Low-coverage vaccination strategies for the
conservation of endangered species
D. T. Haydon1, D. A. Randall2,3, L. Matthews4, D. L. Knobel3,4, L. A. Tallents2,3, M. B. Gravenor5, S. D. Williams2,3,
J. P. Pollinger6, S. Cleaveland4, M. E. J. Woolhouse7, C. Sillero-Zubiri2,3, J. Marino2,3, D. W. Macdonald2
& M. K. Laurenson3,4,8
The conventional objective of vaccination programmes is to
eliminate infection by reducing the reproduction number of an
infectious agent to less than one1, which generally requires
vaccination of the majority of individuals. In populations of
endangered wildlife, the intervention required to deliver such
coverage can be undesirable and impractical2; however, endan-
gered populations are increasingly threatened by outbreaks of
infectious disease for which effective vaccines exist3,4. As an
alternative, wildlife epidemiologists could adopt a vaccination
strategy that protects a population from the consequences of only
the largest outbreaks of disease. Here we provide a successful
example of this strategy in the Ethiopian wolf, the world’s rarest
canid5, which persists in small subpopulations threatened by
repeated outbreaks of rabies introduced by domestic dogs6. On
controls the spread of disease through habitat corridors between
subpopulations and that requires only low vaccination coverage.
This approach reduces the extent of rabies outbreaks and should
significantly enhance the long-term persistence of the population.
the reproduction number of an infectious agent to less than one1.
The use of safe and effective vaccination can have a vital role in
managing infectious disease in wildlife populations. For logistic,
economic and ethical reasons, however, it will always be desirable to
minimize the number of animals to be vaccinated. We distinguish
between two different uses of vaccination: one focused on eliminat-
ing disease from a population, and another focused on protecting an
endangered population from extinction. The design of vaccination
programmes to eliminate infectious disease from populations has
received much attention1,7,8, and usually requires vaccinating a
proportion of the population upward of 1–1/R0, where R0is the
reproduction number of the infectious agent1. A conceptually dis-
tinct approach is to assume that wildlife populations can tolerate
limited outbreaks of disease, but their viability is threatened by large
outbreaks that could reduce their size to below a minimally viable
threshold9,10. Targeted vaccination could be then used to curtail the
largestandmostdamaging outbreaks,whilereducing theproportion
of individuals required to be vaccinated. Here we examine the
effectiveness of such a strategy in conserving populations of Ethio-
viral disease of mammals.
Ethiopian wolves in the Bale Mountains persist in several sub-
populations connected by narrow corridors of habitat11(Fig. 1).
Within these subpopulations, two large outbreaks of rabies were
detected in 1992 and 2003 (ref. 12), and rabies was the suspected
cause of a population crash in 1991 (refs 12–14); canine distemper
was also suspected to have infected wolves in 1993 (ref. 15). These
repeated introductions of infection into the wolf population from a
domestic dog reservoir6, together with permission to mount a
reactive vaccination campaign in response to one such outbreak,
control strategy in an endangered species. Here we briefly review the
Figure 1 | Known distribution of Ethiopian wolf packs in three
subpopulations in the Bale Mountains. Shown is the configuration of
territories in the Web Valley, Sanetti Plateau and Morebawa, including the
linking the Web Valley with Morebawa. Packs are present in Central Peaks,
Raffu, Chafa Delacha and Tullu Deemtu areas, but territory boundaries are
not known with any precision. Filled red circles indicate the location of
of carcass recovery. Filled polygons indicate vaccinated packs. Inset shows
the current distribution of the species throughout Ethiopia.
1Division of Environmental and Evolutionary Biology, University of Glasgow, Glasgow G12 8QQ, UK.2Wildlife Conservation Research Unit, University of Oxford, Tubney House,
Oxford OX13 5QL, UK.3Ethiopian Wolf Conservation Programme, PO Box 215, Robe, Bale, Ethiopia.4Wildlife and Emerging Diseases Section, Royal (Dick) School of Veterinary
Studies, University of Edinburgh, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG, UK.5Institute of Life Science, School of Medicine, Swansea University, Singleton
Park, Swansea, SA2 8PP, UK.6Conservation Genetics Resource Center, University of California, 621 Charles E. Young Drive South, Los Angeles, California 90095, USA.7Centre
for Infectious Diseases, University of Edinburgh, Ashworth Laboratories, Kings Buildings, West Mains Road, Edinburgh EH9 3JF, UK.8Frankfurt Zoological Society, PO Box 14935,
Vol 443|12 October 2006|doi:10.1038/nature05177
© 2006 Nature Publishing Group
parameterize an epidemiological model that predicts the potential
course of the outbreak without vaccination. Finally, we extend the
model to explore the increase in population persistence times
resulting from vaccinating in subpopulations as compared with
vaccinating only in packs occupying corridor habitats between
subpopulations. Additional data and analyses are given in the
In August 2003, the Bale wolf population comprised 200–250
individuals (aged .1yr) in 36 packs associated with known territory
locations. In the Web Valley, roughly 95 wolves resided in ten packs.
Between August 2003 and January 2004, an outbreak of rabies
resulted in the deaths of 72 (76%) of these wolves12. Disease spread
recovered from all territories in the Web Valley, suggesting a carcass
detection probability of ,50% (38/72; Fig. 1). A parenteral reactive
vaccination programme implemented in November was designed to
prevent transmission between subpopulations, while limiting the
number of wolves handled. The campaign initially targeted packs
immediately beyond the corridor connecting the Web Valley and
Morebawa subpopulations (Genale pack, Fig. 1) and then gradually
moved to packs further away from the disease front (Fig. 1). Final
coverage averaged over the whole Morebawa subpopulation was
37.5%. Subsequent monitoring suggested a maximum of seven
deaths out of 105 wolves in the Morebawa subpopulation, all from
a single front-line territory (Weshema pack; Fig. 1). Although
carcasses continued to be recovered in the Web Valley, the epidemic
made no further incursion into the vaccination zone.
To determine whether the vaccination programme itself limited
the severity of the outbreak we developed a spatially explicit demo-
graphically stochastic susceptible–exposed–infectious–removed
(SEIR) model9. The model was parameterized, taking into account
pack and population age structure and the final epidemic size in the
Web Valley (Supplementary Table S2 and Fig. S1). The maximum
likelihood estimate of R0depended on the assumed ratio of between
(bb) to within (bw) pack transmission (Supplementary Table S3).
across packs was most consistent with an intermediate to high ratio
of bb/bw(see Supplementary Information), and here we present
results assuming that bb¼ 0.1bw(further results and genetic data in
support of this choice are reported in the Supplementary Infor-
mation). The maximum likelihood estimate of R0for the 2003
outbreak was 2.4 (95% confidence intervals: 1.7–3.4; Supplementary
Table S3). A risk map constructed from predicted values for R0for
epidemics starting in different packs reflects the importance of pack
composition and configuration (Fig. 2). In Morebawa, pack size
predicted R0values were correspondingly lower.
By repeatedly simulating outbreaks in subpopulations structured
as in 2003, the model suggested that there was a 40% chance that
Figure 2 | R0map of the Web Valley and Morebawa subpopulations.
Mixing is assumed to be intermediate (between pack transmission is 10%
of within pack transmission). R0was predicted from the application of
per capita transmission rates estimated from the fit of the SEIR model to
data from the Web Valley, calculated by direct simulation assuming a single
index case arose within each pack.
Figure 3 | Model projections. a, Probability that a rabies outbreak starting
with a single index case in the central Web Valley goes on to cause a
subsequent outbreak in the subpopulation in Morebawa, assuming the
subpopulation to be fully susceptible and unvaccinated. Results are from an
intermediately mixed model (R0¼ 2.4) in which per capita transmission
rates between packs were 10% of within packs (thick red line). Thin red line
indicates the outcome of an equivalent simulation that includes the effect of
the vaccination programme as implemented during the outbreak. b, Effects
of reactive vaccination in the Web Valley subpopulation on total epidemic
size implemented after the deaths of different numbers of wolves.
Vaccination was assumed to be instantaneously protective when
administered to unexposed individuals (see Supplementary Information),
of catastrophicmetapopulation reductionoccurringovera20-yr periodas a
The model assumes three subpopulations, linked by habitat and migration,
that support a maximum of 100 individuals each. Black line, no control; red
lines, corridor vaccination (CV) that reduces the probability of an epidemic
in one subpopulation spreading to another from 0.25 to a range of possible
values rising in increments of 0.02 from 0.04 to 0.16; blue lines, reactive
details of the model).
NATURE|Vol 443|12 October 2006
© 2006 Nature Publishing Group
rabies epidemics arising from asingle indexcase would fade out with
less than ten (and usually less than four) individuals becoming
infected. Once they exceeded this number, however, the epidemic
would almost certainly go on to be large. Results from the model
indicated that the probability of infection passing through the
corridor into Morebawa increased from 0.08 with vaccination (as
implemented during 2003) to 0.25 in the absence of vaccination
(Supplementary Table S3). The upper ninety-fifth percentile interval
estimates for final outbreak size in the Morebawa subpopulation
started through this route increased from 8 wolves with vaccination
to 41 without (Fig. 3a). Given how close the estimated R0values are
to the unit threshold for packs in this subpopulation (Fig. 2), small
increases in wolf density could result in sharp rises in the probability
of even more damaging outbreaks.
Assuming a randomly located index case in the Web Valley, the
model predictedan average time delay between the firstdeath and an
individual becoming infected in the Genale pack in the corridor to
(21d) and 38d (10d), respectively. This suggests that even if carcass
detection rates fall as low as 20%, there would still be sufficient time
in which to implement a reactive corridor vaccination campaign
triggered by the detection of two carcasses.
By the time permission to vaccinate had been granted, the out-
break was considered too far advanced to protect the Web Valley
subpopulation. Modification of the model to include reactive vacci-
nation in an affected subpopulation triggered after the death of
different numbers of individuals suggested, however, that vacci-
nation would remain beneficial even after 10% of the subpopulation
had died from rabies (Fig. 3b).
These analyses show that the extent of rabies outbreaks can be
action mitigate the risk of extinction in the longer term? We used a
population viability analysis16(PVA) to demonstrate the consider-
able risk posed by rabies outbreaks to these subpopulations and the
substantial benefits of low-coverage reactive vaccination for the
persistence of the metapopulation as a whole. The PVA assumed
that habitat corridors between subpopulations act as conduits for
disease transmission (see Supplementary Information) but also
facilitate migration and recovery after epidemics17. The probability
of catastrophic reduction of the metapopulation (defined here as the
combined metapopulation falling below 20 individuals) at any time
over a 20-yr period fell in response to implementation of reactive
corridor vaccination, and even more quickly as reactive core vacci-
nation strategies were adopted that targeted 40, 30, 20 and as little as
10% of the affected subpopulation (Fig. 3c). For example, the PVA
predicted that, if rabies virus was introduced at an average rate of
once every 5yr into each subpopulation, corridor vaccinationwould
reduce the probability of catastrophic metapopulation reduction
almost fourfold from 0.38 to 0.10, and that reactive core vaccination
of only 10% of individuals would reduce this probability to ,0.001.
Aside from specific recommendations for this population (see
Supplementary Information), our analyses underline the general
importance of baseline ecological data, surveillance and detailed
quantitative contingency planning in the management of epi-
demics18–22. High-quality demographic data enable interventions to
be targeted, effectivemonitoring is essential for the earlydetection of
suspected disease outbreaks, and appropriately calibrated trigger
points minimize unnecessary interventions and facilitate the timing
a programme of minimally invasive but demographically significant
Although preventative vaccination of reservoir hosts can reduce
the frequency of stochastic spill-over infections into wildlife4, the
risk of outbreaks in unvaccinated wildlife populations cannot be
eliminated, particularly when limited resources restrict the extent
and coverage of reservoir vaccination programmes. Vaccination and
handling ofAfrican canidshas generated considerable controversy in
the past2,23, but our analysis provides strong evidence that targeted
low-coverage and less-invasive reactive vaccination strategies can be
effective in curtailing disease outbreaks and enhance the long-term
persistence of endangered populations. However, progress is
required in the development of protocols for more logistically
feasible and cost-effective vaccine delivery methods such as oral
vaccination24–27, and policy-makers and conservation practitioners
must be provided with epidemiologically sound, practical advice
with which to develop contingency plans. Greater knowledge of the
spatialecology andsocial organization ofotherendangeredspecies is
solutions to enduring threats to their persistence posed by infectious
in the Supplementary Information.
Models. We used a conventional SEIR model, which assumes demographically
and an average incubation period of 12d was fitted to match the duration of the
outbreak. The model was fitted to data from the Web Valley, and then used to
predict the probability of spread between subpopulations and the impact of such
outbreaks in these subpopulations. Further details of the epidemiological model
and its parameterization are supplied in the Supplementary Information.
The PVA model used demographic parameters from long-term monitoring
studies, and probabilities of between subpopulation spread of infection and
simulated distributions of outbreak sizes predicted by the model.Furtherdetails
of this model are supplied in the Supplementary Information, together with a
description of the genetic methods and analyses that indicated only limited
movement of infected individuals.
Received 14 June; accepted 15 August 2006.
1. Anderson, R. M. & May, R. M. Infectious Diseases of Humans, Dynamics and
Control (Oxford Univ. Press, Oxford, 1992).
Woodroffe, R. Assessing the risks of intervention: immobilization, radio-
collaring and vaccination of African wild dogs. Oryx 35, 234– -244 (2001).
Woodroffe, R., Cleaveland, S., Courtenay, O., Laurenson, K. & Artois, M. in
Biology and Conservation of Wild Canids (eds Macdonald, D. W. & Sillero-Zubiri,
C.) 123– -142 (Oxford University Press, Oxford, 2004).
Daszak, P., Cunningham, A. A. & Hyatt, A. D. Emerging infectious diseases of
wildlife: threats to biodiversity and human health. Science 287, 443– -449 (2000).
Marino, J. Threatened Ethiopian wolves persist in small isolated Afroalpine
enclaves. Oryx 37, 62– -71 (2003).
Randall, D. A. et al. An integrated disease management strategy for the control
of rabies in Ethiopian wolves. Biol. Conserv. 131, 151– -162 (2006).
Hethcote, H. W. An immunization model for a heterogeneous population.
Theor. Pop. Biol. 14, 338– -349 (1978).
Ball, F. G. & Lyne, O. D. Optimal vaccination policies for stochastic epidemics
among a population of households. Math. Biosci. 177– -78, 333– -354 (2002).
Haydon, D. T., Laurenson, M. K. & Sillero-Zubiri, C. Integrating epidemiology
into population viability analysis: managing the risk posed by rabies and canine
distemper to the Ethiopian wolf. Conserv. Biol. 16, 1372– -1385 (2002).
10. Vial, F., Cleaveland, S., Rasmussen, G. & Haydon, D. T. Development of
vaccination strategies for the management of rabies in African wild dogs. Biol.
Conserv. 131, 180– -192 (2006).
11. Sillero-Zubiri, C. & Marino, J. in Canids: Foxes, Wolves, Jackals and Dogs. Status
Survey and Conservation Action Plan (eds Sillero-Zubiri, C., Hoffmann, M. &
Macdonald, D. W.) 167– -173 (IUCN/SSC Canid Specialist Group, Gland,
Switzerland, and Cambridge, UK, 2004).
12. Randall, D. A. et al. Rabies in endangered Ethiopian wolves. Emerg. Inf. Dis. 10,
2214– -2217 (2004).
13. Sillero-Zubiri, C., King, A. A. & Macdonald, D. W. Rabies and mortality in
Ethiopian wolves (Canis simensis). J. Wildl. Dis. 32, 80– -86 (1996).
14. Marino, J., Sillero-Zubiri, C. & Macdonald, D. W. Trends, dynamics and
resilience of an Ethiopian wolf population. Anim. Conserv. 9, 49– -58 (2006).
15. Laurenson, K. et al. Disease as a threat to endangered species: Ethiopian
wolves, domestic dogs and canine pathogens. Anim. Conserv. 1, 273– -280
16. Morris, W. F. & Doak, D. F. Quantitative Conservation Biology: Theory and
Practice of Population Viability Analysis (Sinauer, Sunderland, MA, 2002).
17. Hess, G. Disease in metapopulation models: implications for conservation.
Ecology 77, 1617– -1632 (1996).
18. Ferguson, N. M. et al. Planning for smallpox outbreaks. Nature 425, 681– -685
NATURE|Vol 443|12 October 2006
© 2006 Nature Publishing Group
19. Keeling, M. J., Woolhouse, M. E. J., May, R. M., Davies, G. & Grenfell, B. T. Download full-text
Modelling vaccination strategies against foot-and-mouth disease. Nature 421,
136– -142 (2003).
20. Keeling, M. J. et al. Dynamics of the 2001 UK foot and mouth epidemic:
Stochastic dispersal in a heterogeneous landscape. Science 294, 813– -817
21. Anderson, R. M. et al. Epidemiology, transmission dynamics and control of
SARS: the 2002– -2003 epidemic. Phil. Trans. R. Soc. Lond. B 359, 1091– -1105
22. Ferguson, N. M., Donnelly, C. A. & Anderson, R. M. The foot-and-mouth
epidemic in Great Britain: pattern of spread and impact of interventions.
Science 292, 1155– -1160 (2001).
23. Morell, V. Wildlife biology—dogfight erupts over animal studies in the
Serengeti. Science 270, 1302– -1303 (1995).
24. Farry, S. C., Henke, S. E., Beasom, S. L. & Fearneyhough, M. G. Efficacy of bait
distributional strategies to deliver canine rabies vaccines to coyotes in
southern Texas. J. Wildl. Dis. 34, 23– -32 (1998).
25. Knobel, D. L., Du Toit, J. T. & Bingham, J. Development of a bait and baiting
system for delivery of oral rabies vaccine to free-ranging African wild dogs
(Lycaon pictus). J. Wildl. Dis. 38, 352– -362 (2002).
26. Knobel, D. L., Liebenberg, A. & Du Toit, J. T. Seroconversion in captive African
wild dogs (Lycaon pictus) following administration of a chicken head bait/SAG-2
oral rabies vaccine combination. Ondersterpoort J. Vet. Res. 70, 73– -77 (2003).
27. Steelman, H. G., Henke, S. E. & Moore, G. M. Gray fox response to baits and
attractants for oral rabies vaccination. J. Wildl. Dis. 34, 764– -770 (1998).
28. Foggin, C. M. Rabies and Rabies Related Viruses in Zimbabwe: Historical,
Virological, and Ecological. PhD thesis, Univ. Zimbabwe (1988).
Supplementary Information is linked to the online version of the paper at
Acknowledgements We thank Ethiopia’s Wildlife Conservation Department, the
Oromiya Rural Land and Natural Resources Administration Authority, and the
Bale Mountains National Park for permission to undertake this work. We thank
the staff of the EWCP for field work; the WCD veterinarians F. Shiferaw and
K. Argaw; C. Rupprecht and staff at CDC; T. Fooks and staff at the VLA;
R. K. Wayne for the work in his genetics laboratory, supported in part by his
grant from the NSF; and the IUCN/SSC Canid and Veterinary Specialist Groups
for advice. D.T.H. acknowledges the award of a MacLagan Travel Grant from the
Royal Society of Edinburgh in support of this research. Funding was provided by
the Born Free Foundation, Frankfurt Zoological Society, the Wellcome Trust,
Wildlife Conservation Network, Morris Animal Foundation, Conservation
International and Siren UK.
Author Contributions D.T.H. undertook much of the model formulation and
analysis, and coordinated the synthesis of ideas and information. D.A.R.
generated and compiled much of the empirical and all of the genetic data. L.M.
assisted with formulation and interpretation of the analyses. D.L.K. managed the
implementation of the vaccination programme. L.A.T. provided data and analysis
for spatial distribution of the wolf population. M.B.G. assisted with early
formulation of the modelling work and interpretation of later model output.
S.D.W. coordinated fieldwork for much of the period over which this study
applies. J.P.P. was instrumental in overseeing the genetic analyses. S.C. assisted
in formulation of the vaccination strategy and in writing the manuscript.
M.E.J.W. suggested analyses and assisted in writing the manuscript. C.S.-Z. was
responsible for funding and coordinating Ethiopian wolf research and
conservation work, and provided data on the 1992 outbreak. J.M. provided
essential data on wolf demography for use in the PVA model. D.W.M. conceived
and supervised several doctoral studies that have underpinned the project since
its inception. M.K.L. was responsible for designing and implementing the disease
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Correspondence and requests for materials should be addressed to D.T.H.
NATURE|Vol 443|12 October 2006
© 2006 Nature Publishing Group