Haydon Et Al - 2002
Haydon Et Al - 2002
Haydon Et Al - 2002
Identifying Reservoirs of
Infection: A Conceptual and
Practical Challenge
Daniel T. Haydon,* Sarah Cleaveland,*Louise H. Taylor,* and M. Karen Laurenson*
Many infectious agents, especially those that cause emerging diseases, infect more than one host spe-
cies. Managing reservoirs of multihost pathogens often plays a crucial role in effective disease control.
However, reservoirs remain variously and loosely defined. We propose that reservoirs can only be under-
stood with reference to defined target populations. Therefore, we define a reservoir as one or more epide-
miologically connected populations or environments in which the pathogen can be permanently
maintained and from which infection is transmitted to the defined target population. Existence of a reser-
voir is confirmed when infection within the target population cannot be sustained after all transmission
between target and nontarget populations has been eliminated. When disease can be controlled solely by
interventions within target populations, little knowledge of potentially complex reservoir infection dynamics
is necessary for effective control. We discuss the practical value of different approaches that may be used
to identify reservoirs in the field.
I nfectious agents that can infect more than one host species
are ubiquitous. Indeed, 62% of all human pathogens are clas-
However, many situations exist in which the role of reservoirs
is less clear; for example, the reservoirs that harbor emerging
sified as zoonoses (1), and 77% of livestock pathogens and 91% viruses such as Ebola and Marburg remain unknown. For
of domestic carnivore pathogens infect multiple hosts. Fifty Mycobacterium bovis in the United Kingdom, a complex res-
seven of the 70 animal diseases considered to be of greatest ervoir system seems most likely, and identification of the most
international importance infect multiple hosts (2). The ability of important source of infection for cattle remains highly contro-
pathogens to infect a wide range of hosts has been demonstrated versial (8). Incomplete understanding of reservoirs has ham-
as a risk factor for disease emergence in both humans (1) and pered control of many diseases in Africa, such as Ebola virus
domestic animals (2). Virtually all recent outbreaks of disease in infection, Buruli ulcer, and rabies (9–13).
endangered wildlife have been caused by pathogens that can Many different and often contradictory definitions of res-
infect other, more abundant host species 3,4. ervoirs exist. Studies stress different characteristics of reser-
Pathogens that infect more than one host species are by voirs, namely, that infections in reservoir hosts are always
definition likely to be encountered in several host populations, nonpathogenic; any natural host is a reservoir host; the reser-
some of which may constitute infection reservoirs. Therefore, voir must be a different species; reservoirs are economically
a key issue in the design of control measures for multihost unimportant hosts; or reservoirs may be primary or secondary
pathogens is defining what is meant by reservoirs of infection hosts (14–18). Some definitions imply that a reservoir com-
and developing guidelines for their identification. prises only one species; other definitions suggest that an eco-
Although many emerging diseases of human, domestic logic system may act as a reservoir (16,18). Confusing,
animal, and wildlife populations are assumed to be maintained conflicting, and often incomplete concepts of what constitutes
in reservoir hosts (4), these reservoirs are rarely identified. In a disease reservoir result. We propose a conceptual framework
recent years, several emerging infectious disease threats to for defining and identifying reservoirs and discuss the practi-
human and animal health have been managed through large- cal value of different approaches that may be used to identify
scale measures directed at suspected reservoirs of infection. reservoirs in the field.
Sometimes action arises from a clearly perceived notion of
where infection resides. For example, approximately 1 million Proposed Framework
pigs were slaughtered in Malaysia in 1999 to control Nipah We propose the following approach, which can be applied
virus (5); several million chickens were slaughtered in Hong to any disease system, for understanding the role of potentially
Kong in 1998 and 2001 to prevent a projected pandemic of relevant reservoirs in that system. Figure 1 illustrates how this
Influenza A virus (6); and several million cows were slaugh- framework might be applied to various systems.
tered in Britain to curtail the epidemic of bovine spongiform
encephalopathy, and its possible transmission to humans (7). Suggested Terminology
The target population is the population of concern or inter-
*University of Edinburgh, Roslin, U.K. est to us. All other potentially susceptible host populations that
can be maintained in a coupled system (illustrated in Figure more precisely that source populations can be identified and
1B). Both satellite and reservoir populations would be compo- the more quantitative data that are available on their relative
nents of our reservoir. contribution to transmission, the more efficient the allocation
In an insightful paper, Ashford recognized many of the of resources is for disease control. Reservoir control tactics
problems in the simplistic use of the term reservoir and pro- require a much more complete understanding of the structure
posed a consistent definition of a reservoir as an “ecological and transmission processes that occur within the reservoir. For
system in which the infectious agent survives indefinitely” example, efforts directed at controlling infection in nonmainte-
(18). This definition differs from ours in that it does not refer- nance components of a reservoir are unlikely to be effective if
ence a target population and thus does not require that a reser- infection in the maintenance component of the reservoir
voir be a source of infection for a target population. Ashford remains uncontrolled.
defined reservoir hosts as those essential to maintenance of the The practical problem of identifying reservoirs of rabies
pathogen. We, however, argue that reservoirs may include for humans in Zimbabwe provides a useful illustration of some
nonessential hosts. Excluding nonessential hosts from a reser- issues involved. After a rise in the incidence of jackal and dog
voir causes two problems. First, populations harboring infec- rabies in the 1990s, debate has centered on whether jackals
tion may be nonessential to maintenance yet play a major role (Canis adustus) are reservoirs of this disease, an issue that has
in transmitting the pathogen to the target population. For important implications for formulating national rabies-control
example, FMDV persists indefinitely in African buffalo herds; programs (10,11). In Zimbabwe, domestic dogs are a mainte-
yet impala may constitute an important source of infection for nance and source population of rabies for humans. However,
the cattle target population (22) (e.g., population Z in Figure jackals account for >25% of all confirmed rabies cases in ani-
1C). Second, as Ashford recognized, the definition of reservoir mals and are also an important source of infection for humans
membership becomes ultimately intractable if each constituent (10,11). Jackals may be important components of the reservoir
population in the reservoir is considered nonmaintenance. as a maintenance or nonmaintenance population (Figure 2).
Under these circumstances, a reservoir could be composed of Because rabies can be maintained in dogs without jackals,
subsets of nonmaintenance populations in a variety of ways jackals are not an essential constituent population of the reser-
(Figure 1E). Although a minimal definition of a reservoir is voir. But can infection persist in jackals without dogs (Figure
clear, a fully inclusive definition is much less so. In Figure 1D, 2B)? Jackals may constitute part of a maintenance community
population V is not an essential host; nonetheless, this popula- in conjunction with an assemblage of other wild carnivores
tion must be considered a component of the reservoir because, (Figure 2A). The question is important because if dogs are the
if infection is eliminated in some other parts of the reservoir, only maintenance population in the reservoir, effective vacci-
eradication would not be achieved. For the same reason, our nation campaigns targeted at dogs should successfully elimi-
concept of a reservoir differs from the notion of a critical spe- nate human rabies from Zimbabwe. If, however, jackals
cies assemblage, which is defined as the minimum set of host comprise all or part of a maintenance community independent
communities in which a parasite can persist (16). of dogs, eliminating rabies will only be successful if jackal
rabies were also controlled (10,11). The recent high incidence
Control of Infection of jackal rabies in Zimbabwe might suggest that jackals are
Practical disease control requires answers to two ques- maintenance populations. A high incidence of disease alone is
tions: 1) Can an acceptable level of control be accomplished neither necessary nor sufficient evidence for this claim, partic-
without consideration of a reservoir? 2) If not, what popula- ularly when wide fluctuations in disease incidence occur (as
tions constitute the reservoir? Given a target-reservoir system, with jackal rabies). Mathematical models suggest that jackals
policies to manage infection may contain elements of three are probably unable to support infection without frequent rein-
broadly different tactics: 1) target control: directing efforts troductions from outside sources (24). However, Bingham et
within the target population with no reference to the reservoir al. (11) argue that spatial patterns are critical and that jackal
(e.g., human vaccination against yellow fever [23]); 2) block- epidemics may be sustained independently within key geo-
ing tactics: directing control efforts at blocking transmission graphic areas. The issue can be resolved unequivocally
between source and target populations (e.g., game fences to through implementation of a mass dog vaccination campaign,
control FMDV in cattle); and 3) reservoir control: controlling which would be a logical first phase of a national program. If
infection within the reservoir (e.g., culling programs, vaccina- jackal rabies persists in the absence of dog rabies, an effective
tion, or treatment of reservoirs). These three approaches program for rabies elimination will likely need to include oral
require progressively increased levels of understanding of res- vaccination of jackals.
ervoir structure and function. Rabies also provides an example of the need to identify a
Target control has the important advantage of requiring no target population when defining reservoirs. In the Serengeti
knowledge of potentially complex reservoir dynamics. A com- Plain in Tanzania, a distinct strain of rabies appears to be
plete understanding of infection dynamics within the reservoir maintained independently in spotted hyenas, without causing
is also not necessary to implement blocking tactics, although them any clinical disease, and with no evidence of spillover
identifying source populations in the reservoir is essential. The infection or disease occurring in any other species (within the
West Africa has been taken as evidence of an animal reservoir infection in the target-reservoir system.
for Gambian sleeping sickness (33). However, as animal-to-
human transmission has never been demonstrated, wildlife Conclusions
remain classified as potential reservoir hosts, and disease con- We have a poorer understanding of the epidemiology of
trol relies on treatment of people. In contrast, for Rhodesian multihost pathogens than simpler single-host systems. This
sleeping sickness, isolation of T. brucei rhodesiense from a dearth of understanding is a particular problem with emerging
single bushbuck in the 1950s (34) led to the assumption that diseases, since most emerging human, domestic animal, and
wildlife was the principal reservoir for human disease and wildlife diseases infect multiple hosts. Reservoirs must be
resulted in widespread culling of wildlife for disease control. defined with reference to particular target populations. Disap-
Only in 1966 were cattle identified as reservoir hosts (35). pearance of the pathogen in the target population after ring-
Current strategies focus on treating cattle with trypanocidal fencing provides categorical evidence of the existence of a res-
drugs (36). ervoir and its possible identity. However, exhaustive identifi-
Detecting a pathogen, particularly its transmission stage, in cation of all constituent populations of a reservoir may be
secretions or tissues provides supportive, but not unequivocal, difficult. This identification need not be a management priority
evidence that transmission to the target population can occur. if disease control is directed at the target population or at
Even where experiments demonstrate that transmission is pos- blocking transmission between reservoir and target. For infec-
sible, it may not occur in nature for a variety of behavioral or tion to be eliminated, however, disease-control measures must
social reasons, because the population is below critical com- be directed at the reservoir. Thus, an understanding of reser-
munity size or because of constraints of pathogen life history. voir infection dynamics is essential.
When the risks and costs of control are low, circumstantial
Genetic/Antigenic Characteristics evidence may be sufficient to justify implementing control
Genetic and antigenic characterization of pathogens iso- measures. Specifically designed intervention studies have ulti-
lated from different populations provides a more powerful tool mately been required to determine whether a particular species
for identifying key components of reservoirs. Antigenic and is a maintenance host, a source of infection, or one that has
genetic variation of pathogens isolated from the target popula- been infected incidentally. Control measures are likely to be
tion within the range observed in the reservoir is consistent ineffective if they are directed at components of the reservoir
with reservoir-target transmission. This pattern can be demon- that are neither maintenance hosts nor transmitters of the
strated by applying phylogenetic methods to sequence, random pathogen to the target population.
amplified polymorphic DNA, or restriction fragment length
polymorphism data, or by using serum cross-reactivity studies. Acknowledgments
Such methods have also been used to rule out important ani- We thank Paul Coleman, Scott McEwen, Eric Fevre, Mark Wool-
mal reservoirs of human disease in studies of Ascaris in Guate- house, and three anonymous referees for commenting on versions of
mala (37) and Cryptosporidium in Australia (38). this manuscript, the rest of the epidemiology group at the Center for
Tropical Veterinary Medicine for valuable discussions during its ges-
Intervention Studies tation, and Kevin Bown, whose thesis sparked the question.
Complete ring-fencing of target populations is the ultimate
Dan Haydon, Louise Taylor, and Karen Laurenson were sup-
step in identifying the existence and structure of reservoirs. In
ported by the Wellcome Trust during this work. Sarah Cleaveland was
practice, however, ring-fencing has rarely been achieved and,
supported by the United Kingdom Department for International
as a result, even those reservoirs we consider to be most fully
Development.
understood are not usually incontrovertibly proven. Despite
this, once a potential reservoir is identified, intervention studies Dr. Haydon is currently conducting postdoctoral research at the
can permit incidental but powerful inferences about the dynam- University of Guelph. His interests include the evolutionary dynamics
ics of infection in target-reservoir systems. In many cases, dis- of picornaviruses, epidemiologic and ecologic modeling, and land-
ease-control programs can effectively act as intervention scape ecology.
studies.
Control in a reservoir host population may be achieved by References
reducing host or vector density (e.g., culling possums to con- 1. Taylor LH, Latham SM, Woolhouse MEJ. Risk factors for human disease
trol tuberculosis in New Zealand [39], mosquito control for emergence. Philos Trans R Soc Lond B Biol Sci 2001;356:983–9.
2. Cleaveland SC, Laurenson MK, Taylor LH. Diseases of humans and their
West Nile fever, or sandfly control for cutaneous leishmaniasis domestic mammals; pathogen characteristics, host range and the risk of
[40]). Alternatively, control measures may focus more directly emergence. Philos Trans R Soc Lond B Biol Sci 2001;356:991–9.
on preventing transmission from the reservoir, e.g., separation 3. Murray DL, Kapke CA, Evermann JF, Fuller TK. Infectious disease and
of cattle and wildebeest to prevent transmission of malignant the conservation of free-ranging large carnivores. Animal Conservation
catarrhal fever in East Africa (41). The success of such inter- 1999;2:241–54.
4. Daszak P, Cunningham AA, Hyatt AD. Wildlife ecology—emerging
ventions often provides reasonable confirmation of the origi- infectious diseases of wildlife: threats to biodiversity and human health.
nal assumptions concerning transmission and maintenance of Science 2000;287:443–9.
5. Enserink M. Epidemiology—new virus fingered in Malaysian epidemic. 26. Monath TP. Ecology of Marburg and Ebola viruses: speculation and
Science 1999;284:407–10. directions for future research. J Infect Dis 1999;179:S127–38.
6. Vogel G. Infectious disease: sequence offers clues to deadly flu. Science 27. Berg AL, Raid-Smith R, Larsson M, Bonnett B. Case control study of
1998;279:324. feline Borna disease in Sweden. Vet Rec 1998;142:715–7.
7. Donnelly CA, Ferguson NM, Ghani AC, Woolhouse MEJ, Watt CJ, 28. Denny GO, Wilesmith JW. Bovine tuberculosis in Northern Ireland: a
Anderson RM. The epidemiology of BSE in cattle herds in Great Brit- case-control study of herd risk factors. Vet Rec 1999;144:305–10.
ain.1. Epidemiological processes, demography of cattle and approaches 29. Bode G, Rothenbacher D, Brenner H, Adler G. Pets are not a risk factor
to control by culling. Philos Trans R Soc Lond B Biol Sci 1997;352:781– for Helicobacter pylori infection in young children: results of a popula-
801. tion-based study in Southern Germany. Pediatr Infect Dis J 1998;17:909–
8. Krebs JR, Anderson RM, Clutton-Brock T, Donnelly CA, Frost S, Morri- 12.
son WI, et al. Policy: biomedicine, badgers and bovine TB: conflicts 30. YaghoobiErshadi MR, Javadian E. Epidemiological study of reservoir
between conservation and health. Science 1998;279:817–8. hosts in an endemic area of zoonotic cutaneous leishmaniasis in Iran. Bull
9. Leirs H, Mills JN, Krebs JW, Childs JE, Akaibe D, Woollen N, et al. World Health Organ 1996;74:587–90.
Search for the Ebola virus reservoir in Kikwit, Democratic Republic of 31. Calderon G, Pini N, Bolpe J, Levis S, Mills J, Segura E, et al. Hantavirus
the Congo: reflections on a vertebrate collection. J Infect Dis reservoir hosts associated with peridomestic habitats in Argentina. Emerg
1999;179:S155–63. Infect Dis 1999;5:792–7.
10. Bingham J, Foggin CM, Wandeler AI, Hill FWG. The epidemiology of 32. Van Bressem M-F, Van Waerebeek K, Jepson PD, Raga JA, Duignan PJ,
rabies in Zimbabwe. 1. Rabies in dogs (Canis familiaris). Onderstepoort J Nielsen O, et al. An insight into the epidemiology of dolphin morbillivi-
Vet Res 1999;66:1–10. rus worldwide. Vet Microbiol 2001;81:287–304.
11. Bingham J, Foggin CM, Wandeler AI, Hill FWG. The epidemiology of 33. Mehlitz D, Zillman U, Scott CM, Godfrey DG. Epidemiological studies
rabies in Zimbabwe. 2. Rabies in jackals (Canis adustus and Canis on the animal reservoir of Gambiense sleeping sickness. 3. Characteriza-
mesomelas). Onderstepoort J Vet Res 1999;66:11–23. tion of Trypanozoon stocks by isoenzymes and sensitivity to human
12. Cleaveland SC, Dye C. Maintenance of a microparasite infecting several serum. Tropenmedizin und Parasitologie 1982;33:113–8.
host species: rabies in the Serengeti. Parasitology 1995;111:S33–47. 34. Heisch RB, McMahon JP, Manson-Bahr PEC. The isolation of Typano-
13. Portaels F, Chemlal K, Elsen P, Johnson PDR, Hayman JA, Hibble J, et al. soma rhodesiense from a bushbuck. BMJ 1958;2:1202–4.
Mycobacterium ulcerans in wild animals. Rev Sci Tech 2001;20:252–64. 35. Onyango RJ, van Hoeve K, de Raadt P. The epidemiology of Trypano-
14. Dorland WAN. Dorland's illustrated medical dictionary. London: W.B. soma rhodesiense sleeping sickness in Alego location, central Nyanza,
Saunders; 1994. Kenya. I. Evidence that cattle may act as reservoir hosts of trypanosomes
15. Henderson IF, Henderson WD. Henderson's dictionary of biological infective to man. Trans R Soc Trop Med Hyg 1966;60:175–82.
terms. London: Longman Scientific & Technical; 1989. 36. Welburn SC, Fevre EM, Coleman PG, Odiit M, Maudlin I. Sleeping sick-
16. Swinton J, Harwood J, Grenfell BT, Gilligan CA. Persistence thresholds ness: a tale of two diseases. Trends Parasitol 2001;17:19–24.
for phocine distemper virus infection in harbour seal Phoca vitulina 37. Anderson TJC, Jaenike J. Host specificity, evolutionary relationships and
metapopulations. Journal of Animal Ecology 1998;67:54–68. macrogeographic differentiation among Ascaris populations from
17. Swinton J, Woolhouse MEJ, Begon ME, Dobson AP, Ferroglio E, Gren- humans and pigs. Parasitology 1997;115:325–42.
fell BT, et al. Microparasite transmission and persistence. In: Hudson PJ, 38. Morgan UM, Constantine CC, ODonoghue P, Meloni BP, OBrien PA,
editor. The ecology of wildlife diseases. Oxford: Oxford University Press; Thompson RCA. Molecular characterization of Cryptosporidium isolates
2001. from humans and other animals using random amplified polymorphic
18. Ashford RW. What it takes to be a reservoir host. Belgian Journal of DNA analysis. Am J Trop Med Hyg 1995;52:559–64.
Zoology 1997;127:85–90. 39. Roberts MG. The dynamics of bovine tuberculosis in possum populations,
19. Mosier DA, Oberst RD. Cryptosporidiosis—a global challenge. Ann N Y and its eradication or control by culling or vaccination. Journal of Animal
Acad Sci 2000;916:102–11. Ecology 1996;65:451–64.
20. Bartlett MS. The critical community size for measles in the United States. 40. Davies CR, Llanos-Cuentas EA, Campos P, Monge J, Leon E, Canales J.
Journal of the Royal Statistical Society 1960;123:37–44. Spraying houses in the Peruvian Andes with lambda-cyhalothrin protects
21. Thomson GR. Overview of foot-and-mouth disease in southern Africa. residents against cutaneous leishmaniasis. Trans R Soc Trop Med Hyg
Rev Sci Tech 1995;14:503–20. 2000;94:631–6.
22. Bastos ADS, Boshoff CI, Keet DF, Bengis RG, Thomson GR. Natural 41. Machange J. Livestock and wildlife interactions. In: Thompson DM, edi-
transmission of foot-and-mouth disease virus between African buffalo tor. Multiple land-use: the experience of the Ngorongoro Conservation
(Syncerus caffer) and impala (Aepyceros melampus) in the Kruger Area, Tanzania. Gland, Switzerland: IUCN-The World Conservation
National Park, South Africa. Epidemiol Infect 2000;124:591–8. Union; 1997. p. 127–41.
23. Robertson SE, Hull BP, Tomori O, Bele O, LeDuc JW, Esteves K. Yellow
fever—a decade of reemergence. JAMA 1996;276:1157–62.
24. Rhodes CJ, Atkinson RPD, Anderson RM, Macdonald DW. Rabies in Address for correspondence: Daniel T. Haydon, Department of Zoology, Uni-
Zimbabwe: reservoir dogs and the implications for disease control. Philos versity of Guelph, Guelph Ontario, Canada N1G 2W1; fax: 519-767-1656; e-
Trans R Soc Lond B Biol Sci 1998;353:999–1010. mail: dhaydon@uoguelph.ca
25. East ML, Hofer H, Cox JH, Wulle U, Wiik H, Pitra C. Regular exposure
to rabies virus and lack of symptomatic disease in Serengeti spotted hye-
nas. Proc Natl Acad Sci U S A 2001;98:15026–31.
Use of trade names is for identification only and does not imply endorse- The opinions expressed by authors contributing to this journal do not
ment by the Public Health Service or by the U.S. Department of Health necessarily reflect the opinions of the Centers for Disease Control and
and Human Services. Prevention or the institutions with which the authors are affiliated.