**1. Introduction**

Frogs and toads (Anura) comprise more than 7000 species worldwide [1]. Special attention has been given to this group due to the large number of new species described each year as well as due to the increasing number of endangered species [2,3]. According to the IUCN Red List criteria [4,5], there are 1825 species of anurans at risk of extinction (25% of all species), making Anura the vertebrate order with the highest proportion of endangered species [5]. Since 1980, there have been records of a rapid population decline of nearly 450 anuran species [6–8]. The decline of these species can be mainly attributed to habitat loss and pathogens, such as chytrid fungi and Ranavirus [6,7,9–11]. Recently, Ranavirus has been reported in natural populations of frogs in South America, but the effects in wild anuran populations are still unknown [11]. Unlike Ranavirus, chytrid fungi (*Batrachochytrium dendrobatidis*) has been commonly reported as a cause of population decline in high altitude locations in Costa Rica and Panama [9]. Due to the rapid rate of the description of a new species, the proportion of endangered species, and sensitivity, Anura is the priority order for a conservation assessment, particularly in countries with a high level of deforestation, such as in Brazil [3].

The Atlantic Rainforest, a biodiversity hotspot [12], is the largest in area after the Amazon forest, with its original extent covering more than 1.3 million km<sup>2</sup> [13,14]. It is located on the eastern coast of South America, stretching from northeastern to southern Brazil, with inland extensions to the east of Paraguay, northeast of Argentina, and central Brazil. This biome has been experiencing massive habitat loss due to agricultural expansion, urbanization, and historic loss of natural habitats [15]. Currently, only 28% of the original extent remains if secondary forests and forests affected by the edge effects are included [15]. The Atlantic Rainforest houses nearly 2500 species of vertebrates, including 550 anurans, of which 323 are endemic (63%) and 15 are currently considered to be threatened by extinction [1,5,16].

The genus *Brachycephalus* (Fitzinger, 1826) is endemic to the Atlantic Rainforest and includes small (less than 2.5 cm in snout-vent length) diurnal toadlets with a reduced number of digits, bright colors, neurotoxins in the skin, and direct development, and they live in leaf litter, specifically that of montane forests [17–23]. There are currently 36 recognized species of *Brachycephalus* [1], of which 22 have been described in the last decade [1]. Most have extremely restricted geographical distributions of less than 100 ha [12,24,25]. *Brachycephalus* is divided into three phenetic groups [26], two of which (*B. ephippiumsi* and *B. pernix* groups) are montane with few records at lower altitudes, whereas the remaining group (*B. didactylus* group) includes more ecologically plastic species that occur from the sea level up to high altitudes [23,27]. The dependence on a colder climate and isolation in the mountains as sky islands have been hypothesized as the reason that montane groups have diverged into so many species (19 of *B. pernix* and 12 of *B. ephippiumsi* groups), whereas the *B. didactylus* group includes only four species [23,28,29]. Another species (*B. atelopoide*) cannot be compared to any of the groups due to the unavailability of the holotype [23,30].

Species descriptions of *Brachycephalus* have not been accompanied by corresponding assessments of the conservation status. Only 11 species have been assessed for the IUCN Red List to date [31–41]: eight as Data Deficient (DD) and three as Least Concern (LC). The Ministério do Meio Ambiente (MMA, the Ministry of the Environment of the Brazilian government) evaluated only four species and categorized one as Critically Endangered (CR), two as DD, and one as Near Threatened (NT) [42–45]. The absence of conservation status assessments of most species and the evaluation of some of them as DD highlight the need for a comprehensive effort to assess the risk of extinction of the *Brachycephalus* species, most notably the microendemic taxa found in the *B. pernix* and *B. ephippiumsi* species groups (*sensu* [26]). Species evaluated as DD should be prioritized to generate enough data to properly classify them into a conservation category [46,47].

One way to direct effective initiatives for conservation species is through a prior assessment of their conservation status [3]. There is a widely adopted IUCN methodology for proposing a conservation status [3], which serves an important role in allowing for comparisons and for classifying conservation actions as well the proposition of public policies. The objectives of the study were (1) to review data on occurrence, altitudinal distribution, density, and threats to the *Brachycephalus* species, (2) to compile new data from the literature and unpublished observations, (3) to generate systematized data on geographic distribution, population sizes, and threats to place them into IUCN conservation categories, and (4) to discuss conservation priorities and future managemen<sup>t</sup> actions.

## **2. Material and Methods**

All available occurrence records of *Brachycephalus* spp. were compiled from the literature up to the time of compilation (June 2019). The data encompassed toponymy, species identification, geographical coordinates of the occurrence record, and altitude of the corresponding site. Data on altitudinal range were also considered when available. The process began with the latest compilation of locality and altitude data for *Brachycephalus* provided by Bornschein et al. [23], and the same selection criteria were adopted for subsequent records. For example, those associated with precise localities were retained, and records that included only municipality names as occurrence information were discarded. Finally, the authors' previously unpublished data were included.

Occurrence records were plotted using Google Earth Pro v. 7.1.4.1529 and connected to form a polygon using the Minimum Convex Polygon approach (MCP; [48]) with modifications suggested by Reinert et al. [49] and adopted by Bornschein et al. [23]. These modifications allow for the exclusion of inappropriate habitats, such as bodies of water, pastures, silvicultures, urban areas, rock areas, and/or forest areas, beyond the altitudinal range of occurrence of the species.

Polygon delimitation required three or more occurrence records. For species with one or two records, polygons encompassing the altitudinal range of the species were created [23]. A continuous topography inside the polygon was considered a location (*sensu* IUCN and as IUCN [48]) that could potentially contain one or more records of a given species. The topography was considered discontinuous if it was isolated by altitudes beyond the altitudinal range of the respective species.

The MCP and altitudinal polygons were measured using GEPath v. 1.4.5 to obtain the extent of occurrence (EO; IUCN [48]; see also [23,25,50]) of each species. Because some species have such reduced EO, they could potentially also be ranked by area of occupancy (AO), although AO was not measured in this study; however, species with less than 1000 ha of EO could also be categorized based on the criterion of an AO of less than 1,000 ha (criteria B2, for CR [48]) as well as species with an AO less than 50,000 ha (criteria B2, for EN [48]) because AO is always smaller than EO and is located within the EO polygon [48].

Population size was inferred for each species based on the estimates of area in m<sup>2</sup> inhabited by one individual compiled by Bornschein et al. [24]. Based on estimates of the number of calling males [24], a sex ratio of one female per male [24] was assumed. In cases with distinct estimates of densities per species [24], the mean density was used. The mean area in which one individual per species can be found and its respective EO was then used to calculate the population size.

Data on EO, number of locations, population size, and threats of the species were integrated to evaluate and to categorize its conservation status according to the IUCN Red List and Criteria [48]. For the recognition of threats, data from the literature, personal field experience of the authors collected in the EO of 29 species, and information on land use, forest quality, and trends of deforestation over the previous 10 years were considered. For temporal trends in land use, the time series of satellite images of Google Earth Pro v. 7.1.4.1529 was analyzed.

In the treatment of the data in relation to the IUCN criteria, the flow chart presented in Figure 1 was used. Six pathways were developed beginning with the evaluation of the number of localities (one to two; three or more). If the species had up to two recorded localities, its altitudinal range was calculated. If an altitudinal range was not associated with the record, this prevented creating a polygon and estimating the EO. The species was then considered DD (pathway 3 of Figure 1). If the records were associated with altitudinal range, an EO was created based on the lower and upper altitudinal limits. It was not always possible to infer the EO without encompassing inland areas far west of the record and outside the assumed natural range, sometimes nearly reaching Argentina, which is clearly unrealistic. In these situations, the species were considered DD (pathway 2 of Figure 1). When there were up to two records associated with an altitudinal range that encompassed a realistic polygon for EO (as indicated), the status of the species was evaluated (pathway 1 of Figure 1). Further pathways related to the procedure can be observed in Figure 1.

**Figure 1.** Flow chart indicating the approach to creating polygons of the extent of occurrence to compare the results with IUCN's species extinction risk classification criteria [48].
