Study area

Serra do Quiriri (c. 26°00’S, 48°57’W) is a mountain range located on the border of the states of Paraná and Santa Catarina, southern Brazil, which reaches a maximum altitude of 1,507 m above sea level (a.s.l.). The dominant vegetation is the highland grasslands (see the black polygons in Fig 1 of Nadaline et al. [19]), which are bordered by cloud forest (see [20]), while this forest transitions to montane forests on the slopes with lower altitudes. The Serra do Quiriri is also inhabited by B. quiririensis and B. auroguttatus (see Fig 1 of Bornschein et al. [21]), the former known from two localities and the second only from the type locality [2]. Towards the north, Serra do Quiriri is separated from Serra do Araçatuba, in the state of Paraná, by a valley 1,080 m a.s.l. In the Serra do Araçatuba, which reaches 1,656 m a.s.l., occurs B. leopardus, being known from two localities [2].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Holotype of Brachycephalus lulai sp. nov. (MHNCI 11592), male.

(A) Dorsal view of the body. (B) Lateral view of the head. (C) Ventral view of right hand. (D) Ventral view of right foot. The specimen’s image was projected using a stereomicroscope with a camera lucida, and the illustration was rendered in black ink using the pointillism technique Drawing by Verônica R. Apolônio.

https://doi.org/10.1371/journal.pone.0334746.g001

According to Köppen’s climate classification, the eastern slopes of the Serra do Quiriri have a Cfa climate: a humid subtropical zone (C) with oceanic climate, without dry season (Cf), and with a hot summer (Cfa; [22]. At higher altitudes, the Serra do Quiriri exhibits a Cfb climate, characterized by a temperate summer [22]. The mean annual temperature ranges from 20–22 °C on the eastern slopes of the Serra do Quiriri to 12–14 °C at higher altitudes, with intermediate altitudes averaging between 16–18 °C [22]. Annual rainfall varies from 1,600–1,900 mm across most of the Serra do Quiriri, increasing to 1,900–2,200 mm in the highest areas [22].

Specimen collection and processing

The Instituto Chico Mendes de Conservação da Biodiversidade (ICMBioc) granted the license to collect the specimens for this study (permission number #20416−2). No additional approval was necessary, given that the study follows the general guidelines for taxonomic work in herpetology. Populations of the new species were identified in the field through the recognition of the advertising call emitted by males. Specimens were collected through active search and during the day. The advertisement call aided the researchers in locating the males, whereas the females were collected haphazardly. Collection locations were recorded with coordinates using Garmin GPSmap 60CSx (DATUM WGS84) and the vegetation was classified according to the criteria of Veloso et al. [23]. Specimens collected in the field were placed alive in individual plastic bags, containing damp leaf litter. The plastic bags were placed inside Styrofoam boxes to maintain a mild temperature. The specimens were taken to the morphology laboratory, anesthetized and killed painlessly using 2% lidocaine hydrochloride. Each one was dissected to obtain a tissue sample, fixed with 10% formalin for 24 h and then preserved in 70% ethanol. All specimens, after being analyzed and measured, were deposited at the Museu de História Natural Capão da Imbuia (MHNCI), Curitiba, state of Paraná, Brazil. Other specimens of various congeneric species were analyzed, belonging to the following collections: Célio F. B. Haddad collection, Departamento de Zoologia, Universidade Estadual Paulista, Campus Rio Claro, state of São Paulo (CFBH); Coleção Herpetológica do Departamento de Zoologia, Universidade Federal do Paraná, Curitiba, state of Paraná (DZUP); Museu de História Natural Capão da Imbuia, state of Paraná (MHNCI); Museu de História Natural, Universidade Estadual de Campinas, Campinas, state of São Paulo (ZUEC); Museu de Zoologia da Universidade de São Paulo, São Paulo, state of São Paulo (MZUSP); and Museu Nacional, Rio de Janeiro, state of Rio de Janeiro (MNRJ). All specimens examined in the study are listed in S1 Appendix. Fifteen morphometric measurements [24] were used, according to Pie et al. [6], obtained using a micrometric eyepiece attached to a Zeiss Stemi 2000 stereomicroscope. These measurements are presented in millimeters (mm) and were taken by a single researcher (Júnior Nadaline). The sex of each specimen was determined based on the presence of the linea masculinea in males as described in Pie et al. [6].

We describe the skin texture of examined specimens of Brachycephalus as smooth, moderately rough, and densely rough (S1 Fig). Smooth texture refers to skin with small warts that are barely or not at all prominent, giving a homogeneously smooth appearance (S1A Fig). Moderately rough texture refers to skin with protruding warts that maintain spacing between them (S1B Fig). Lastly, densely rough texture refers to skin with prominently protruding warts that are densely packed, with little or no space between them (S1C Fig). We prefer not to create a fourth state of skin texture for a condition observed in individuals with few prominent warts on smooth skin, a condition we then characterize as smooth skin.

To produce the drawing of the holotype, a stereomicroscope equipped with a camera lucida was employed for direct observation of the specimen. Initially, a preliminary sketch was created using graphite. A black ink pen (Faber-Castell ECCO Pigment Gray 0.05 mm) with pointillism technique was then applied to achieve the final rendering. A millimeter scale was included adjacent to the specimen to ensure accurate proportions. Magnification levels ranging from 1x to 4.5x were utilized. Finally, the drawing was digitized. The design was executed by Verônica R. Apolônio.

Computed tomography

High resolution computed tomography (CT) scans were produced for one paratype of Brachycephalus sp. nov. at LAMIR – Laboratório de Análise de Minerais e Rochas, Universidade Federal do Paraná, using a Skyscan 1172 desktop MicroCT with a 100 kV X-ray tube following settings: 59 kV, 167 uA, a 1.4 s exposure time, four images per rotation and an image pixel size of 5.99 µm. Raw X-ray data were processed using NRecon reconstruction software (Micro Photonics Inc.) to produce a series of tomogram images. These MicroCT image stacks were then viewed, sectioned, measured, and analyzed using ITKSnap 3.2 and Meshlab 2020.12. Final figures were prepared with Adobe Photoshop and Illustrator (CS6; Adobe, San Jose, CA, USA).

Skin histology

We examined three adult individuals from each locality where the new species was recorded. Two transverse skin cuts were taken from the medial and caudal regions in the dorsal surface of the body. Samples were dehydrated in increasing series from 70% to 100% ethanol, and cleared in xylol. Subsequently, samples were embedded in paraffin and cut into 5-μm thickness cross-sections, which were stained with the Hematoxylin and Eosin (HE) technique for measurement of the thickness of dermis and the von Kossa technique for measurement of the mineralized dermal layer (MDL) [11]. To obtain the measurements, each section was photomicrographed (200x magnification) and three thickness measurements (µm) were taken (one in each lateral portion of the body and one in the middle portion) using Image-Pro-Plus 3.0.1 software.

Bioacoustics

Calls were obtained using the digital recorder Tascam DR-44WL with a Sennheiser ME 67/K6 microphone and digital recorder Sony PCM-D50 with a Sennheiser ME 66/K6 microphone, with sampling frequency rate of 44.1 kHz and 16-bit resolution. We deposited the recordings in the sound collections of the herpetological collection of the MHNCI. We evaluated calls under note-centered approach [25], as in Bornschein et al. [26–28] and Pie et al. [6]. We described the parameters based on Bornschein et al. [26,28] and Pie et al. [6], with modifications. For Bornschein et al. [28], we considered the parameter named attenuated notes. Attenuated notes generally precede the emission of notes with relatively higher sound energy/amplitude. In sum, our description of the call of the new species includes the following parameters, divided into the following sections: entire call (nine parameters), isolated notes (five parameters), note groups (eight parameters), and attenuated notes (seven parameters). Entire call: (1) call duration (s); (2) note rate (notes per minute; excluding attenuated notes); (3) number of notes per call (including attenuated notes); (4) number of notes per call (excluding attenuated notes); (5) note duration (s) (of isolated notes and notes within note groups); (6) number of pulses per notes (of isolated notes and notes of note groups); (7) note dominant frequency (kHz); (8) highest frequency (kHz); and (9) lowest frequency (kHz). Isolated notes: (10) duration of the call including only isolated notes (s) (when the calls present note groups among isolated notes, only the longest part with isolated notes was counted); (11) note rate of the call including only isolated notes (notes per minute); (12) number of isolated notes per call (excluding attenuated notes); (13) number of pulses per isolated notes; and (14) inter-note interval in isolated notes (s) (time from the end of one isolated note to the beginning of the next isolated note). Note groups: (15) duration of the call including only note groups (s) (only the longest part was counted, when the calls present isolated notes among note groups); (16) note rate of the call including only note groups (notes per minute; calculated only when at least two note groups were present); (17) number of note groups per call; (18) number of notes in each note group; (19) duration of note groups (s); (20) inter-note group interval (s) (time from the end of one note group to the beginning of the next note group); (21) inter-note interval within note groups (s) (time from the end of the first note to the beginning of the next note of the same note group); and (22) number of pulses per note in note groups. Attenuated notes: (23) number of notes per call associated with attenuated notes; (24) number of attenuated notes associated with each note of the call; (25) number of pulses in attenuated notes; (26) shortest interval between an attenuated note and its associated note (s); (27) attenuated note dominant frequency (kHz); (28) attenuated note highest frequency (kHz); and (29) attenuated note lowest frequency (kHz). For rate calculation, we counted the time between the beginning of the first note to the beginning of the last note under consideration and the number of notes in this section [26].

Advertisement calls were analyzed in Raven Pro 1.6.3 with a 256-point Fast Fourier Transform and a 3-dB Filter bandwidth of 492 Hz, Hann window, and 50% overlap. Raven table selections were imported to the R environment using Rraven [29] and summarized using Summarytools [30]. Spectrograms for figures were generated using the Seewave package, v. 2.2.0 [31] of the R environment, v. 4.2.0 [32] using the same preset parameters as in Raven Pro. We examined calls from other 22 species in the sound collections of MHNCI, Coleção Audiovisual do Semiárido (CASA), Mossoró, state of Rio Grande do Norte, Brazil, and Fonoteca Neotropical Jacques Vielliard (FNJV), Campinas, state of São Paulo, Brazil, for comparative purposes (S2 Appendix). In the B. pernix group, there is still no available data on the calls of B. mariaeterezae.

Molecular phylogeny

To determine the phylogenetic position of the new species, we analyzed DNA sequences from four paratypes (MHNCI 11596, MHNCI 11598–11600). The choice of species to be analyzed alongside the new species was based on previous work on the phylogeny of the group (i.e., [33,34]). Preliminary analyses indicated that the new species was closely related to a well-supported clade that included B. auroguttatus, B. quiririensis, B. ferruginus, B. pernix, and B. pombali. The monophyly of this clade was first suggested by Firkowski et al. [33] using Sanger sequencing and was later strongly supported by a phylogenomic analysis using UCEs by Pie et al. [34]. We therefore chose those species for exploring the phylogenetic position of the new species. The resulting tree was rooted according to the phylogenomic work by Pie et al. [34].

Whole genomic DNA was extracted using SPRI beads [35,36]. We amplified one mitochondrial locus (16S rRNA) and three nuclear loci (β-fibrinogen, ribosomal Protein L3, and tyrosinase exon 1) via polymerase chain reaction (PCR). Each PCR reaction had a total volume of 25 μL, containing 2 U AmpliTaq DNA polymerase, 1 × PCR buffer, 1.5 mM MgCl2, 0.5 mM dNTPs, 1.0 μM each primer, and approximately 40 ng of template DNA. The thermocycling protocol included an initial denaturation at 94 °C for 5 min, followed by 35 cycles of 94 °C for 1 min, 48–62 °C for 50 s, and 72 °C for 35–50 s, with a final extension at 72 °C for 5 min (S1 Table). PCR products were run on 1.5% agarose gels, and successful amplifications were purified using PEG 8000.

Sequencing reactions were prepared in a 10 μL volume, comprising 0.7 μL ABI Prism® BigDye™ v3.1 (Applied Biosystems Inc., Foster City, CA, USA), 1.0 μL 5 × buffer, one μL each primer (3.2 pmol), and about 30 ng of template DNA. The cycle sequencing conditions involved an initial denaturation at 96 °C for 1 min, followed by 35 cycles at 96 °C for 15 s, annealing at 50 °C for 15 s, and extension at 60 °C for 4 min. Each locus was sequenced in both directions using an ABI 3500 sequencer.

Sequences were aligned for each locus with those of other species in the B. pernix group (B. auroguttatus, B. ferruginus, B. pombali, B. pernix, and B. quiririensis, see [33]) using MUSCLE v3.8.31 [37] under default settings. The sequences were then concatenated, resulting in a final alignment length of 1828 bp. The optimal partitioning scheme was selected using PartitionFinder2 [38] based on the AICc criterion. The best scheme included six partitions: (1) HKY + I for 603 sites (16S and L3), (2) HKY + Γ for 133 sites (β-fibrinogen), (3) K80 model for 133 sites (β-fibrinogen), (4) F81 for 322 sites (β-fibrinogen and Tyr), (5) HKY + I + Γ for 448 sites (L3, L3, and Tyr), and (6) HKY + I + Γ model for 189 sites (Tyr).

A Bayesian phylogeny was inferred using MrBayes 3.2.7a [39]. Each analysis included two independent runs with four chains, run for 20 million generations, sampling every 1,000th generation. After confirming convergence of the chains, data sets were combined. Stationary distribution and effective sample sizes (ESS) for all parameters were assessed using Tracer v1.5 [40]. We discarded the initial 20% of trees as burn-in and used the remaining trees to estimate the maximum clade credibility consensus topology in TreeAnnotator v1.7.5 [41]. GenBank accession numbers are provided in S2 Table. Additional analyses using maximum likelihood provided qualitatively identical results and will not be shown for the sake of brevity.

Conservation

We evaluated the conservation status of the new species and compared it with assessments of other species in the B. pernix group to better discuss and adjust recommended conservation measure. Although reassessing the conservation status of the others was not a primary objective, we also reassessed them because previous assessments for these species had some particularities in the spatial metrics based on their distribution pattern [2] that could complicate the consistency across taxa. To guide our evaluations, we followed the criteria and guidelines of the IUCN [42], IUCN SSC Red List Technical Working Group [43], and IUCN Standards and Petitions Committee [44].

We used the IUCN’s geographical distribution criteria to assess the extinction risk of the species [42], as we lacked available data on population estimates and trends for the new species. Geographical distribution was delineated by constructing polygons based on the mapping of suitable habitat within and around the altitudinal range isolines of known records, in Google Earth Pro 7.3.6.10201, following and updating the mapping method of Bornschein et al. [2]. Polygons of potential habitat that extended several kilometers westward were not considered.

After this preliminary delimitation of the geographical distribution polygons, we connected their outermost points to generate an external polygon representing the extent of occurrence (EOO), using the Minimum Convex Polygon method [42–44] in Google Earth. Subsequently, we reviewed the entire EOO surface to identify and map additional potential habitats within the species’ altitudinal range that may have been overlooked during the preliminary delimitation, also using Google Earth. If the resulting EOO was lower than the area of occupancy (AOO, see below), we equaled them to ensure consistency in the definition [43,44].

The resulting polygons of mapped habitat within the EOO represented the potential occupied habitat for the species, also referred to as limits of distribution, area of habitat, or field guide map [43]. To calculate the upper bound AOO, we overlapped these polygons of mapped habitat with a grid of 2 km x 2 km cells (= 4 km² of area), and then summed the area of overlapped cells [44] in QGIS 3.30.1. To calculate the lower bound AOO, we overlapped only the current records with the same grid [44]. We adjusted the grid to obtain the smallest number of occupied cells for each species under consideration [44], using the same position for upper and lower bound. We evaluated the range between the upper and lower bound values to potentially access the species as Data Deficient (DD) due to uncertainty [44]. For species in which we have not mapped the habitat because of the absence of an associated altitudinal range or because the mapping was unrealistic given the suitable habitat and altitude, we considered the lower bound AOO as the best estimate.

We assessed the IUCN Green Status of the new species and remaining species of the B. pernix group according to IUCN Green Status of Species Working Group [45], and IUCN [46], in order to compare the benefits of conservation measures. We considered the locations (sensu [42]) as conservation units for the assessment [46]. We used the conservation status (see above) in the unit to assess its states, and also the field information to assess its functionality. We used fine-resolution weights values (0, 1.5, 2.5, 3.5, 4.5, 5.5, 6.5, or10) to better calculate the conservation impact [45]. Some species, despite their naturally small geographic distribution, do not seem to be threatened and occupy environments that are little or not at all disturbed, and, therefore, have been classified as functional at baseline levels in certain spatial units [45]. We assessed the Current and also Long-term potential Green Scores to calculate the Recovery Potential (percentage) of each species [46]. For the Long-term potential scenario, we considered stopping deforestation, promoting forest regeneration and mitigate wildfires as potential conservation actions. We created a new metric, called Strategic Weight (points), to allow numerical comparisons of species’ conservation priorities (higher priorities for higher point values). It is calculated by adding the values we attribute to each conservation status (CR = 20; EN = 15; VU = 5; DD = 5; NT = 2; LC = 1) with the species’ Recovery Potential value. Finally, we proposed the Potential Weight of a Conservation Unit (points) for a potential conservation unit to be created to protect each species or group of species with distributions close to each other, outside conservation units, adding Strategic Weight of each species occurring in this planned conservation unit.

Nomenclatural acts

The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix “http://zoobank.org/”. The LSID for this publication is: urn:lsid:zoobank.org:pub:DD6AC76F-9D25-443B-A6B9-8AFC196DAD64. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the digital repository DARE (Institutional Research Data Repository of the Carl von Ossietzky Universität Oldenburg).

Brachycephalus lulai sp. nov.

(Figs 1–4)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Holotype of Brachycephalus lulai sp. nov. (MHNCI 11592), male, in life.

(A) Anterolateral view. (B) Ventral view. In B, white arrow indicates the presence of the linea masculinea. Photographs by Luiz F. Ribeiro.

https://doi.org/10.1371/journal.pone.0334746.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Holotype of Brachycephalus lulai sp. nov.

(MHNCI 11592), male, in preservative. (A) Dorsal view. (B) Ventral view.

https://doi.org/10.1371/journal.pone.0334746.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Variation in coloration of paratypes of Brachycephalus lulai sp. nov. Column one and column three show specimens in dorsal view in life and in preservative, respectively.

Column two and column four show specimens in ventral view in life and in preservative, respectively. A1–A4 = MHNCI 11612. B1–B4 = MHNCI 11598. C1–C4 = MHNCI 11596. D1–D4 = MHNCI 11599. E1–E4 = MHNCI 11600. F1–F4 = MHNCI 11594. Scale bars equal 5 mm. Photographs by Luiz F. Ribeiro.

https://doi.org/10.1371/journal.pone.0334746.g004

urn:lsid:zoobank.org:act:2A4472A8-55B9-4478-B65C-632D7B31B2A9

Holotype

MHNCI 11592 (Figs 1–3), male, Pico Garuva (26°02’32”S, 48°53’27”W; 635 m a.s.l.), municipality of Garuva, state of Santa Catarina, southern Brazil, on 2 November 2016 by Luiz F. Ribeiro, André E. Confetti, Júnior Nadaline, and Marcio R. Pie.

Paratopotypes

MHNCI 11593, male, and MHNCI 10847, female, both collected with the holotype; MHNCI 11594–8 and MHNCI 11600, males, and MHNCI 11599, female, 20 January 2017, Luiz F. Ribeiro, André E. Confetti, Júnior Nadaline, and Marcio R. Pie; MHNCI 11601 and MHNCI 11603–10, males, and MHNCI 11602, female, 15 November 2018, Júnior Nadaline, André Luiz Ferreira Silva, and Marcio R. Pie. Specimens were collected in same locality as the holotype, between 635–990 m a.s.l.

Paratypes

Specimens were collected at Monte Crista (26°05’33”S, 48°55’16”W), municipality of Garuva, state of Santa Catarina, southern Brazil, between 435–895 m a.s.l. MHNCI 11611 and MHNCI 11613, males, and MHNCI 11612, female, 15 November 2016, Luiz F. Ribeiro, Liliane Pires, and Marcio R. Pie; MHNCI 11614–5, MHNCI 11617, MHNCI 11619, and MHNCI 11621–2, males, and MHNCI 11616, MHNCI 11618, and MHNCI 11620, females, 26 February 2019, Júnior Nadaline, Philippe Fumaneri Teixeira, Tainara Thais Jory, and Marcio R. Pie.

Diagnosis

Brachycephalus lulai sp. nov. is identified as a member of the B. pernix group (sensu Pie et al. [6] and Ribeiro et al. [13]) by having a bufoniform body shape (Figs 1–4), absence of dermal co–ossification, and presence of linea masculinea (Fig 2B). Brachycephalus lulai sp. nov. is distinguished from all of the species in the genus by the following combination of characters: 1) body shape bufoniform; 2) snout shape in dorsal view rounded; 3) SVL 8.9–11.3 mm for males and 11.7–13.4 mm for females (Table 1); 4) proportion of HL/SVL 32.5–41.4% for males and 31.8–34.9% for females; 5) presence of linea masculinea in males; 6) absence of dermal co-ossification; 7) dorsum with smooth texture; 8) sides of the body with densely rough texture; 9) tip of fingers I rounded, II rounded, and III pointed; 10) toe V externally absent; 11) outer metacarpal tubercle present; 12) iris black; 13) general color bright orange with small green irregular dots on sides of the body and belly and sometimes with brown dots on sides of the body; 14) general color in preservative pale cream with small light gray to dark gray irregular dots on sides of the body and belly and sometimes with dark spots on sides of the body; 15) advertisement call including note group; 16) two notes per note group; 17) advertisement call including attenuated notes; and 18) up to four pulses per note. The comparison with our analysis of other 20 species of the B. pernix group is summarized in Table 2, including the coloration in life based on specimens we collected, which sometimes differs from the color variation given in the original species description (Fig 5). Brachycephalus tabuleiro was assigned to the B. pernix group in its original description [47]. Brachycephalus lulai sp. nov. can be distinguished from the seven flea toad species (B. clarissae, B. dacnis, B. didactylus, B. hermogenesi, B. pulex, B. puri, B. sulfuratus) by having bufoniform body shape instead leptodactyliform body shape (but see below) and from 15 species of the B. ephippium group (sensu Ribeiro et al. [13]; B. alipioi, B. bufonoides, B. crispus, B. darkside, B. ephippium, B. garbeanus, B. guarani, B. herculeus, B. ibitinga, B. margaritatus, B. nodoterga, B. pitanga, B. rotenbergae, B. toby, B. vertebralis) by absence of dermal co-ossification [13,17,48–53]. Brachycephalus rotenbergae, B. ibitinga, and B. herculeus belong to the B. ephippium group due to the presence of dermal co-ossifications and a bufoniform body shape [49–51]. Brachycephalus puri was originally treated as a flea toad due to its leptodactyliform body shape and the absence of dermal co-ossification [48]. Brachycephalus clarissae, with its like-bufoniform body shape and lack of dermal co-ossification [17], was treated as a flea toad by Toledo et al. [54]. Brachycephalus dacnis was assigned to the B. didactylus group by Bornschein et al. [28], prior to its formal description by Toledo et al. [54], who treated it as a flea toad. We cannot provide distinguishing traits of B. lulai sp. nov. from B. atelopoide because its holotype is apparently missing [55] and there are no additional collections [56]. However, reanalysis of the original description [57] suggests that the holotype of B. atelopoide likely had a bufoniform body shape, dermal co-ossification, and warts [56]—the latter two features being absent in B. lulai sp. nov.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Measurements of the 15 variables (in mm) of the type series of Brachycephalus lulai sp. nov.

https://doi.org/10.1371/journal.pone.0334746.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Comparison of the diagnostic features of Brachycephalus lulai sp. nov. in relation to remaining species of B. pernix group. In bold we highlight the variables that are diagnostic in relation to B. lulai sp. nov. Sample size refers to the number of individuals in which the authors observed each variable or number of individuals from which one or more calls where recorded and analyzed by the authors.

https://doi.org/10.1371/journal.pone.0334746.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Divergent color variation of Brachycephalus species.

A1–A3 = B. actaeus MHNCI 11625 (Forte Marechal Luz, Ilha de São Francisco, municipality of São Francisco do Sul, Santa Catarina). B1–B3 = B. auroguttatus MHNCI 11768 (trail to Pedra da Tartaruga, municipality of Garuva, Santa Catarina). C1–C3 = B. fuscolineatus MHNCI 11599 (Morro do Baú, municipality of Ilhota, Santa Catarina [type locality]). D1–D3 and E1–E3 = B. tridactylus MHNCI 10852 and MHNCI 11767, respectively (Torre Embratel, Parque Estadual do Rio Turvo, municipality of Cajati, São Paulo). Abbreviation: MHNCI = Museu de História Natural Capão da Imbuia, Curitiba, Paraná. Scale bars equal 5 mm. Photographs: A1–D3 = Luiz F. Ribeiro; E1–E3 = Marcos R. Bornschein.

https://doi.org/10.1371/journal.pone.0334746.g005

Description of the holotype

Body shape bufoniform (Figs 1–4); head wider than long (proportion HW/HL = 120%); head length 34% of SVL; snout short (proportion SL/HL = 31%), with length almost equal to eye diameter (proportion SL/ED = 92%), rounded in dorsal and lateral views; nostrils not protuberant, directed anterolaterally; canthus rostralis not distinct; mouth approximately sigmoid in shape; loreal region slightly concave; eye slightly protuberant in dorsal and lateral views; ED 32% of HL; tympanum indistinct; vocal sac not expanded externally; vocal slits present; tongue longer than wide, with posterior half not attached to floor of mouth; choanae small and ovoid, anterior to eyes; vomerine odontophores absent. Arm and forearm relatively slender; arm approximately as long as forearm; finger IV not visible externally; tip of fingers I and II rounded, tip of finger III pointed; relative lengths of fingers (in external view) I < II < III; subarticular tubercles absent; inner and outer metacarpal tubercles absent. Legs short, moderately robust; THL 36% of SVL; shank length 87% of thigh length; toes II–III short and distinct; toes I and V not visible externally; relative lengths of toes (in external view) II < III < IV; subarticular tubercles and inner metatarsal tubercles absent; outer metatarsal tubercle present and discreet. Dorsum with smooth texture, without visible co-ossifications; head, chin and arms with smooth texture; dorsal region of legs and sides of the body with densely rough texture; ventral surface (except chin) with densely rough texture.

Coloration of the holotype

In life (Fig 2), overall dorsal coloration uniform bright orange from head to pelvic region; granular warts on lateral regions of body greenish yellow, with orange between warts; arms and legs orange, with discrete irregular greenish or brown spots. Chin and chest orange; belly and ventral region of thighs greenish yellow; iris completely black. In preservative after four months (Fig 3), dorsum of head cream; orange regions on the dorsum becomes pale cream, greenish yellow region of the flanks becomes gray, and discreet irregular greenish spots becomes brown.

Measurements of holotype (in mm)

SVL = 10.6; HL = 3.6; HW = 4.3; ED = 1.2; IOD = 2.1; IND = 1.2; EN = 0.6; SL = 1.1; UEW = 0.7; FLL = 2.2; HAL = 1.6; THL = 4.0; TL = 3.4; TSL = 1.8; FL = 2.5.

Variation in the type series

Morphometric variation is given in Table 1. Species sexually dimorphic, with the females generally bigger (SVL = 11.7–13.4 mm) than males (SVL = 8.9–11.3 mm). Although the finger IV of most specimens is absent on an external examination, some specimens have a small vestigial trace of the finger IV (e.g., Fig 4A4, E3, F3). There are slight differences in coloration among specimens (Fig 4), particularly in the presence of irregular, green spots in some specimens on the arms, legs, side of the body, and near the head (e.g., Fig 4E1, F1). Though most of the bright orange and yellow coloration fades in preservation, the green spots remain, becoming dark gray (e.g., Fig 4D3, E3, F3).

Description of general osteology

Based on microCT scan of MHNCI 11601 (paratype; male; Fig 6). The skull is short and slightly broader than long. Vomers and nasals are distinct and not synostosed with other bones (Fig 6C). Neopalatines are distinct and minute. Frontoparietals, prootics, exoccipitals, sphenethmoids, and parasphenoid are synostosed, resulting in a skull shape similar to other species in the genus (e.g., B. coloratus and B. curupira). The frontoparietals are partially fused across the midline. The premaxillae are broad, widely separated, and have weakly developed odontoids; each has a robust pars dentalis, and a robust alary process that is taller than wide and widely separated from adjacent nasal. The quadratojugal has a broad articulation with the maxilla, which is thin and nearly straight. The pterygoids are slender, each with a long anterior ramus that approaches but does not articulate with the adjacent maxilla. The squamosal are robust, and each bears a large rectangular zygomatic ramus which is directed towards the maxillae; they also have a long, slender, and somewhat flattened posterior ramus that has a broad connection to the prootic (Fig 6C). Small, curved sphenethmoids are present at the anterior margin of nasal capsule. The parasphenoid is broad and robust. The vertebral column of B. lulai sp. nov. has seven presacral, procoelous, non-imbricate vertebrae, with the eighth presacral fused to the sacral vertebra. First presacral vertebra (atlas) lacks transverse process and has widely separated cotyles. Lengths of the transverse process of presacral vertebrae along with that of the sacral diapophyses: III > IV ≅ SD > V ≅ VI > II ≅ VII (SD = sacral diapophyses). Transverse processes of presacral vertebrae II and III perpendicular to the notochordal axis, IV and V oriented posteriorly and those of VI and VII oriented slightly anteriorly in relation to the notochordal axis. The sacrum (a composite including the fused presacral VIII) has robust transverse processes and a large sesamoid is found near the articulation with each ilium. The urostyle is robust and has a high dorsal ridge that decreases along the length of the bone (Fig 6A). The pectoral girdle is arciferal, robust, and lacks a sternum and omosternum (Fig 6F). The scapula is robust, featuring a prominent anterior process. The suprascapula is weakly ossified with a well-ossified cleithrum. The ilium is robust, with a fully ossified acetabulum with a ventral acetabular expansion composed of ilium, pubis, and ischium. The ilium has a well-developed and prominent dorsal crest (Fig 6E). The radioulna is slightly shorter than the humerus. The femur and tibiofibula are similar in length. The distal carpals (Element Y and II–V) are fused. The radiale and ulnare are large and subequal in size. The phalangeal formula for the manus is 1–2–3–1 and there is both a single ossified prepollex and a small palmar sesamoid. The tips of the terminal manual phalanges are blunt (Fig 6H). There are two large distal tarsals. The phalangeal formula for the pes is 1–2–3–4–1 and there are two plantar sesamoids but no obvious distinct prehallux. The tips of the terminal pedal phalanges are blunt (Fig 6G).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. High-resolution computed tomography (CT) scans of a paratype of Brachycephalus lulai sp. nov. (MHNCI 11601) showing key osteological features.

(A) Dorsal view of the skeleton; (B) dorsal, (C) lateral (without the lower jaw), and (D) ventral views of the skull; (E) pectoral girdle in ventral view; (F) ilium in lateral view; (G) left hand in palmar view; (H) left foot in plantar view. While the vomer is fused to surrounding elements, we have highlighted its approximate boundaries. Abbreviations: qj = quadratojugal; sq = squamosal; v = vomer. Scale bars equal 2 mm.

https://doi.org/10.1371/journal.pone.0334746.g006

Description of skin histology

A conspicuous and homogeneously distributed MDL was detected in the skin of the dorsal region of the body of B. lulai sp. nov. (Fig 7). In specimens from Pico Garuva, the MDL represented, on average, 22% and 30% of the dermis thickness in the medial and caudal portions of the dorsal region, respectively, while in specimens from Monte Crista, the MDL represented 26% and 27% of the dermis thickness in the medial and caudal of the dorsal region, respectively (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Measurement (µm) of the dermis thickness and mineralized dermal layer (MDL) of the skin of the medial and caudal regions of the dorsal body of Brachycephalus lulai sp. nov.

https://doi.org/10.1371/journal.pone.0334746.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Photomicrography of the dorsal skin of Brachycephalus lulai sp. nov.

(A and B) Specimens from Pico Garuva. (C and D) Specimens from Monte Crista. The red arrows indicate the mineralized dermal layer in the dermis. The black line indicates the extent of the dermis. In A and C, Hematoxylin and Eosin staining; in B and D, von Kossa staining.

https://doi.org/10.1371/journal.pone.0334746.g007

Bioacoustics

We analyzed 13 advertisement calls (Table 4), each from a distinct individual. Recordings were made at Pico Garuva (MHNCI 224–32, vouchers MHNCI 11597–8, MHNCI 11600) and at Monte Crista (MHNCI 233–6; Table 4; S2 Appendix). Values represent the range and, if relevant, mean ± SD between parentheses (see Table 6 for results of all parameters). The advertisement calls of the species include isolated notes and note groups (Table 4, Fig 8).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Structure of the advertisements calls (AC) of Brachycephalus lulai sp. nov. and other members of the B. pernix species group.

https://doi.org/10.1371/journal.pone.0334746.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Call parameters described in the advertisement call of Brachycephalus lulai sp. nov.

https://doi.org/10.1371/journal.pone.0334746.t005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Example of advertisement call parameters of Brachycephalus lulai sp. nov.

(A) Entire advertisement call (MHNCI 226). (B) Isolated note with one pulse (MHNCI 225). (C) Isolated note with three pulses (MHNCI 224). (D) Note group with three and two pulses (MHNCI 232). (E) Attenuated note (black circle) preceding an isolated note with three pulses (MHNCI 226). Spectrograms are produced with a FFT size of 8192 points, Hann window, and overlap of 90% in A and FFT 512 points, Hann window, and overlap of 90% in B–E.

https://doi.org/10.1371/journal.pone.0334746.g008

Entire call

We did not record an advertisement call from the beginning, because calls were already in progress when recording started (Table 4). This prevented us from measuring the complete duration of the call, but the call duration from recorded calls varies from 61.7–201.8 s (116.5 ± 45.8 s; N = 13 calls; Fig 8A). Note rate was 5.2–10.6 notes per minute (7.5 ± 1.7 note per minute; N = 13 calls). The number of notes per call (excluding attenuated notes) varies from 7–30 notes (15.8 ± 7.0 notes; N = 13 calls). Note duration (of isolated notes and notes of note groups) was 0.010–0.061 s (0.033 ± 0.009 s; N = 205 notes). Note dominant frequency was 5.2–6.6 kHz (6.2 ± 0.2 kHz; N = 205 notes).

Isolated notes

Eleven calls present only isolated notes and the total number of notes are not possible to determine since these calls were not recorded from the beginning. However, the duration of the call including only isolated notes based on the recorded part of these call lasts from 48.9–201.8 s (106.8 ± 50.6 s; N = 11 calls) and they included 7–30 isolated notes (14.9 ± 7.2 notes; N = 11 calls). Note rate of the call including only isolated notes was 5.2–16.1 notes per minute (8.0 ± 2.8 notes per minute; N = 13 calls). Isolated notes present 1–4 pulses (2.5 ± 0.6 pulses; N = 189 isolated notes; Fig 8). The inter-note interval in isolated notes was 4.6–15.0 s (8.2 ± 2.2 s; N = 173 inter-note intervals).

Note groups

Two calls present note group, one with three note groups and the other with five note groups. The duration of the call including only note groups was 18.4–33.7 s (26.0 ± 10.8 s; N = 2 calls). Note rate of the call including only note groups was 16.3–17.8 notes per minute (17.1 ± 1.0 notes per minute; N = 2 calls). All note groups present two notes composing the group (N = 8 note groups; Fig 8). Duration of note groups was 0.361–0.416 s (0.396 ± 0.020 s; N = 7 note groups). The inter-note interval within note groups was 0.290–0.346 s (0.325 ± 0.021 s; N = 8 inter-note interval within note groups). Note groups present 2–3 pulses per note (2.3 ± 0.5 pulses; N = 16 notes; Fig 8).

Attenuated notes

All advertisement calls recorded presented attenuated notes (N = 13 calls; Fig 8E and 9), in the number of 7–21 notes per call associated with attenuated notes (14.1 ± 5.1 notes). There were 181 notes preceded by a single attenuated note and three notes preceded by two attenuated notes. All note groups were preceded by attenuated notes (N = 8 note groups). However, only the first note comprising the note group was preceded by the attenuated note (Table 4). Most of the isolated notes were preceded by attenuated notes (N = 176 notes), with only 13 that were not (Table 4). Attenuated notes present 1–2 pulses (1.0 ± 0.2 pulse; N = 186 attenuated notes). Note dominant frequency for attenuated notes was 5.1–7.7 kHz (5.7 ± 0.4 kHz; N = 186 attenuated notes).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Example of an attenuated note (smaller, on the left) of Brachycephalus lulai sp. nov. followed by an isolated note with three pulses (larger, on the right).

Sound taken from MHNCI 227 and generated with 90 overlap and 256 FFT size by R package Soundshape.

https://doi.org/10.1371/journal.pone.0334746.g009

Phylogenetic relationships

The inferred phylogenetic relationships between B. lulai sp. nov. and other species of the B. pernix species group are shown Fig 10 and the inferred phylogenetic distances between these species are shown in Table 6. Minimum and maximum 16S distances between the tested species are shown in S3 Table. All specimens of B. lulai sp. nov. form a clade, with strong support, which itself forms a larger clade together with B. auroguttatus and B. quiririensis, also with strong support. The relationships between these three species was not resolved.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Pairwise genetic distance (%) between the studied Brachycephalus species for all loci (upper diagonal) and 16S alone (lower diagonal).

https://doi.org/10.1371/journal.pone.0334746.t006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Relationships between Brachycephalus lulai sp. nov. and some species of the B. pernix group based on three loci.

Phylogenetic analysis was carried out using Bayesian inference and values above branches correspond to node posterior probabilities. Nodes with posterior probabilities lower than 50% were collapsed.

https://doi.org/10.1371/journal.pone.0334746.g010

Habitat, geographical distribution, and natural history

Brachycephalus lulai sp. nov. was recorded in montane Atlantic Forest (Floresta Ombrófila Densa Montana; Fig 11), where it lives in the leaf litter. It seems to be locally abundant in the two localities with records, on the southeastern hillside from Serra do Quiriri, municipality of Garuva, southern Brazil: Pico Garuva and Monte Crista, at altitudes between 435–990 m a.s.l. As these localities are 6.3 km apart, it is likely that B. lulai sp. nov. populations occur across the intervening forested hillside (Fig 12). We recorded individuals calling throughout the day, but vocal activity decreased during the hottest hours (c. 11:00 h – 15:00 h).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 11. Habitat at the type locality of Brachycephalus lulai sp. nov., Pico Garuva, municipality of Garuva, Santa Catarina, southern Brazil.

(A) Middle and low strata of the forest (Floresta Ombrófila Densa Montana) at 750 m above sea level. (B) East slope of Pico Garuva, 1,260 m above sea level. Photographs by Luiz F. Ribeiro.

https://doi.org/10.1371/journal.pone.0334746.g011

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 12. Location of specimens examined of species from Brachycephalus pernix group, in southeastern and southern Brazil.

(A) All specimens. (B) Detail of the distribution of B. lulai sp. nov. Abbreviations: SP = São Paulo; PR = Paraná; SC = Santa Catarina. Urban areas are displayed in gray, open vegetation areas in light green, and dense vegetation in dark green. Basemaps: OpenStreetMap, under Open Database License (ODbL), available at https://www.openstreetmap.org/.

https://doi.org/10.1371/journal.pone.0334746.g012

We found two parasites located subcutaneously at the left thigh and the gular region of one individual of B. lulai sp. nov. (MHNCI 11593) (Fig 13). We tentatively identified these as Ophiotaenia sp., a proteocephalid tapeworm, due to the morphological similarity with previously recorded parasites found under similar conditions [58], without apparent diagnostic morphological structures and at a young stage. The known phylogenetic proximity between species of the genus Brachycephalus, evidenced in this study and in other previous studies [33,34,59], presumably makes it possible for parasitism by Ophiotaenia to occur in all species of the genus. Indeed, Ophiotaenia spp. are generalist parasites that use not only amphibians, but also snakes and turtles as hosts [58]. Finally, Ophiotaenia parasites have been found in other species of the genus, such as B. brunneus and B. izechsohni, under the same conditions as those reported in the present study (L.F. Ribeiro, pers. obs).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 13. Brachycephalus lulai sp. nov. (MHNCI 11593) parasitized by tapeworms (Ophiotaenia sp).

(A) The arrows point to the locations of the parasites at the gular region and the left thigh. (B) The arrow points to the location of the parasite on the left thigh. Edema regions can also be seen in these images, such as the left leg and right thigh. (C) Ophiotaenia sp. soon after removal from the gular region of the collected frog. (D) Ophiotaenia sp. isolated from the B. lulai sp. nov. Photographs by Luiz F. Ribeiro.

https://doi.org/10.1371/journal.pone.0334746.g013

Conservation status and Green Status

Although we have altitudinal data for B. lulai sp. nov., the polygons of mapped habitat based on isolines and suitable habitat are unrealistic, as they extends for many kilometers beyond the two current records in areas that are highly unlikely to be occupied. Therefore, its geographical distribution was estimated by overlapping the grid and current records (lower bound and best estimate), resulting in two occupied cells (S13 Fig) and a total area of occupancy of 8 km² (Table 7).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Conservation status of species of the Brachycephalus pernix group, southeastern and southern Brazil.

https://doi.org/10.1371/journal.pone.0334746.t007

Considering the current environmental quality of its habitat and absence of significant threats in the region of the records, as well as the absence of plausible threats that could affect the species and qualify it for a threatened category in the near future, we propose for it to be considered as Least Concern (LC) (Table 7), even considering its small distribution area. Of the 20 remaining species of B. pernix group, six are also assessed as LC, four as Vulnerable (VU), five as Endangered (EN), and five as Critically Endangered (CR) (Table 7; S2–S22 Figs).

Based on field observations of threats and ecology and on its conservation status (see above), we consider that the species is performing its ecological functions at baseline levels, even being naturally small. Therefore, the Current and Long-term Potential Green Scores of B. lulai sp. nov. are 100% (classified as Non-Depleted; Table 8). Of the 20 remaining species of B. pernix group, nine are also Non-Depleted, two are Moderately Depleted, two are Largely Depleted, and seven are Critically Depleted (Table 8).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Green Status of species of the Brachycephalus pernix group, southeastern and southern Brazil.

https://doi.org/10.1371/journal.pone.0334746.t008

A conservation unit could be proposed for the protection of B. lulai sp. nov., together with B. auroguttatus and B. quiririensis, as they are distributed close to each other (Fig 14; see S1 File). This conservation unit would have a Weight of 107.5 points (Table 8), a value higher than any other conservation unit that could be created to protect other species of the B. pernix group in Paraná and Santa Catarina states (Table 8).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 14. Limits of the Refúgio de Vida Silvestre Serra do Quiriri, integral protection conservation unit proposed for Brachycephalus lulai sp. nov., B. auroguttatus, and B. quiririensis.

Abbreviations: PR = Paraná; SC = Santa Catarina; RVS = Refúgio de Vida Silvestre. Basemaps: Natural Earth, under Public Domain, available at https://www.naturalearthdata.com, and OpenStreetMap, under Open Database License (ODbL), available at https://www.openstreetmap.org/.

https://doi.org/10.1371/journal.pone.0334746.g014

Etymology

The specific epithet honors Luiz Inácio Lula da Silva, who has been elected President of Brazil on three occasions. Through this tribute, we seek to encourage the expansion of conservation initiatives focused on the Atlantic Forest as a whole, and on Brazil’s highly endemic miniaturized frogs in particular.

Phylogenetic relationships

Although the new species is clearly a member of the B. pernix species group, the available phylogenetic evidence indicated in Fig 2 does not allow for inferring whether B. lulai sp. nov. is more closely-related to B. quiririensis or B. auroguttatus and it is likely that a more extensive analysis will be necessary to address this issue. Nevertheless, the phylogeny conclusively shows three important findings: (1) all sequenced individuals within each species are monophyletic; (2) B. lulai sp. nov., B. quiririensis and B. auroguttatus are reciprocally monophyletic (even though only one sample of B. auroguttatus was used); and (3) the level of intraspecific genetic variability within B. lulai sp. nov. is comparable of that found in B. quiririensis. These findings reinforce our interpretation that genetic evidence supports the recognition of B. lulai sp. nov. as a new taxon.

Osteology

All species of the B. ephippium group have eight presacral vertebrae, in general with presence of ornamented spinal plates (lacking ornamentation in B. alipioi), with presence of ornamented paravertebral plates and with presacrals IV–V and VI–VII fused in B. ephippium [52,53]. Most species in the B. pernix group with their osteology described have eight presacral vertebrae, viz B. izecksohni and B. brunneus [60], B. pombali and B. ferruginus [61], B. coloratus and B. curupira [4], B. albolineatus [62], and B. actaeus [63]. Like B. lulai sp. nov., the VIII presacral vertebra is fused in B. albolineatus, and it is possible that B. brunneus and B. pombali also presents this fusion [60]. However, more specimens should be analyzed, since there may be variation in the number of vertebrae and malformations. Manuella Folly (personal communication to LF Ribeiro, 2022) reported that among 13 analyzed species of B. pernix group none showed this degree of fusion and all have eight distinct presacral vertebrae. In addition, she said also that some specimens showed malformations, which could be fusion or an extra fused vertebra.

Bioacoustics

In addition to B. lulai sp. nov., we describe aspects of the advertisement call of 16 other species (Table 4). These descriptions, all under a note-centered approach, reveal important aspects of the call structure that are overlooked in a call-centered approach [28], for example presence of note groups, attenuated notes, and warming notes (see below). These descriptions reinforce that calls increase in complexity along note emissions during a given call [26], with the incorporation of higher number of pulses and, in some cases, note groups during the call emission. In the call-centered approach (see below), these differences in call emissions would be perceived as variation in the number of notes per advertisement call, in the number of pulses per notes, and in the rate of call emissions [28]. This last parameter is exclusive of the advertisement call description under call-centered approach because the intervals between calls under the note-centered approach, as [26], represent a long time of silence of about 20–40 min [6,26,27] which, usually, is not measured.

The note-centered approach to advertisement call description in Brachycephalus [6, 26, 27, 28, this study] had not been adopted by other research groups working with the genus until recently [47,54]. However, in both studies, the authors treated the advertisement call as a phenomenon distinct from that considered in other works [6,26,27,28], this study]. Mângia et al. [47] interpret the call of B. tabuleiro as what other words have described as “note groups” (see Table 4), whereas Toledo et al. [54] interpret the call of B. dacnis as corresponding to what other works have described as “note groups” (the “complex” call) and “isolated notes” (the “simple” call). Mângia et al. [47] stated that they used the call-centered approach to describe the advertisement call of B. tabuleiro, but, considering two notes as its call they have worked, in fact, with the note-centered approach. In addition to this methodological issue (see below), the authors also used two different terms (“call duration” and “note duration”) to refer to the same parameter, and missed the opportunity to clarify what the isolated notes of B. tabuleiro actually represent (see Table 4). Toledo et al. [54] adopted the note-centered approach in the description of B. dacnis, following Bornschein et al. [28], who demonstrated that it was impossible to distinguish the call of that, at the time undescribed, species from that of B. hermogenesi using the call-centered approach. These authors also treated “call duration” and “note duration” as equivalent parameters. However, by doing so, either “call duration” would represent a longer time span than “note duration,” or “note duration” would necessarily include the interval between the two notes of the so-called “complex call” described by Toledo et al. [54].

For some species in the B. pernix group, the first notes emitted during an advertisement call, under a note-centered approach, are difficult to hear in both the field and in recordings, which is why we rarely record the first emissions. We named these weak starting notes of an advertisement call as “warming notes” [26], assuming that these reflect the individual’s preparation process building up to the “typical” strongest notes. Nevertheless, for other species that we analyzed, we recorded some complete advertisement calls by individuals and for which the first notes were similar to others in structure and intensity. It is possible for individuals to begin calling with warming notes if they are “cold” and without them if they are “warmed up” by the previous emission of one or more calls.

Like warming notes, attenuated notes, also under a note-centered approach, could prepare the individual to emit the immediately subsequent notes with relatively higher sound energy/amplitude. Of those species that show attenuated notes, we demonstrate that not all notes are preceded by attenuated notes (see Table 4). Attenuated notes are perceived more clearly in spectrograms than in oscillograms. The quality of the recordings may influence the perception of these notes. “Typical” notes—when recorded far away from the emitter—have a spectral quality that does not allow for the detection of pulses and, in these conditions, attenuated notes might have been diluted by background noise (see [27]). The accuracy of the researchers can also influence the perception of the attenuated notes. For example, there are attenuated notes in the advertisement calls of two species, B. olivaceus and B. quiririensis (Table 4), for which previous descriptions of their calls did not mention this feature [64]. However, the oscillogram of Fig 2A of Monteiro et al. [64] does reveal that attenuated notes were emitted in the advertisement call of B. quiririensis; prior to the notes shown in that figure, there is a discrete intensity of energy related to the attenuated notes.

Intraspecific variation in the level of complexity in advertisement call, under a note-centered approach, could reflect different functions to the different parts of the calls (with or without note groups), i.e., defense vs. mating [26]. However, later it was demonstrated that not all Brachycephalus advertisement calls present note groups [27]. In fact, we still know little about the daily, seasonal, and behavioral variation of Brachycephalus calls and their functions.

Finally, Goutte et al. [9] demonstrated that some species in the B. ephippium group are insensitive to their own advertisement call. However, Monteiro et al. [63] described responses by B. actaeus to playback experiments, perhaps perceiving call emissions through vibrations in other body receptors [28].

Taxonomy

The first described species of the B. pernix group stood out for its unusual morphology (see [65]), including the absence of dermal co-ossification that is common to the B. ephippium group and a body shape that differs from the leptodactyliform shape common to the flea toads (see [13]). Considering all 21 species in the B. pernix group described to date, there are few diagnostic features distinguishing them—sometimes limited to coloration in life—though genetic evidence supports these as distinct lineages [33,59]. Relatively few additional characters contribute to distinguishing species within the B. pernix group, in part due to the scarcity of osteological and vocal descriptions. For example, the presence of the neopalatine ( [60]; see also [61]) and presence of note groups in the advertisement calls [6,26,27]. Only eight species of the B. pernix group had their advertisement call completely described so far [6, 26, 27, 47, 63, 64, this study] and only four were also diagnosed by call features [6, 47, 63, this study].

There are two additional difficulties in delimiting species of the B. pernix group: the procedures used in call descriptions and the use of color characters based on the literature alone. Two approaches are recognized for describing anuran calls: the call-centered approach and the note-centered approach [25]. Brachycephalus actaeus [63], B. olivaceus [64], B. quiririensis [64], and B. tabuleiro (see below; [47]) had their advertisement calls initially described in a call-centered approach (but see Table 4 for partial description of these species under note-centered approach), whereas B. albolineatus [26], B. mirissimus [6], B. tridactylus [27], B. lulai sp. nov. and all remaining species of the B. pernix group, except B. mariaeterezae, were described in a note-centered approach (this study). The primary difference between these approaches is that a set of notes issued in sequence is considered the advertisement call in the note-centered approach, whereas the call-centered approach treats a single note as the entire call [25]. Mângia et al. [47] introduced an unprecedented variation, likely not supported by the reference literature (see above). While based on Köhler et al.’s [25] call-centered approach, they suggest that the call of B. tabuleiro consists of two notes [47]. For a simple comparison, under the note-centered approach, as [26], the call of B. tabuleiro of Mângia et al. [47] with one note would be an example of an isolated note, and the call of this species composed of two notes would be an example of a note group (see Table 4). As demonstrated previously [28], the call-centered approach prevents the detection of some diagnostic call parameters because different parameters and substantial differences across them along the duration of a call are treated as simple variation of the call of the species. In conclusion, the use of the call-centered approach for species of the B. pernix group masks heterogeneity by treating distinct elements, perhaps with distinct functions, as mere variations of a single entity. We recommend that researchers employing the call-centered approach to describe Brachycephalus advertisement calls prioritize obtaining long-duration recordings of individuals, made at close range (< 2 m from the caller; see [27]). Moreover, we encourage the investigation of the potential biological significance of any heterogeneity observed in the calls.

As for the coloration in life, it has typically relied on qualitative interpretation of colors when descriptions were not made using a standardized color guide, as is the case of all described Brachycephalus species. This prevents robust comparisons in the descriptions of new species. However, the effect of personal interpretation can be minimized in the future literature if works limited the inclusion of live colors comparisons of the species that have been described by a single researcher.

The scarcity of information on features to distinguish species and the extent of intraspecific variation has led to debate about the validity of species in the B. pernix group [54,66,67]. No species, however, has been proposed as a synonym thus far, despite the accumulation of new specimens and expanding the knowledge on intraspecific variation. We emphasize that a species diagnosis can be revised through redescription, as was recently done for the flea toads B. sulfuratus in relation to B. hermogenesi [68]. A reassessment indicated that the diagnostic characters of these species varied across individuals, but also revealed other differences [28].

We have summarized the commonly used diagnostic traits and how these vary across the 21 species of the B. pernix group (Table 2). This is based on our observations of both living and preserved specimens in each species. All species can be distinguished by one or a combination of traits. There are also many similarities across these species, including in their advertisement calls and in coloration, such as the several orange or predominantly green species. Our summary shows that the green species such as B. olivaceus and B. albolineatus present distinct advertisement calls, the first with attenuated notes and B. albolineatus without them (Table 2). The advertisement calls also help to distinguish certain orange or brown species, such as B. tridactylus (see Table 2 and Fig 5D1–D3, E1–E3) without note groups and without attenuated notes, in comparison to the orange B. lulai sp. nov. that has both. The brown species B. brunneus and B. curupira can be distinguished by the maximum number of notes in a note group and maximum number of pulses per note, 3 and 4 versus 5 and 3, respectively, in addition to the iris having golden spots in B. curupira when alive (Table 2).

We have taken care to present the sample size in our summary of comparisons. Unfortunately, there are some species with less than 10 individuals collected, which limits our understanding of intraspecific variation in coloration and anatomical traits. Further sampling of Brachycephalus species is expected to significantly enhance our understanding of intraspecific variation, while also contributing to the revision of species boundaries. Both the lack of funds supporting field research and the difficulty of accessing some montane locations remain a problem to increasing sampling, which is sometimes achieved only after opening many kilometers of trails in dense forests [15].

Biogeography

Geomorphological and sedimentological studies in Serra do Quiriri have shown that there were at least three semi-arid periods in the region, interspersed with wet periods. The most recent of these semi-arid periods dates from the beginning of the Pleistocene [69,70]. We hypothesize that the biogeography of Brachycephalus reflects a history in which their forested habitat occurred continuously at the base of the mountain complex in semi-arid periods and that these forests extended to higher altitudes during more humid periods [14,33,34]. For the B. lulai sp. nov. clade to diverge into three species that occur in Serra do Quiriri (Fig 10 and 12; see Fig 1 of Bornschein et al. [21]), these forests likely occurred as isolated patches (microrefugia) among the dominant grasslands, which could have led to speciation by vicariance. In contrast, vicariance among other species in the B. pernix group likely occurred through isolation on mountains separated by forested valleys.

The southern arboreal vegetations of the Atlantic Forest are richer in “forest types” than the arboreal vegetations of the regions in the northern portion of this domain [71], but little is known, and even modeled, about the relationship and interaction between fauna and vegetation during the hypothesized dispersals of vegetation over time. The expansion of arboreal vegetation over mountaintop grasslands is documented by palaeoecological studies (e.g., [72,73]) and still occurs today, potentially in all mountains where grasslands are still present. This expansion typically begins with the establishment of shrubs and small trees (M.R. Bornschein, pers. obs.). This early-stage “arboreal” vegetation is less than 1 m tall and is characterized by an herbaceous-shrubby stratum. It gradually develops in structure and increases in height, transitioning into more complex arboreal formations. In areas where grasslands persist, the advance of this arboreal stratum continues, leading to the formation of a mosaic of vegetation types in various stages of ecological succession. When shrubs and trees reach approximately 3–4 m in height, the vegetation becomes stratified into two strata: an herbaceous-shrub stratum and an arboreal stratum. From this point up to around 10 m in height, the vegetation structure typically includes three strata (herbaceous, shrub, and arboreal). Despite its relatively low stature and simplified vertical structure, this vegetation is classified as highland dense ombrophylous forest (Floresta Ombrófila Densa Altomontana), also referred to as cloud forest [74–81], based on an adapted version of the nomenclatural criteria proposed by Veloso et al. [23]. According to these criteria, true forests are expected to feature taller vegetation structured into four distinct strata ( [23]; see also [82]), a condition not met by these formations. Indeed, the recognition that highland dense ombrophylous forest is a structurally simplified vegetation type is well established. When composed of three strata, this environment is treated as a transitional formation toward montane dense ombrophylous forest (Floresta Ombrófila Densa Montana; [79]), which typically occurs at lower elevations. In contrast, any arboreal formation exhibiting four strata in the study region, above 400 m a.s.l., is consistently classified as montane dense ombrophylous forest [79].

It is crucial to recognize that the colder-climate arboreal vegetation occupying mountaintops within the geographic range of Brachycephalus does not exhibit four strata. This indicates that such vegetation functions similarly to pioneer formations in ecological succession (sensu [23]) or to regenerating vegetation following disturbances such as logging or fire (see [83]). Pioneer and regenerating formations undergo structural changes over time [23], often evolving substantially within just a few decades [84,85], unlike mature forests, which tend to remain “static” in their structural configuration over many decades or even centuries.

The growth rate of highland dense ombrophylous forest following its initial colonization of mountaintop grasslands remains unknown. However, since 1988, M.R. Bornschein has documented the advance of this forest over grasslands in the Serra dos Órgãos, Paraná (not to be confused with the homonymous region in Rio de Janeiro), reaching higher elevations and subsequently being colonized by B. brunneus and an undescribed Brachycephalus species [2]. These observations demonstrate the upward expansion of the geographic range of species in the genus Brachycephalus, reaching elevations of up to 1,878 m a.s.l. These species were recorded in 2018 in areas where they had not previously been observed, occupying vegetation estimated to be less than 30 years old, based on surveys conducted in the 1980s, 1990s, and 2000s. In contrast, forest advancement has been interrupted in some mountains where grassland fires are frequent, such as in Serra do Araçatuba, Paraná ( [72]; M.R. Bornschein, pers. obs.), where B. leopardus is found [2].

We hypothesize that highland dense ombrophylous forest served as the initial environment that enabled the historical upward expansion of forest ecosystems, subsequently facilitating the dispersal of Brachycephalus species and, later, their allopatric speciation via vicariance [14,33,34]. Furthermore, our findings indicate that Brachycephalus species continue to disperse into recently formed environments, demonstrating a degree of ecological plasticity. At least B. brunneus, B. leopardus, B. izecksohni, and Brachycephalus sp. currently occupy 1-meter-tall highland dense ombrophylous forest (M.R. Bornschein, pers. obs.), while B. leopardus and B. izecksohni may also occur in grassland habitats [21].

The same speciation processes that likely acted within Brachycephalus probably also characterize other organisms with limited dispersal capacity (see [86]). In addition to the 21 species of the B. pernix group that are endemic to the mountains of southern Brazil, this region is also home to four endemic phytotelm-breeding anurans in the genus Melanophryniscus [87,88] and two small bird species related to Scytalopus speluncae that are awaiting description ( [89]; taxonomy according to Maurício et al. [90]).

Conservation

The many recently described species of anurans endemic to the state of Santa Catarina [6,13,62,63,88,91] motivated technical meetings between the Environmental Institute of Santa Catarina (Instituto de Meio Ambiente – IMA), the Fundação Grupo Boticário de Proteção à Natureza, and researchers from the Mater Natura – Instituto de Estudos Ambientais. As a result, IMA instituted the program called “Conservação dos anfíbios microendêmicos e de distribuição restrita de Santa Catarina”, which aims to support and develop research on the distribution and conservation status of these species, and to develop conservation strategies to them (Portaria IMA nº 283/2018). The present study is an effort aimed at all of the objectives of that program.

Our Green Status assessments follow the recommendation to perform this analysis for other endangered species of Brachycephalus [92]. This third Brazilian study evaluating Green Status again demonstrated the efficiency of this tool in assessing the species’ current situation and the impacts of conservation strategies to be developed [85,86]. Additionally, the general assessment of the species made it possible to rank priorities and to define target species and key areas for conservation.

None of the 12 species of the B. pernix group from Santa Catarina state were recorded in Integral Protection Conservation Units (IPCUs; [2, this study]), a type of Brazilian conservation unit that necessitates the acquisition of land by the government. Although B. tabuleiro has been recorded as occurring in the IPCU Parque Estadual da Serra do Tabuleiro [47], our assessment suggests that the records are outside the limits of that park (M.R. Bornschein, pers. obs.). On the other hand, in the state of Paraná, eight out of nine species occur at least partially within this type of protected areas. In this state, most species are classified as Least Concern (N = 4) or Vulnerable (N = 3), mainly due to the absence of ongoing population declines and the lack of plausible future threats, contrasting with Santa Catarina, where most of the species are Critically Endangered (N = 5). These findings highlight the critical role of IPCU areas in safeguarding the microendemic Brachycephalus species.

There is a particular type of IPCU established in Brazil law that does not imply expropriation, unless there are uses in the area incompatible with the protection of natural resources. This IPCU is the Wildlife Refuge (Refúgio de Vida Silvestre; see the National System of Conservation Units – SNUC, Law Nº 9.985/2000). The original objective of Refúgio de Vida Silvestre was the protection of environments (SNUC, Law Nº 9.985/2000), but is has also incorporated the objective to protect particular species of the fauna and fauna [93]. Thus, we propose the creation of an IPCU called Refúgio de Vida Silvestre Serra do Quiriri, with 6,600 ha located in Serra do Quiriri, municipalities of Garuva and Campo Alegre, Santa Catarina (Fig 14). Its objectives would be the protection of B. lulai sp. nov., but also B. quiririensis, B. auroguttatus, and Melanophryniscus biancae, in addition to cloud forests and grasslands (campos de altitude). The limits of this conservation unit include all points of occurrence of B. lulai sp. nov. (this study) and B. auroguttatus as well as likely the entire potential habitat of B. quiririensis [2]. To comply with these purposes, we delimitated the conservation unit based on the altitude records of the three species. Melanophryniscus biancae is endemic to the highland grasslands of Serra do Quiriri and the adjacent Serra do Araçatuba, in Paraná state [19,88], and was recently recognized as endangered (EN) by Brazilian government (Portaria MMA Nº 148/2022) and also globally by IUCN [94].

The delimitation of the Refúgio de Vida Silvestre Serra do Quiriri encompasses a relatively small area with a low human population density. This refuge would have a greater strategic importance for conservation than any other unit that might be created for the other species of Brachycephalus in Santa Catarina (Table 8). Its creation and implementation would maintain the Green Score of B. lulai sp. nov. and B. auroguttatus, and, especially, change the status of B. quiririensis from Critically Depleted to Fully Recovered and performing in its baseline potential.

Among the anthropic impacts in the Serra do Quiriri are the regular burning of grasslands (that also affect adjacent forests), cattle grazing, invasion of Pinus spp., and kaolin mining [88,95]. Additional impacts include erosion along roads (MR Bornschein, per. obs.) and touristic trails [96]. Part of the area proposed as a conservation unit contains many cattle that negatively impact cloud forests by creating deep holes formed along cattle paths and, secondarily, erosion by rainwater. Under these conditions, in 2004, we still recorded B. quiririensis vocalizing in the vegetation between hollows, but we were not able to return to the area after 2010 and to confirm its permanence in the site (MR Bornschein, pers. obs.).

Possible management actions in the Refúgio de Vida Silvestre Serra do Quiriri could include a diagnosis of the impact of fire, management of grasslands with plots for controlled and rotational burning, establishment of a low-impact cattle stocking rate, reassessment of the impact and mitigation actions for kaolin mining, the cutting of Pinus, and the establishment of a plan to regulate tourism activities. Environmental compensation using governmental or municipal resources should be considered to reward landowners for the efforts to adapt to the management plan and to encourage alternatives to native wood used in the construction of barbed-wire fences. Due to its scenic beauty, the region attracts many tourists annually. Tourism can be expanded and provide funding to the region, also with a possible financial return to landowners.