Integrative taxonomic revision of mantellid frogs of the genus Aglyptodactylus (Anura: Mantellidae) Author Köhler, Jörn Author Glaw, Frank Author Pabijan, Maciej Author Vences, Miguel text Zootaxa 2015 4006 3 401 438 journal article 10.11646/zootaxa.4006.3.1 9b8d3194-5435-46fb-92ee-301fd23b0782 1175-5326 240836 136F727A-1628-449F-B762-254371D27360 Aglyptodactylus madagascariensis ( Duméril, 1853 ) Identity. Described as Limnodytes madagascariensis from Madagascar , without precise locality data ( Duméril 1853 ). This is the earliest nomen in the genus and the type species of Aglyptodactylus and therefore certainly is valid. Due to the scarcity of clear morphological data it is difficult to assign this nomen without ambiguity to any of the distinct species previously subsumed under A. madagascariensis , although the size of the metatarsal tubercle clearly demonstrates it does not refer to A. laticeps or A. securifer (Glaw et al. 1998) . We here assign the type specimens of madagascarienis to the highland populations to which this name has usually been applied (e.g., Blommers-Schlösser 1979 ), rather than to any of the coastal populations, using the following rationale: (1) Duméril (1853) mentions a faint light middorsal line, a character that is common in the highland populations (e.g., in four out of 10 ZSM specimens from the Besariaka/Andasibe area; see male specimen in Fig. 9 d), a character we did not detect in any of at least 12 specimens examined from coastal populations (i.e., specimens from Table 3 plus some others for which only photographs are available; here assigned to A. inguinalis and A. chorus , see below). (2) Of the morphometric values differentiating the highland from the coastal populations (especially relative HL and END ) the values of the type (0.385 and 0.092) are closer to the values of the highland populations (see section on morphometric differentiation above). For morphological measurements see Table 3. (3) Together with A. madagascariensis, Duméril (1853) also described two other Malagasy frogs, Mantidactylus lugubris and Boophis tephraeomystax . While the latter is mainly a lowland species (but also occurs at midelevation sites such as Andasibe), the former is relatively rare at low elevations, suggesting that maybe also the type of A. madagascariensis was collected at mid-elevations (here for brevity referred to as highland). FIGURE 9 . Photographs in life of Aglyptodactylus madagascariensis . Specimens from Montagne d'Ambre (a,b), Fierenana (c,f), and Andasibe (d,e), with (d) showing a pair in amplexus. Photographed individuals not unequivocally assignable to preserved voucher specimens. A further distinction between the highland versus the coastal lineage is the lack of clear marbling on posterior surfaces of thighs in the latter ( Fig. 5 ). Unfortunately no such pattern is recognizable in the type upon examination in 2014, and also Duméril (1853) does not mention it in his original description. However, also in some other highland specimens (e.g. ZSM 350/2000) this pattern is poorly expressed. Given that A. madagascariensis is very common at highland areas such as Andasibe or Ranomafana and the name has regularly been applied to specimens from these sites (e.g., Blommers-Schlösser 1979 ) we propose to continue common usage and apply the name A. madagascariensis to the most widespread Aglyptodactylus lineage which occurs in the highlands (mid-elevation rainforest) and ranges from Montagne d'Ambre in the north to Andringitra in the south-east ( Fig. 2 ). Remark. We here continue considering the nomen Mantidactylus purpureus Ahl, 1929 as a junior synonym of A. madagascarienis (see Guibé 1978 ). This taxon was described on the basis of a syntype series purportedly from "NW Madagaskar " ( Ahl 1929 ). However, the syntypes differ in multiple morphological characters from A. securifer , the species commonly encountered in north-western Madagascar (Glaw et al. 1998), such as relative finger length, webbing, size of inner metatarsal tubercle, and hindlimb length. As discussed previously (e.g., Vences & Glaw 2004 ), it is likely that these specimens were collected during Hildebrandt’s last trip on his 1881 travel route in central and eastern Madagascar (see Beentje 1998 ), along with several other frog species typical for mid- and high-elevation localities in this region. Furthermore, most of the syntypes (especially the largest female, ZMB 53728) have a distinct marbling on the posteroventral portion of thigh. It is therefore almost certain that the type series of M. purpureus belongs to the highland lineage for which the earlier name A. madagascariensis is available. Call. The advertisement call of populations referable to A. madagascariensis has been described by Glaw & Vences (1994) from three different localities (Andasibe, Andringitra, Mandraka). The call consists of a series of pulsed notes repeated at regular intervals, with inter-note intervals being shorter than note duration. Within notes, amplitude almost constantly increases from beginning to the end of notes. Our analysis of more recent recordings from Andasibe (Vences et al. 2006: CD1/Track 2; Fig. 10 ) largely confirmed the characteristics provided by Glaw & Vences (1994) with regard to pulse rate within notes, inter-note intervals, and frequency range. However, we found a higher number of notes per call (3‒9 vs. 3‒5) in calls from Andasibe ( Fig. 10 ) compared to the previous report ( Glaw & Vences 1994 ), most likely a result of different motivational state in recorded males. Calls from Maharira (Ranomafana region) and Andringitra agree in parameters with calls from Andasibe ( Fig. 3 , Table 2). Some remarkable differences are evident when comparing our results with the call parameters of earlier recordings from Mandraka published by Blommers-Schlösser (1979) . Mandraka calls were described as having longer note duration (280‒320 vs. 140‒211 ms) and a higher number of pulses per note (36 vs. 14‒21). However, we re-analysed the respective Mandraka call recordings (provided by R. Blommers-Schlösser) and our analysis revealed call characters in agreement with other calls referred to A. madagascariensis (see Table 2). The numerical parameters provided by Blommers-Schlösser (1979) were likely based on the analysis of analogous sonagrams with high frequency resolution, resulting in a poor resolution of temporal parameters. Despite differences, A. madagascariensis calls within the genus are most similar to those of A. australis (see respective section for A. australis below). For comparative parameters of A. madagascariensis calls from Andasibe, Mandraka and Andringitra see Table 2. FIGURE 10 . Audiospectrogram (above) and oscillogram (below) of an advertisement call of Aglyptodactylus madagascariensis , recorded at Andasibe. Air temperature 22.6 °C. Natural history. Stomach content of specimens from different highland localities has been analysed by Vences et al. (1999a) . Females contain up to 1325 eggs ( Vences et al. 1999b ). The tadpole has been described by Blommers-Schlösser (1979) and more recently, Randrianiaina et al. (2011) provided figures of A. madagascariensis tadpoles and their mouthparts. The endocommensal “protist” Protoopalina perantonii was described from A. madagascariensis from Andasibe ( Delvinquier et al. 1998 ). An undescribed species of polystome flatworm, Metapolystoma sp. 1, has been recovered from the bladder of a specimen probably assignable to A. madagascariensis ( Verneau et al. 2009 ) . Distribution. As defined herein, A. madagascariensis is distributed in rainforests at elevations between approximately 650‒1510 m a.s.l. Based on genetic and/or bioacoustic data reported herein, we consider the following localities as confirmed for A. madagascariensis : Andringitra (Imaitso forest, 22°08'25'' S , 46°56'49'' E , 1509 m a.s.l.—highest and southernmost locality), Andasibe, Besariaka, Fierenana, Forêt d’Ambre, Makira, Mandraka, Montagne d’Ambre, Tsinjoarivo, Ranomafana and Vohiparara ( Fig. 2 ; see also Blommers-Schlösser 1979 ; D’Cruze et al. 2008).