Hidden by the name: A new fluorescent pumpkin toadlet from the Brachycephalus ephippium group (Anura: Brachycephalidae)

Species of Brachycephalus has been having taxonomical issues due its morphological similarity and genetic conservatism. Herein, we describe a new species of Brachycephalus from the south Mantiqueira mountain range and semidecidual forests in the municipalities of Mogi das Cruzes, Campinas and Jundiaí, state of São Paulo, Brazil, based on an integrative approach. It can be distinguished from all species of the B. ephippium species group based on morphological characters (especially osteology and head shape), advertisement call and divergence in partial mitochondrial DNA gene sequences (16S). The new species is genetically similar to B. margaritatus and morphologically similar to B. ephippium. It can be differentiated from B. ephippium by the presence of dark faded spots on skull and post-cranial plates, presence of black connective tissue connective tissue scattered over dorsal musculature, parotic plate morphology, smaller snout-vent length (adult SVL: males 13.46–15.92 mm; females 16.04–17.69 mm) and 3% genetic distance. We also present natural history data and discuss the robustness of the integrative approach, geographic distribution, genetic data, behaviour, fluorescence in ontogeny, and conservation status.


Introduction
The genus Brachycephalus Fitzinger, 1826, commonly known as pumpkin toadlets [1], is composed of miniaturized frogs with cryptic and aposematic species that live in the forest leaf litter and are most active during daylight [2,3]. The specimens are endemic to the Brazilian Atlantic Forest, eastern Brazil, and are spread along almost 1700 km, from Santa Catarina to southern Bahia states [3,4]. Regardless of genus distribution, most Brachycephalus species occur in restricted lowlands or mountains areas, where some representatives seem to occupy less than 100 ha of area (sky islands; see [5]). During the last ten years, researchers have looked deeper into these frogs' diversification and biogeography [6]. Consequently, the number of endemic Brachycephalus species increased significantly, especially with 15 species described for the last five years (e.g., B. crispus [7]; B. albolineatus [8]; B. sulfuratus [9]; B. darkside [10] do Meio Ambiente" (MMA), that also evaluates protocols for our collection and research. All collecting, holding and storage (scientific collection) procedures followed Brazilian animal care guidelines and were previously approved by the Universidade Estadual Paulista (UNESP) animal care committee (registration number CEUA IB/CLP #03/2020). All specimens were euthanized using anaesthetic overdose (Lidocaine 5%). The anaesthetic was applied to specimens' dorsum and venter. All procedures were made after breathing movements ceased. The procedures followed the recommendations of AVMA [29] and CFMV [30].

Collections and specimens examined
The specimens and audio records examined were deposited in the following Brazilian zoologi-

Morphometry
Morphological terminology of the snout shape follows Heyer et al. [32]. We used 15 measurements in the account, recorded in millimetres. Seven measurements follow Duellman [33]: Snout-vent length (SVL), head length (HL), head width (HW), eye diameter (ED), interorbital distance (IOD), internarial distance (IND) and tibia length (TL). Upper arm length (UAL), forearm length (FAL), hand length (HAL), thigh length (THL) and shank length (SHL), follows Heyer et al. [32]. Foot length (FL) was taken including the tarsus. Nostril diameter (ND) and eye-nostril distance (END) follows Napoli [34]. Snout-vent length (SVL) was taken with a calliper (precision 0.02 mm). All others were taken using the ZEISS ZEN Digital Imaging Software and a ZEISS Discovery V8 stereomicroscope (precision 0.01 mm) with an attached ZEISS Axiocam ERc 5s (5 MP) camera. ND was taken by the axis of largest diameter of the structure. Individuals HCLP-A 078, 079 and 081 could not be measured properly since they were damaged (dehydration) prior fixation. Sex of individuals was determined by jointly evaluating the presence of vocal slits and testicles in dissected males, and the presence of oocytes in dissected females, which is also externally evidenced by transparency and their overall body shape. We also took sexual size dimorphism into account. Besides their smaller size, juvenile individuals lack hyperossification or present less developed bony elements when compared with adults [10,35].

Osteology
Osteological descriptions were based on cleared-and-double-stained specimens with Alizarin red and Alcian blue (to distinguish bone and cartilage, respectively) following the developed protocol of Taylor & Van Dyke [43]. We prepared three specimens (ZUFMS-AMP 13647-13649; one juvenile and two male adults, respectively) under this procedure. Terminology of the bone elements follows Silva et al. [44] and Campos et al. [35]. We removed the skin of the specimens after the result of cleared-and-double-stained (and there were no osteoderms on the skin).

Genetic distances
We extracted whole genomic DNA from ethanol-preserved muscle tissue according to the following protocol: approximately 1 mm of tissue sample was put in a sterile Eppendorf and were added 50 μL of EAR buffer (100 mM of TrisHCl pH 8.5; 5 mM of EDTA pH 8.0; 0.2% of SDS and 200 mM of NaCl) and 4 μL of proteinase K (20 mg/mL). The sample was incubated overnight at 55˚C, under 300 rpm shaking. Then, the sample was incubated for 10 minutes at 95˚C without agitation, followed by the addition of 500 μL of TE buffer (1 mM of EDTA; 10 mM of TrisHCl, pH 7.5). Finally, 5 μL of RNAse A (3mg/mL in TE buffer) was added and incubation was done for 10 minutes at 37˚C. The final sample was stored at -20˚C. Primers (An16SF 5'-ACCGTGCGAAGGTAGCGTAATC-3'; and An16SR 5'-CCTGATCCAACATCGAGGTCG-3') and DNA amplification (PCR) procedures followed Lyra et al. [45]. Sequencing was performed at BPI Biotecnologia Ltda. (Botucatu, state of São Paulo, Brazil). A fragment of the 16S rRNA gene was sequenced for three specimens of the new species (HCLP-A 078-080, GenBank accession numbers: MT396578, MT396579 and MT396580, respectively) resulting in 416 bp.
We edited and aligned with sequences of seven species from Brachycephalus ephippium species group available in GenBank in Geneious v.7.1.3 with the MUSCLE algorithm using default parameters [46]. We removed the gaps using Gblocks [47,48]. The final dataset was 410 base pairs (bp; 98% of the original 416 positions). We have calculated sequence divergence (uncorrected p-distances) among species/individuals for the 16S mtDNA using MEGA v.7.0.26 [49]. To assess the sequence divergence of the new species within the genus Brachycephalus ephippium species group, homologous sequences of remaining species of this group available in GenBank were downloaded (S1 Table).

Natural history
We studied the natural history of the new species from October 2017 to September 2019 in two different areas in São Francisco Xavier subdistrict, São José dos Campos, state of São Paulo, Brazil (Projeto Dacnis Reserve and Reserva das Araucárias). São Francisco Xavier is located in the Mantiqueira Mountains with typical montane and submontane Atlantic Forest formations (750 to 2000 m a.s.l.) with patches of rocky outcrops and many rocky streams. The area has a seasonality with the dry season occurring from April to August (average temperature in July is 15.5˚C) and the wet season between September and March (average temperature in January is 21.6˚C) (http://climate-data.org). In 2018, the pluviometric index ranged from 26 mm in June to 444 mm in November (private pluviometer located near Projeto Dacnis area; brand Acurite 01079 Pro Weather Station). To statistically analyse the influence of rainfall on the abundance of sighted individuals of Brachycephalus sp. nov. monthly, we used a generalized linear model (GLM) with a Poisson regression in software R [39]. This test is suitable for parametric data and count data.
During the study we conducted 76 active search field surveys between October 2018 and September 2019, totalling 94 hours/man. We also used additional data from fortuitous observations made from October 2017 to September 2018. Our search effort was made during daylight hours, mostly between 08:00 and 11:00 am. We collected data on seasonal and daily activity, microhabitat use, breeding activity, and defensive behaviours of the specimens. For the purpose of discussing conservation issues, we have calculated the Extent of Occurrence (EOO; [50]) using the tool "Polygon" on Google Earth Pro © (Google LLC). EOO is calculated by applying a Minimum Convex Polygon (MCP), the smallest polygon in which no internal angle exceeds 180 degrees, and which contains all the sites of occurrence [50].

Registration of nomenclatural act
The electronic version of this article in Portable Document Format (PDF) will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank Life Science Identifiers (LSIDs) 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:08CB9413-3F67-4B23-8D71-214B604B46FD. The online version of this work is archived and available from the following digital repositories: PLOS One, PubMed Central, and CLOCKSS.
Brachycephalus rotenbergae sp. nov. also exhibits black connective tissue scattered over dorsal musculature, which in most specimens also forms two dorsally visible paired marks on the sacral region (Fig 2E and 2F). This feature distinguishes this species from B. ephippium, B. izecksohni, B. pernix, and B. pitanga, all lacking this pigmentation. In B. garbeanus and B. margaritatus, some specimens were reported to present few scattered areas with pigmentation in the dorsolateral region adjacent to the dorsal musculature. While in B. vertebralis there is some internal dark pigmentation near the region of the spinal vertebrae [10].  [9,26,56,57]. The parotic plates of B. rotenbergae sp. nov. are expanded laterally, making the posterior region of head wider than long, and partially concealing the posterior tip of the squamosal. In B. ephippium, the parotic plates are not expanded laterally, making the head longer than wide, while they are more developed anteriorly (CSG pers. obs.; also see [10,35] [10,13,55,64]. Moreover, the paravertebral plates of B. rotenbergae sp. nov., although fused with the spinal plates forming a shield, is not projected over the vertebral spine. This characteristic further distinguishes it from B. garbeanus, that exhibits paravertebral plates that are projected over the vertebral spine [64]. The absence of warts on body also differentiates the new species from B. atelopoide [13,55]. Snout shape rounded in dorsal and lateral views in Brachycephalus rotenbergae sp. nov. differentiates it from B. didactylus, B. hermogenesi and B. pulex (snout pointed in dorsal view [26,56,57]), B. brunneus (snout slightly mucronate in dorsal view [2]), B. leopardus (snout slightly truncate in dorsal and lateral views [2]), and also B. quiririensis (snout mucronate in dorsal view [62]). The protuberant nostrils of the new species also distinguishes it from B. alipioi, B.
Arm and forearm relatively slender; arm slightly shorter than forearm (UAL 22.6% and FAL 23.1% of SVL); forearm slightly hypertrophied; hand shorter than forearm. Hands with all fingers distinct; fingers I and IV reduced, vestigial; II and III robust; tips of fingers I and IV rounded, II and III pointed; relative finger length IV < I < II < III; subarticular tubercles absent, inner and outer metacarpal tubercles absent (Fig 1D). Leg relatively short and moderately robust; shank slightly shorter than thigh (SHL/SVL = 38.0% and THL/SVL = 44.3%); foot longer than tarsus and shorter than shank, with toes I and V reduced, vestigial; toes II, III and IV distinct and with pointed tips; relative toe length I < V < II < III < IV; subarticular, inner and outer metatarsal tubercles absent (Fig 1E).
Skin on top of head and dorsum rough due to surface ornamentation of skull and post-cranial plates; dorsal posterior region of body with scattered warts; skin on belly and limbs smooth; skin on lateral surface of the body and around cloacal opening granular. SVL

Colour in life
Overall colour of the body is bright orange. Skull and post-cranial plates darker than the background of the dorsum, with faded dark spots or shades. Eyes completely black. Like other Brachycephalus species, Brachycephalus rotenbergae sp. nov. presented fluorescence when backlighted by a UV flashlight (Ultrafire1, 395 nm wavelength) (Fig 4).

Colour in preservative
Overall colour of the body is pale, yellowish cream. Belly with dark background. Tips of FIV and TIV with dark spots. Skull and dorsal plates greyish yellow with faded dark spots. Eyes completely black.

Variation
Measurements of the type series are given in Table 2. The overall texture of the skin varies among individuals, with some specimens presenting an entirely smooth body, while others    exhibit scattered warts or granular texture on the lateral portions of the trunk. Ontogenetic variation in co-ossification of skull and post-cranial plates is present, with juveniles exhibiting less ossification and ornamentation. The fluorescence at the skull and post-cranial plates cannot be seen by the naked eye in juvenile specimens (SVL = 7.0 mm). The relative size of paravertebral plates may vary between individuals with the same body size (Fig 2A-2D). Spinal plates of presacral vertebrae I and II may be separated from each other, although always in close contact. The spinal plate of presacral III may be fused to the dorsal shield. The spinal plate of presacral VIII and sacral vertebrae may be separated from the dorsal shield. The amount of black connective tissue covering the dorsal musculature may vary from just a few scattered spots ( Fig 2F) to a more intense coverage (Fig 2E). Dark spots may be present or absent below the tip of FIV or TIV. Colour in life is the same among adults, but different between adults and juveniles: the juvenile specimens are slightly darker, as reported for other Brachycephalus species.

Etymology
The epithet rotenbergae is a patronym in honor to Elsie Laura K. Rotenberg, founder, and leader of the Projeto Dacnis, a Brazilian NGO dedicated to research and conservation of Atlantic Forest domain.

Advertisement call
Our recordings featured a single type of call from Brachycephalus rotenbergae sp. nov., which is considered herein as the species advertisement call (sensu [67]). This call is characterized by pulsed notes (note = call) without frequency modulation and emitted in long sequences (i.e. call series, sensu [67]; n = 5 call series, 408 notes; Fig 5; Table 1). Our longest recording featured more than 265 notes in a series but did not encompass the beginning of such series. Note duration is 0. 13

Osteology
The two cleared and double stained specimens of Brachycephalus rotenbergae sp. nov. showed a well-developed skull and post-cranial plates, both laterally expanded. Parotic, spinal and paravertebral plates ornamented with dermal roofing bones ( Fig 6A). The specimens ZUFM-S-AMP 13648-13649 (males) have, respectively SVL 15.9, and 15.8 mm. Body skin of these specimens without osteoderms. Dorsomedially, the parotic plate is fused with the post-orbital crest, and it is expanded laterally, making the posterior region of head wide; this plate partially conceals the posterior edge of squamosal and optic capsule, laterally ( Fig 6A). Nasal, sphenethmoid, frontoparietals, prootics, and exoccipital fused, forming a dorsal cranial plate. All frontoparietals (periorbital, parietal and central regions) ornamented with dermal roofing bones; anterior and parietal regions of frontoparietals fused medially; lateral and posterior orbital flaps present and conspicuous; post-orbital crest present and ornamented with dermal roofing bones (Fig 6C). Premaxillae broad, not fused medially; pars dentalis present, but odontoids absent. Posterior end of the maxilla tapered; posterior end of angulosplenial with squamosal is projected anteriorly than the posterior line of the skull. Quadratojugal and palatine absents. Pterygoid slender, dorsoventrally oriented, anterior ramus long articulating with maxillary arch; posterior ramus short, articulating with ventral ramus of squamosal ( Fig 6D). Squamosal broad in lateral view; zygomatic extremely short, and optic ramus short, both not ornamented. Nasal capsule and sphenethmoid (nasal region) completely ossified. Nasal bones hyperossified with ornamentation; anteromedial region of nasals concave and posteromedial fused. Parasphenoid robust and fused with sphenethmoid. Vomer distinct with rounded medial edges (6E). Tympanic annulus absent. Mandible edentate. Arytenoid cartilages not mineralized (Fig 6E).
Ventral column composed of eight non-imbricate vertebrae and hyperossified spinal processes of vertebrae (Fig 6B and 6I). Only the first presacral vertebra, atlas, lacks transverse process. Presence of spinal and paravertebral plates well developed and ornamented (Fig 6A). Spinal plates (spl) associated with the spinal process of all presacral (I-VIII) and sacral vertebrae. The spl I and II fused, forming broad plate as kidney shaped; spl III small with irregular shaped, separated of principal group of dorsal plates; spl IV and V fused; spl VI-VIII slightly separated each other's with irregular shaped, and size decreasing towards the sacral region; spl of sacral vertebral fin as dagger (Fig 6A). Paravertebral plates broad with rounded edges. Dorsally, the paravertebral plates are medially fused with spinal plates IV and V ( Fig 6A); anteriorly conceals (completely or partially) the transverse process of the vertebra III; medially conceals completely the transverse process of vertebra IV and V, which are fused with the paravertebral plates ventrally (Fig 6I); and posteriorly conceals (completely or partially) the transverse process of the vertebra VII (Fig 6B and 6I).

Genetic distances
The genetic distances varied from 1-7% of divergence and depict B. rotenbergae sp. nov. as most similar to B. margaritatus (1-2%

Geographic distribution
Currently, a monophyletic clade of four populations named Brachycephalus sp. 5. in the cities of Campinas, Mogi das Cruzes, Atibaia and Jundiaí, all in the state of São Paulo; and two populations from the municipality of Resende, state of Rio de Janeiro [6]. Based on molecular distance, B. sp. 5 is synonymous with B. rotenbergae sp. nov., known for 10 locations in the southern portion of Serra da Mantiqueira (Fig 7). The known distribution of B. rotenbergae sp. nov. is shown on the map, as well as the other type localities for species of the B. ephippium group. In the case of B. ephippium, we indicate the type locality [56,63,66] and the locations pointed out by Condez et al. [6].

Natural history
Microhabitat structure. The field surveys were conducted in two different areas within the subdistrict. These areas present some differences in the slope of the terrain, vegetation structure, and altitude.
Projeto Dacnis (22˚53'S 45˚56'W, 842m a.s.l.). This area is composed of two land margins separated by a small stream that is at most 1.5m wide and runs slowly through the landscape. This stream is shallow, its bottom is composed mainly of sand and small rocks, and there are many logs, tree branches and plants that can be used as bridges by small animals. On one margin there is a swamp area where most of the soil is permanently wet. The canopy cover of the swamp is mostly composed of banana trees (Musa sp.), and light incidence is very reduced. It is surrounded by a sloping dry area with medium-high canopy cover and bushes. On this side of the stream, most Brachycephalus rotenbergae sp. nov. specimens were using the slopes, perched on bushes, twigs or roots. However, we also found other specimens moving across the muddy area of the swamp, using fallen and non-fallen banana trees as perching sites.
On the opposite margin of the stream there is a heavily sloped terrain with medium-high canopy cover formed by juçara palms (Euterpe edulis), bushes and some sparse trees (Fig 8A). The light incidence is medium-high. There is a shallow leaf litter cover with many branches and roots emerging from the soil. These are used by the specimens as perch sites during their activity, especially the dry leaves of juçara palms. Most of the specimens in this area were found perched at small heights, although others were found under leaf litter. specimens also used the root formations of juçara palms as shelter, and we also observed a reproductive activity inside these structures (see below). The exposure of the specimens on the perches seems to be linked to their seasonal activity and to the weather conditions. In dry conditions, most specimens were found under the leaf litter instead of perched.
Our observations also suggest that there is a local movement of the specimens on the slope that is linked to abiotic factors, since during dry periods we found most specimens on the lower parts of the slope (closer to the stream). However, when the wet season begins, and the soil gets wet, activity seems to occur all along the slope. The sloped terrain, the large number of perches, the presence of leaf litter and the proximity to the stream seem to influence the specimens concentration in this spot when compared to other sites in the vicinity where specimens are present only in small numbers.
Reserva das Araucárias (22˚54.1130'S 46˚0.0590' W, 1299 m a.s.l.). This is a relatively flat, forested area with some pronounced rocky outcrops. The canopy is dense and high, and light incidence on the soil is medium-low. In the understory there are countless lianas, small bushes and many fallen twigs and logs (Fig 8B). A deep leaf litter layer lies on a muddy soil and most of the trees have abundant moss. Small streams surround the area but do not cut into it as in the Projeto Dacnis area. In Reserva das Araucárias we found more Brachycephalus rotenbergae sp. nov. specimens per day than in Projeto Dacnis area. specimens were found both perched and moving across the leaf litter. specimens used rock outcrops, fallen tree trunks, roots and twigs as perching sites. Even though the streams are not so close in Reserva das Araucárias, the environment is also very humid during the wet season. In fact, this area is usually covered in dense fog at night due to the high altitude.
In both areas, specimens were found using perches and leaf litter. Most perches were no more than 30cm high, although a male was 140 cm from the ground. Males and females share perching habits, and it seems to be more related to weather conditions than to vocalization activity. During all field surveys, we noticed that perched specimens detected our presence very quickly. We could see them trying to escape or hide under the leaf litter. Thus, even without any rigorous scientific test, we believe that Brachycephalus rotenbergae sp. nov. has good eyesight and maybe this could be linked to the perching behaviour. Activity and spatial use. The monthly abundance of individuals is positively correlated with monthly rainfall (R 2 = 0.25; p = 0.005; Fig 9). During the study we performed 370 observations. In both areas, we were able to detect specimens almost all along the way, but in much higher numbers in specific points. Brachycephalus rotenbergae sp. nov. activity is seasonal and related to rain regime (Fig 9). specimens were found all year long, except from May to July (Table 3). The activity peak seems to occur from September to February. During the active season, specimens were found in most field surveys, but activity peaks seem to be related both to the season and to rainy weather. Thus, in the same month specimens were present in high or low numbers according to the weather conditions. In September 2019, for instance, we were able to find 38 specimens during a single field survey that occurred after almost a week of weak rain, but during successive dry days such activity decreased abruptly, and no specimens were found in this same area. Months with less than 200 mm of precipitation did not seem to be favourable to the activity of Brachycephalus rotenbergae sp. nov., however, it is important to notice that months with low precipitation but with constant drizzling rains can present a huge specimen's activity (Fig 9). During these months (especially at the beginning of the wet season), most specimens were found after short rain periods, when air and soil humidity increased for a little while. Although soil humidity was not measured during our study, specimens' activity seemed to be linked to it. During extended rain periods, soil was moist, and activity seemed to occur more continuously, but during sparse rainy periods, soil moisture did not last long, and only brief peaks of activity were observed. Temperature may also affect activity, yet probably secondary to humidity, as during our field surveys we found active specimens at temperatures below 16˚C coupled with drizzling rain, but no activity in warm but dry weeks.
The species is diurnal and only seven of 370 observations were made at night. These seven specimens were inactive and hidden under leaf litter, and no calls were heard. In our study we found Brachycephalus rotenbergae sp. nov. specimens during all daytime length, including late afternoon. We did not measure the activity of the specimens in different daylight hours, but we recorded calling activity and amplexus both in the morning and in the afternoon.
In both areas, specimens were found using perches and leaf litter. Most perches were no more than 30 cm high, although a male was 140 cm from the ground. Males and females share perching habits, and it seems to be more related to weather conditions than to vocalization activity. During all field surveys, we noticed that perched specimens detected our presence very quickly. We could see them trying to escape or hide under the leaf litter. Thus, even without any rigorous scientific test, we believe that Brachycephalus rotenbergae sp. nov. has good eyesight and maybe this could be linked to the perching behaviour.
Breeding activity. Calling activity seems to be seasonal and related to weather, it was recorded less than 15 times during all field surveys (Table 3). Males were found calling mostly under the leaf litter, but we also saw males perched with their legs extended and heads pointing up, in calling activity (S1 Video).
During this study, we found four amplexi, all of them the inguinal type ( Fig 10A). In November 2018, we followed an amplected couple in the Projeto Dacnis area for five consecutive hours. During this time, the amplected couple climbed the slope terrain and stopped from time to time under some leaves. After a while, the couple entered under a juçara palm root system ( Fig 10B). They went so deep into this shelter they were barely visible. After a few hours, four individuals emerged from the E. edulis root shelter: three distinctive bold females and one male. We searched for eggs at the site but did not find any. Considering that we observed this site during all the action, we strongly believe that the other two females were already inside the  shelter before the couple arrived. Since we did not detect any clutch inside the shelter, we cannot say if the root system of juçara palm is used as a breeding site. Nonetheless, the presence of four different specimens under it is a strong evidence of its use at least as a shelter. We followed a second amplected couple on 5 February 2019, in Reserva das Araucárias area. The couple was moving through the landscape until it disappeared inside a huge root system. We waited to see if the couple would leave the site, without success. A third amplected couple was followed on the same day in the same area, but they separated after a few minutes of observation. The fourth amplexus occurred after two days of constant drizzling rain on 26 September 2019. Sadly, we unintentionally disturbed the couple, interrupting the amplexus. Despite the field work effort, we did not find any clutch deposition. Defensive behaviour. We detected gaping defensive behaviour in three specimens that were manipulated, and we saw them opening their mouths widely (Fig 10C). The gaping and the flee away strategies were the two defensive displays that Brachycephalus rotenbergae sp. nov. presented during our study. The mouth-gaping behaviour was previously reported for B. pitanga and B. ephippium by Goutte et al. [68]. Despite the aposematic colours of Brachycephalus rotenbergae sp. nov., it is important to note that their colours may also work as a camouflage in their microhabitat, since there are great amounts of tiny yellow and orange leaves, mushrooms and seeds on the ground (Fig 10D), especially during the active season. Together, these elements make the specimens' bright colours not as conspicuous as they may seem. During our field surveys, it was quite common to mistake a specimen for other yellow/orange small things on the landscape.
In order to identify possible predators of Brachycephalus rotenbergae sp. nov., we modelled tiny frog shapes using yellow and non-yellow atoxic modelling clay. We placed the frog models in some areas, recording any interaction with potential predators. During this assay, we recorded a slaty-breasted wood-rail (Aramides saracura) grasping a yellow frog model in its bill. The bird started running and dropped the model but picked it up from the ground. Although we cannot be sure if the bird recognized our model as a Brachycephalus, we suggest that yellow is not an efficient warning colour for A. saracura. During the study period, no predatory event on Brachycephalus rotenbergae sp. nov. was recorded.
Additional information. All study areas in São Francisco Xavier have problems with the invasive and exotic wild boar (Sus scrofa). Wild boar numbers are increasing in the subdistrict, and due to their foraging methods could negatively impact the Brachycephalus rotenbergae sp. nov. microhabitat, since these mammals excavate the soil and leaf litter in search of food. Despite no scientific evidence, we highlight the importance of future studies in evaluating the impact of this invasive species on frog communities.

Discussion
The description of Brachycephalus rotenbergae sp. nov. is based on an integrative approach, coupling external morphology, coloration, osteology, vocalization, and genetic data. The Brachycephalus species has been taxonomically problematic owing to its morphological similarity and genetic conservatism [6,22]. Currently, the literature has listed osteology, advertisement call and genetics as the main taxonomic informative data to describe new species for the genus Brachycephalus [23,24]. The new species here described is genetically close to B. margaritatus. However, coloration with faded dark spots, absence of osteoderms, flat paravertebral plates, can easily diagnose B. rotenbergae sp. nov. from the latter. Furthermore, Brachycephalus margaritatus are restricted to the state of Rio de Janeiro [6,64], as supported by our comparisons with congeners.
Brachycephalus rotenbergae sp. nov. has been referred to as B. ephippium for the state of São Paulo in the literature (e.g., [53,69] for the type locality of the former), and scientific collection records. However, coloration with faded dark spots, presence of black connective tissue, parotic plate morphology, head wider than long, smaller size and 3% of genetic distance diagnose B. rotenbergae sp. nov. from B. ephippium. Also, B. ephippium is restricted to the state of Rio de Janeiro [6] with Mountains of Rio de Janeiro city suggested as type locality for the species; and we also have in the present study a robust number of specimens from several localities to support our comparisons with congeners.
Additionally, the duration of notes of the advertisement call of Brachycephalus rotenbergae sp. nov. seems to present slight differences for other localities of the State of São Paulo (cited as Brachycephalus ephippium [51]), with the "note B" showing higher values of pulses/s than the higher values of our recordings for this same acoustic parameter (100 pulses/s according to Fig 1B of [51]). The temporal parameters can be influenced by abiotic factors and the differences here seems to not be remarkable. Thus, further integrative investigation is needed to give proper description to the other vocalization records of Brachycephalus rotenbergae sp. nov. from the state of São Paulo. Moreover, the advertisement call of Brachycephalus ephippium (populations from the State of Rio de Janeiro), which would be valuable for further comparative studies, remains undescribed.
The genetic distance among the 16S rRNA mtDNA sequences of B. ephippium species group showed variation that helped us solve a puzzle. Considering both the 1% to 7% of variation for this fragment (present study) and the tree topology presented by Condez et al. [6], some conclusions are warranted. Brachycephalus rotenbergae sp. nov. is genetically divergent from all other species of the B. ephippium group, except B. margaritatus, with the phenotypic differences corroborating the specific status of the former. The low genetic divergence from B. rotenbergae sp. nov. to B. margaritatus can be due to (1) real genetic similarity, or (2) the 16S barcode is not effective to show the divergence between these populations. Several factors involved in the divergences at the population level can act in the magnitude of differences in intraspecific genetic divergences [70]. Thus, one gene fragment (here the 16S rRNA mtDNA fragment) for a recently derived species may not be very informative [71]. Recent splits have low support when studied by non-informative data (even for ultraconserved elements), as for Brachycephalus and Melanophryniscus [22]. For instance, Firkowski et al. [22] pointed out the differences in species delimitation according to the loci and methods and reinforced the need of several genes to have good resolutions for Brachycephalus. In addition, B. rotenbergae sp. nov. (called as Brachyephalus sp. 5, see [6]) and B. margaritatus, despite of the genetic distance, are reciprocally monophyletic.
The limits of the distribution of Brachycephalus rotenbergae sp. nov. overlaps the type locality of B. atelopoide. Brachycephalus atelopoide was described by Miranda-Ribeiro [55] based on only one specimen from municipality of Piquete, state of São Paulo, Brazil [4]. According to the original description (see [55], for the discussion of the information presented below), this would be a species without dorsal bone plates or bony projections on the spine, but with warts on the body. The holotype (by implication, not original designation) was collected among 31 specimens of B. ephippium. However, none of these remaining specimens agreed with the B. atelopoide description and were originally considered to be B. ephippium. Today, B. atelopoide cannot be associated with any of the populations associated with B. ephippium (by morphological similarity) or other Brachycephalus species (by distribution). Here, we have only one taxonomic information provided by the original description: the absence of dorsal bone plates. The first question that can be raised regards the ontogeny of dorsal bone plates. Miranda-Ribeiro [55] did not provide the size of the specimen. However, he also did not make any observations about the B. atelopoide specimen being smaller than the others or being a juvenile. In addition, the absence of these developed warts on the body in B. rotenbergae sp. nov. is a significant diagnosis from the original description of B. atelopoide. Thus, there is no information that would lead us to consider the lost holotype of B. atelopoide as a juvenile with absence of dorsal plates due to ontogeny. According to Pombal [13], among the specimens available, there are no individuals who agreed with the original description. The absence of information cannot be used here as a fact for the specimen developmental hypothesis. Therefore, the type cannot be traced, and we consider B. atelopoide a species inquirenda.
The fluorescence of the skull and post-cranial plates cannot be seen by the naked eye in juvenile specimens (less than SVL = 7.0 mm). The presence of different colour phases is not uncommon for anurans, including fluorescence (see [72,73]). For instance, some species of the Boana albopunctata group have two colour phases, a remarkable ontogenetic change, with juveniles exhibiting a green background (potentially fluorescent), while adults have a brown background [73]. We suggest that absence of visible fluorescence in juveniles of Brachycephalus rotengergae sp. nov. is linked to the development of the co-ossified skull and post-cranial plates in adults. However, this statement is contingent on further analyses.
The natural history of Brachycephalus rotenbergae sp. nov. seems to show some conservative behaviours for the genus, as how specimens position themselves perched for vocalization (see [51,73]), which brings some questions. Brachycephalus specimens are predominantly active by day on the forest floor and, occasionally, on low perches up to 50 cm [6,51]. Males usually hold a territory for advertisements by acoustical and visual signals [51]. It has recently been discovered that some pumpkin toadlets are unable to listen to their own congeners and visual communication can be an important selective force for song being conserved in the lineages [68]. As these frogs have striking colours, they are diurnal and are probably well oriented visually, using higher perches helps intraspecific communication and consequently influences reproductive success in Brachycephalus. This would probably explain the use of higher perches as a vocalization site for species of the genus.
The inguinal amplectic position registered here for B. rotenbergae sp. nov. converges with data already presented in the literature (cited as B. ephippium; [51]), but we were not able to find oviposition sites. Brachycephalus rotenbergae sp. nov. and toads of the genus Alytes can shift to a more dorsal position during oviposition [51,74]. This change to new position probably was the reason a cephalic amplexus was previously reported for Brachycephalus (see [75]), but not corroborated by the subsequent literature ( [51,76]; this study). The ancestral condition for the anuran amplexus seems to be inguinal, with the male clasping the female around the waist [74]. Considering most recent phylogenies for Anura [77], Brachycephalidae is the sister group of Eleutherodactylidae, and both are the sister clade of Craugastoridae. Thus, given that the axillary amplexus is widespread in the families Eleutherodactylidae and Craugastoridae [78,79], we propose that Brachycephalus presents this form of amplexus by secondary derivation of this character, not by the retention of a plesiomorphic state. However, this statement needs further evolutionary analyses. The pumpkin toadlet can be locally abundant in many places, but natural history reports are scarce [51]. These reports are still scarce after 26 years, as previously observed by Wells [74], reinforcing that the natural history information presented herein is of notable importance for the knowledge of the group and future conservation efforts.
The geographic records for B. rotenbergae sp. nov. are here delimited to several localities at the south Mantiqueira mountain range and semidecidual forests in the municipalities of Mogi das Cruzes, Campinas and Jundiaí, with the Extent of Occurrence (EOO; [50]) being~583.600 ha. It is common to the majority of the species of Brachycephalus to present a restricted distribution due to the geographic pattern for the genus [5,80]. The sky islands scenario is supported by diversification-by-isolation and suggests that allopatric speciation due to climatic gradients is an important mechanism for generating species diversity and endemism in these regions [5,7,11,22]. However, not all of the species of this genus presents such restricted distributions, like B. nodoterga. Despite the wider distribution for these two species, the geographic range is restricted to high places, like the other members of Brachycephalus, and it is expected that species with low dispersal capacity and poor plasticity regarding microhabitat and/or microclimate conditions are especially prone to diversify in these conditions [3,5,7]. Moreover, there are preserved areas near the type-locality and along the distribution polygon that could shelter the new species and possibly amplify its distribution.
Herein, we also presented several biological data. Despite the absence of population status, we have easily encountered this species in the type-locality and surrounding areas year by year. The habitat system is terrestrial, and the habitat type is forest. The district of São Francisco Xavier is mostly inserted in a Governmental Protected Area (Portuguese acronym APA), called São Francisco Xavier, and the Projeto Dacnis (and neighbours), has played an important role in hindering the continued decline in habitat area, extent and/or quality due to the residential and agricultural development. However, we call attention that the species was last found in the municipality of Campinas, São Paulo state, in February 1995 and is suggested to be locality extinct in that area (L.F. Toledo pers. comm.). Thus, we suggest the classification of B. rotenbergae sp. nov. as Least Concern (LC) category based on the IUCN Red List Categories and Criteria (Version 3.1; [81]), but highlight the sensitivity of these organisms to environmental disturbances. These data are significant to be used on the IUCN's criteria and future lists of Brazilian threatened species.
Supporting information S1 File. List of examined specimens. (DOCX) S1  Table. Record and respective sources for the geographic distribution map (Fig 5).
(XLSX) S1 Video. Video recording of a male Brachyecephalus rotenbergae sp. nov. in calling activity. (MP4) Laboratório de Aquicultura Sustentável (LAS) for the use of the stereomicroscope coupled to a camera. Douglas Alves Lopes for helping with diaphanization protocol, and Francisco Severo Neto for photographs of skeletons. Lena Rotenberg and Paulo Varchavtchik for supporting our work and always motivating us.