Rosie Woodroffe, Department of Wildlife, Fish and Conservation Biology, University of California, One Shields Avenue, Davis, CA 95616, USA (fax +1 530 752 4154; e-mail email@example.com).
1Control of zoonotic disease is difficult to achieve when populations of multiple hosts, particularly wildlife, become persistently infected. Bovine tuberculosis (TB) is one such disease: its causative agent, Mycobacterium bovis, infects cattle, humans and multiple wildlife species including European badgers Meles meles.
2In Britain, from 1974 to 1998 various strategies for the control of cattle TB involved culling badgers in the immediate vicinity of TB-affected herds. However, patterns of association between cattle and badgers had not been investigated at a local scale.
3Using data from the Randomized Badger Culling Trial, an ongoing large-scale study of TB dynamics and control, we investigated local geographical associations between M. bovis infection in badgers and cattle.
4Mycobacterium bovis infections were locally clustered within both badger and cattle populations.
5We show, for the first time, that M. bovis infections in badgers and cattle are spatially associated at a scale of 1–2 km. Badgers and cattle infected with the same strain type of M. bovis are particularly closely correlated. These observational data support the hypothesis that transmission occurs between the two host species; however, they cannot be used to evaluate the relative importance of badger-to-cattle and cattle-to-badger transmission.
6Synthesis and applications. The close associations between M. bovis infections in cattle and badgers suggest that localized badger culling could reasonably be expected to reduce the risks of cattle TB infection; however, experimental culls have found no such beneficial effects over the time-scale on which they were tested. This demonstrates the difficulty of predicting the outcome of management interventions, and reinforces the need for well-designed empirical assessments of future control strategies.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Many pathogens of concern regarding the protection of human and domestic animal health can also infect wildlife. Infection may be sustained in a wildlife ‘reservoir’, with occasional ‘spillover’ to people and livestock (Dobson & Meagher 1996; Gordon et al. 2004), but transmission can also occur in the opposite direction, with domestic animals themselves contributing to the persistence of infection (Rhodes et al. 1998). The relative importance of transmission to and from wildlife is very difficult to assess without experimental manipulation (Haydon et al. 2002).
Mycobacterium bovis, the causative agent of bovine tuberculosis (TB), has a wide host range including cattle, humans and multiple wildlife species (Morris, Pfeiffer & Jackson 1994). Regular testing of cattle, with slaughter of those testing positive, has successfully controlled the infection across much of the developed world. However, control has not been achieved where wildlife populations have become persistently infected (Morris, Pfeiffer & Jackson 1994). In Britain, failure to control TB in cattle has been linked to persistent infection in populations of badgers Meles meles L., a widespread, although protected, wildlife species that thrives in landscapes where cattle are farmed (Neal & Cheeseman 1996).
Policies to control bovine TB in Britain have been based on the assumption that cattle acquire infection both from badgers and from other cattle. Control measures include restrictions on the movement of cattle from herds confirmed infected, and testing of cattle on farms that either adjoin the affected farm or have recently received animals originating from the restricted herd. Since 1974, these measures have been supplemented by various forms of badger culling on, and sometimes around, farms that have experienced recent TB outbreaks in cattle (Zuckerman 1980; Dunnet, Jones & McInerney 1986; Krebs et al. 1997).
Control strategies involving localized badger culling are based on the assumption that infections in cattle and badgers are associated, and that cattle may therefore act as a sentinel for infection in badgers. If this assumption is correct, then removing badgers that are spatially associated with infected cattle herds would be expected to reduce the risks of future badger-related outbreaks in the same herd, and also in neighbouring herds likely to be in contact with the same badgers. Despite this plausible logic, the only experimental test of localized badger culling showed that this was not associated with any reduction in the incidence of TB in cattle over the time-scale studied (Donnelly et al. 2003).
There is evidence to suggest that M. bovis infection is spatially clustered in both cattle (Krebs et al. 1997) and badger populations (Cheeseman, Wilesmith & Stuart 1989; Delahay et al. 2000; Olea-Popelka et al. 2003), at regional scales (e.g. counties) as well as at a more localized level (e.g. a few farms or badger territories). Regional clusters (sometimes termed hotspots) are geographically associated in the two species (Krebs et al. 1997). Moreover, at a regional level the prevalence of M. bovis infection has been found to be higher among badgers culled in association with cattle TB incidents than among those killed in road traffic accidents (Krebs et al. 1997), suggesting a local association between infections in the two species. However, detailed spatial comparisons have not been attempted.
Spatial associations between infection in cattle and badgers provide no information on whether transmission occurs from badgers to cattle, from cattle to badgers, from some other host (or hosts) to both species or some combination of these scenarios. Experimental studies of badger culling suggest that badgers do play a role in transmitting infection to cattle (Donnelly et al. 2003; Griffin et al. 2005), but other transmission scenarios have not been tested experimentally in the field. Hence, their potential importance, if any, to the maintenance of infection cannot be determined.
We investigated geographical associations between M. bovis infection in badgers and cattle, and evaluated indirect evidence for badger-to-cattle and cattle-to-badger transmission, using data from an ongoing large-scale study of bovine TB dynamics and control.
Materials and methods
This study draws upon data from the Randomized Badger Culling Trial (RBCT), a study initiated in 1998 to evaluate badger culling as a strategy to control TB in cattle. The RBCT's aims and methods are described in detail elsewhere (Bourne et al. 1999; Donnelly et al. 2003) but, in summary, the study was designed to compare TB incidence in cattle under three conditions of badger culling: proactive culling, which aimed to maintain low badger densities across large areas for the duration of the RBCT; reactive culling, which aimed to cull only those badgers spatially associated with farms that had experienced recent TB outbreaks in cattle; and no culling, an experimental control. Each of these treatments was replicated 10 times in trial areas of approximately 100 km2 each, to give a total of 30 trial areas (3000 km2) grouped into 10 ‘triplets’.
We investigated spatial associations between M. bovis infection in badgers and cattle in the 10 proactive culling areas of the RBCT. Because badgers were culled across entire trial areas, this allowed sampling of badgers in the vicinity of farms that experienced TB incidents in cattle, as well as nearby areas with lower cattle TB risk. To avoid possible distorting effects of recent badger culling on the distribution of infection (Swinton et al. 1997; Tuyttens et al. 2000), we restricted our analyses to the initial proactive cull carried out in each trial area. The locations of the 10 proactively culled trial areas (A–J) are shown in Fig. 1.
badger culling operations
Badgers and their setts (dens) are protected by law in Britain under The Badger Act 1992. All RBCT culling was undertaken by staff of the UK Department for Environment, Food and Rural Affairs (Defra; formerly the Ministry of Agriculture, Fisheries and Food), working under Crown Immunity. Before trial areas were randomly allocated to culling treatments they were surveyed for signs of badger activity. Surveys recorded, inter alia, the locations of all badger setts, including ‘main setts’ judged to be occupied year round.
Within each trial area, badger culls were carried out simultaneously on all land where landowners or occupiers had given Defra staff permission to cull (approximately 79% of the land area on average; Bourne et al. 2005). Following a period of pre-baiting, badgers were captured in cage traps, set mainly in the vicinity of badger setts. Traps were checked in the early morning and any badgers caught were immediately despatched by gunshot. Independent audits have deemed this method of despatch ‘humane’ (Kirkwood 2000; Ewbank 2003); confinement in the trap caused no detectable injury in the majority of badgers (Woodroffe et al. 2005b). No culling was carried out in February–April to avoid killing the mothers of unweaned cubs still confined to the sett (Woodroffe et al. 2005a).
diagnostic procedures for badgers
All badger carcasses were chilled and then subjected to necropsy according to standard operating procedures, usually within 72 h of dispatch. Necropsies were carried out at nine laboratories, which each received between 25 and 1497 badger carcasses. Sixty-one badgers had either no necropsy laboratory entered in the database or an invalid entry. Veterinarians carrying out the necropsies recorded basic information on sex, body size and age (distinguishing adults from cubs, and also recording tooth wear which gives an indication of the relative age of adults; Neal & Cheeseman 1996), and collected a standard set of lymph node samples (retropharyngeal, bronchial and mediastinal) as well as sampling any lesions that might be indicative of TB. These tissues were cultured for evidence of M. bovis infection. Badgers were considered infected if M. bovis was cultured from tissue samples or if acid-fast bacteria were detected by Ziehl Neelsen staining (Gallagher & Clifton-Hadley 2000). These diagnostic procedures were independently audited for quality assurance (Corbel 2003; Hall 2004). Cultures were carried out at three laboratories, with two laboratories carrying out the majority of cultures (3091 from a total of 3139 badgers culled). Eleven badgers had no valid culture laboratory recorded. The proportions of badgers found to be infected did not vary between laboratories, after adjusting for other significant predictors of prevalence, notably sex and trial area (Table 1). Restricting the data to include only badgers that were submitted to the most heavily used laboratories did not affect the results.
Table 1. Results for logistic regressions comparing M. bovis infections in badgers with hypothesized covariates
Odds ratio (95% CI)
Animals with missing or invalid database entries for laboratories are included in a ‘missing’ category.
Adding age only (adult vs. cub)
1·51 (1·01, 2·27)
Adding sex only (male vs. female)
1·39 (1·10, 1·77)
Adding age (with sex also in model)
1·51 (1·23, 1·86)
Adding sex (with age also in model)
1·40 (1·10, 1·78)
Model adjusted for trial area, age and sex
Age × sex
Trial area × age
Trial area × sex
Age-specific models adjusted for trial area
Adults: sex (male vs. female)
1·43 (1·26, 1·63)
Cubs: sex (male vs. female)
1·15 (0·77, 1·70)
Additions to age-specific model with trial area and sex
In total 3139 badgers were culled on initial proactive culls. These included 2699 adults, 434 cubs and six badgers of undetermined age. Badgers without age data were excluded from analyses, as were those without data on sex or M. bovis infection status (seven adults and six cubs in total), giving a total of 3120 badgers in the following analyses.
Isolates of M. bovis cultured from trial badgers were strain typed by spacer oligonucleotide typing (spoligotyping; Kamerbeek et al. 1997) by the Veterinary Laboratories Agency, Weybridge, UK. Spoligotype data were available for 338 (95%) of the 357 badgers with M. bovis infections confirmed by culture.
cattle tb data
Data on M. bovis infections in cattle came primarily from routine tuberculin testing. This involves giving two separate intradermal injections, of purified protein derivative (PPD) from M. bovis and Mycobacterium avium, respectively, and examining the injection sites 3 days later. Any herds containing cattle that show substantially greater reaction to M. bovis than to M. avium PPD are considered ‘skin test-positive’. Such herds are put under movement restrictions, and all skin test-positive animals are compulsorily slaughtered and subjected to necropsy. If lesions characteristic of TB are identified at necropsy, the herd is considered ‘lesion-positive’. Within trial areas Defra policy is to culture tissue samples from all compulsorily slaughtered cattle. Otherwise, samples are routinely cultured from up to three slaughtered cattle from herds with multiple visibly lesioned cattle, up to five slaughtered cattle from herds with one bovine with a single lesion, or up to 10 slaughtered cattle if no lesions are detected. If M. bovis is cultured from any animal, the herd is considered ‘culture-positive’. In addition, the Meat Hygiene Service inspects all cattle sent for slaughter and, if suspected cases of TB are identified, samples are also collected and cultured; confirmed slaughterhouse cases trigger a test in the herd of origin.
We examined the distribution of cattle, relative to individual badgers, during two time periods: the 12 months prior to (the last day of) each badger cull, and the 12 months following (the last day of) each cull. All cattle within the RBCT were required to have annual TB tests throughout the course of the trial, so each of these 12-month periods should represent complete testing of the herds in each trial area. However, very little routine TB testing was undertaken during the epidemic of foot and mouth disease (FMD) that occurred from 20th February to 28th November 2001 (Bourne et al. 2005). This period of suspended testing included part of the 12 months following the initial culls in trial areas E, F, G and H, and part of the 12 months preceding initial culls in trial areas I and J. Hence, to ensure data on the distribution of infection in badgers was compared with data on tests of all cattle herds, we extended the comparison periods to include 12 months of routine testing. A small number of test results from the FMD period were also included but were too few to influence the outcome of analyses. As an example, for a badger culled on 18 October 2002, in trial area J, the comparison period prior to culling was 9th January 2001–18th October 2002.
Isolates of M. bovis cultured from cattle in the trial areas were subjected to spoligotyping by the Veterinary Laboratories Agency. Spoligotype data were available for 5469 (95%) of the 5745 cattle with M. bovis infections confirmed by culture that were tested within 10 km of badger capture locations, in the 12 months before or after badger culling. Spoligotype data were available from one or more animals in 98% of culture-positive herds.
Data on the locations of cattle herds within trial areas, and badger capture locations, were taken from the RBCT trial database. The locations of herds outside core trial areas were obtained from VETNET, the Defra animal health information system. The locations of both badger capture sites and cattle herds were recorded to an accuracy of 100 m.
Our analyses of spatial associations of infection were based on nearest neighbour distances. We used ArcGIS version 9.0 (ESRI, Redlands, CA) to calculate nearest-neighbour distances from badgers to other badgers, from cattle herds to other cattle herds, and from badgers to cattle. Statistical analyses were carried out in SAS (SAS Institute, Cary, NC). We used non-parametric Wilcoxon–Mann–Whitney rank sum tests to compare the distributions of measures of distance from infected and uninfected animals, and the non-parametric sign test to compare paired measures of distance from the same animal.
For analyses of distances from badgers to cattle, we condensed data from badgers trapped at the same location (which shared the same distances to the nearest infected cattle herd); a single location could contribute data both as M. bovis infected (if one or more infected badgers were trapped there) and as uninfected (if one or more uninfected badgers were trapped there). In spoligotype-related calculations, a capture location contributed one observation for each spoligotype, as multiple spoligotypes were occasionally found at the same location. These procedures were undertaken to avoid spurious precision in the resulting statistics that could be caused if similar badgers trapped at a single location were treated as producing independent observations.
Evidence of clustering of infection in previous studies (Cheeseman et al. 1981; Delahay et al. 2000) suggests that the infection status of badgers from the same social group may also not be independent. Thus, to be conservative, we applied a further correction to tests comparing distances from badgers to cattle. Rather than attempting to assign individual badgers to social groups, we estimated the number of social groups based on the number of main setts identified on initial field surveys. The correction factor, denoted R, was based on the ratio of the number of main setts to the number of locations at which badgers were trapped. The standard normal deviate z-statistics corresponding to the rank sum tests are proportional to the standard deviation of a difference and thus can be scaled down by the square root of R. A similar correction was made to the reported significance of the sign tests by transforming the sign-test P-values into their corresponding standard normal deviate z-statistics, scaling them down by the square root of R, and then transforming back to the corresponding P-values.
In total, badgers were captured at 1246 locations, including captures at 392 main setts, giving a ratio R of 0·315. Tests comparing distances from badgers to cattle were adjusted using this ratio. Note that comparisons of distances between badgers were not adjusted in this way, because these analyses were designed to seek evidence of spatial correlation of infection among badgers. The number of main setts at which badgers were captured under-estimates the number of social groups represented, because some main setts were not culled because of lack of landowner consent or interference by animal rights activists. On these initial culls, badgers were captured within 500 m of 438 active main setts; this would give an alternate ratio R of 0·351. To be conservative, we used the lower estimate of R; however, using the alternate figure did not qualitatively affect the results (data not shown).
In addition to these spatial analyses, we used logistic regressions and odds ratios to examine associations between the infection status of badgers and covariates such as age and sex.
Table 2 presents details of the numbers of cattle and badgers sampled in each trial area, and the prevalence of M. bovis infection in both species.
Table 2. Summary statistics for each of the 10 trial areas in the proactive treatment of the RBCT, describing cattle populations in the 12-month periods preceding the initial proactive badger culls (as defined in the text) and badgers sampled during initial culls
TB-affected herds are defined as those containing cattle that are skin test-positive, culture-positive or TB-lesioned.
Numbers of cattle estimated from RBCT database.
Numbers of positive cattle identified in both new and ongoing cattle TB incidents.
Not all compulsorily slaughtered cattle had tissues samples cultured for M. bovis (see the Materials and Methods for details).
Numbers of badgers exclude 19 animals for which age, sex or infection status could not be determined.
Twenty spoligotypes were detected in total, seven shared and 13 in cattle only; spoligotypes found locally in badgers but not cattle were detected in cattle from other trial areas.
Of 3120 culled badgers analysed, 428 (13·7%) were cubs born in the previous 12 months. The proportion of badgers recorded as cubs varied substantially between areas, from roughly 25% for areas culled early in the season to approximately 5% for areas culled late in the season.
Of the 2692 adults culled, 1209 (44·9%) were males. However, there was substantial variation between trial areas, with the adult sex ratio being female biased in some areas and male biased in others (Table 3). Of the 428 cubs culled, 192 (44·9%) were males.
Table 3. Numbers, sex ratio and M. bovis prevalence of badgers taken on initial proactive culls
M. bovis prevalence
M. bovis prevalence
prevalence of m. bovis infection in badgers
Table 3 presents the prevalence of M. bovis infection in different age and sex classes of badgers. Prevalence varied substantially between trial areas.
After accounting for variation between trial areas, the overall prevalence of infection in adults (12·0%, n= 2692) was significantly higher than that recorded in cubs (7·9%, n= 428; Table 1). Among adults, male badgers were at significantly higher risk of infection (overall prevalence 14·4%) than females (overall prevalence 10·0%; Table 1) but there was no difference in the risk of infection of male and female cubs (overall prevalence: males 8·9%, females 7·2%; Table 1). We found no significant effect of the interaction between age class and sex, trial area and age class, or trial area and sex on risk of M. bovis infection (Table 1).
Among adults there was no association between tooth wear, a measure of age, and the risk of M. bovis infection; excluding badgers with a missing tooth wear score did not affect this result (Table 1). In contrast, in cubs there was a strong association between cub age, measured as days from 1 February, and the risk of M. bovis infection (Table 1).
clustering of infection within the badger population
Our analysis of spatial clustering of M. bovis infection within the badger populations sampled was influenced by the observation that infected badgers were greatly outnumbered by those showing no evidence of infection (Table 3). Thus, if infection were randomly distributed, the median distance from any badger to an uninfected badger would be smaller than the distance to a (less common) infected animal. To account for this expected pattern, we calculated, for each badger, the distance to the nearest infected badger and the distance to the nearest uninfected badger, and used the ratio between these two distances as the basis of our analysis. Using ratios also avoided biases that might occur if true nearest neighbours were not identified, for example for badgers sampled near trial area boundaries or close to areas without landowner consent for culling. To avoid generating infinite ratios for badgers captured at the same location, we added 1 m to all distances between pairs of badgers.
The median ratio between the distance to the nearest infected badger, and the distance to nearest uninfected badger, was much lower for infected animals (adults, 1·00; cubs, 1·00) than for uninfected animals (adults, 641·31, rank sum test P < 0·001; cubs, 807·23, rank sum test P < 0·001), indicating that infected badgers were closer to other infected badgers than would be expected if the distribution of infection within the population was random. This pattern was observed in nine of the 10 trial areas (Fig. 2). Trial area I, the only one lacking evidence of clustering, also had the highest prevalence (Table 3). Hence in this area all badgers were close to infected badgers, irrespective of their own infection status, and the median ratio was 1 for both infected and uninfected adults (Fig. 2).
clustering of infection within the cattle population
We analysed clustering of infection in the cattle population using an approach similar to that adopted for badgers, except that nearest neighbouring herds could fall outside trial area boundaries. We defined herds as ‘TB affected’ if they had any evidence of M. bovis infection (i.e. were skin test-positive, culture-positive or TB lesioned). In the 12-month period prior to badger culling, the ratio between the distance to the nearest TB-affected herd, and the distance to the nearest unaffected herd, was lower for TB-affected herds (median 1·38) than for unaffected herds (2·92; rank sum test P < 0·001). This indicated that TB-affected herds were closer to other affected herds than would be expected if the distribution of infection within the population was random. Similar results were obtained in the 12-month period following badger culling. Clustering remained significant (rank sum test P < 0·001) when only routine and whole herd testing results were analysed, demonstrating that the observed clustering of infection was not the result of increased testing of contiguous herds following a TB diagnosis. This pattern was detected in nine trial areas (Fig. 2); the tenth area, E, had no obvious characteristics that might explain the failure to detect clustering.
spatial associations between infected badgers and cattle
Infected and uninfected badgers had comparable opportunities for contact with cattle: in the 12 months prior to badger culling, distances to the nearest tested cattle herd were similar for infected and uninfected adults and cubs (Fig. 3; rank sum tests P > 0·05). However, over the same time period the ratio between the distance to the nearest TB-affected cattle herd and the distance to the nearest unaffected herd was significantly smaller for infected than uninfected badgers. This indicated that infected badgers were spatially associated with infected cattle herds, whether such herds were defined as those which responded to the tuberculin test (rank sum tests: adults P < 0·001, cubs P= 0·046), those in which infection was confirmed by culture (adults P < 0·001, cubs P= 0·036) or those in which TB lesions were found (adults P < 0·001, cubs P= 0·053). As shown in Fig. 3, these associations occurred at a scale of 1–2 km. Similar spatial associations were found with cattle tested in the 12 months following the badger cull. These associations were found in eight trial areas, the exceptions being I and J (Fig. 3).
In addition to this spatial association between infected badgers and infected cattle herds, there were strong associations between infected badgers and the numbers of individual infected cattle detected. Similar numbers of herds were tested within 1 km of infected and uninfected badgers, and herd sizes were also comparable (P > 0·1 for adults and cubs, both in the 12 months before and the 12 months following badger culling). However, larger numbers of infected cattle were detected within 1 km of sites where infected badgers were caught. For adults, this association was found consistently across all definitions of infected cattle (rank sum tests: skin test-positive cattle, P < 0·001; culture-positive cattle P < 0·001; TB-lesioned cattle P < 0·001), although the effects were not significant for cubs (P > 0·1 for all three definitions of infected cattle). Similar associations were found with infected cattle tested in the 12 months following the culls.
associations between strain types of m. bovis
In all 10 trial areas, one or more spoligotypes of M. bovis were found to infect both cattle and badgers (Table 2). Additionally, in all trial areas spoligotypes were found in cattle but not in the badgers sampled on these initial culls. In two trial areas, spoligotypes were found in badgers that were not detected locally in cattle during the period under consideration, although these spoligotypes were found in cattle in other trial areas.
We investigated more local spatial associations between spoligotypes detected in cattle and badgers by calculating, for each infected badger, the distance to the nearest bovine with the same spoligotype and to the nearest bovine infected with a different spoligotype, and compared these distances using a paired sign test. Overall, infected badgers were closer to cattle infected with the same spoligotype as themselves (Fig. 4; sign tests adults P < 0·001, cubs P= 0·031), with the same pattern detected in all trial areas except E (adults) and F (cubs). Similar associations were found with infected cattle tested in the 12 months following the culls.
spatial associations between lesioned cattle and badgers
If badgers transmit M. bovis infection to cattle, a particularly close association might be expected between infected cattle and badgers with tuberculous lesions, which are expected to excrete more infectious particles (Gallagher & Clifton-Hadley 2000; Delahay, Cheeseman & Clifton-Hadley 2001). To test this hypothesis, we calculated, for each infected badger, the distance to the nearest skin test-positive herd and the distance to the nearest unaffected herd. The ratio between these two distances was comparable for lesioned (127 badgers at 106 capture locations) and unlesioned adult badgers (196 badgers at 159 capture locations), both in the 12 months before the cull (rank sum test P= 0·59) and in the 12 months following the cull (rank sum test P= 0·36). There was likewise no difference in the number of skin test-positive cattle found within 1 km of lesioned and unlesioned adult badgers infected with M. bovis, either in the 12 months before the cull (rank sum test P= 0·85) or in the 12 months after (rank sum test P= 0·70). Note that there were too few lesioned cubs to allow meaningful comparisons.
The same approach can be used to investigate possible transmission of infection from cattle to badgers. If one assumes that infected cattle with tuberculous lesions are more infectious than those without, then, if cattle transmit M. bovis to badgers, infected badgers should be relatively closer to lesioned cattle than badgers showing no evidence of infection. The ratio between the distance to the nearest lesioned infected bovine, and the distance to the nearest unlesioned infected bovine, gave a measure that could be meaningfully compared between infected and uninfected badgers. This analysis suggested that infected badgers might indeed be relatively closer to lesioned cattle than uninfected badgers; however, the effect was not strong, with one significant difference when considering cattle tested in the 12 months following the cull (rank sum tests: adults P= 0·022, cubs P= 0·081) but no effect (although a trend in the same direction) for the 12 months preceding the cull (rank sum tests: adults P= 0·65, cubs P= 0·094).
In summary, our data provide clear evidence of an association between M. bovis infection in cattle and badgers. Not only are patterns of infection in the two species spatially correlated, there are also close linkages in the distribution of M. bovis strain types in the two species. Our data do not, however, allow an assessment of the relative importance of badger-to-cattle and cattle-to-badger transmission.
The pattern of M. bovis infection prevalence in different age and sex classes of badgers is likewise similar to that observed in other studies (Cheeseman et al. 1988). We found no association between M. bovis prevalence and tooth wear; this is somewhat surprising as the cumulative probability of infection would be expected to increase with age. We observed substantial variation in prevalence between trial areas and between cubs sampled at different times of year. However, because each trial area was sampled only once (at the initial proactive cull), it is impossible to attribute these findings to regional variation, seasonal differences or changes in prevalence over time as the trial areas were recruited.
Our data reveal evidence of spatial clustering of M. bovis infection within badger populations at a scale of a few kilometres. This is consistent with patterns observed on smaller scale culls, which have found that prevalence may be high in particular social groups while groups in neighbouring territories show no evidence of infection (Cheeseman et al. 1981). Longitudinal studies suggest that these clusters may persist for many years if left undisturbed (Delahay et al. 2000).
We also found clear evidence of clustering of infection within the cattle population, even when analyses excluded any tests that might have been prompted by detection of TB in a contiguous herd. Such clustering has been recognized for a long time, and is the reason why contiguous tests are carried out, as well as being the justification for determining cattle TB-testing regimes at a local (parish) rather than a regional (e.g. county) level.
Unlike previous studies, our data allowed us to detect a spatial association between M. bovis infection in badgers and cattle, again at a scale of 1–2 kilometres. At the start of the trial, badger home range sizes were likely to be on the order of 1 km2 (Woodroffe & Macdonald 1993) and average cattle herd density was approximately 0·75 herds km−2; hence the scale at which associations were detected is similar to the scale at which badgers and cattle could be expected to move and, potentially, to interact. It is difficult to be more precise about the scale at which the association occurs, as both badger and cattle locations were, for the purposes of analysis, necessarily represented as point locations. In reality, badgers move around their home ranges and cattle move within farms that may comprise multiple parcels of land (Johnston et al. 2005). Thus, each point location can be regarded as an approximation of reality measured with error. Such errors, assuming they are random, will have the effect of diminishing the estimated spatial associations, biasing the results toward the null hypothesis of no association. Detecting a spatial association with such imprecise location data supports the hypothesis that M. bovis infections in cattle and badgers are linked.
Spatial associations between cattle and badgers infected with the same spoligotype of M. bovis provide further evidence of a link between infections in the two host species. Simple spatial associations between infections in the two species might be generated by localized environmental conditions that predispose both host species to intraspecific transmission (e.g. microclimatic or micronutrient effects on susceptibility). However, our finding that badgers and cattle are infected with the same spoligotypes at a very localized scale suggests that, to the contrary, transmission is likely to occur between the species. The observation that some spoligotypes were detected only in one or other host species within a trial area (Table 2) might reflect the existence of transmission occurring solely within the cattle or badger population. However, it is also possible that these spoligotypes were undetected because not every infected animal, of either host species, was sampled in these cross-sectional datasets.
Clustering of infection within host populations, and associations between infection in cattle and badgers, were both detected across multiple trial areas. While the strength of the effects appeared to vary between trial areas, there was no consistent pattern in which areas showed particularly strong or weak effects (Figs 2, 3 and 4). In particular, there was no consistent difference in patterns observed in trial areas recruited before and after the FMD epidemic, suggesting that the temporary disruptions to cattle controls had no obvious effect on the associations we describe.
Tests for associations between the distribution of lesioned animals of one host species, and patterns of infection in the other species, provide only very limited information on the direction of transmission. Infected badgers were more closely associated with infected cattle that had tuberculous lesions than with cattle confirmed to be infected, but with no visible lesions. This association might be interpreted as indicating that cattle are involved in transmitting infection to badgers. However, the evidence is indirect and based on an assumption (that lesioned cattle are more infectious) that may be incorrect (Neill, Bryson & Pollock 2001). Moreover, the analysis is based on observational rather than experimental data, so it is impossible to exclude alternative explanations (e.g. that cattle infections acquired from badgers result in more severe pathology than infections acquired from other cattle). Experimental manipulations would be the only way to demonstrate cattle-to-badger transmission under field conditions, and to estimate the relative importance of such transmission. We found no significant differences between the patterns of cattle infections in the vicinity of lesioned badgers relative to infected badgers without visible lesions. However, as the relationships between lesion status and infectiousness are unknown, and are likely to vary between host species, these findings provide no insights into the relative importance of badger-to-cattle vs. cattle-to-badger transmission.
Our findings of clustered infection within both badger and cattle populations, and close spatial associations between M. bovis infections in cattle and badgers, suggest that control measures based on localized removal of badgers in the vicinity of cattle TB incidents, such as the ‘clean ring’ and ‘interim’ strategies, and the RBCT reactive treatment (Dunnet, Jones & McInerney 1986; Krebs et al. 1997) could have been expected to prove effective. However, the reactive strategy was not associated with any reduction in cattle TB incidence over the time-scale on which it was tested (Donnelly et al. 2003) and, while the other strategies were never tested experimentally, national TB incidence rose during the period when they were part of TB control policy (Krebs et al. 1997). Empirical experimental approaches, such as that taken in the RBCT, are required to estimate the potential value (or otherwise) of any culling strategy, despite the strong spatial associations observed between M. bovis infections in cattle and badgers.
One possible explanation for the failure of the reactive strategy to reduce cattle TB incidence is that, in addition to reducing badger density, culling disrupted clusters of infection within badger populations, potentially spreading infection to larger numbers of cattle herds. Comparisons of clustering on these initial culls, with patterns seen on subsequent ‘follow-up’ culls, will allow us to evaluate this hypothesis in future. We note, however, that evidence of clustering was equally strong, and patterns of association with cattle were equally close, in trial areas with varying histories of prior culling, ranging from no culling before the RBCT (trial area G), limited culling (e.g. trial areas D and H) and persistent culling over two decades (e.g. trial areas B and F).
Our finding that cattle might be involved in transmitting infection to badgers, as well as vice versa, would also have relevance to TB control policy if substantiated by further studies. This possibility suggests that aggressive control of TB within cattle populations (e.g. by improved testing regimes and movement restrictions) might help to reduce the risks of developing persistent infection in local badger populations. Such measures would be a particularly high priority in areas with little previous history of TB infection in cattle, where infection may not yet be established in badger populations.
All of the data presented here were collected by staff of the Department for Environment, Food and Rural Affairs, and the Veterinary Laboratories Agency; their assistance is gratefully acknowledged. We also wish to thank the many land owners and occupiers for consent to work on their land. This manuscript was greatly improved by three anonymous referees.