4.1 Spatial distribution of the roadkill
The mammals are the vertebrate group that is the most frequent focus of studies in road ecology (Cherem et al. 2007), due not only to their relatively large home ranges, but also their relatively active foraging behavior. The Crab-eating fox, C. thous, is one of the mammal species most impacted by highway traffic (Sobanski 2017).
During a year-long monitoring period of four stretches of highway in Brazil, Rezini (2010) collected 40 specimens of C. thous. Orlandin et al. (2015), collected 11 specimens of C. thous on 100-km stretches of a number of different highways, while Castro (2017) collected 122 samples of C. thous, of which, 27 were roadkilled. These data reflect the high frequency of collisions involving this species, and the abundance of specimens in the samples.
Relatively few incidents were observed in areas with vegetation of medium stature, which may be related to the reduced abundance of feeding resources for C. thous in these areas in comparison with areas of denser vegetation, which are assumed to have more abundant resources, which may lead to a reduction in the ranging of the animals during foraging. In areas of sparse and low-lying vegetation, the foxes may increase their ranging behavior in search of relatively scarce resources, and become more generalist, even exploiting carrion from roadkill on the margin of the highway, which makes the foxes more vulnerable to collisions.
The analysis of the rainfall of the different routes revealed many areas of low rainfall (annual mean of 700–800 mm), which are typical of the Caatinga dry forest domain. In other areas, annual rainfall is much higher, however, reflecting specific features of the environment, including the altitude, for example. The Borborema Plateau has a major influence here, and includes areas of cloud forest. Cloud forests are relatively humid environments located at high altitudes, typically more than 600 m above sea level (Freitas et al. 2019), forming ecoregions that act as refuges for species found in the Atlantic Forest biome (Lima 1982, Barbosa et al. 2004). The roadkill rates recorded in C. thous may also be related to the availability of feeding resources. In more arid areas, where food tends to be scarcer, the foxes are forced to range more widely in search of scant resources. Areas with a mean rainfall of between 600 mm and 700 mm also tend to have a reduced abundance of feeding resources, although, as this precipitation level is typical of the Caatinga biome, to which C. thous is well adapted, the foxes tend to be more active in these areas.
Overall, roadkill events were relatively infrequent in urban areas and within the areas with intense anthropogenic impacts (quarries, wind farms). Otherwise, C. thous appears to be well adapted to modified landscapes, such as pastures, sugarcane plantations, other crops, and suburban areas (Dotta and Verdade, 2007; Lemos et al. 2011). In general, the distribution of the roadkill events documented in the present study study is consistent with the pattern observed in previous studies, in particular, the ecological characteristics of the habitats most occupied by the species.
The crab-eating fox prefers edge habitats and open areas over dense forest, and is often observed in pastures and moving along the margins of highways (Dotta and Verdade, 2007; Lemos et al. 2011). This is consistent with the high frequency of roadkill in areas where the vegetation is sparse or shrubby. The preference of C. thous for edge habitats brings these animals into potential contact with traffic more frequently.
Most of the area traversed by the study highways are anthropogenic landscapes, with a predominance of pasture and some residencial infrastructure, characteristics that are highly similar to the landscape surveyed by Guimarães (2017), where the areas adjacent to the study highways included farmland and pasture, as well as urban infrastructure. This author noted that the intensity of land use did not decline with increasing distance from the margin of the highways.
Foraging behavior is one other aspect of the ecology of C. thous that should be considered here. Most of the roadkill locations coincide with areas of low rainfall rates, which are associated with typical Caatinga vegetation (Drumond et al. 2000). These areas tend to have a reduced availability of feeding resources, especially during the dry season, when the foxes are obliged to range more widely in search of food. The composition of the diet of these carnivores may also shift in intensively impacted environments, such as the most anthropogenic environments, where the animals are obliged to consume the resources available in the local area (Castro 2017). As the supply of food is greatly reduced, the foxes tend to become more cosmopolitan, ranging more widely, and thus being more likely to come into contact with the margins of the highways, in particular given that roadkill carrion may become an increasingly attractive resource (Gatti et al. 2006).
Reis et al. (2011) identified C. thous as an opportunist, which adjusts its diet according to the seasonal conditions, feeding more on fruit during the rainy season, and tending to prey on small animals more frequently during the dry season. Qualitative observations during the present study indicate that roadkill events were more frequent during the dry season, and that the C. thous specimens were often encountered in the vicinity of other carcasses, which is consistent with the data of Bueno and Motta-Junior (2004), who concluded that this carrion is part of the diet of this fox.
4.2 Roadkill Hotspots
Highways are the principal links between rural municipalities in Brazil, and are one of the most important mechanisms that facilitate the colonization of the interior of the continent (Barat 1969; Bertussi and Ellery Junior 2012). In general, highway networks tend to have a major impact on the biodiversity of the areas they traverse, primarily due to the fragmentation of habitats (Bager et al. 2007). One of the principal objectives of road ecology is to determine spatial and temporal patterns in roadkill rates, in an attempt to identify potential measures to mitigate the negative effects of the presence of highways (Bager and Fontoura 2012).
To this end, it is essential to verify the areas with the greatest concentration of roadkill incidents for the target species. Roadkill hotspots were identified in the present study, in particular on the highway that links Garanhuns to Palmeira dos Índios. As a monogamic species, C. thous is typically found in pairs or small packs (MacDonald and Courtenay 1996), but if one member of the pair dies, the survivor will often seek a new partner in a distant area. The breeding pairs share a home range, which may often overlap with those of its neighbors (MacDonald and Courtenay 1996). The behavioral characteristics of the species thus do not appear to influence the spatial concentration of roadkill, although the location of the highways in relation to the dispersal and foraging routes adopted by the foxes may lead to a greater potential vulnerability to roadkill in certain specific areas.
In the specific case of the Garanhuns–Palmeira dos Índios route, the hotspot is located within an area of low-lying vegetation and open fields, surrounded by denser vegetation. In this context, as C. thous prefers edge habitats and more open vegetation (Lemos et al. 2011), the roadkill hotspot may coincide with dispersal routes, which the foxes use to avoid areas of denser vegetation. This is supported by the plots (Figs. 12 and 13), which show the predominant vegetation and rainfall rates of the study area. While C. thous tends to be resident within a relatively stable home range, it tends to share its range with other individuals, all of which will likely use the same dispersal route in search of similar resources. Assis (2014) concluded that each animal species will perceive the permeability of highways differently, according to its ecological relationship with the environment, leading to variation among the species in the exact location of their roadkill hotspots.
4.3 Sexing of the specimens
A total of 51 female and 50 male C. thous were collected in the present study, effectively, a 1:1 sex ratio, which is typical of this species (Montgomery and Lubin 1978). Castro (2017) recorded a slight male bias (63 males and 59 females) in the roadkilled C. thous in central and southeastern Brazil. These findings are consistent with the known behavioral characteristics of the species, which is monogamic and tends to form pairs, even though foraging is typically an individual behavior (de Mello Beisiegel et al. 2013). While C. thous does not appear to hunt cooperatively, males and females may forage together (Montgomery and Lubin 1978), which may account for the balanced sex ratio observed in the present study. Overall, then, as the males and females have similar foraging and ranging habits, they presumably run a similar risk of being the victim of roadkill.
4.4 Genetic Characteristics
The genetic diversity of C. thous in the study area was low. The He was 0.35 for the study area as a whole. Castro (2017) recorded a He value of 0.686 in the São Paulo, Mato Grosso do Sul, and Goiás, in central and southeastern Brazil. Tchaicka (2006) estimated a mean He of 0.64 for the C. thous populations found in Brazil, which further reinforces the low value recorded in the present study, which indicates that only a single population exists in the study area, in the southern backlands (Agreste) of Pernambuco and the backlands of Alagoas.
The Bayesian analyses indicated that the population sampled in the present study had a likely maximum (K) of two genetic clusters. Castro (2017) found a maximum probability of the presence of three genetic clusters in the populations analyzed in this study. The present study is thus consistent with that of Castro (2017), in that no differences were found in the structure of the possible populations sampled. Here, all the individuals from all four study highways present both genetic clusters identified using the method of Evanno et al. (2005).
However, while Castro (2017) identified three distinct populations, one from the state of São Paulo, the second from Mato Grosso do Sul, and the third from the state of Goiás, only one population was identified in the present study. Analyzing microsatellites of specimens representing almost the whole of Brazil, Tchaicka (2006) found no clear differentiation of the populations, except in the case of Northeastern Brazil, which was distinct from the rest of the Brazilian populations.
The lack of differentiation of the individuals collected in the present study may be related in part to the inclusion of non-resident animals in the specimens sampled (Castro 2017). Roadkill events may be more likely in certain areas that coincide with dispersal routes, which may lead the individuals to share and exchange their genetic characteristics more widely in these environments.
The home range of a male C. thous may cover an area of as much as 11.5 km2, while females tend to occupy ranges half this size. The area occupied by a crab-eating fox may also vary seasonally, according to the availability of feeding resources (Reis et al. 2011). These ample home ranges, together with the feeding ecology of the species, and the extensive overlap with the ranges of different individuals, may bring individuals from different areas together. Based on the trapping of individuals, Faria-Corrêa (2009) calculated a population density of 0.78 ind/km2 in Itapuã State Park in the Brazilian state of Rio Grande do Sul, while Desbiez et al. (2010) recorded a density of 0.55 ind/km2 in areas of the Brazilian Cerrado savanna. No density estimates are yet available for Caatinga populations of C. thous, but the available data on population density and home range sizes implies the frequent contact among individuals from different social nuclei, increasing the potential for gene flow within a single general population. Overall, then, the low genetic diversity of C. thous in the study area, together with the ranging and foraging habits of the species, appears to indicate the existence of a metapopulation in the present study area (Hanski and Simberloff 1997; Hanski 1998).
The low levels of genetic differentiation and variation, together with the Fst indices, which reinforces the conclusion that there is no isolation among the groups identified in the study. Overall, this evidence indicates that the population is organized in groups of individuals that tend to avoid urban areas and other environments that may threaten their survival.
One other important factor in this scenario may be the influence of anthropogenic pressures, which may have determined the recent fragmentation of the populations, even though this was not perceptible in the indices of genetic differentiation. Anthropogenic environments, such as urban areas may have irreparable effects on well-structured populations, reducing their genetic diversity, due to the impacts of the environment on the species (Castro 2017). While the species may be tolerant of anthropogenic environments, and be able to survive in these areas, it may depend on the availability of more permeable habitats, such as pasture and farmland (Ferraz et al. 2010). In the study region, the rapid, ongoing expansion of human populations, and, in particular, the growth of urban environments may further reduce the local C. thous population, with an associated reduction in the genetic variability of the species.