The bush dog Speothos venaticus and the short-eared dog Atelocynus microtis are categorized as Near Threatened on the IUCN Red List (IUCN, 2009), in particular because of habitat loss. Data on the distribution and ecology of the bush dog are scarce (Eisenberg, Reference Eisenberg1989; Redford & Eisenberg, Reference Redford and Eisenberg1992; Silveira et al., Reference Silveira, Jacomo, Rodrigues and Diniz-Filho1998; Michalski & Peres, Reference Michalski and Peres2005; DeMatteo & Loiselle, Reference DeMatteo and Loiselle2008; Oliveira, Reference Oliveira2009) and mostly based on opportunistic sightings (Peres, Reference Peres1991; Strahl et al., Reference Strahl, Silva and Goldstein1992; Silveira et al., Reference Silveira, Jacomo, Rodrigues and Diniz-Filho1998; Barnett et al., Reference Barnett, Shapley and Engstrom2001). Even less data are available on the distribution and status of the short-eared dog (Peres, Reference Peres1991; Eisenberg & Redford, Reference Eisenberg and Redford1999; Koester et al., Reference Koester, de Azevedo, Vogliotti and Duarte2008; Leite & Williams, Reference Leite and Williams2008).

The southern Brazilian Amazon has been subjected to large-scale deforestation (INPE, 2008) that has produced a fragmented landscape comprising forest remnants of varying size, shape and connectivity (Michalski et al., Reference Michalski, Peres and Lake2008). These forest fragments are experiencing multiple disturbance regimes (Michalski & Peres, Reference Michalski and Peres2005). Here, I document the occurrence of bush and short-eared dogs in the Alta Floresta region, northern Mato Grosso state, in the southern Brazilian Amazon, based on interview data and camera trapping.

During June–September 2001 and May–July 2002 I obtained data on occupancy of carnivores using interviews with local landowners in 129 forest fragments and 15 sites in continuous forest in a 4,648-km² area (see Michalski & Peres, Reference Michalski and Peres2005, for details; Fig. 1). I then conducted two camera trap surveys in a 2,850 km² area of the same region. In the first survey (June 2003–December 2004, total trap days = 3,086) CamTrakker phototraps (Camtrakker, Watkinsville, USA) were distributed across three habitat types on a hexagonal grid (c. 500 m apart): forest fragments < 1,000 ha (n = 18, 1,599 trap days), forest fragments > 1,000 ha (n = 3, 916 trap days) and continuous forest (n = 2, 571 trap days). In the second survey (October 2007–December 2008, total trap days = 3,635) Tigrinus phototraps (Tigrinus, Santa Catarina, Brazil) were used in four habitat types: continuous forest (n = 4, 1,028 trap days), forest fragments > 1,000 ha (n = 20, 1,102 trap days), riparian corridors (n = 12, 660 trap days) and pasture (n = 12, 845 trap days). The distribution of camera traps in the latter three habitats followed perennial streams (c. 500 m apart). In both surveys cameras were placed 30–40 cm above ground, deployed for 30 days, and checked every 3–4 days to renew the scent lure (Hawbaker’s Wild Cat Lure 2; Minnesota Trapline Products, Pennock, USA), batteries and film as necessary. Consecutive photos of the same species were defined as independent occurrences if individuals could be unambiguously distinguished or if the interval between photographs was > 30 minutes.

Fig. 1 The study area in Alta Floresta, northern Mato Grosso state, Brazil (see inset for location), showing the 144 forest patches and continuous forest sites where interviews with landowners were conducted (solid circles) and the location of the two camera trap surveys in 2003–2004 (open circles) and 2007–2008 (open rectangles). Grey and white areas represent forest and non-forest cover, respectively.

In the interviews bush and short-eared dogs were reported less frequently than all other carnivores (Michalski & Peres, Reference Michalski and Peres2005). Bush dogs were reported in only two forest fragments and six continuous forest sites, and short-eared dogs in only four forest fragments and 10 continuous forest sites (Michalski & Peres, Reference Michalski and Peres2005).

I obtained 1,167 independent photographs in 3,086 trap days during 2003–2004 and 1,845 independent photographs in 3,635 trap days during 2007–2008. From this total I recorded bush dogs in two photographs and short-eared dogs in seven photographs, in the 2007–2008 survey only (Table 1). These photographs were obtained in three continuous forest sites. The two photographs of bush dogs were from abandoned logging access tracks (5 and 8 years after selective logging), six photographs of the short-eared dogs were obtained along perennial rivers within continuous forest sites, and one photograph was obtained in a continuous terra firme forest. All photographs detected single individuals apart from one of bush dog that recorded two adults and one juvenile (K.E. DeMatteo, pers. comm.). Photographs of short-eared dogs were confirmed by carnivore experts with > 10 years of experience working with these canids (M.R.P. Leite-Pitman, pers. comm.; T.G. de Oliveira, pers. comm.). Landowners of two of the three continuous forest sites where these canids were photographed reported the presence of the two species in 2001–2002. Based on the number of independent photographs obtained and the sampling effort in all habitat types, one photograph required 3,361 trap days for the bush dog and 960 trap days for the short-eared dog. Considering only camera-trapping in forest fragments > 1,000 ha and continuous forest sites, these numbers decrease to 1,065 and 304 trap days per photograph for bush and short-eared dogs, respectively.

Table 1 Summary of photographs obtained during camera-trapping in 2003–2004 and 2007–2008 for bush dog Speothos venaticus and short-eared dog Atelocynus microtis in the Alta Floresta region, northern Mato Grosso, Brazil (Fig. 1).

¹ Daytime photographs were considered those between 05.00 and 18.00

² Obtained in the same day at the same camera-trap station within a 30-minute interval

³ Number and distance between camera-trap stations

Despite its broad distribution and occurrence in a variety of habitats (Silveira et al., Reference Silveira, Jacomo, Rodrigues and Diniz-Filho1998; DeMatteo & Loiselle, Reference DeMatteo and Loiselle2008; Oliveira, Reference Oliveira2009) the bush dog seems to be naturally rare throughout its range. Previous data based on tracks recorded the minimum home range for a group to be c. 1,600 ha, with a density estimate of c. 0.04 km^-2 (Beisiegel & Ades, Reference Beisiegel and Ades2004; Zuercher et al., Reference Zuercher, Swarner, Silveira and Carrillo2008). The record of only two photographs of bush dogs (three individuals in one and one individual in another) in an area of 2,850 km² gives an estimate of 0.001 km^-2 for the study area. Although the interview data reported the occurrence of bush dogs in 1.6% of the fragments and 40% of the continuous forests surveyed, I never recorded this species by camera traps in pasture, in riparian forests or in forest fragments < 1,000 ha. This contrasts with the finding that 20% of historical bush dog locations were associated with fragmented or altered habitat (DeMatteo & Loiselle, Reference DeMatteo and Loiselle2008).

Records of short-eared dog show a discontinuous distribution across its range (Emmons & Feer, Reference Emmons and Feer1997) and the species is characterized as rare and uncommon (Leite & Williams, Reference Leite and Williams2008). Recent studies (M.R.P. Leite-Pitman, unpubl. data) followed five individuals of this species in Cocha Cashu, Manu National Park, Peru, and estimated density to be 0.5 km^-2 (Leite & Williams, Reference Leite and Williams2008). Although recorded in seven independent photographs in Alta Floresta, all were obtained in a single continuous forest area. This may confirm the discontinuous distribution described by Emmons & Feer (Reference Emmons and Feer1997). This species was more frequently reported in forest fragments (3.1%) and continuous forests (66.7%) in interviews, which may suggest that a greater effort is required to record the presence of short-eared dogs in the study area. Another camera-trapping study recorded this species only twice in 905 trap days in 220,000 ha in Rondônia state, Brazil (Koester et al., Reference Koester, de Azevedo, Vogliotti and Duarte2008).

With intensive sampling across several years and seasons, my results highlight the low density of bush and short-eared dogs in a fragmented region in southern Amazonia. This is also supported by the interview data, which demonstrated < 40% probability of occurrence for both species in forest areas < 10,000 ha (Michalski & Peres, Reference Michalski and Peres2005). Although camera-trapping can produce unpredictable results for rare species, the fact that I surveyed a variety of habitat types and recorded bush and short-eared dogs in only 1.6 and 3.1%, respectively, of forest fragments (through interviews) and did not record either species at camera traps in sites other than continuous forest indicates that these species may avoid disturbed areas or at least use such areas less frequently compared to undisturbed areas. An evaluation of the distribution of bush dogs in northern Brazil also reported that the majority of records of this species are from undisturbed lowland terra firme forest (Oliveira, Reference Oliveira2009). My data provide information about these two elusive canids in southern Amazonia and this information is of particular importance for long-term conservation initiatives in this region and in other Neotropical agricultural frontiers.