New distribution reports of Rhagomys rufescens (Rodentia: Sigmodontinae) Thomas, 1886

Rhagomys rufescens is a rare, arboreal sigmodontine rodent endemic to the Brazilian Atlantic Forest biome. This species is known from eight localities in Brazil. Here we present a new report based on four individuals of this species registered in Serra do Ouro Branco, municipality of Ouro Branco (Minas Gerais, Brazil). One juvenile male, one adult male and two juvenile females were captured in pitfall traps during the rainy season, in a study of small mammal ecology. This is the first record for the Espinhaco Mountain range and the northernmost report for the species in this state, 85 km northeast of the last registered location.

Forest domain, at the south of the Espinhaço mountain range. Our report occurred in an area of vegetation characterized as semi-deciduous mountain forest. A portion of the sampled locality is now inside a conservation area, the State Park of Serra do Ouro Branco. This is the first record of this species in the Espinhaço mountain range, and the fourth and northernmost report for the species in the state of Minas Gerais. It is located about 85 km northeast of the last registered location in the Estação de Pesquisa, Treinamento e Educação Ambiental Mata do Paraíso (EPTEA Mata do Paraíso), in the municipality of Viçosa (Percequillo et al. 2004). The other known report for Minas Gerais is in the municipality of Poços de Caldas (Percequillo et al. 2004) and Parque Estadual Serra do Papagaio, municipality of Baependi (Passamani et al. 2011).
We carried a long term study on small mammals of the Serra do Ouro Branco from December 2006 to December 2010. In this study we used pitfall traps with drift fences, composed of buckets of 60L. The sampling was conducted for four nights per month. The traps were organized in two different ways along the study time.
From December 2006 to December 2009, the field work was conducted in one locality using three linear transects of 10 buckets each, perfoming 4440 traps / night. The transects were installed in three different habitats and distanced at least 500 m from each other. One of these habitats corresponded to the forest interior, another to an edge of the forest fragment near a cornfield, and the last one, to an edge near an area with a particular physiognomy of the brazilian savanna biome, known as "campo rupestre".
From January to December 2010, we sampled two other localities within the same forest spot. In each of these we placed three transects, one in the interior of the forest fragment, another at an edge near the road and the last one at an edge near a cornfield. Thus, the edge near a "campo rupestre" physiognomy was excluded in this second phase. All sampled localities distanced at least 1.5 km from each other. In this period we also changed the conformation of the pitfall traps from linear to Y-shaped transects. Each Y-shaped transect was composed of four buckets, one at each end, and one at the juncture of the Y. These transects were always installed in sets of three in each sampled habitat. This second phase of the study totaled 5184 traps/night, corresponding to 1728 in each locality, and 576 in each habitat type.
The collects were authorized by IBAMA NUFAS license under the number 482/06 and ICMBIO 21543-1. The animals were deposited in the Mammal Collection of the Laboratory of Vertebrate Zoology of the Universidade Federal de Ouro Preto (LZV-UFOP), with accession numbers 163R, 213R, 214R and 215R.
Rhagomys rufescens was captured only in two localities, and always in the interior of the forest fragment, over 40 m away from the nearest edge. These two localities were approximately 2.5 km distant to each other. In one of these localities, the canopy reaches 25 meters in height and the traps were located within ten meters of a stream (20°29'43 .16" S, 43°36'26.71" W, 1069 masl). In the other locality, there was no water source nearby and the canopy was lower, with 18 meters on average (20°29'46.36" S and 43°37'23.97" W, 935 masl). This report of R. rufescens corroborates the pattern predicted by Steiner-Souza et al. (2008), in which this species seems to occur at altitudes above 500m, in transitional vegetation zones of montane forests.  We captured one juvenile male in October 2010, one adult male in September 2010 and two juvenile females in February and December 2007. The morphometric data of these specimens followed Pardiñas (2008) and the age classes followed Voss (1991); they are shown in Table 1. The cranial measures seem to agree with the description of Percequillo et al. (2004), but our sample size and the high proportion of juveniles hindered us from making statistic analyses.
All captures of R. rufescens occurred on nights of heavy rain, with the animals being drowned in the water accumulated inside the buckets. The buckets were provided with a piece of Styrofoam in order to help the animals float, which unfortunately was not effective for this species. Umetsu et al. (2006) observed that the efficiency of the capture of rodents using pitfalls increases with precipitation. This seems to be true for R. rufenscens. In our study, the rain may have acted by stunning the animals, thereby allowing them to fall into the bucket full of water, and making it impossible to escape by jumping out of it. Steiner-Souza et al. (2008) raise the question of whether R. rufescens is really a rare species or it is subsampled due to the type of trap used. The authors believe that pitfall traps are more efficient at capturing this species, because almost all specimens known were captured this way. However, in our four years of work, with a total effort of 9624 traps /night, only four individuals were collected, which leads us to believe that the species is really rare. If pitfall traps are the best kind to collect this species, maybe the buckets must be bigger to prevent them from escaping, because the specimens were captured only when the buckets were full of water. Thus, adults may have the ability to jump out of a 50 cm high bucket, which would be more difficult for young individuals, which correspond to the majority of the specimens captured in this study.
These records in a protection area are very important to R. rufescens conservation. Currently there are only three records of this species for Minas Gerais, and one of them, in Poços de Caldas, corresponds to an area under great impact due to bauxite mining (Percequillo et al. 2004).
More studies need to be made in montane forests aimed at finding new populations of R. rufescens in Minas Gerais, so we can monitor it and better understand the requirements and particularities of this species.