INTRODUCTION
Diseases transmitted between wildlife and livestock are increasingly challenging animal health authorities and methods used to mitigate the spread of pathogens. In North America it is estimated that at least 79% of diseases reportable to the World Organisation for Animal Health (OIE) have a wildlife component [Reference Miller, Farnsworth and Malmberg1]. In North America, establishment of Mycobacterium bovis, the causative agent of bovine tuberculosis (bTB), in wildlife is recognized as an impediment to eradication in cattle [Reference Sweeney, Miller, Portacci and Lombard2]. M. bovis has one of the broadest host ranges of all known pathogens, affecting many groups of mammals [Reference O'Reilly and Daborn3]. In some parts of the world, this bacterial pathogen has spilled over from cattle into wildlife, where it has persisted as a reservoir of infection and thwarted efforts to eradicate the disease from farm animal populations. Major free-ranging hosts of endemic M. bovis infection include the European badger (Meles meles) in the UK and Ireland [Reference Cheeseman, Wilesmith and Stuart4–Reference Woodroffe7], brushtail possum (Trichosurus vulpecula) in New Zealand [Reference Morris and Pfeiffer8, Reference Palmer, Childs, Mackenzie and Richt9], Cape buffalo (Syncerus caffer) and greater kudu (Tragelaphus strepsiceros) in southern Africa [Reference Bengis10–Reference De12], elk (Cervus canadensis) and American bison (Bison bison) in Canada [Reference Wobeser13], white-tailed deer (Odocoileus virginianus) in the USA [Reference O'Brien14, Reference Schmitt15], and wild boar (Sus scrofa) in Spain [Reference Naranjo16, Reference Parra17]. Many other species, including humans, are susceptible to M. bovis but are not believed to sustain transmission in the absence of infection from another species [Reference O'Reilly and Daborn3] or a change in population or behaviour that enhances disease spread [Reference Cousins and Florisson18]. The host status of a species with regard to M. bovis may differ between regions or may change over time depending on population density or management regimen [Reference Cousins and Florisson18]. This, in turn, is the basis for efforts to manage or eradicate M. bovis in wildlife and domestic animals.
Here we present a systematic review of the English scientific literature (n = 163) to evaluate the status of M. bovis in free-ranging wildlife of North America. We had three goals: first, to evaluate the known status of M. bovis in wildlife populations of North America; second, to identify potential risks for establishment of M. bovis in uninfected North American wildlife populations where eradication or control would be difficult and costly; and third, to highlight critical issues faced in managing M. bovis at the livestock–wildlife interface.
METHODS
We used a systematic literature review to identify and characterize studies on the status of M. bovis in free-ranging wildlife populations in North America [Reference Khan19, Reference Okoli20]. Our review focused on literature published since 1900. All literature relating to the microbiology, epidemiology, or occurrence of M. bovis in wildlife were considered eligible and are reflected in the use of the broad search terms: bovine tuberculosis, Mycobacterium bovis, M. bovis, and wildlife. The review focused only on the scientific peer-reviewed literature, edited book chapters, and government technical reports in the public domain. We expect that this search will have captured the majority of the scientific publications on M. bovis in the wildlife of North America.
The literature review involved three steps. First, we identified key words for use in the search process. Second, we conducted a systematic review of PubMed, Scopus, and Web of Science according to the search terms. Finally, once all relevant sources were identified and retrieved, we reviewed and categorized each paper. Categories were identified a-priori and included (1) status of M. bovis in wildlife which included case reports, pathology findings, surveillance reports, and molecular epidemiology studies; (2) epidemiology characterization which included analysis of factors associated with transmission or drivers of infection in the wildlife population; (3) mitigation studies evaluating or reporting the use of vaccination development, eradication tools, or other control methods. To identify potential risk factors, studies were reviewed and common themes related to the occurrence of M. bovis in free-ranging wildlife were identified. Risk factors identified in each study were enumerated in table format for easy comparison. The frequency with which each risk factor was reported across all studies was assessed and tallied. We report the risk factors most commonly reported in the literature. Due to differing methodologies and often incomplete reporting of results, meta-analysis was not applicable for assessment of risk factors.
RESULTS
Status of M. bovis infection in North American and Hawaiian wildlife species
Cervidae
Historically bTB has been identified and confirmed in nine geographically distinct wildlife populations in North America and Hawaii and is thought to be endemic in three cervid populations (Fig. 1, Table 1). Before the 1990s, M. bovis had only rarely been reported in free-ranging Cervidae in North America. In Canada, Hadwen [Reference Hadwen21] confirmed M. bovis infection in elk, moose (Alces alces), and mule deer (O. hemionus) that ranged with M. bovis-infected bison in the Buffalo National Park in east-central Alberta. Belli [Reference Belli22] reported M. bovis in a white-tailed deer in Ontario, Canada. In the USA, Levine [Reference Levine23] and Friend et al. [Reference Friend, Kroll and Gruft24] each reported two cases of M. bovis in white-tailed deer in New York, and Ferris et al. [Reference Ferris25] reported two cases in white-tailed deer in Illinois. In Michigan, a M. bovis-infected white-tailed deer was documented in 1975 [Reference Schmitt15] and in Montana, a free-ranging mule deer living near a M. bovis-infected elk ranch was diagnosed with M. bovis [Reference Rhyan26].
* Reports for Mexico represent unconfirmed reports only. They are included here for completeness as they are the only reports of M. bovis in wildlife in Mexico. Prevalence reported is for the presence of serum antibodies for bovine purified protein derivative (PPD) and have not been confirmed using bacterial culture.
More recently, hunter harvest surveillance by the Michigan Department of Natural Resources in 1995 identified via bacterial culture an endemic focus of M. bovis in free-ranging white-tailed deer within a five-county area of the northeastern Lower Peninsula after M. bovis was diagnosed in a hunter-killed white-tailed deer [Reference Schmitt15, Reference O'Brien27–Reference Palmer29]. In response, the state established a special deer management unit that encompassed the core area of infected deer. Despite previous reports of M. bovis-infected white-tailed deer in New York and Ontario, this focus of infection represented the first known persistent reservoir of M. bovis in white-tailed deer in the North America [Reference Palmer, Childs, Mackenzie and Richt9]. It is believed that M. bovis was transmitted from cattle to deer in Michigan sometime during the early to mid-1900s when bTB was widespread in cattle in the state. Epidemiological models estimate that transmission from cattle to deer occurred around 1955 [Reference Palmer, Childs, Mackenzie and Richt9].
To manage the epizootic, Michigan adopted two principal management strategies in the affected five-county area [Reference O'Brien14]. First, the state increased hunter harvests to reduce deer population density. Second, it placed restrictions on supplemental feeding and baiting of deer. Voluntary restrictions were initially sought from the public and hunt clubs, followed by a regulatory ban on feeding and baiting within the bTB area and the rest of the state. In addition, Michigan adopted a moratorium on establishing new captive cervid facilities. Hunter harvest has reduced the five-county deer population by about 50% and, concurrent with these management actions, M. bovis prevalence has declined from 4·9% to about 1·7% in adult deer [Reference O'Brien14, Reference O'Brien30]. Rocky Mountain elk (C. canadensis) have also periodically been found to be infected, however at a low apparent prevalence of 0·3% [Reference O'Brien31].
In 2006, M. bovis was discovered and confirmed using bacterial culture in white-tailed deer in Minnesota in conjunction with an outbreak of M. bovis in beef cattle [Reference Portacci32]. One of 474 hunter-harvested deer tested in the vicinity of infected cattle herds was positive for M. bovis, and targeted culling and surveillance identified another positive deer. Epidemiological linkages between M. bovis-infected deer and cattle were supported by the proximity of deer and cattle cases and M. bovis strain identity between cattle and deer. The state responded by initiating more rigorous sampling protocols to estimate prevalence of M. bovis in the deer population and conducted widespread culling to reduce deer densities. Hunter-harvested deer surveillance during the autumn hunting season is conducted and if M. bovis-infected deer are identified, targeted culling is conducted in the spring around areas where infected deer have been found. Based on autumn hunter-harvested sampling, disease prevalence in deer has decreased from 1·2% in 2005 to an undetectable level in 2010 [Reference Portacci32, Reference Carstensen and DonCarlos33].
A similar situation exists in Manitoba, Canada, where a herd of 2500–4000 elk (C. c. manitobensis) was implicated in an outbreak of M. bovis in 11 cattle herds surrounding Riding Mountain National Park (RMNP) [Reference Lees, Copeland and Rousseau34]. Commingling of elk and cattle feeding on the same hay bales was considered the most likely mode of transmission between species. Typing of M. bovis isolates from this area indicated the presence of a unique strain seemingly unrelated to other strains previously identified in Canada [Reference Wobeser13, Reference Lutze-Wallace35]. Management to help reduce transmission of M. bovis between wildlife and cattle around RMNP has included increased M. bovis surveillance in wildlife and livestock, extended hunting seasons to reduce elk and deer populations, barrier fencing to protect hay-storage yards from free-ranging Cervidae, legislation to prevent baiting and concentration of cervids, and prescribed burning to improve wildlife habitat in RMNP [Reference Wobeser13, Reference Nishi, Shury and Elkin36].
The status of M. bovis infection in cervid populations of Mexico is uncertain. Two studies have evaluated the presence of M. bovis in free-ranging white-tailed deer in northern Mexico. A serosurvey of white-tailed deer conducted from 2004 to 2009 reported antibody responses to bovine PPD in 8·9% of deer (n = 347) on six ranches in the states of Tamaulipas, Nuevo León, and Coahuila [Reference Medrano37]. A follow-up study, conducted during the 2009–2010 hunting season, evaluated mandibular lymph nodes and tonsils collected from white-tailed deer for presence of Mycobacterium using histopathology, Ziehl–Neelsen staining, mycobacterial isolation and PCR. The study evaluated 44 deer and found that 4·5% had gross changes. Histopathology revealed morphological changes suggestive of tuberculosis such as macrophage aggregation, necrosis, giant cells, mineralization and bacilli acid-alcohol resistance (BAAR). The presence of M. tuberculosis complex by amplification of DNA from a tissue sample by PCR was confirmed. However, M. bovis has not yet been confirmed in white-tailed deer using bacterial culture, which is considered the ‘gold standard’ for M. bovis identification [Reference Barrios-García38].
Bovidae
Hadwen [Reference Hadwen21] documented a severe and prolonged outbreak of M. bovis in bison in Canada's Buffalo National Park (BNP) in Wainwright, Alberta during the early to mid-1900s. The bison herd was maintained in a semi free-ranging condition, within a fenced natural area co-inhabited by elk, deer, and moose. More than one-half of about 12 000 bison culled between 1923 and 1939 had M. bovis lesions at meat inspection, as did about 5–6% of elk and moose and less than 1% of mule deer culled in 1939 and 1940. Because of the number of source herds for these bison, the exact origin(s) of M. bovis in BNP cannot be stated with any certainty. Possible sources of M. bovis infection included domestic cattle that were pastured with bison before the bison were transported to BNP, cattle in areas adjacent to or within the park, or cattle-bison hybrids in one or more of the source herds [Reference Wobeser13]. BNP was disbanded in 1939 and the herd was destroyed [Reference Wobeser13].
Canada's largest remaining reservoir of M. bovis is the free-ranging bison population in and around Wood Buffalo National Park (WBNP), bordering northern Alberta and southern Northwest Territories. M. bovis infection was introduced to the park between 1925 and 1928 when more than 6600 plains bison (B. b. bison) were imported from the infected herd at BNP [Reference O'Reilly and Daborn3]. The bison population in WBNP grew to an estimated 12 000–15 000 animals in the late 1940s and then declined to about 5000 by 1968 [Reference Tessaro39]. Despite a further decline in bison numbers, prevalence of M. bovis has remained high; 39% of 3400 bison necropsied in the park between 1950 and 1967 had M. bovis lesions [Reference Broughton, Reynolds and Hawley40] and 49% of 342 bison captured at WBNP between 1997 and 1999 were positive on either the caudal fold test or a fluorescent polarization assay [Reference Jolley41]. Tessaro et al. [Reference Tessaro, Forbes and Turcotte42] isolated M. bovis from 21% of 72 bison found in and around WBNP during 1983–1985 and concluded that M. bovis was endemic in the bison population and represented a growing threat to uninfected bison and cattle in the region.
Suidae
M. bovis has not been isolated from feral swine (Sus scrofa) on the mainland of North America; however, it has been isolated from feral swine on Molokai Island, Hawaii [Reference De Lisle, Mackintosh and Bengis11]. The epidemiological role of feral swine was investigated in a multispecies outbreak of M. bovis and although M. bovis was efficiently transmitted among swine (estimated prevalence of 20%), the disease was controlled in cattle by depopulating infected cattle herds and culling infected swine and deer [Reference Essey43–Reference Essey45]. Post-culling M. bovis prevalence in feral swine was estimated to be 3·2% [Reference Essey45]. More recently feral swine were investigated having been suspected of transmitting M. bovis to a cow on the island. Subsequent wildlife surveys for M. bovis identified five positive swine out of 482 tested, with an estimated apparent prevalence of 1% [46]. Risk of transmission to cattle on the island has been mitigated by maintaining cattle in an area of the island where feral swine are not known to exist.
Other species
There have been relatively few North American surveys for M. bovis in host species other than Cervidae and Bovidae. Rhyan et al. [Reference Rhyan26] reported culture and histopathological evidence of M. bovis in 2/16 free-ranging coyotes (Canis latrans) collected adjacent to a confined M. bovis-infected elk herd in southwestern Montana. Large-scale investigations of M. bovis infection rates in carnivores were not reported until Bruning-Fann et al. [Reference Bruning-Fann47] conducted necropsies of 294 carnivores from the M. bovis-endemic area of Michigan. Seven animals had microscopic lesions suggestive of M. bovis and nine had lymph node cultures positive for M. bovis – six coyotes, two raccoons (Procyon lotor), one red fox (Vulpes vulpes), and one black bear (Ursus americanus). Restriction fragment length polymorphism patterns of M. bovis isolates were identical in carnivores and deer, indicating that both groups were infected with the same strain. Moreover, most of the lesions in affected carnivores were in mesenteric lymph nodes, suggesting exposure through ingestion of scavenged material. The location of lesions, variety of species involved, and widely dispersed cases of M. bovis (i.e. no foci of infection) were indicative of disease spillover (i.e. the incidental infection of a host species that cannot maintain the infection without continued exposure to a competent host) rather than endemic M. bovis in these carnivores [Reference Bruning-Fann47]. A broader survey of 175 coyotes in Michigan found that 33% were either culture-positive for M. bovis or had granulomatous lesions suggestive of M. bovis infection [Reference VerCauteren48]. Prevalence of M. bovis in coyotes varied regionally from 19% to 52%. Lesions occurred most commonly in the gastrointestinal tract; however, one coyote had advanced disease with lesions occurring in the lung and liver [Reference VerCauteren48]. A study by Johnson et al. [Reference Johnson49] found that captive coyotes orally inoculated with 1 × 105 c.f.u. of M. bovis did not become infected or shed M. bovis in faeces or oronasally supporting the presumption that coyotes were spillover hosts.
A follow-up study of raccoons in Michigan, reported an estimated apparent prevalence of 2·5%, with 5/199 raccoons testing positive [Reference Witmer50]. Another study reported a similar prevalence of M. bovis in raccoons, with eight (2·5%) positive out of 333 sampled [Reference O'Brien31]. Follow-up investigations documented excretion of M. bovis in saliva or nasal secretions in raccoons but not in urine or faeces [Reference Palmer, Waters and Whipple51]. Current evidence suggests that raccoons are likely a spillover host for M. bovis with no significant role in the maintenance of M. bovis in livestock [Reference Witmer50, Reference Palmer, Waters and Whipple51].
North American opossums (Didelphis virginiana) have also been surveyed with an estimated apparent prevalence of 0·5% (2/379) [Reference O'Brien31] to 3·0% (4/134) [Reference Witmer50]. Shedding of M. bovis has been reported for North American opossums and experimental inoculation has demonstrated that they are relatively susceptible to M. bovis infection. However, opossums do not typically develop generalized disease, and therefore, may not shed the bacteria sufficiently to serve as a reservoir for infecting livestock [Reference Fitzgerald52, Reference Diegel53].
Avian species may also be involved in the transmission cycle of M. bovis, but their relative importance is not well understood [Reference Butler54–Reference Clarke56]. In the El Paso Texas milkshed, pigeons (Columba livia), blackbirds (Agelius sp.), and other species were thought to have introduced M. bovis into US dairies from dairies in Ciudad Juárez, Mexico, either through mechanical transport of contaminated material or infected birds shedding M. bovis into cattle feed [Reference Pillai57]. However, follow-up surveys of tissues from 252 pigeons, nine European starlings (Sturnus vulgaris), and one common grackle (Quiscalus quiscula) from the 14 infected dairies in El Paso failed to identify M. bovis [Reference Pillai57]. Blackbirds were not sampled, and the study did not address the potential of birds to mechanically transport M. bovis-contaminated material. Currently, this is the only reported survey of wild avian species for M. bovis in North America.
RISK FACTORS FOR TRANSMISSION OF M. BOVIS TO NORTH AMERICAN WILDLIFE SPECIES
Four risk factors were consistently and most frequently reported as important to the establishment of M. bovis in uninfected wildlife populations in North America: (1) commingling of infected cattle with susceptible wildlife, (2) supplemental feeding of wildlife, (3) inadequate surveillance of at-risk wildlife, and (4) unrecognized emergence of alternate wildlife species as successful maintenance hosts.
Commingling of infected cattle with susceptible wildlife
Commingling of livestock, particularly cattle, with wildlife has been associated with the introduction of several pathogens into wildlife populations [Reference Cross58, Reference Maichak59]. In North America, examples include Brucella abortus in elk and bison and multiple respiratory pathogens in bighorn sheep [Reference Wolfe60, Reference Olsen61]. In the six wildlife populations in North America – Michigan, Minnesota, Hawaii, RMNP, BNP, and WBNP – known to currently have or historically had endemic M. bovis infection, all are thought to have been established via initial transmission of M. bovis from cattle to wildlife. In Manitoba, Canada, commingling of elk and cattle feeding on the same hay bales was considered the most likely mode of transmission between species [Reference Lees, Copeland and Rousseau34].
Commingling of cattle and cervid species is common throughout North America where domestic and wild ruminants share pasture resources. In many regions of North America commingling of cattle with cervid species has increased as wildlife populations have increased [Reference Côté62]. This is expected to continue with growing conservation efforts that preserve valuable wildlife habitat [Reference Rooney63]. In some regions of North America, commingling has increased as a result of game ranching which is a rapidly growing segment of the US animal agriculture industry, particularly in parts of the South and Midwestern United States [Reference Teer, Renecker and Hudson64]. This often controversial industry is thought by some to pose a risk for introducing and establishing additional foci of endemic wildlife diseases, including M. bovis [Reference Butler65]. In some locations, exotic hoofstock or native Cervidae are released or baited into a semi free-ranging environment for recreational hunting. There, the animals often mix with cattle and feral swine (Fig. 2), creating an environment for pathogen exchange [Reference Cooper66]. Some livestock producers have boosted incomes by combining exotic and native hoofstock game ranching with traditional cattle production; others have replaced cattle with exotic or native hoofstock ranching as a new form of animal agriculture.
Supplemental feeding of wildlife
Research indicates that supplemental feeding and baiting have been major factors in the propagation and persistence of M. bovis in Michigan's white-tailed deer population [Reference O'Brien14]. Deer populations with the highest prevalence of M. bovis were clustered on private land where feeding and baiting were common practices. M. bovis can survive for months on foodstuffs commonly used by deer [Reference O'Brien14] and transmission between deer via shared feed has been documented [Reference Palmer, Waters and Whipple67]. Shared feed also appears to be the primary route of M. bovis transmission from deer to cattle [Reference O'Brien14, Reference Palmer, Waters and Whipple67]. Furthermore in Michigan the presence of supplemental feeding has been identified as a significant predictor of M. bovis prevalence in white-tailed deer [Reference Hickling68]. Deer density, which is often increased by supplemental feeding, has also been associated with increased prevalence of M. bovis [Reference O'Brien14]. Collectively these data strongly suggest that supplemental feeding and baiting of deer has been a critical factor in the persistence of M. bovis in Michigan's deer population [Reference O'Brien14]. States that employ this practice would also be at risk if M. bovis were introduced [Reference Schmitt15, Reference Palmer, Waters and Whipple67–Reference Miller69]. Recreational feeding of wildlife for viewing or hunting is common throughout North America and has been banned or restricted in only 16 states. The remaining states have few or no regulations regarding feeding of wildlife.
The federal and state feeding grounds in Wyoming, where brucellosis is an ongoing problem in elk, are a particular worry with regard to the spread of wildlife diseases [Reference Cross58, Reference Maichak59] including M. bovis [Reference Peterson70]. Winter feeding of elk in Wyoming has a long history. The first feedground, the federally operated National Elk Refuge, was created in 1912 to mitigate damage to hay stored for wintering cattle. Over the past century the number of feeding grounds has grown in Wyoming; the state now manages 22 feedgrounds in three counties. On average, these sites support 20500 elk comprising an estimated 80% of the state's total elk population during winter [Reference Boyce71–Reference Smith73]. Despite increasing recognition of wildlife disease risks, social and political constraints have allowed for the continued operation of the winter feedgrounds in Wyoming [Reference Smith73]. Several other western states also periodically feed populations of elk, deer, and bighorn sheep during severe winters when animal mortality is expected to be high [Reference Smith73]. Although less of a hazard due to the infrequent nature of these feeding programmes, these practices also pose a risk of spreading of M. bovis, if it were introduced.
Inadequate surveillance of at-risk wildlife
There is evidence that M. bovis infections in free-ranging wildlife may be silent, existing for years or even decades before being detected in hunter-killed animals or emerging or re-emerging in local cattle populations. Examples include Michigan, where M. bovis is thought to have persisted for 40–50 years before identification, Hawaii's Molokai Island, where M. bovis persisted for years among feral swine before being identified in cattle, and RMNP where bTB was first identified in 1937 but re-emerged in cattle and elk in 1991 more than 50 years after first being discovered in wildlife [Reference Carbyn, Harrington and Paquet74, Reference Tabulenas75]. Lack of surveillance, poor diagnostic tests, and the extended latency of M. bovis pose risks for re-emergence of this pathogen in areas where it has historically existed in wildlife or cattle.
In addition there is likely to be bias associated with historic accounts of M. bovis in wildlife. Wobeser [Reference Wobeser13] discussed the scarcity of documented bTB cases in wild Cervidae in North America during the early to mid-1900s suggesting several factors that may have contributed to the lack of such reports, including the fact that deer were less abundant 40–50 years ago, disease surveillance in wildlife was not a routine practice, and M. bovis lesions in Cervidae may not have been recognized as tuberculosis. The few early cases of M. bovis in cervids in North America were regarded as spillover events from cattle, in which M. bovis was quite common at the time. The potential for M. bovis infections in free-ranging wildlife to exist undetected for years or even decades necessitates robust surveillance systems in wildlife, particularly in regions where M. bovis has been historically identified. In addition, surveillance around M. bovis-infected cattle herds is needed to identify and address infection in wildlife.
Alternate wildlife species as maintenance hosts
Traditionally, it was assumed that M. bovis would not persist in populations of free-ranging deer or elk unless they had contact with infected bison, cattle herds, or captive deer [Reference Essey, Vantiem and Thoen44, Reference Bleem76]. This initial presumption that affected wildlife cannot successfully maintain M. bovis in the population has been a consistent theme associated with nearly all of the established wildlife reservoirs in North America. As a result animal health authorities have often limited surveillance in potentially affected wildlife or disregarded findings of M. bovis in wildlife. This may contribute as a risk factor for wildlife species that are currently considered to be purely spillover species and thought unable to maintain persistent infection in the population. In contrast to this paradigm Minnesota presumed that deer could become a wildlife reservoir and aggressively reduced the deer population in the affected area by 55% and banned recreational feeding [Reference Carstensen and DonCarlos33]. Currently M. bovis infection in the affected area has declined to undetectable levels.
Feral swine may pose a significant risk for spread of M. bovis. Historically, feral swine were considered spillover hosts, unable to maintain infection at the population level [Reference Corner77, Reference McInerney, Small and Caley78]. However scientific evidence regarding the role of feral swine in the epidemiology of M. bovis is shifting this view. Naranjo et al. [Reference Naranjo16] reported that wild boar in Mediterranean ecosystems sustained M. bovis infection and appeared capable of transmitting the disease to other species. Supporting evidence included high prevalence of M. bovis in wild boar fenced from contact with other species, lesions in thoracic lymph nodes and lungs suggesting respiratory infection and transmission, and extensive lesions in juvenile boar that likely represented the main shedders of M. bovis. Likewise, Aranaz et al. [Reference Aranaz79], de Mendoza et al. [Reference de Mendoza80], and Santos et al. [Reference Santos81] concluded that wild boar in Spain and Portugal were maintenance hosts of M. bovis based on ecological factors and lesion characteristics. Circumstances favouring M. bovis transmission between wildlife and livestock included artificial increases in wild game populations stimulated by a robust hunting industry, lack of natural predators, and intensive cattle grazing in game preserves with susceptible wildlife hosts [Reference de Mendoza80].
Given the range expansion of feral swine in North America (Fig. 3) there is enormous concern over their capacity for transmitting diseases that impact animal agriculture and human health. Free-ranging populations of swine have been reported in at least 38 US states and three Canadian provinces, nearly doubling the number of states occupied since 1988 [Reference Hutton82]. Particularly worrisome is the recent appearance of feral swine in Michigan, where the potential exists for interaction with M. bovis-infected white-tailed deer and cattle.
OPPORTUNITIES FOR MANAGEMENT OF M. BOVIS IN WILDLIFE
The need for a structured approach to management of diseases shared between wildlife and livestock has been identified as a critical need in North America (1). Concepts for integrated adaptive management of diseases at the livestock–wildlife interface have been proposed by multiple authors [Reference Miller, Farnsworth and Malmberg1, Reference Wasserberg83, Reference Thirgood84]. Furthermore, many countries have developed surveillance systems for disease events in wildlife [Reference Morner85]. Many of these systems implement integrated adaptive disease management using risk assessments and disease monitoring systems for wildlife [Reference Hartley and Gill86–Reference Sainsbury89]. These systems have common themes, which may be adaptable to the management of M. bovis in North America. Disease management systems that integrate methods to identify newly emerging science related to M. bovis conduct risk evaluations of wildlife, implement risk mitigations to prevent transmission between cattle and wildlife, and conduct disease surveillance in at-risk wildlife, would be the most successful at preventing new foci of M. bovis infection in wildlife [Reference Miller, Farnsworth and Malmberg1]. Integrating these components allows adaptive management of the disease system and may provide the most success for management of M. bovis. Adaptive management has been well described in the ecological and wildlife management literature [Reference Kendall90, Reference McCarthy and Possingham91], but has only recently been proposed as a method for managing disease systems [Reference Miller, Farnsworth and Malmberg1, Reference Wasserberg83, Reference Thirgood84].
Rapid identification of emerging science of M. bovis in wildlife is critical to successful management and prevention of additional foci of infection in North America. Even well-documented disease systems, such as bTB, can rapidly change with new issues emerging. Furthermore, policies to address the newly emerging science are often slowly adopted, detrimentally influencing agricultural systems and wildlife management in North America [Reference Olsen61, Reference O'Brien92]. Risk analysis can then be used to understand risks posed these emerging issues [Reference Suter93]. However, for diseases at the livestock–wildlife interface, quantitative risk assessments are often limited by the availability of data describing contact between wildlife and livestock. Historically, risk assessments have often assessed the risk of pathogen transmission from wildlife to livestock [Reference Daniels, Hutchings and Greig94]. However, for M. bovis in North America the initial transmission event is often from livestock to wildlife [Reference Tessaro95, Reference Sweeney, Miller, Portacci and Lombard96]. The most successful and useful risk analyses for M. bovis should consider the bi-directional nature of transmission. This then allows for the implementation of risk mitigations to prevent transmission between livestock and wildlife.
Mitigating risks of pathogen transmission between livestock and wildlife has received considerable attention [Reference Wasserberg83, Reference VerCauteren, Gehring and Landry97, Reference VerCauteren, Lavelle and Hygnstrom98]. The ability to eliminate M. bovis from North American wildlife populations has been rare and when successful required extensive culling of wildlife. The only example of success is the likely eradication of bTB from Minnesota where over 9700 deer were tested and the deer population in the affected area was reduced by an estimated 55% in 4 years [Reference Carstensen and DonCarlos33, Reference Carstensen, O'Brien and Schmitt99]. This effort cost an estimated US $86 million in federal and state government expenditures [Reference Carstensen and DonCarlos33].
Wildlife removal strategies can have unintended consequences, as exemplified in the UK where badger behaviour was changed as a result of culling, increasing the risk of M. bovis transmission to cattle [Reference Woodroffe100]. Eradication efforts requiring culling of large numbers of wildlife are costly and often publicly untenable, thus preventing establishment of M. bovis in wildlife populations is a central pillar of long-term risk mitigation strategies [Reference Meyer and Meagher101]. Implementing risk mitigations that integrate wildlife surveillance, modify livestock husbandry practices, incorporate barriers between livestock and wildlife, and use other wildlife deterrents may offer the greatest potential for reducing economic and social impacts resulting from newly infected wildlife populations [Reference VerCauteren, Gehring and Landry97, Reference VerCauteren, Lavelle and Hygnstrom98, Reference VerCauteren, Shivik and Lavelle102].
The need to develop comprehensive surveillance systems that integrate livestock and wildlife components has been suggested [Reference Miller, Farnsworth and Malmberg1, Reference Mörner103]. Robust surveillance systems in wildlife at the livestock–wildlife interface to provide early detection of M. bovis when transmission from livestock to wildlife occurs is essential. Recent epidemiological models suggest that once M. bovis is introduced, the probability of becoming established in a wildlife population is at least 10% [Reference Ramsey104]. Furthermore, once established M. bovis can be fiscally impossible to eradicate, recent analysis found that to achieve eradication in Michigan would cost at least US $1.5 million annually over the next 30 years [Reference Cosgrove105]. As a result identification and rapid response when M. bovis is transmitted from livestock to wildlife has proven the most successful tool for preventing establishment of M. bovis in wildlife [Reference Carstensen and DonCarlos33]. Early detection in an outbreak is essential for successful control of the pathogen; however, this requires proactive testing of wildlife by animal health authorities. Furthermore, surveillance of potentially affected wildlife must be done at levels necessary to identify M. bovis in the susceptible wildlife population.
Surveillance should include species known to be at risk for maintaining M. bovis (i.e. white-tailed deer, elk, bison) and also species that have been identified as potential emerging maintenance hosts such as free-ranging swine. Creative surveillance programmes that sample both known maintenance species and use sentinel species (e.g. coyotes) may provide the best solution for conducting surveillance at detection levels while balancing what is fiscally feasible [Reference VerCauteren48, Reference Berentsen106]. Developing a comprehensive national monitoring system for M. bovis in wildlife that is logistically and fiscally sustainable could yield economic benefits for livestock health management by helping prevent transmission from livestock to wildlife through early detection.
CONCLUSION
M. bovis has been identified in nine geographically distinct wildlife populations in North America and is endemic in at least three of these populations. Furthermore, the status of M. bovis in an additional cervid population in Northern Mexico is uncertain requiring additional investigation. A structured approach to management is needed to prevent the establishment of M. bovis in additional susceptible wildlife populations. Theoretical and practical methods for integrated adaptive wildlife disease management have been proposed by multiple authors [Reference Miller, Farnsworth and Malmberg1, Reference Wasserberg83, Reference Thirgood84]. Integration of these concepts to directly address critical risk factors associated with establishment of M. bovis in wildlife offers the best opportunity to successfully manage this pathogen in North American wildlife. Adaptive disease management principles can help integrate and prioritize surveillance activities along with the development and deployment of effective, practical, and economical mitigation measures designed to lower bTB transmission risk. Furthermore, control tools such as intensive culling or hunting, specifically for reducing disease transmission, can be incorporated when appropriate in the adaptive management framework. This would allow for incorporation of uncertainty such as the limited understanding of the effect culling might have on deer populations and the dynamics of M. bovis infection in these populations. Strategies such as selective culling of high-risk individuals or groups can be tested as part of the management process to improve the development of practical and effective culling programmes.
ACKNOWLEDGEMENTS
We appreciate the encouragement of Dr Katie Portacci and Dr Jerry Freier. We thank Dr Kurt VerCauteren, Dr Mike Dunbar, Dr Thomas Gidlewski, Dr Colleen Bruning-Fann, and Dr Thomas Kasari for detailed review of our work and providing suggestions. We also thank numerous other reviewers for their thoughts on the manuscript.
DECLARATION OF INTEREST
None.