Contributions to Zoology, 86 (4) – 2017Samuel G. Penny; Angelica Crottini; Franco Andreone; Adriana Bellati; Lovasoa M.S. Rakotozafy; Marc W. Holderied; Christoph Schwitzer; Gonçalo M. Rosa: Combining old and new evidence to increase the known biodiversity value of the Sahamalaza Peninsula, Northwest Madagascar

To refer to this article use this url:

Molecular taxonomic identification

Tissue samples were collected with a maximum of five individuals per species-level taxon per population. If individuals appeared to belong to new and undescribed species, a limited number of voucher specimens were collected, as advised by the Code of Zoological Nomenclature (ICZN 1999). These were anaesthetised (by immersion in MS222), and fixed in 10% buffered formalin or 90% ethanol, and later transferred in 65-75% ethanol. Voucher specimens were deposited in the Museo Regionale di Scienze Naturali, Torino, Italy, the Parc Botanique et Zoologique de Tsimbazaza (PBZT), Antananarivo, Madagascar, and Mention Zoologie et Biodiversité Animale, Faculté des Sciences, Université d’Antananarivo, Madagascar (UADBA). Most of the tissue samples were collected in the 2013 expedition and only a small number of tissue samples were collected in the 2011-2012 surveys.

Total genomic DNA was extracted from the tissue samples using proteinase K digestion (10 mg/ml concentration) followed by a standard salt extraction protocol (Bruford et al., 1992). A fragment of ca. 550 bp of the 3’ terminus of the mitochondrial 16S rRNA gene (16S), proven to be suitable for amphibian identification (Vences et al., 2005a), was amplified for 78 amphibian tissue samples, while a fragment of around 650 bp of the standard barcoding region of the cytochrome c oxidase subunit I gene (COI) (Nagy et al., 2012) was amplified for 42 reptile tissue samples and one amphibian (Table S1). In reptiles the molecular taxonomic identification using the mitochondrial COI fragment was not possible for some taxa. In these instances, the mitochondrial gene fragments 16S or NADH dehydrogenase subunits 1, 2 and 4 (ND1, ND2, ND4) were amplified and sequenced for a selected number of samples to allow a finer taxonomic identification (see Table S1). For primers and cycling protocols see Table 1. All fragments were sequenced using an ABI 3730XL automated sequencer by Macrogen Inc.


Table 1. Primer information (gene fragment, primer name, sequence, literature source) and PCR conditions used for the present study.

Chromatographs were checked and sequences were edited, where necessary, using the BioEdit sequence alignment editor (version; Hall, 1999). To assess the species attribution and the genetic distinctness of each taxa, sequences of each morphological taxa were compared among each other and each sequence was than compared using the BLAST algorithm in GenBank.

Some specimens could not be assigned to any described or identified candidate species as in Vieites et al. (2009), Perl et al. (2014) or Nagy et al. (2012). For these taxa we applied the terms and abbreviations, confirmed candidate species (CCS), unconfirmed candidate species (UCS) and deep conspecific lineage (DCL) as defined by Vieites et al. (2009). Working names of the already identified candidate species follow Perl et al. (2014) for amphibians and Nagy et al. (2012) for reptiles. Additionally, when available, we used the names proposed by Glaw and Vences (2007) which usually prefix the species epithet with ‘‘sp. aff.’’ of the morphologically closest described species or a descriptor that is either geographic or refers to a characteristic trait of the candidate species. Candidate species of amphibians were identified based on a threshold of 5% minimum divergence for the 16S fragment (Vences et al., 2005a; Fouquet et al., 2007; Vieites et al., 2009), whereas candidate species of reptiles were identified following the different thresholds proposed for the different groups as in Nagy et al. (2012). Obtained sequences were submitted to GenBank (Accession Numbers are available in Table S1) and reptile COI sequences were associated to the BOLD database.

Automated acoustic recording took place at 37 locations. Recordings were made with a single Song Meter SM2 digital recorder (Wildlife Acoustics Inc, Concord, USA) at a 16-bit resolution and 16 kHz sampling rate using two side-mounted SMX-II microphones. The digital recorder was placed one to two metres above the ground/water by securing it with bungee leads to deadwood or a protruding branch. Acoustic recordings were made between sunset and sunrise over 60 nights, when frog activity is greatest (Glaw and Vences, 2007). Continuous recordings split into sections of 120 minutes each were saved in the standard uncompressed .WAV format. Preceding analysis recordings were split using a custom-written MATLAB (The Mathworks, Natick, USA, V7.14.0.739) script into minute long segments to allow for more efficient analysis. Spectrograms were viewed individually as a dual channel output using Avisoft SASlab Pro (Berlin, Germany, V5.2.06); a Hamming window with FFT window size of 512, with 100% frame, and an intensity threshold of 50% were used to create spectrograms. Species were distinguished by matching their temporal and spectral patterns with that of known reference recordings (S. Penny) and an acoustic library of Malagasy frogs (Vences et al., 2006; Rosa et al., 2011). This was achieved by both ear and through taking parameter measurements with Avisoft SASLab Pro (Avisoft SASlab Pro; Berlin, Germany; V5.2.06).