 145% of BH, snout almost truncate. Eyes large, ED 16.0% of BL, slightly bulging, not visible in ventral view, positioned dorsolaterally but directed almost laterally, situated at about the anterior third of the body. Nares moderately small, round, rimmed with a very small and flat mediodorsal projection, positioned dorsally, directed anterolaterally and dorsally, much closer to anterior edge of eyes than to snout, RN 169% of NP; NN 44% of PP. Spiracle sinistral, narrow and long, slightly visible from dorsal view, its tip free from body; spiracular opening orientated posterodorsally, much closer to end of body than to snout, SS 76% of BL, situated at the height of the lower part of tail musculature. Vent tube small, short, medial, opening just before the beginning of ventral fin, opening medial, directed more posteriorly than posteroventrally, linked to caudal muscle. Caudal musculature strong, TMH 66% of BH and 59% of MTH, TMW 48% of BW, especially in the anterior half of the tail; height of caudal musculature almost half of total tail height at mid-length of tail, almost reaching tail tip. Caudal fins regular with straight edges, very shallow, MTH 111% of BH; dorsal fin originating next to body-tail junction, convex in its medial part; ventral fin beginning at the level of the ventral terminal of the body, following the caudal muscle; tail tip fine.
Oral disc (Figure 6c) enlarged, ODW 41% of BL and 63% of BW, positioned and directed ventrally, not emarginated. Several uninterrupted rows of marginal papillae around oral disc; no medial gaps in rows of marginal papillae on upper or lower labium. Papillae of internal row round and of moderate size, papillae of external rows small, elongate, and cylindrical. No denticulate papillae. Keratodont row formula 4:3+3/3. About 203 keratodonts on A3 (32-33 per mm). The length of interrupted anterior keratodont rows (A5, A6, and A7) decreases gradually towards the centre of the disc, keratodont rows of lower labium subequal in length. Upper jaw not serrated and weakly developed, median part straight, a black halo on its distal part; lower jaw more developed, narrow V-shaped, ribbed, composed of two parts connected by a less keratinized median area, coarsely serrated.
Coloration in preservative. Dorsally: part anterior to eyes covered with sparse black dots; dark pigmentations concentrated dorsally forming dark patches between eyes and beside nares; black colour on either side of lateral body obscuring intestine area; caudal musculature with scattered dark pigmentation. Laterally: intestinal coils invisible; caudal musculature with sparse black spots, showing some reticulations at its posterior end; dorsal fin with small black spots, ventral fin clear. Ventrally: branchial and cardial region translucent, well visible through ventral body wall; intestinal coils invisible.

Boophis albipunctatus group
Boophis sibilans Glaw and Thiesmeier, 1993 ( Figure 7) The following description is based on one tadpole in developmental stage 25 (ZSM 557/ 2004, field number LR 269, TL 25.2 mm, BL 9.0 mm), from locality 4. In dorsal view (Figure 7a), body elliptical, snout rounded. In lateral view (Figure 7b), body depressed, BW 129% of BH, snout rounded. Eyes moderately large, ED 12.2% of BL, bulging, not visible in ventral view, positioned more dorsally than dorsolaterally but directed dorsolaterally, situated at about the anterior third of the body. Nares moderately sized, elliptical, rimmed with a flat mediodorsal projection, positioned dorsally, directed anterolaterally and opening almost dorsally, much closer to anterior edge of eyes than to snout, RN 148% of NP; NN 50% of PP. Spiracle sinistral, narrow and long, slightly visible from dorsal view, its tip free from body; spiracular opening orientated almost posteriorly, much closer to end of body than to snout, SS 72% of BL, situated below the lower part of tail musculature. Vent tube small, medial, directed posteriorly, linked to caudal muscle, opening posterolateral. Caudal musculature strong, TMH 77% of BH and 75% of MTH, TMW 59% of BW, at mid-length of tail height of caudal musculature almost half of total tail height, parallel in its proximal third then gradually tapering, almost reaching tail tip. Caudal fins moderately shallow, MTH 103% of BH; dorsal fin originating next to body-tail junction, shallow in its anterior part then becoming convex towards mid-tail, ventral fin beginning just behind body and reaching its maximum height more posteriorly than dorsal fin; tail tip rounded.
Oral disc (Figure 7c) enlarged, ODW 44% of BL and 89% of BW, positioned and directed ventrally, not emarginated. Several rows of marginal papillae around oral disc, interrupted by a large median gap on the upper labium (DG 53% of ODW); no gap on lower labium. Papillae small, conical with a more or less pointed tip. No denticulate papillae. Keratodont row formula 4:3+3/3. Keratodonts on continuous rows A1, A2, and A3 very small and difficult to count; estimation of keratodonts done on A4 with a total of about 208 (ca 70 per mm). The length of interrupted upper keratodont rows (A5, A6, and A7) decreases gradually towards the centre of the disc, keratodont rows of lower labium subequal in length. Upper jaw not serrated and weakly developed, black, median part straight; lower jaw more developed, V-shaped, ribbed, coarsely serrated, partially pigmented.
Coloration in preservative. Dorsally: body and tail musculature brownish with scattered dark pigmentation; either side of dorsal body black coloured with white stripes (the blood vessels); some dark pigment concentrated dorsally forming dark patches between eyes and beside nares; tail muscle barred black and white. Laterally: intestinal coils not visible through lateral body wall; spiracle not pigmented; caudal fins clear; lower part of caudal musculature less pigmented anteriorly; musculature junctions very distinct anteriorly.
Ventrally: opaque branchial and cardial organs slightly visible through ventral body wall; intestinal coils not visible.

Boophis luteus group
Boophis luteus (Boulenger, 1882) ( Figure 8) The following description is based on one tadpole in developmental stage 33 (ZSM uncatalogued, field number LR 218, TL 33.1 mm, BL 13.1 mm), from locality 3. In dorsal view (Figure 8a), body ovoid, snout rounded. In lateral view (Figure 8b), body depressed, BW 125% of BH, snout rounded. Eyes moderately large, ED 14.5% of BL, bulging, not visible in ventral view, positioned dorsolaterally but directed laterally, situated at about one-third of the body. Nares moderately sized, nearly oval, not rimmed, positioned and directed dorsally, closer to the anterior edge of eyes than to tip of snout, RN 133% of NP; NN 64% of PP. Spiracle sinistral, moderately small, slightly tubular, well visible from dorsal view, its tip free from body; spiracular opening oval, orientated posterodorsally, positioned at about three-quarters of body length, SS 76% of BL, and situated well below the longitudinal axis of tail musculature. Vent tube short, dextral, opening at ventral edge of fin, opening directed posterolaterally, linked to caudal muscle, its right wall displaced anteriorly. Caudal musculature moderately strong, TMH 54% of BH and 65% of MTH, TMW 48% of BW, height of caudal musculature about two-fifths of the total height at mid-tail, reaching tail tip. Caudal fins shallow anteriorly, deepest at about half of their length, their height decreasing progressively up to tail tip, MTH 82% of BH; dorsal fin originating well posterior to the dorsal tail-body junction, ventral fin starting just behind the body; tail tip obtuse.
Oral disc (Figure 8c) large, ODW 37% of BL and 63% of BW, positioned anteroventrally and directed ventrally, not emarginated. Oral disc bordered by one or two rows of marginal papillae interrupted by a large median gap on the upper labium (DG 94% of ODW), and by a small median gap on the lower labium. Submarginal papillae clustered in the corners of labia. Papillae of moderate size, cylindrical elongate with rounded tip. No denticulate papillae. Keratodont row formula 1:5+5/1+1:2. About 238 keratodonts on A1 (ca 63 per mm). The length of the interrupted anterior rows A2, A3, and A4 decreases towards centre of disc; keratodont rows of lower labium subequal in size. Both jaw sheaths serrated and fully black pigmented; upper jaw a wide flattened arch convex medially; lower jaw Vshaped.
Coloration in preservative. Dorsally: body and tail musculature dark pigmented; dark pigmentation concentrated dorsally forming obvious dark patches between eyes and immediately next to nares. Laterally: intestinal coils well visible; dorsal fin pigmented; ventral fin almost clear, musculature mottled. Ventrally: branchial and cardial organs apparent through the transparent surface of body.
Oral disc (Figure 9c) moderately large, ODW 31% of BL and 56% of BW, positioned anteroventrally and directed ventrally, not emarginated. A total number of 164 marginal papillae disposed in two rows around the oral disc, interrupted by a large median gap on the upper labium (DG 67% of ODW); gap on the lower labium not distinct. Submarginal papillae clustered in the corners of labia. Papillae of moderate size, elongate with more or less rounded tip. No denticulate papillae. Keratodont row formula 1:3+3/1+1:2. About 194 keratodonts on A1 (ca 54 per mm). The length of the interrupted anterior rows A2, A3, and A4 decreases towards centre of disc; P2 and P3 subequal in size, P1 slightly shorter. Both jaw sheaths serrated and fully black pigmented; upper jaw with a large medial serration surrounded by smaller serrations on each side forming a slight convexity; lower jaw Vshaped.
Coloration in preservative. Dorsally: body and tail musculature dark brown with scattered black spots; some pigmentation forming a distinctive dark patch between eyes and also on mid-dorsal area of body. Laterally: body and caudal musculature covered by scattered dark pigmentation; musculature junctions obvious in the anterior half of tail musculature; intestinal coils completely visible; fins transparent with scattered pigmentation; dorsal fin showing some reticulations. Ventrally: whole surface of ventral body transparent; branchial and cardial organs visible through translucent ventral surface. Boophis rappiodes group Boophis tasymena Vences and Glaw, 2002 ( Figure 10) The following description is based on one tadpole in developmental stage 25 (ZSM 527/ 2004, field number LR 96, TL 17.5 mm, BL 6.7 mm), from locality 1.
In dorsal view (Figure 10a), body ovoid, snout rounded. In lateral view (Figure 10b), body greatly depressed, BW 156% of BH, snout rounded. Eyes of moderate size, ED 10.5% of BL, not visible in ventral view, positioned dorsolaterally, situated at about the anterior third of the body. Nares moderately sized, round, rimmed with a small flat mediodorsal projection, positioned quasi-dorsally and directed anterodorsolaterally, equidistant between tip of snout and anterior edge of eyes, RN 100% of NP; NN 60% of PP. Spiracle sinistral, moderately large, well visible from dorsal view, tubular, inner wall free from body; spiracular opening oval, orientated more posteriorly than posterodorsally, closer to end of body than to snout, SS 69% of BL, situated at the height of the lower part of tail musculature. Vent tube moderately sized, short, dextral, opening directed posterolaterally, its right wall displaced anteriorly, entirely linked to ventral tail fin. Caudal musculature moderately developed, TMH 44% of BH and 42% of MTH, TMW 38% of BW, its height almost half of total height at mid-length of tail. Caudal fins regular with straight edges, MTH 104% of BH; dorsal fin originating next to body-tail junction, reduced in the anterior part and convex at about half of tail length; point of maximum tail height reached towards mid-tail; ventral fin moderately shallow; tail tip roughly rounded.
Oral disc (Figure 10c) of moderate size, ODW 25% of BL and 44% of BW, positioned and directed anteroventrally, emarginated. A single row of marginal papillae interrupted by a large median gap on the upper labium (DG 59% of ODW), and by a short gap on the lower labium; total number of marginal papillae about 48 (23 on the left side and 25 on the right side). No distinct submarginal papillae present. Papillae of moderate size, cylindrical elongate with rounded tip. No denticulate papillae. Keratodont row formula 1:3+3/1+1:2. About 58 keratodonts on A1 (55 per mm). The length of interrupted anterior keratodont rows decreases gradually towards the centre of the disc, P1 and P2 subequal in length, P3 shorter. Jaw sheaths moderately coarsely serrated; upper jaw as a large arch with a median convexity, totally coloured by black pigment; lower jaw V-shaped.
Coloration in preservative. Dorsally: body and tail musculature generally brownish, uniformly pigmented by black dots. Laterally: intestinal coils partially visible on the lower part of the body; dorsal fin faintly pigmented on the anterior part near its starting point; ventral fin clear; body and caudal musculature with black dots. Ventrally: region of the branchial and cardial organs somewhat obscure but still visible through the ventral surface of the body.
Boophis viridis Blommers-Schlö sser, 1979 ( Figure 11) The following description refers to one voucher specimen in developmental stage 36 (ZSM 574/2004, field number LR 222d, TL 30.5 mm, BL 11.5 mm), from locality 3. In dorsal view (Figure 11a), body ovoid, snout rounded. In lateral view (Figure 11b), body depressed, BW 118% of BH, snout rounded. Eyes of moderate size, ED 10.0% of BL, slightly bulging, not visible in ventral view, positioned almost dorsally but directed anterolaterally and dorsolaterally, situated at about the anterior quarter of the body. Nares oval, of moderate size, rimmed, positioned dorsolaterally, directed anterolaterally and dorsolaterally, nearer to anterior edge of eyes than to tip of snout, RN 133% of NP; NN 49% of PP. Spiracle sinistral, small, conical, inner wall free and formed such that aperture opens laterally instead of posteriorly; closer to end of body than to snout, SS 67% of BL, situated a little below the longitudinal axis of tail musculature. Vent tube short, dextral, opening dextral, its right wall displaced anterodorsally, directed posterodorsally, linked to ventral tail fin except its tip. Tail relatively long. Caudal musculature moderately developed, TMH 50% of BH and 67% of MTH, TMW 37% of BW, at mid-length of tail its height about half of the total tail height, reaching almost to tail tip. Caudal fins moderately weakly developed, MTH 75% of BH; dorsal fin originating at the dorsal tailbody junction and slightly taller than ventral fin at mid-length of tail, ventral fin originating immediately behind body; tail tip pointed.
Oral disc (Figure 11c) moderately small, ODW 19% of BL and 31% of BW, positioned and directed anteroventrally, emarginated. Oral disc bordered by a single row of about 50 marginal papillae interrupted by a large median gap on the upper labium (DG 57% of ODW), and by a small medial gap on the lower labium. A few submarginal papillae positioned in the lateral parts of the anterior and posterior labia. Papillae moderately large, stout, conical with a rounded tip. No denticulate papillae. Keratodont row formula 1:2+2/ 1+1:2. P1 interrupted by less than 0.1 mm; P2 the longest keratodont row of the oral disc. About 79 keratodonts on A1 (ca 49 per mm); P1 and P2 of about similar length, P3 about two-thirds of P2. Upper jaw finely serrated, black with a larger medial serration surrounded by smaller serrations on each side forming a slight convexity; lower jaw black, moderately serrated, V-shaped.
Coloration in preservative. Dorsally: whole dorsum body and caudal musculature uniformly mottled; eyes partially sunk into orbital sockets; nares mixed with the markings and difficult to see. Laterally: intestinal coils well visible; musculature junctions slightly distinct in the anterior part of tail musculature; dorsal fin moderately mottled; ventral fin almost clear. Ventrally: branchial and cardial region slightly visible through ventral body wall; intestinal coils dextral, visible ventrally with regular spiral shape.

Identity of Boophis tadpoles
In this paper we identified tadpoles of 11 Boophis species using DNA barcoding, all from a single area around Andasibe village in central-eastern Madagascar, which is known to be a biodiversity hotspot (Lees et al. 1999) with a large number of species of frogs recorded in sympatry. The molecular identification of the tadpoles described herein was, despite this large diversity, unequivocal, since a large comparative database of homologous sequences from adult frogs collected at the same site exists  and could be used for comparison. In all cases, the larval DNA sequences were fully identical to those of adult frogs. The only exception is Boophis aff. elenae: although we have often heard the typical calls of this species at Andasibe, we have not yet been able to collect adults and, hence, to obtain DNA sequences from a well-identified specimen from this site. However, we have many such sequences from a second locality, Ranomafana, and our tadpole sequence from Andasibe clearly clusters with these, despite a considerable pairwise sequence divergence of 3.7%. In general, none of the species studied here belongs to a complex of sibling species among which instances of hybridization or haplotype sharing have been reported, which increases our confidence in the identity of the tadpoles.
Of the Boophis tadpoles described previous to the present paper, several can be considered as reliably identified. The tadpole of B. laurenti was identified by DNA barcoding ) and is extremely similar to that of B. microtympanum as described by Blommers-Schlö sser (1979b) based on reared juveniles from the Ankaratra massif. This and the tadpole of the syntopic B. williamsi originate from montane habitats of low species diversity, and their large juveniles are very similar in colour and size to the adults, which is why we believe their identification is correct. Based on the large size of juveniles, which already bear characters of the morphology of the adults, we also consider the B. goudoti tadpoles described by Blommers-Schlö sser (1979b) as likely to be correctly identified. The tadpole of B. occidentalis, described by Andreone et al. (2002), was reared from eggs deposited by a large aggregation of breeding adults, and its identification is therefore beyond doubt. The tadpole of B. jaegeri as described by Glaw and Vences (1994) was collected at the type locality of this species, Nosy Be, where only one additional stream-breeding Boophis occurs (B. brachychir). The B. jaegeri tadpole is similar in morphology to those of B. luteus and B. aff. elenae as described herein, which are its close relatives according to molecular data , and therefore it is likely to be correctly assigned. The B. ankaratra tadpole as described by Glaw and Vences (1994) originates from a montane site where it is the only bright green-coloured Boophis species, increasing the confidence in a correct assignation of the green juvenile that was reared through metamorphosis. Furthermore, we have recently identified further tadpoles of this species using DNA barcoding, and these specimens fully conform to the previous description. The tadpole of Boophis mandraka, described by Blommers-Schlö sser (1979b) from the type locality of this species, agrees in morphology to that of a close relative of this species, B. sambirano Vences and Glaw, 2005, which is currently being described by R.-D. Randrianiaina; this makes it likely that the B. mandraka tadpole has also been correctly identified.
Two other tadpole descriptions of Blommers-Schlö sser (1979b), however, are equivocal. The larva of B. majori described by this author from the locality Mandraka almost certainly belongs instead to B. marojezensis which was described subsequent to Blommers-Schlö sser's works (Glaw and Vences 1994). Indeed, adults from Mandraka collected by Blommers-Schlö sser (1979b) were re-determined as B. marojezensis (Glaw et al. 2001), and the tadpole description conforms to that of B. marojezensis herein. More complex is the case of the B. erythrodactylus tadpole from the same locality, Mandraka (Blommers-Schlö sser 1979b). This species clearly belongs in the B. rappiodes group according to molecular data, and is sister to B. tasymena as B. sp. aff. erythrodactylus). Nevertheless, the tadpole described by Blommers-Schlö sser (1979b) is highly different from that of B. tasymena and two other species of the B. rappiodes group (B. rappiodes and B. viridis; Blommers-Schlö sser, 1979b and description herein), instead resembling those of B. mandraka and B. ankaratra, which are known from the same locality as well. Because of this indirect evidence we consider the identity of the tadpoles assigned by Blommers-Schlö sser (1979b) to B. erythrodactylus as in need of confirmation.

Lotic tadpole adaptations and Boophis phylogeny
Our data, in combination with published tadpole descriptions of other Boophis species (Blommers-Schlö sser 1979b; Vences 1992, 1994;Andreone et al. 2002;Grosjean et al. 2006), provide evidence for varying degrees of larval adaptation to running water in the stream-breeding clade of this genus. In the field we observed many times how species with large oral discs and strong caudal musculature were mainly distributed in fastflowing parts of streams, and how these tadpoles attached to the surface of rocks and, later, of collecting bags or buckets with their oral discs. We are therefore convinced that these morphological features, large oral discs with high numbers of smaller papillae and keratodont rows, and stronger caudal musculature with reduced height of dorsal fins, are to be interpreted as adaptations to the current in the lotic environment of these tadpoles, conforming to Altig and Johnston (1989). More extreme evolution of these features is observed in the members of the gastromyzophorous guild such as the unrelated toadlets of the genus Atelopus, and the stream-breeding ranids of the genus Amolops, which have developed large ventral suctorial discs (Inger 1966;Lö tters 1996).
For the purpose of generalization, we think that three guilds of lotic Boophis tadpoles can be distinguished, although intermediate adaptive states certainly exist and assignation is therefore certainly not unequivocal.  Glaw and Vences (1994) and Grosjean et al. (2006), B. rappiodes, B. brachychir, and B. goudoti can also be tentatively assigned to this guild. 2. Guild B contains intermediate forms, such as Boophis luteus, B. aff. elenae, and possibly Boophis albilabris and B. madagascariensis, although these latter species may also be included in Guild A. These species show a distinct enlargement of the caudal musculature, a reduction of the upper fin in its anterior part (especially in the two former species), and a rather large oral disc with an increase in the number of keratodont rows, although the size of the keratodont rows and the size and number of the oral papillae, and the horny beak, are not conspicuously modified. Their ODW is 43-63% of BW, they have one or two rows of 69-164 marginal papillae, and the first upper keratodont row has 164-238 keratodonts. The tadpoles of B. microtympanum and B. laurenti may also be assigned to this guild, although these montane larvae are deviant in being rather large, and in having high numbers of keratodont rows. According to the descriptions of Glaw and Vences (1994), the tadpole of B. jaegeri can also be assigned to this guild, as can possibly the larva of B. occidentalis which is similar to that of B. albilabris . 3. Guild C contains the most highly modified tadpoles, with a very strong caudal muscle, a flat body ventrally, a large ventral suctorial oral disc, enlarged number of keratodont rows and high density of keratodonts, a large number of small oral papillae which in some cases tend to close the dorsal papilla gap, and often with a reduced size of the horny beak, reduction or absence of serrations on the upper beak, and a ''ribbed'' lower beak (one species with a less keratinized median area) characteristic of the stream-dwelling gastromyzophorous tadpoles of the genus Meristogenys (Inger 1966). The tadpoles of Boophis marojezensis and of B. sibilans clearly belong in this guild; their ODW is 63-89% of BW, there are multiple rows of many small marginal papillae, and the first upper keratodont row has many small keratodonts which are difficult to count, but certainly are more than 200 (203-208 in the third or fourth upper keratodont row). The previously described larva of B. mandraka (Blommers-Schlö sser 1979b) belongs in this guild as well, as probably does the tadpole of B. ankaratra that was briefly described by Glaw and Vences (1994). The montane specialist B. williamsi may also be assigned to this guild, although its large tadpoles do not show any obvious reduction of the horny beak (Blommers-Schlö sser 1979b; and our own data).
Using these guild designations as a tentative yardstick to measure adaptations to fastflowing waters in stream-breeding Boophis, several correlates of phylogenetic relationships are obvious, but in other cases convergence is to be invoked to explain morphological similarities (see Table II). The species of Guild A belong in three Boophis species groups, the B. goudoti group (B. boehmei, B. brachychir, B. reticulatus, B. goudoti), the B. rappiodes group (B. rappiodes, B. tasymena, B. viridis), and the B. majori group (B. pyrrhus). Of these, the B. goudoti group seems to be phylogenetically fairly well-defined by molecular data (e.g. Vences et al. 2002), and although a tadpole of Guild B is included, Guild A larvae may be a common (though probably plesiomorphic) feature of all taxa in this group, which to a large extent are known to breed in slow-flowing or near-stagnant stretches of streams (Blommers-Schlö sser 1979b; Glaw and Vences 1994). An evaluation of the pattern in the B. rappiodes group is complicated by the description of a Guild C larva in B. erythrodactylus (Blommers-Schlö sser 1979b). After exclusion of the species of the B. mandraka group, the B. rappiodes group is highly likely to be monophyletic , and a Guild C larva in B. erythrodactylus would be a clear instance of convergent evolution. However, as discussed above, this record is in need of confirmation. Although several species of these small treefrogs often call along large and rather fast-flowing waters, their larvae may indeed be adapted to more slow-flowing sections of these streams. The discovery of a Guild A larva in a species of the B. majori group (B. pyrrhus) appears surprising, since the only other known tadpole of this group belongs in Guild C (B. marojezensis). However, in light of the fact that this species group is probably not monophyletic , larval morphology may help to reconcile phylogeny and classification.
The species of Guild B belong in four species groups, the B. albilabris group (B. albilabris, B. occidentalis), the B. goudoti group (B. madagascariensis), the B. luteus group (B. jaegeri, B. luteus, B. sp. aff. elenae), and the B. microtympanum group (B. laurenti, B. microtympanum). All these groups are likely to be monophyletic if the B. albipunctatus group is excluded from the B. luteus group . In the case of the B. luteus and B. albilabris groups, it is likely that most or all included species will have a Guild B morphology. The third species of the B. microtympanum group (B. williamsi) is here included in Guild C, but its morphological and genetic relationships to B. laurenti and B. microtympanum are wellsupported, and it shows just a slightly more extreme expression of their montane larval morphology. Likewise, B. madagascariensis probably just shows a slightly more expressed adaptation to stream-life compared to other members of the B. goudoti species group, which have Guild A tadpoles.
The species of Guild C belong in four species groups, the B. mandraka group (B. mandraka), B. majori group (B. marojezensis), B. albipunctatus group (B. ankaratra, B. sibilans), and the B. microtympanum group (B. williamsi). Of these, the B. mandraka group is monophyletic . All species of this group were found along fastflowing streams, and we consider it as likely that all of them have Guild C tadpoles. The B. albipunctatus group was recently split from the B. luteus group  Table II. Summary of published tadpole descriptions in stream-breeding species of Boophis (all species except those of the Boophis tephraeomystax group) (the table lists descriptions according to the name used in the respective publication, grouped according to current classification in phenetic species groups; tadpole guilds are used as defined in the Discussion).
Species and species group based on molecular phylogenetic data, and may also be characterized by Guild C tadpoles in its entirety. The B. majori group is probably not monophyletic (see discussion of B. pyrrhus under Guild A above), and more data are necessary to understand relationships among species of this assemblage. Finally, for B. williamsi, see discussion under Guild B above.
As a conclusion, the distribution of larval morphologies among stream-breeding Boophis conforms rather well to current phylogenetic knowledge. Where species groups are known to be monophyletic, they appear to be characterized by a generally similar larval morphology, and where more than one larval morphology occurs in a species group, its non-monophyly has already been ascertained (like the B. majori group) (Table II). Hence, rapid and frequent adaptive shifts expressed in larval morphology seem to be uncommon in this genus of frogs. Nevertheless, it is highly probable that convergent adaptation to fastflowing aquatic environments has taken place repeatedly. This is indicated by (1) the presence of Guild C morphologies in representatives of different species groups that do not have obvious phylogenetic relationships, and (2) the fact that several species of Guild C, although having apparently similar ecomorphological adaptations, show important morphological differences: for example, in Guild C, B. ankaratra and B. mandraka have a (plesiomorphic) dorsal gap in the rows of marginal papillae which is closed in B. marojezensis; and B. williamsi has no obvious reduction of the horny beak which is apparent in other species such as B. marojezensis. A precise analysis of these convergences requires a stable phylogeny on which morphological changes can be plotted, but unfortunately the available data so far are not conclusive with respect to the basal relationships among species groups Vences and Glaw 2005). Obtaining more reliable phylogenies and completing the taxonomic inventory of Boophis larvae will allow further understanding of the importance of reproductive and larval traits as possible key innovations in the evolution of mantellid frogs.