Conceived and designed the experiments: RW KCP LM JAKM. Performed the experiments: RW KCP PAC EJD. Analyzed the data: RW. Contributed reagents/materials/analysis tools: PAC EJD. Wrote the paper: RW KCP PAC EJD JAKM.
This work was partly funded by conservation funds associated with commercial operations (Denver Zoo, Columbus Zoo, Philadelphia Zoo). This does not alter the authors' adherence to all the PLoS ONE policies on sharing data and materials.
Infectious diseases have contributed to the decline and local extinction of several wildlife species, including African wild dogs (
We investigated interspecific and intraspecific transmission routes, by comparing African wild dogs' exposure to six canine pathogens with behavioural measures of their opportunities for contact with domestic dogs and with other wild dogs. Domestic dog contact was associated with exposure to canine parvovirus,
These findings, combined with other data, suggest that management directed at domestic dogs might help to protect wild dog populations from rabies virus, but not from canine distemper virus. However, further analyses are needed to determine the management approaches – including no intervention – which are most appropriate for each pathogen.
Infectious diseases have contributed to the decline and local extinction of several wildlife species
Serious impacts on host populations most frequently occur when generalist pathogens are maintained within populations of abundant “reservoir” hosts and “spill-over” into less abundant, and potentially more susceptible, host species
Population crashes and local extinctions of endangered African wild dogs (
The dynamics of infectious diseases are also likely to be strongly influenced by wild dogs' behaviour and ecology. Wild dogs occur at low population densities, living in highly cohesive territorial social groups (packs)
Here, we assess the importance of within- and between-species transmission of pathogens in African wild dogs coexisting with domestic dogs. We describe patterns of exposure to six canine pathogens with diverse transmission mechanisms: rabies virus and canine distemper virus, which are transmitted by direct contact
Animals were captured and handled in collaboration with the Kenya Wildlife Service, with permission from the Kenyan Ministry of Science and Technology (permit MOEST 13/001/32C 47) as well as private and community landholders, according to guidelines of the IUCN/SSC Canid Specialist Group, and following a protocol approved by the Animal Care and Use Committee of the University of California, Davis (protocol 10813), and the Ethics Committee of the Zoological Society of London (project BPE/0510).
The study was conducted in 2001–9 in northern Kenya, in Laikipia District (37° 2′ E, 0° 6′ N, 1800m ASL), and parts of neighbouring Samburu and Isiolo Districts. The core study area is 4,500 km2 of semi-arid bush land (mean annual rainfall 590mm), used for subsistence pastoralism, commercial ranching, and tourism.
Wild dogs disappeared from the study area in the 1980s, but recolonised naturally in the late 1990s
The study area comprises two main land use types. Privately-owned commercial ranches form a contiguous block in the south-west of the study area, with the remainder being community lands occupied by Samburu and Masai pastoralists. Human and livestock densities are substantially higher on community lands than on commercial ranches, with densities of wild dog prey correspondingly lower
In 2001–9, 90 wild dogs in 19 packs were immobilized to fit radio-collars for monitoring purposes. Capture methods are detailed in ref
While wild dogs were immobilized, blood samples were collected from the jugular vein, into 10ml evacuated serum separator tubes (Vacutainers, Becton-Dickinson, Oxford, UK). Blood samples were allowed to clot before being centrifuged; serum was then removed and aliquots were stored at −20°C.
Wild dogs fitted with radio-collars were monitored using aerial and ground-based telemetry. Aerial telemetry was conducted approximately weekly, usually between 0700–0830h (during wild dogs' morning hunting period), and provided locations with an accuracy of around 200m. Packs including radio-collared animals were visited regularly on the ground to monitor pack size, membership, and reproductive state.
Of 90 wild dogs captured, 33 were of known age, having been previously identified (from their unique pelage patterns) as pups (<1 year) with birth dates known from regular pack monitoring with a precision of a few days. The ages of a further 12 wild dogs could be confidently estimated, having been first identified as pups or yearlings (≥1 year, <2 years, recognisable during handling based on body dimensions and tooth wear), with birth dates estimated as 12 months prior to their pack's next recorded breeding attempt (since wild dogs breed approximately annually). The ages of the remaining 45 wild dogs, first identified as adults (≥2 years), were estimated using a combination of tooth wear, pelage characteristics, reproductive state and social status.
None of the wild dogs in this study had been vaccinated against any pathogen.
During 2001–9, blood samples were collected from 184 domestic dogs which owners reported had no history of vaccination against any pathogen. Of these domestic dogs, 121 were sampled at 75 bomas throughout the study area, and 63 were sampled at 14 locations on community lands within the study area where free rabies vaccination was being provided annually. Domestic dogs were manually restrained and blood was collected from the cephalic vein into 10ml evacuated tubes (Vacutainers, Becton-Dickinson, Oxford, UK). Blood samples were allowed to clot before being centrifuged; serum was then removed and aliquots were stored at −20°C.
Data on the densities and distribution of domestic dogs were available from a survey of 639 bomas, described in ref
For all serological analyses, threshold titres interpreted as evidence of prior pathogen exposure were selected to optimise test sensitivity and consistency with other studies. Additionally, the effects of choosing different threshold titres were explored in statistical analyses (see below).
Serum samples were screened for antibodies to rabies virus, using a rapid fluorescent focus inhibition test (RFFIT) at the Centers for Disease Control and Prevention, Atlanta, GA
Serum samples were screened for antibodies to canine distemper virus, using a serum neutralisation (SN) test and the Onderstepoort virus strain, at the Animal Health Diagnostic Center at Cornell University
Serum samples were screened for antibodies to canine parvovirus, using a haemagglutination inhibition (HAI) test, again at the Animal Health Diagnostic Center at Cornell University
Serum samples were screened for antibodies to
We investigated three potential measures of contact among wild dogs. First, we used pack size (measured as numbers of adults and yearlings, and numbers of pups, at the time of sampling) as an index of individual wild dogs' day-to-day probability of contact with conspecifics. Second, we used the date of sampling (measured in days since 1 Jan 2001) as a proxy to estimate contact probability, since wild dog population size, density, and home range overlap all increased over time
Since domestic dog densities are higher on community lands than on commercial ranches
In addition to this simple binary measure, we derived a continuous measure of domestic dog contact risk for each wild dog sampled. Full details of this method are given in ref
Primary analyses of serological data considered the proportions of wild dogs showing evidence of exposure to the particular pathogens, as indicated by having antibodies detectable at dilutions greater than the thresholds indicated above. These analyses were conducted using mixed logistic regression models, including pack identity as a random effect, using the
Continuous variables which were not normally distributed were log-transformed. To avoid problems associated with zero values, a number equivalent to half the lowest non-zero value was added to all values before taking the (natural) logarithms. Where animals had been sampled on more than one occasion, data from the most recent date were used in statistical analyses, to maximise available data on the animals' history prior to sampling. In addition, data on repeated sampling of the same animals were used to investigate seroconversion (by comparing data from the first and last sampling event).
Preliminary analyses used mixed logistic regression models (including pack identity as a random effect) to investigate, separately, the potential effects on pathogen exposure of (i) individual characteristics (sex, age in months); (ii) risk of contacting other wild dogs (measured as pack size, sampling date {a proxy for overall population density}, and mean risk of contacting another wild dog pack in the previous 12 months); and (iii) risk of domestic dog contact (measured as land use type {commercial ranch/community land}, and as domestic dog density experienced in the previous 12 months). For
Exploratory analyses were also conducted to investigate the effects of using alternative threshold values for considering animals exposed (using mixed logistic regression), and, alternatively, using the (log-transformed) antibody titres as continuous outcome variables (using generalised linear mixed models {GLMM}, including pack identity as a random effect, using the
The three measures of intraspecific contact among wild dogs were correlated with one another. Pack size increased over time, whether measured as the number of adults and yearlings (GLMM, effect of days since 1 Jan 2001, p = 0.002), or as the number of pups (GLMM, effect of days since 1 Jan 2001, p = 0.006). The estimated risk of contacting another wild dog pack likewise increased over time (GLMM, effect of days since 1 Jan 2001, p<0.001). However, after adjusting for the effects of time, the estimated risk of contacting another wild dog pack was not significantly related to pack size, whether measured as the number of adults and yearlings (GLMM including days since 1 Jan 2001, effect of adult and yearling number p = 0.16) or as the number of pups (p = 0.11).
There was likewise evidence of correlation between the two measures of wild dogs' risk of contact with domestic dogs. The domestic dog density experienced in the 12 months prior to sampling was higher for 20 wild dogs living mainly on community lands, than for 37 living mainly on commercial ranches (
Data show the mean (and SD) estimated density of domestic dogs at points where wild dogs were located by aerial radio-telemetry in the 12 months prior to sampling for pathogen exposure. Figures along the top of the graphs indicate the numbers of wild dogs sampled from each pack (over periods of 1–7 years).
Error bars indicate exact binomial confidence intervals. Figures along the tops of the graphs indicate the numbers of samples screened in each year; the sums of these figures exceed the denominators in
Maps show, for each pathogen, the sampling locations for animals with (filled symbols) and without (open symbols) evidence of exposure. Shading indicates commercial ranch land. Note that multiple animals were sampled at some locations.
Pathogen | Wild dog | Domestic dog | ||
|
|
|
|
|
rabies virus | 13/88 | 15% | 24/82 | 29% |
canine distemper virus | 14/88 | 16% | 88/184 | 48% |
canine parvovirus | 22/89 | 25% | 117/183 | 64% |
canine coronavirus | 21/83 | 25% | 28/184 | 15% |
|
70/88 | 80% | 56/65 | 86% |
|
45/87 | 52% | 12/65 | 18% |
None of these animals had any history of vaccination to any pathogen.
unchanged | seroconverted | ||||
Pathogen |
|
|
|
|
total |
rabies virus | 11 | 2 | 0 | 4 | 17 |
canine distemper virus | 12 | 2 | 1 | 4 | 19 |
canine parvovirus | 13 | 0 | 4 | 2 | 19 |
canine coronavirus | 4 | 4 | 2 | 3 | 13 |
|
0 | 13 | 5 | 1 | 19 |
|
8 | 8 | 3 | 0 | 19 |
Animals which seroconverted were considered negative when first sampled, but positive subsequently, or
There was a non-significant trend suggesting that the proportion of wild dogs exposed to rabies virus may have been higher among wild dogs with greater opportunities for contact with domestic dogs (
Pathogen | Risk factor | OR (95% CI) | P |
rabies virus | domestic dog contact | 1.95 (0.91–4.18) | 0.088 |
canine distemper virus | year 2003 |
32.0 (2.8–360.1) | <0.001 |
canine parvovirus | adult pack size | 0.23 (0.08–0.67) | 0.007 |
domestic dog contact | 8.66 (1.59–219.02) | 0.020 | |
canine coronavirus | age (in months) | 1.07 (1.02–1.13) | 0.009 |
days since 1 Jan 2001 | 0.997 (0.995–1.000) | 0.036 | |
|
adult pack size | 0.77 (0.63–0.95) | 0.016 |
domestic dog contact | 3.43 (1.01–11.70) | 0.049 | |
|
adult pack size | 0.27(0.09–0.81) | 0.019 |
domestic dog contact | 8.53 (3.40–67.7) | 0.015 |
All models also include pack identity as a random effect. For risk factors measured as continuous variables, odds ratios describe the effects of a doubling in value.
The proportion of wild dogs exposed to canine distemper virus declined over time (mixed logistic regression model, effect of days since 1 Jan 2001, OR = 0.999, CI = 0.998–1.000, p = 0.036). However, the temporal pattern suggested that prevalence peaked in 2003 (
The proportion of wild dogs with evidence of exposure to parvovirus was greater in small packs, with greater opportunities for contact with domestic dogs (
The proportion of wild dogs with evidence of exposure to canine coronavirus was higher in older animals, and also declined over time (
The proportion of wild dogs with evidence of exposure to
The proportion of wild dogs with evidence of exposure to
Our findings are consistent with the hypothesis that domestic dogs transmit canine pathogens to wild dogs. Wild dogs with greater opportunities for contact with domestic dogs were at greater risk of exposure to canine parvovirus,
We found a clearer link between pathogen exposure and domestic dog density within this population than in a parallel study which compared exposure across populations, using protected area status as a proxy for domestic dog contact
The lack of any association between wild dogs' exposure to canine distemper virus and their contact with domestic dogs is consistent with recent evidence suggesting that wildlife may play an important role in maintaining this pathogen, both in this study area and elsewhere
Our findings revealed no evidence that wild dogs' contact with other members of the same species increased their exposure to canine pathogens. Contrary to expectation, exposure to parvovirus,
Our finding that wild dogs' exposure to several pathogens was elevated by contact with domestic dogs is consistent with, though not sufficient to confirm, the hypothesis that domestic dogs function as reservoir hosts for these pathogens. Empirical data on both contact rates
Since the pathogens associated with domestic dog contact have all been linked to wild dog mortality, the viability of wild dog populations might be undermined by contact with domestic dogs. However, exposure to these pathogens – and by extension contact with domestic dogs – is not necessarily harmful in all cases. Indeed, the fact that we detected evidence of exposure to these pathogens in apparently healthy wild dogs, in a growing population, shows that exposure is not invariably fatal, and did not cause population decline at the level of domestic dog contact observed here. Exposure to these pathogens may maintain some level of immunity in the population, helping to prevent large outbreaks from occurring or mitigating their lethality, and may also help to maintain selection pressure for disease resistance. The balance between these positive and negative effects is likely to vary between pathogens, suggesting that a single management strategy (e.g., domestic dog removal) might not be appropriate for all pathogens. However, substantial increases in contact between wild dogs and domestic dogs – as might occur through land use change
Our findings help to identify the most appropriate pathogen-specific control measures. Among the pathogens we studied, rabies virus has the clearest record of causing mortality in wild dogs
This study is one of a very small number of empirical attempts to quantify rates of contact between wild and domestic mammals
We thank Kayna Chapman, Stephen Chege, Andrew Francombe, Evans Lemusana, Peter Lindsey, Symon ole Ranah, Stephanie Romañach and Andrew Stein for assistance in the field, Charles Rupprecht and his staff at the Centers for Disease Control for conducting rabies screening, Christl Donnelly and Nathalie Pettorelli for statistical advice, and two anonymous reviewers whose comments improved the manuscript.