Frequent contact with domestic dogs increases the exposure and disease risk for wild carnivores [8, 36]. Therefore, local pet management practices allowing dog’s predominantly free-roaming habits, poor veterinary assistance, along with recent dog death and disease reports and the low multiple vaccination coverage detected per se place the wild carnivores at the study sites in a potentially dangerous scenario of disease spill over (or spillback) from dogs.
Exposure patterns and antibody titre profile
Exposure to most pathogens tested is widespread throughout the study sites, and prevalence is widespread and moderate to high, particularly for CPV. It must be noted that CDV and CPV have higher fatality rates than do CAV and CCV milder infections [31, 32, 37]. Fatality rate may impact prevalence by removing exposed individuals from the population and this can result in the low prevalence observed here, for instance, for CDV. Also, prevalence of CDV might be low in these areas because a ‘wave of infection’ could have passed through the area in the past, and recent circulation of this virus may not have occurred regularly .
Regarding antibody titre frequency and duration of immunity, mostly high levels of antibodies against CPV and CAV were found. The duration of antibodies to the viral agents studied here is longer than two years , and such titres may indicate that the exposure to these agents is mostly recent. However, as said before, it is not possible to accurately determine the timing of exposure from antibody titres, and our cross-sectional serological approach does not permit deeper inferences on temporal-spatial dynamics of exposure neither detects pathogen introduction or seroconversion events.
Prevalence comparisons between small and large protected areas
Proportionally more CDV and CPV-exposed dogs were found in small areas. Perhaps the reduced perimeter of these areas allows less space between properties and households, and ensures higher host densities, contact rates and exposure (including environmental) to these agents. Therefore, small areas should be prioritized in health improvement efforts for rural dogs aiming also to prevent wildlife disease-associated mortality in the Atlantic Forest, which is currently mostly composed by relatively small remnants .
Risk factor modelling
Of fifteen initially assessed variables, thirteen entered multivariate tests, and eleven remained as significant associations with exposure for at least one pathogen in the best models: sex, age, body score, mobility, veterinary assistance, number of people per household, interaction with wildlife, recent disease in dogs, area size class and access to villages and forests. However, one must account for associated bias due to issues with missing controls in an observational prevalence study when interpreting data for risk factor detection, and there might be biologically relevant putative causalities out of the full set of factors presented. Moreover, the statistical associations found do not necessarily imply direct causation.
High levels of circulation indicated by the high prevalence of exposure indicate that CPV is perhaps one of the most dangerous agents in this scenario. It is one of the most commonly reported canine viral agents worldwide [3, 31], including in South American wild canids (e.g.) [16, 18, 19]. High prevalence was concomitantly observed in wild and domestic canids from two protected areas [18, 19] in the same state of the present study, although in the Cerrado Biome. CPV infection has proven capable to cause serious population impacts (mostly through pup mortality) in wild carnivores, for instance, grey wolves (Canis lupus) . However, our cross-sectional approach does not distinguish epidemic from endemic states. For instance, a CDV epidemic that might blow through these mostly susceptible dog populations might also cause high dog mortality and interspecific transmission that can be dangerous for wild carnivores as well.
The modelling shows that unassisted dogs that roam freely have more chances of exposure for CPV. This is explained by higher contact rates with diseased animals and enhanced environmental exposure due to the great environmental resistance of this virus , therefore dog restriction must be enforced around protected areas. Dogs that never received veterinary care were also more prone to exposure, and this emphasizes the role of veterinarians in health promotion and maintenance at human/wildlife interfaces. Other factors appearing in candidate models (recent dog disease and area size class) indicate that CPV may be causing significant morbidity in the study sites, and that small reserves require priority for the control of the disease.
Canine distemper is a systemic highly fatal disease, representing a major conservation concern around the globe [3, 32]. Evidence of infection in dogs is widespread in and around South American protected areas [18, 20, 22, 23]. Antibodies against distemper were already found in Brazilian wild felids and canids (e.g.)[17, 19], and there are reports of distemper-induced mortality in two Brazilian fox species, the crab-eating fox C. thous and the hoary fox Lycalopex vetulus [14, 15]. In Chile, domestic dogs have proven blamed for the transmission of CDV to wild canids . However, CDV transmission may be, in some cases, predominantly maintained by wild reservoirs [4, 32, 36].
In our rural settings, age, body score, number of people in the household and access to forests were mostly associated with CDV prevalence. However, only age and the number of people appeared in all candidate models. Older animals might have had more chances of exposure events throughout their lives. However, since CDV is a highly lethal disease , the low prevalence and the age effect observed suggest that CDV seropositive animals may be survivors of a past wave of infection in some of the study areas. The number of people cohabiting households was positively associated with CDV seropositivity. This raises the possibility of fomite transmission, mentioned in the literature as a mode of spread of this highly infectious virus which is mainly transmitted by contact or aerosols . Male dogs may be at higher risk of exposure, probably because of their more aggressive and roaming behaviours . Higher body scores may also favour dog free-roaming behaviours and ensuring both exposure and antibody response to the virus, which, according to the presence of the access to forests among risk factors, might be occurring when dogs walk inside forested areas. As said before, CDV can be maintained in wild reservoir species alone [4, 36].
Mostly unstudied, CAV is also capable to cause damage to wildlife populations , although its impact is still unknown. This directly transmitted virus may cause severe respiratory disease being of concern for domestic dog health, and evidence of exposure was found in many wild species  including in South American wild carnivores and sympatric dogs from Bolivia [16, 23] and Brazil [18, 19].
For CAV, female sex, age, number of people and access to villages entered all selected models. Females are more exposed to CAV, and this is probably related to behavioural differences, as females tend to display less roaming and aggressive behaviours , and thus, spend relatively more time around their homes in rural and urban settings, leading to host aggregation and enhancement of the density-dependent transmission of CAV. The explanation for increasing prevalence with age is that older dogs probably had moved more through time and faced more opportunities for contact and exposure to CAV, and this is expected since older animals tend to have more opportunities of infection when disease-induced mortality is low or infections are often fatal . The association with the number of people in the households reinforces the questioning about the fomite spreading of canine viruses in rural environments. However, this is not mentioned as a CAV transmission mode . The access to villages was a strongly associated factor, appearing in all candidate models. Urban settings hold higher-density dog populations , what enhances the maintenance and transmission of the density-dependent CAV . Therefore, such freedom of rural dog movement must be prevented in order to reduce exposure and infection by CAV in near urban areas and subsequent spillover to wild species. Recently, data from the same state showed that free-roaming behaviours (e.g. hunting) are among risk factors for canine neosporosis . Additionally, recent mortality in dogs may be associated with CAV infection in the study sites. Dogs with higher body scores may have survived past infection waves and sustained the mostly low antibody levels found. Interactions with wildlife may also be increasing the risk for CAV. However we cannot infer whether the main reservoirs are domestic or wild animals. Nevertheless, the free-ranging behaviour of dogs must be inhibited through fence or leash restraining in order to reduce contact rates and opportunities for general pathogen exposure and transmission in rural/wild interface areas.
Although we did not test the samples against rabies, our survey revealed that despite the apparently good previous vaccination coverage (more than 80 %), several owners reported the total absence or the periodic lack of visits of health agencies promoting vaccination against rabies in their households in some years. Canine-mediated rabies is a multi-species highly fatal disease, representing a major problem for carnivore conservation particularly in Africa . In Brazil there are domestic animal rabies control programmes through vaccination since the 1980’s, and reports of wildlife mortality have been attributed to the disease. Additionally, serological evidence of exposure was already found in Brazilian carnivore species . Therefore, more attention should be given to rabies in wildlife-domestic animal interfaces, and the vaccination programme should be reinforced so as to continuously warrant good coverage in these areas.
In our scenario, dogs with a history of vaccination against agents other than rabies, which are not cost-free, may be an indicator of increased owner care. Better care for the dog would result in better physical and immunologic condition, and better supportive care and veterinary care in times of illness. Those healthier animals will be more likely to survive an illness, and a higher survival rate would lead to a greater proportion of seropositive dogs. Unfortunately, the mostly unvaccinated dogs in these populations may be less likely to receive a high level of owner care. Lack of owner care may result in lower probability of supportive or veterinary care when ill, and an overall decrease in animal health. Poor health and lack of veterinary care would result in a higher death rate in the result of an illness. A higher death rate would remove exposed animals from the population. Therefore, unvaccinated dogs may have less supportive care from owners, which would lead to higher death rates in the face of illness, which would lead to artificial decline in the measured population proportion of seropositives, as may be the case for CDV in this study. However, our data set does not permit such distinction, and we acknowledge the uncertainty about these relationships.
Commercial multiple vaccines against CDV, CPV, CAV, CCV and other canine pathogens are available in the region. However, owner unawareness added to the relatively prohibitive costs of vaccines, and the fact that such intervention may seem directed solely to protect the health of dogs make multiple-agent vaccination of dogs a lesser priority for the mostly low-income rural families, as shown by the low percentage of vaccinated dogs in this study. We are unaware of the use of other possible cross-reactive vaccines, such as those using adenovirus as a vector, in the area. This means that multiple vaccination have to be reinforced, even though with a more flexible and viable interval in these areas , if the aim is to induce protective herd immunity against other dangerous pathogens in dogs from wildlife-rich areas. Vaccination schemes for dogs around protected areas, directing to protect wild carnivore and human welfare has proven successful , notably with concomitant low coverage vaccination of wildlife species . In our case, the dog population living in proximity of protected area borders should be targeted in comprehensive multiple-agent vaccination schemes. Vaccination of dogs can improve the collective immune status or herd immunity necessary to avoid or decrease the transmission of some agents. However, as pointed out by some authors, vaccination of naturally exposed populations may not be of great value for highly prevalent pathogens , and this may be the case for CPV in our study area. Nonetheless, the transmission of lower local prevalence pathogens such as CDV and CAV may be successfully limited by dog vaccination in this case. Decreasing dog numbers is also necessary to keep the population below transmission thresholds of most directly transmitted canine pathogens, and this can be achieved through sterilization and education campaigns. According to our results, prohibiting dogs from moving across the interface and spreading or acquiring pathogens is essential for disease prevention, and responsible ownership reinforcement alongside with legal penalties for irresponsible owners is highly recommended. Health monitoring should, afterwards, be continuously performed in both domestic and wildlife species, to assess the efficacy of the proposed measures.