Bioacoustics of Mantidactylus katae

Identity and justification.— This lineage has been considered as confirmed candidate species M. sp. 28 by Vieites et al. (2009) and M. sp. Ca28 by Perl et al. (2014) . It was reported as ‘ Mantidactylus sp. aff. betsileanus “slow calls”’ by Glaw and Vences (2007) , and has previously been considered as M. multiplicatus (e.g. Poth et al. 2012 , 2013 ); however, DNA barcodes obtained in the present study have shown that this assignment was wrong and that the holotype of M. multiplicatus is conspecific with M. betsileanus , and the nomen therefore a junior synonym of M. betsileanus (see account of that species). Mantidactylus katae is a widespread species with a unique call, characterized by a very low pulse repetition rate (or rather, consisting of a single-pulse note repeated regularly). It also is genetically distinct in mitochondrial DNA and does not appear to share Rag-1 haplotypes with other similar species, including the sympatric (and syntopic) M. betsileanus . Unfortunately, due to failure of other samples, M. katae sp. nov. is only represented by two individuals from the South East of Madagascar in our phylogenomic tree, from where no reliable call recordings are available. The samples (FGZC 163 and ZCMV 14842) cluster closely with M. tripunctatus . However, these samples correspond to those alleles that also in the Rag-1 haplotype network cluster close to M. tripunctatus , and we cannot exclude that they represent individuals of M. tripunctatus with an introgressed M. katae mitochondrial genome. The exact phylogenetic position of M. katae therefore remains in need of confirmation; the phylogenetic tree of Wollenberg et al. (2011) , based on multiple mitochondrial genes, placed it (as M. sp. 28) in a clade with M. noralottae and M. tripunctatus (as M. sp. 29), which would agree with the affinities suggested by our phylogenomic tree. Typical M. katae from the Southern Central East and Northern Central East of Madagascar differ from both M. noralottae and M. tripunctatus very strongly in advertisement call structure, by femoral gland morphology, and do not share Rag-1 haplotypes with these two species, thus leaving no doubt about the species status of this lineage.

Diagnosis .— Mantidactylus katae sp. nov. is a member of the M. betsileanus clade, related to M. noralottae and M. tripunctatus . See Table 4 for a list of diagnostic morphological characters. The combination of a relatively small body size (male SVL 22–27 mm ), slightly tubercular dorsal skin with distinct continuous dorsolateral ridges, absence of white spots on flanks, presence of a white marking on snout tip, large femoral glands (FGW up to 13% of SVL) that contact each other medially, and advertisement call consisting of a single pulse repeated at a slow rate of 10–16 calls per second distinguishes M. katae sp. nov. from species of all other clades.The only other species with a similar advertisement call structure is M. fergusoni sp. nov. which has a larger size and more tubercular dorsal skin (see account of that species below). Within the M. betsileanus clade, the new species can be distinguished from all other species by its unique call structure and larger femoral glands ( Table 4 ); furthermore from M. noralottae by smaller size of males and presence of a distinct white marking on snout. For a field diagnosis from the syntopic species M. betsileanus , according to our measurements, M . katae differs by a smaller dot on the snout tip (dot on the snout tip is 9–14% of head width vs 13–24%), a smaller distance between the femoral glands (distance between the femoral glands is 0– 14% of SVL vs 13–22%), larger femoral glands (femoral gland length is 19–31% of SVL vs 13–22%), and a higher number of granules per femoral gland (5–8 vs 1–5). For a detailed distinction from other new species described herein, see the respective species accounts. A full list of molecular diagnostic sites in the 16S gene of M. katae sp. nov. in pairwise comparisons to all other Brygoomantis species is provided as Supplementary appendix.

Calls .—The advertisement call of M. katae , recorded on 17 December 1994 at Andasibe, ca 20°C air temperature ( Vences et al. 2006 : CD2, track 64, cut 2), consists of a simple, very short, single pulse ‘click’ note, emitted in long series at regular intervals and fast succession ( Fig. 43 ). The duration of the call series analysed from Andasibe was 7568 ms. Numerical parameters of 47 analysed calls were as follows: call duration (= note duration) 2–5 ms (3.6 ± 0.8 ms); 1 pulse per note (1.0 ± 0.0); pulse duration = note duration = call duration; dominant frequency 3036–3165 Hz (3108 ± 46 Hz); prevalent bandwidth 1250–3540 Hz; call repetition rate (= note repetition rate = pulse repetition rate) within regular series 696–971 calls/min (840 ± 106 calls/min). Calls recorded on 24 January 2004 at Maharira forest, Ranomafana National Park, 18.4°C air temperature ( Vences et al. 2006 : CD2, track 64, cut 1) are in agreement with the calls from Andasibe in all spectral and temporal parameters, except for slightly lower dominant frequency ranging from 2713–2993 Hz (2833 ± 123 Hz), and slightly lower call repetition rate, ranging from 515–731 calls/min (626 ± 109 calls/min). Both slight differences can be explained by potentially larger body size of the calling male and lower temperature during recording and leave no doubt about the conspecificity of the calls from the two localities. Another call recording from Andasibe, obtained on 20 March 1995 , 23.4°C air temperature ( Vences et al. 2006 : CD2, track 63) is similar to the calls described above, but differs by shorter call series (2226–2520 ms), slightly longer call duration (= note duration = pulse duration) of 8–10 ms, and lower dominant frequency of 1323–1433 Hz (1365 ± 52 Hz). However, the latter might be due to recording equipment and/or recording conditions, as the calls described above have a second frequency peak that roughly corresponds to the lower dominant frequency range described here. Call repetition rate (= note repetition rate = pulse repetition rate) is in the same range compared to the calls described above, ranging from 630–857 calls/ min (741 ± 90 calls/min). Tadpoles .—The tadpole of M. katae was described by Knoll et al. (2007) under the name ‘ M. sp. aff. betsileanus “very slow calls”’.

Natural history .—Occurs along running water bodies in rainforest, where specimens can often be seen calling at night from muddy banks of somewhat larger, slow-moving streams. It often occurs in syntopy with M. betsileanus in rainforest, including forest fragments ( Riemann et al. 2015 ), and along streams with at least a narrow gallery forest band surrounded by degraded or unforested areas. However, in contrast to M. betsileanus it is not found in rice fields or other plantations ( Ndriantsoa et al. 2017 ). Usually sitting in shallow water or along the shore, hiding in the leaf litter or perched on low vegetation up to 0.5 m hight in the vicinity of a stream. Found at an elevational range between 450–1142 m a.s.l. in Ranomafana and surrounds. Females with visible eggs (transparent abdominal wall) were observed in May 2010 and from January to June in 2011 in the Ranomafana area.