Blommersia wittei

Blommersia wittei (Guibé, 1974) Figures 3, 8 Identity and type material. Clarifying the taxonomy of the B. wittei complex requires first ascertaining the identity of B. wittei sensu stricto. The species was described by Guibé (1974) as Mantidactylus wittei, and type specimens are deposited in the Paris museum (acronym MNHN, in earlier times written MNHNP), with the following verbatim information in the original description: “ Holotype: no 1953-60 MNHN Paris. Environs d’Ambanja. J. Guibé (11- 1951). Paratypes: nos 1953-60 A à 1953-60 L MNHN Paris. Même provenance. Nos A.682 à A.684: Ampijora , station forestière à 45 km de Marovaoy (Ouest). Nos 1973-944 à 1973-951 MNHN Paris: forêt d’Ankarafantsika. Ch. P. Blanc (11-1973). ” Due to some imprecise information in the handwritten MNHN catalogue and extensive re-labelling of several of these specimens, there has been confusion on the type series in subsequent publications and in the current digital MNHN catalogue. Frost (2023) correctly states that the holotype is MNHNP 1953.60, by original designation, but merges two separate sites in the type locality account: “ ‘ Environs d’Ambanja’, Forest Ankarafantsika, Madagascar. ” In contrast, the digital type catalogue of the MNHN (downloaded from gbif.org in 2022) as well as the MNHN online catalogue (https://science.mnhn.fr/institution/mnhn/collection/ra/; accessed 11 February 2023) state that the holotype originated from “ Vallée de la Tsiribihina ” and furthermore provide the following list of paratypes: MNHN 1973.944–1973.951 (8 specimens) from Ankarafantsika, and 1993.682–1993.684 (3 specimens) from Ampijoroa. Vences et al. (2010) provided measurements of part of the type series of B. wittei, purportedly all from Ambanja: the holotype MNHN 1953.60, and the following paratypes: MNHN 1991.2529–1991.2533 (5 specimens; previously labelled MNHN 1953.60 A to E) and MNHN 1991.2536–1991.2539 (4 specimens, previously labelled MNHN 1953.60 H to K). After inspecting the original (handwritten) MNHN catalogues (partly available from https://science.mnhn.fr/ catalogues; accessed 11 February 2023) we conclude that: (1) The holotype of Mantidactylus wittei Guibé, 1974 (Fig. 8) unambiguously is the specimen MNHN 1953.60 for which morphometric measurements were provided by Vences et al. (2010) and whose locality, according to the original description and original catalogue entry is “ Environs d’Ambanja”, with the addition in the catalogue: “Cacaoyéres” (= cacao trees, thus indicating the specimen was collected in a cacao plantation). (2) A series of 11 paratypes (originally labelled MNHN 1953.60 A to 1953 K, currently re-labelled as MNHN 1991.2529–1991.2539) originates from the same locality as the holotype. (3) A series of 8 paratypes (MNHN 1973.944–1973.951) originate from Ankarafantsika, (4) A series of 3 paratypes (MNHN 1993.682–1993.684) originate from Ampijoroa, which represents a forestry station (currently local headquarters of Madagascar National Parks) within the Ankarafantsika forest. As explained in the Taxonomic conclusion account above, based on the provenance of the holotype from Ambanja, we continue assigning the name B. wittei to the genetic lineage occurring across much of northern Madagascar and collected by us at Ambanja and nearby localities. The paratype series is however mixed: specimens from Ankarafantsika (MNHN 1973.944–1973.951) and Ampijoroa (MNHN 1993.682–1993.684) are here assigned to B. sp. Ca 5. Only the 11 paratypes from the type locality are likely conspecific with the holotype. Morphology. Measurements of the type series of B. wittei have been published by Vences et al. (2010) and measurements of additional specimens by Pabijan et al. (2011). Measurements of further specimens are included in Table 1 herein.A full set of measurements including those from the previous publications is given as Supplementary Table 2 (available from the Zenodo repository under DOI 10.5281/zenodo.8049142). Based on these data, male SVL is 21.0–27.0 mm and female SVL is 20.7–25.0 mm. In the specimens examined for the present study (listed in Table 1), vomerine teeth are usually recognizable (clearly visible in specimens from the type locality Ambanja) but weakly expressed in several specimens from Montagne d’Ambre. In ZSM 563/2000 from Sambava, vomerine teeth are weakly recognizable, and this specimen apparently has more developed webbing than many other individuals examined. Vocalizations.Advertisement calls recorded in February 1991 at the type locality Ambanja (recording temperature not taken) consist of a single short pulsed note repeated in call series at regular intervals and very fast succession (Fig. 5). Recording quality is poor and the detailed call structure is probably partly masked by background noises. However, each call (= note) seems to contain two pulse groups which are separated from each other, with the first pulse group being of lower amplitude. Pulse structure in not very obvious in the recording and most pulses appear basally fused. Maximum call energy is distributed towards the middle of the call’s duration. Numerical parameters of 16 analyzed calls of one male are as follows: call duration (= note duration) 62–77 ms (66.8 ± 5.3 ms); inter-call intervals within regular call series 15–31 ms (18.6 ± 5.6 ms); pulses/note 9–18 (13.8 ± 3.3); duration of regular call series 1275 ms (n = 1); call rate within call series approximately 740 calls/minute; dominant frequency 4694–4886 Hz (4797 ± 82 Hz); second frequency peak around 2500 Hz; prevalent bandwidth 1800–6000 Hz. Advertisement calls recorded on 7 February 1992 from north of Andoany, Nosy Be (air temperature 25°C), consist of a single short pulsed note repeated in short call series at regular intervals and very fast succession (Fig. 5). Each call (= note) exhibits two pulse groups which are clearly separated from each other, with the first pulse group being of lower amplitude. Pulses are partly fused, but countable. Maximum call energy is distributed in the middle of the call, namely the beginning of the second pulse group. Numerical parameters of 10 analyzed calls of one male are as follows: call duration (= note duration) 65–83 ms (74.8 ± 6.3 ms); inter-call intervals within regular call series 16–32 ms (21.5 ± 5.6 ms); pulses/note 14–19 (16.3 ± 1.9); duration of regular call series 490 and 500 ms (n = 2); call rate within call series approximately 640 calls/minute; dominant frequency 4823–5240 Hz (5033 ± 163 Hz); second frequency peak around 2500 Hz; prevalent bandwidth 2000–8200 Hz. Advertisement calls recorded on 15 March 1994 on Montagne d’Ambre (air temperature 22°C), consist of a single short, distinctly pulsed note repeated in call series at regular intervals and very fast succession (Fig. 5). Each call (= note) exhibits two pulse groups which are clearly separated from each other, with the first pulse group being of lower amplitude. Pulses are barely fused and rather distinctly separated. Maximum call energy is distributed in the middle of the call, namely the beginning of the second pulse group. Numerical parameters of 19 analyzed calls of one individual are as follows: call duration (= note duration) 59–73 ms (65.4 ± 4.2 ms); inter-call intervals within regular call series 22–39 ms (29.6 ± 7.1 ms); pulses/note 7–11 (8.9 ± 1.3); duration of regular call series 1821 ms (n = 1); call rate within call series approximately 600 calls/minute; dominant frequency 4489–4597 Hz (4542 ± 41 Hz); prevalent bandwidth 1800–7500 Hz. Advertisement calls recorded from specimen ZSM 50/2018 (MSZC 521) on 25 December 2017 on Montagne d’Ambre (air temperature ca 19°C), consist of a single short, distinctly pulsed note repeated in call series at regular intervals and very fast succession (Fig. 5). Each call (= note) exhibits clearly separated pulses. Intervals between pulses are somewhat irregular, in some calls resulting in two pulse groups separated by a larger interval. Maximum call energy is evident in the middle of the call. Numerical parameters of 16 analyzed calls of one male are as follows: call duration (= note duration) 68–88 ms (78.2 ± 5.7 ms); inter-call intervals within regular call series 28–50 ms (41.4 ± 4.2 ms); pulses/note 7–11 (9.0 ± 1.3); duration of regular call series 1590 ms (n = 1); call rate within call series approximately 500 calls/minute; dominant frequency 5066–5232 Hz (5137 ± 80 Hz); prevalent bandwidth 2000–10000 Hz. Advertisement calls recorded on 20 March 2000 in Sambava (air temperature 24.7°C), here tentatively allocated to B. wittei, consist of a single, short, pulsed note repeated in call series at regular intervals within series. Calls (= notes) exhibit a pulsed structure, but most pulses are largely fused, resulting in an irregular pulse pattern and varying number of countable pulses (Fig. 7). Maximum call energy is distributed among the first two thirds of the call’s duration. Numerical parameters of 22 analyzed calls of one male are as follows: call duration (= note duration) 27– 59 ms (39.4 ± 11.8 ms); inter-call intervals within regular call series 54–76 ms (67.9 ± 7.7 ms); pulses/note 2–6 (4.1 ± 1.0); duration of regular call series 2209 ms (n = 1); call rate within call series approximately 600 calls/minute; dominant frequency 5316–5598 Hz (5440 ± 145 Hz); second frequency peak around 2600 Hz; prevalent bandwidth 2100–8600 Hz.