**3. Results**

We recovered a phylogenetic tree (Figure 2) that was largely congruen<sup>t</sup> with previous analyses [2,24]. However, our tree recovered three species of *Bryophryne* not previously included in phylogenetic analyses (*B. gymnotis*, *B. flammiventris*, and *B. mancoinca*) as a clade that is sister to the clade containing all species of *Microkayla*. Thus, species of *Microkayla*, instead of other species of *Byrophryne*, share the most common shared ancestor with *B. gymnotis*, *B. flammiventris*, and *B. mancoinca*. The presence of large, external tympanic membrane and annulus, and males with a median subgular vocal sac and production of vocalizations, immediately distinguishes the newly recognized genus from all other species of *Bryophryne*. At least four species of *Bryophryne* were described as having small, barely visible (under the skin surface) tympanic membranes and annuli (*B. bustamantei*, *B. quellokunka*, *B. tocra*, *B. wilakunka*), but their external appearance does not look that different from the other species of *Bryophryne* known to lack a visible tympanic membrane [2,14,18]. One of these species, *B. bustamantei* was described as producing a short whistle, but there is no recording of the call nor voucher associated with a call [18]. The distribution range of *B. bustamantei* overlaps with that of *B. gymnotis* in the cloud forest near Abra Málaga [14,18,40], and thus it is possible that the call of *B. gymnotis* was erroneously associated with males of *B. bustamantei*. There also seems to be some problems identifying specimens of this species, as shown by our phylogeny where specimens identified as *B. bustamantei* by one of us do not group with sequences from one of the paratypes of *B. bustamantei* (MHNC 6019).

**Figure 2.** Bayesian maximum clade-credibility tree for 106 species of Holoadeninae (Terrarana) based on a 2646-bp concatenated partitioned dataset (fragments of genes 16S, 12S, COI, RAG1, and Tyr), highlighting the relationships of the three genera *Bryophryne*, *Microkayla* and *Qosqophryne* gen. n. Posterior probabilities are indicated at each node. The frog illustrated here is *Qosqophryne gymnotis*, paratype MUSM 24542 (photograph by A. Catenazzi).

We propose to erect the new genus *Qosqophryne* gen. n. to accommodate *Bryophryne gymnotis*, *B. flammiventris*, and *B. mancoinca*. Several lines of evidence support the idea that *Qosqophryne* is distinct from its sister genus *Microkayla*. The molecular phylogeny indicates there is a degree of divergence comparable to that observed between other genera of strabomantid frogs (Figure 2). Our molecular analyses show strong support for the divergence of *Microkayla* and *Qosqophryne* gen. n. The lack of geographic overlap between the two genera, with a gap region of ~320 km by airline where both genera are absent, further supports this divergence by preventing recent gene flow among species of both

genera (Figure 3). Furthermore, several glaciated peaks, including the massive Ausangate mountains and associates peaks of the Cordillera de Vilcanota, are interspersed along this gap region of 320 km.

**Figure 3.** Type localities of frogs in the genera *Bryophryne* (white circles, species details not shown), *Microkayla* (squares) and *Qosqophryne* gen. n. (red asterisks) in southern Peru and northern Bolivia. The known distribution range of these frogs is limited to the type locality and immediate surroundings. For species of *Microkayla*: (1) *M. boettgeri*; (2) *M. chilina*; (3) *M. chapi*; (4) *M. katantika*; (5) *M. chaupi*; (6) *M. melanocheira*; (7) *M. colla*; (8) *M. kallawaya*; (9) *M. guillei*; (10) *M. saltator*; (11) *M. iani*; (12) *M. illampu*; (13) *M. ankohuma*; (14) *M. condoriri*; (15) *M. teqta*; (16) *M. huayna*; (17) *M. chacaltaya*; (18) *M. wettsteini*. The map does not include seven species of *Microkayla* distributed in central and southern Bolivia (type localities outside the limits of this map).

Similarly to recent phylogenies [28,41], we found that *Noblella* is not monophyletic: the species from southern Peru along with species of *Psychrophrynella* form a clade that is sister taxon to *Microkayla* +*Qosqophryne*, whereas the species of *Noblella* from northern Peru and Ecuador are closely related to "*Eleutherodactylus bilineatus*" and *Barycholos* (Figure 2). Because the type species *N. peruviana* occurs in southern Peru, and the most similar species sequenced to date *N. thiuni* is part of the *Noblella*/*Psychrophrynella* clade [28], our findings support the hypothesis that *Noblella* occurs only in southern Peru and northern Bolivia, and that species from northern Peru and Ecuador belong to a di fferent genus [28,41]. Furthermore, our tree suggests that species of *Noblella* and *Psychrophrynella* belong to the same lineage, as supported by the respective type genera sharing several morphological traits [2,5,20,28,42]. Therefore, the two possibilities are that some species of *Noblella* have been misidentified as *Psychrophrynella* (and vice versa), or that *Psychrophrynella* is a junior synonym of *Noblella*. We will not be able to resolve the taxonomic uncertainty associated with *Noblella* and *Psychrophrynella* until we obtain DNA sequences from the respective type species *N. peruviana* and *P. bagrecito* [2,19,20,28].

Finally, our inferred phylogeny suggests that there are at least seven additional putative new species of *Bryophryne*, *Noblella*, and *Psychrophrynella* (Figure 2), and confirms previous findings of cryptic species diversity particularly in leaf litter, cloud forest frogs in the *Noblella*/*Psychrophrynella* clade [22]. These putative new species, similarly to most known species of high-elevation Holoadeninae [4], are highly endemic and known from single localities (or, around those localities, from within a narrow elevational range in the same valley, [22]). Of special interest among the putative new species, *Psychrophrynella* MUSM 27619 is the first specimen of the *Noblella*/*Psychrophrynella* lineage known from the Vilcabamba range.
