Next Article in Journal
Monitoring Mycoplasma bovis Diversity and Antimicrobial Susceptibility in Calf Feedlots Undergoing a Respiratory Disease Outbreak
Next Article in Special Issue
Bovine Tuberculosis (Mycobacterium bovis) Outbreak Duration in Cattle Herds in Ireland: A Retrospective Observational Study
Previous Article in Journal
Concurrent Infection of Skunk Adenovirus-1, Listeria monocytogenes, and a Regionally Specific Clade of Canine Distemper Virus in One Gray Fox (Urocyon cinereoargenteus) and Concurrent Listeriosis and Canine Distemper in a Second Gray Fox
Previous Article in Special Issue
Development and Challenges in Animal Tuberculosis Vaccination
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Pathogens
Volume 9
Issue 7
10.3390/pathogens9070592
pathogens-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Mycobacterium bovis Population Structure in Cattle and Local Badgers: Co-Localisation and Variation by Farm Type

by
Georgina Milne
1,*,
Adrian Allen
1,
Jordon Graham
1,
Raymond Kirke
2,
Carl McCormick
3,
Eleanor Presho
1,
Robin Skuce
1,4,† and
Andrew W. Byrne
4,5,†
1
Veterinary Sciences Division, Agri-Food and Biosciences Institute (AFBI), Belfast BT4 3SD, UK
2
Veterinary Epidemiology Unit, Department of Agriculture, Environment and Rural Affairs, Belfast BT4 3SB, UK
3
Department of Agriculture, Environment and Rural Affairs, Veterinary Service Animal Health, Coleraine BT52 2AJ, UK
4
School of Biological Sciences, Queen’s University Belfast, Belfast BT9 5DL, UK
5
One-Health Unit, Department of Agriculture, Food and the Marine, Agriculture House, Dublin, Ireland
*
Author to whom correspondence should be addressed.
†
Both were PI’s on this project with shared responsibility, so retain Joint senior authorship.
Pathogens 2020, 9(7), 592; https://doi.org/10.3390/pathogens9070592
Submission received: 25 May 2020 / Revised: 3 July 2020 / Accepted: 8 July 2020 / Published: 21 July 2020
(This article belongs to the Special Issue Tuberculosis Epidemiology and Control in Multi-Host Systems)
Browse Figures
Versions Notes

Abstract

:
Bovine tuberculosis surveillance in Northern Ireland includes Multiple-Locus Variable number tandem repeat Analysis (MLVA) to determine the Mycobacterium bovis genetic type present in both cattle and the predominant wildlife host, the European badger (Meles meles). These data are useful for investigating clusters of infection and understanding the scale at which interspecific transmission may occur. We utilised a comprehensive dataset of routinely sampled isolates from infected cattle and from badgers killed in road-traffic accidents to investigate the spatial co-location of MLVA types in, and between, the badger and cattle populations. Furthermore, we investigated the hypothesis that the type of farming enterprise might explain some variation in this relationship. MLVA types were spatially co-localised in cattle and road-traffic accident (RTA) badger hosts, indicative of a shared epidemic. Dairy herds were more likely to have at least one MLVA type in common with nearby RTA badgers, compared to non-dairy herd types. Marginally more MLVA spatial clustering was observed in non-dairy herds, which may be a consequence of relatively more between-herd movements. For the cattle population, local transmission mechanisms such as infection from contiguous herds, infectious wildlife and short-range between-herd cattle movements appear primarily to drive the epidemic: there appears to be a more limited role for long-range movements. Animal management practices are likely to be the driving force behind this observation, as beef rearing is associated with elevated numbers of animal movements compared to dairy herds.

1. Introduction

The wildlife-livestock interface presents a conduit through which pathogens can be exchanged [1]. In the UK and Ireland, the presence of wildlife reservoirs is implicated in the persistence of Mycobacterium bovis, the principal causative agent of bovine tuberculosis (bTB) in cattle [2,3]. M. bovis can infect a wide range of hosts, both wild and domesticated [4]. In the United Kingdom (UK) and the Republic of Ireland (ROI), the most important wildlife maintenance host is the European badger, Meles meles [5,6,7,8]. Infection transmission may be as a result of direct contact between species [9], or potentially through contaminated shared environments and fomites [10,11].
In Northern Ireland (NI), the costs of bTB control in cattle have exceeded £365 million over a recent twelve year period [12] and, despite an intensive and costly state-led programme focusing on the cattle population, eradication has not yet been achieved [13]. NI contains a relatively small area (approx. 13,500 km2), yet sustains a badger population of approximately 33,500 individuals (95% CI 26,000–41,200); [14]. A passive road-traffic accident (RTA) surveillance programme for M. bovis in badgers has been ongoing since 1998. This survey estimated M. bovis prevalence in sampled badgers to be 15.3% (95% CI 13.1–17.5%) [15,16], and revealed elevated bTB risk in cattle herds in close proximity to infected RTA badgers, compared to herds proximal to uninfected badgers [15]. Routine surveillance efforts for M. bovis in cattle and badger hosts also include spoligotyping and Multiple-Locus Variable number tandem repeat Analysis (MLVA) typing [17,18]. These data revealed clear spatial structuring (i.e., clustering) of M. bovis genetic types in cattle herds [18,19], and have shown that both cattle and badger hosts with the same MLVA type tend to be closer together than hosts with a different MLVA type [20,21]. This observed structure in the M. bovis population in NI indicates that the bTB epidemic is co-localised between both wild and domestic hosts, consistent with some degree of transmission between wild and domestic species [7]. Nevertheless, cattle to cattle transmission also contributes to the maintenance of infection at both the local and national scales, e.g., via within-herd amplification (regardless of source), contact with nearby infected herds, or between-herd cattle purchases involving infected animals [7,22,23,24].
However, the influence of cattle management and trade, as a risk factor, on the M. bovis population structure in both cattle and badgers has been given little attention to date. Beef and dairy production systems differ across a number of factors which may influence transmission. For example, we showed recently that beef herds in NI were generally associated with more between-herd cattle movements than dairy herds, and also experience elevated MLVA richness at the herd-level [25]. Different herd types have specific between-herd contact patterns potentially linked to different infection pathways [22], with beef fattening herds appearing more susceptible to infection introduced by bought-in cattle than dairy herds, and indeed, in NI, there is elevated risk of bTB infection associated with the purchase of beef animals [26]. These differences in herd management may subsequently manifest in different spatial relationships in the clustering of M. bovis in, and between, infected cattle and badger populations. We therefore aimed to analyse spatial relationships in the M. bovis MLVA types found both in RTA badgers and cattle herds in NI to ultimately gain insight into the spread of the epidemic, both within and between hosts.

2. Results

2.1. Summary Statistics

The final cattle dataset contained information on 9208 bTB breakdowns occurring between 2008 and 2016, in 6594 herds (herds with a milk license = 1822, 27.6%; without a milk license = 4772, 72.4%; see Figure 1). In total, 364 M. bovis MLVA types were isolated from the cattle population, 135 were found in both herds with and without milk licenses, 62 were found only in herds with milk licenses, and 167 were isolated only from herds without milk licenses. At herd level (i.e., the yearly herd-level incidence per MLVA type), the 12 most common MLVA types represented 77.8% of the total cattle isolates (Figure 2a; 001, 002, 003, 004, 005, 006, 007, 009, 010, 027, 117 and 122). The final RTA badger dataset contained data on 271 RTA badgers collected between 2008 and 2016, inclusive. Thirty different MLVA types were identified in this population, with the 12 most common MLVA types representing 90% of the total (Figure 2b). The spatial distribution of the most common herd-level M. bovis MLVA types present in both cattle and RTA badger hosts is shown in Figure 2c. In total, 26 (83.3%) of the 30 MLVA types found in badgers were also found in cattle. Herds with and without milk licenses differed across a number of epidemiologically relevant criteria, including herd size (herd size of herds with a milk license, median = 213; inter-quartile range (IQR): 123–338; without a milk license = 69; IQR: 33–137; wilcoxon signed rank test p < 0.001), outward movements (herds with a milk license = 23; IQR: 41–68; without a milk license = 21; IQR: 8–57; p < 0.001), and inwards movements (herds with a milk license = 1; IQR: 0–6; n without a milk license = 8; IQR: 0–43; p < 0.001).

2.2. Assessment of MLVA Clustering within the Cattle Population

Figure 3 illustrates the distribution of the most common MLVA type (002) in cattle in: (i) all cattle herds; Figure 3a, (ii) herds with milk licenses; Figure 3b, and (iii) herds without milk licenses; Figure 3c (See Supplementary Material 1, Figures S1–S11 for the remaining plots). The results of permutation analyses are visualized in Figure 4, which shows the actual median distance between herds infected with each MVLA type (vertical lines) compared to the distances between herds from the sampling distribution (histograms). For each MLVA type in all herds and non-dairy herds, the median distance between infected herds was significantly different from the distances in the sampling distribution (p < 0.05). In all cases, the actual, observed median distances were smaller than those from the sampling distribution, indicating that herds infected with the same MLVA type tended to be closer together than a random selection of herds. Similar results were obtained for herds with a milk license, with the exception of herds infected with types 001, 002, 010 and 027. In these cases, the actual median distance between infected herds did not differ significantly from the median distances obtained in the sampling distribution (p > 0.05). This can be observed for these MLVA types in Figure 4, where the values at the red vertical lines (distances between herds infected with each MLVA type) were not distinguishable from the red sampling distribution.
Table 1 shows the values (in km) of the 25th, 50th, 75th and 100th percentiles for between-herd pairwise distances for all herds, herds with a milk license, and herds without a milk license (see also See Supplementary Material 2, Figures S12–S23). The values for the 25th, 50th and 75th percentiles were furthermore reported as a percentage of the 100th percentile. The maximal distance between any two herds sharing the same MLVA type could be considerable, with observed distances of 167.3 km between infected herds (e.g., type 005, all herds). In all but one MLVA type (type 006), the overall maximal extent was smaller in herds with milk licenses than in herds without. Infection was considered spatially clustered if 50% of pairwise distances were found within 50% of the maximal pairwise distance. Indeed, the results show that, for all herds, 50% were found between 11.1% (type 022, all herds) and 35.5% (type 001, all herds) of the maximal pairwise distance between any two herds infected with the same MLVA type. This reflects a general trend of localised spatial clustering, with the majority of infected herds within relatively close proximity to each other, and a smaller number of infected herds disproportionally widely distributed. For only those herds with a milk license, 50% of herds were within 15.9% (type 009, dairy herds) to 44.8% (type 027, dairy herds) of the maximal pairwise distance between any two herds with a milk license infected with the same MLVA type. For eight MLVA types (001, 002, 003, 004, 122, 010, 027 and 117), marginally less localised clustering was observed in dairy herds compared to non-dairy herds.

2.3. Intra and Interspecific Nearest Neighbour (NN) Analysis

The median distance between NN RTA badgers which shared an MLVA genotype was 2.44 km (Inter-Quartile Range; IQR: 1.22–4.04 km), whilst the median distance between nearest neighbour RTA badgers which did not share an MLVA genotype was 3.33 km (IQR: 1.94–5.22 km; Wilcoxon signed rank test; V = 2853, p = 0.011). The distance between an RTA badger and the NN cattle herd in which the MLVA type was not isolated (1.49 km; IQR: 0.83–2.25 km) was almost 45% larger than the distance to a NN herd with the same MLVA type (0.82km; IQR: 0.50–1.64 km; V = 9469, p < 0.001). Similar findings were obtained for RTA badgers in proximity to both beef and dairy herds, see Table 2. The NN distance for cattle herds sharing an MLVA type was over 25% greater (1.12 km; IQR: 0.61–2.32 km) than the NN distance between herds which did not share the same MLVA type (0.82km; IQR: 0.51–1.27 km). This was also observed in non-dairy herds and, to a lesser extent, in dairy herds; see Table 2. All findings were also replicated in the sensitivity analysis (all p < 0.05), indicating that the spatial resolution of the data has no substantive impact on the interpretation of the results.

2.4. Distance-Based Similarlty Analysis

Each RTA badger was surrounded by an average of four other badgers (IQR: 2–7) within a 7 km radius. The probability of two RTA badgers sharing the same MLVA type dropped by 17% for every km increase in distance between them (Odds Ratio; OR: 0.86; 95% lower and upper confidence limits; 95% CI: 0.80–0.92; Inverse OR: 1.17); Figure 5a. There were 63 cattle herds (IQR: 39–81) in the 7 km radius around each RTA badger, and the probability of RTA badgers and herds sharing MLVA types fell by 9% with every km (OR: 0.91; 95% CI: 0.90–0.92; Inverse OR: 1.09); Figure 5b. When stratified by herd type, we found that each RTA badger was surrounded by 17 dairy herds (IQR: 10–27) and 41 (IQR: 23–59) non-dairy herds, respectively. The probability of RTA badgers and dairy herds sharing MLVA types fell by 7% per km increase in distance (OR: 0.94; 95% CI: 0.92–0.95; Inverse OR: 1.07, and by 11% for non-dairy herds (OR: 0.90; 95% CI: 0.89–0.91; Inverse OR: 1.11). In the cattle-cattle context, each cattle herd was surrounded by 63 others (IQR: 40–88), and we found that the probability of two cattle herds sharing the same MLVA type dropped by 9% for every km increase in distance (OR: 0.91; 95% CI: 0.91-0.92; Inverse OR: 1.09); Figure 5c. There was a slight decrease when considering the dairy herd population independently (OR: 0.93; 95% CI: 0.93–0.94; Inverse OR: 1.07) from non-dairy herds (OR: 0.92; 95% CI: 0.91–0.92; Inverse OR: 1.09).

2.5. Determinng the Factors Associated with RTA Badgers and Nearby Cattle Herds Sharing MLVA Types

In the final dataset, there were 9471 RTA badger-cattle herd pairs with at least one MLVA type in common, and 5964 RTA badger-cattle herd pairs with no MLVA types in common. The univariable analysis revealed that the ‘number of MLVA types isolated from a bTB breakdown’ (OR per every additional type: 1.98; 95% CI: 1.83–2.13) and ‘herd size’ (OR per every additional ten animals: 1.01; 95% CI: 1.01–1.02) were positively associated with RTA badgers and nearby cattle herds sharing MLVA types. ‘Distance between hosts’ (OR per km: 0.89, 95% CI: 0.87–0.91), ‘the absence of a milk license’ (OR: 0.72, 95% CI: 0.67–0.77), ‘inwards movements’ (OR per every additional ten animals: 0.99; 95% CI: 0.99–0.99) and ‘outwards movements’ (OR per every additional ten animals: 0.98; 95% CI: 0.98–0.99) were negatively associated with RTA badgers and nearby cattle herds sharing MLVA types. The final model is presented in Table 3, and shows that the factors associated with RTA badgers and cattle sharing MLVA types were increasing ‘number of MLVA types during a breakdown’, ‘decreasing distance between hosts’, ‘the absence of a milk license’ and ‘fewer inwards movements’. No significant, biologically relevant interactions were identified. Herd size was correlated with the number of inwards movements (r = 0.63), was confounded with herd type and was therefore omitted from the final model; this did not impact Akaike information criterion (AIC) scores or model coefficients, notwithstanding that inwards movements was deemed to be the more relevant factor associated with the outcome.

3. Discussion

Spatial clustering in M. bovis molecular types, at various genetic and geographic scales, has previously identified co-localisation of infection between infected livestock and wildlife hosts [20,27,28,29,30,31]; however, little attention has been given to what the patterns in spatial distribution of M. bovis genetic types in NI reveal about the processes driving the epidemic in, and between, hosts. It is already understood that the M. bovis population in Northern Irish cattle herds is characterized by marked spatial structuring and spatial clusters of MVLA types at the herd level [18,19]. Here, we additionally show that the distribution of infection within clusters is not homogeneous, and that clusters consist of central foci, where 50% of infected herds lie within 35.5% of the cluster extent. This is consistent with “anchoring” influences in driving spatially restricted epidemics [32]; in Great Britain (GB), some 75% of infection was attributed to local spread [33]. Such processes act over relatively short distances, and can include infection from contiguous herds [26,34,35], infected wildlife [6,36,37], or the predominance of short-range, between-herd movements over longer-range movements, as observed in GB [38] and Ethiopia [39] (but not in Uruguay, where infection clusters change location year on year, suggesting long-distance spread of disease [40]). Spatial correlation has been reported in disease transmission coefficients at scales < 14 km, suggesting that a highly localised contact network is an important epidemiological driver of bTB [32]. However, we also identified pairwise distances of 167.3 km in herds infected with the same M. bovis MLVA type, and thus the spatial distribution of MLVA types also has an expansive element. Long-range cattle movements, or moving cattle between distal land-parcels, may drive the wider dissemination of infection [41,42], but our evidence is consistent with such processes being relatively less important than local factors investigated in driving the overall epidemic. This is in agreement with results from France, where spatial proximity to another infected herd was more strongly associated with bTB infection than inwards movements [43].
In our study, there was little compelling evidence of differences in the spatial dissemination of MLVA types between herds with and without a milk license, notwithstanding differences in inwards and outwards movements in both herd types. This observation may further reflect the diminished role of long-distance cattle movements in disease spread, compared to other sources. However, the data do tentatively indicate slightly less localised clustering in MLVA types in dairy herds, which could be explained by the fewer short-range, between-herd movements in dairy production. However, as yet the frequencies and Euclidian distances associated with the full cattle movement network in different herd types in NI are unknown [44]; whilst this study confirms that there are more inwards animal movements in non-dairy production, the distribution of trading distances between herds (as in Vernon, 2011 [38]) has not yet been derived.
While there was greater richness in MLVA types in cattle herds compared to RTA badgers, the cattle population of 1.6 million is over 40 times larger than the badger population, estimated at 33,500 individuals (95% CI 26,000–41,200) [14], and may therefore be able to harbor a larger, more diverse microbial population [30]. Whilst the drivers of this are not yet clear, super-spreading, or historical expansions of the M. bovis population, may be implicated [45]. Indeed, this within-herd MLVA type diversity was revealed in the cattle-cattle nearest neighbor analysis, where it appeared that the neighboring herd which did not share an MLVA type was closer than the herd that did. This observation arose because more than one MLVA type can be present in cattle herds, and thus nearby herds can both share MLVA types, as well as exhibit differences. The nearest neighbor analysis was unable to account for this, whereas the dissimilarity analysis could better model the MLVA type diversity in the cattle population.
Over 80% of the M. bovis MLVA types identified in the RTA badger isolates were also found in cattle, and furthermore the infection was spatially co-localised in both hosts. This is consistent with previous findings using more limited herd-level data from NI [20], and data from GB and the Republic of Ireland (ROI) [27,28,29]. This association was clear, despite accepted limitations of both the badger and cattle data; farmstead locations are unlikely to represent actual land-parcel (or herd) locations, badgers killed in RTAs may not represent the background badger population, and the RTA dataset is spatially biased to the south-east of NI, and under-sampled in the north-west [16]. Given this, detecting associations despite these disruptive factors means that the actual spatial associations may be even stronger than observed. However, this would require a more thorough systematic sample of M. bovis infection in the extant NI badger population. While there is some work ongoing to collect such data [46], that study is limited to a small geographic area of NI (~100 km2). The probability of RTA badgers and cattle herds sharing M. bovis MLVA types decreased by approximately 9% for every km between hosts, up to a 7 km cut-off. Not only does this further confirm the co-localisation of infection in both host systems, but the relatively small rate of change also alludes to the strength of localised influenced in maintaining the M. bovis population structure, and by extension, the epidemic. There was greater similarity when looking at the likelihood of RTA badgers and herds with milk licenses sharing MLVA types (decrease of 7% per km) compared to RTA badgers and herds without milk licenses (decrease of 11% per km). Whilst this difference is small, it suggests greater between-herd homogeneity in the M. bovis population in herds with milk licenses, compared to herds without, possibly due to inwards movements driving accumulations of within-herd MLVA type diversity in non-dairy herds [25].
RTA badgers were more likely to have at least one MLVA type in common with herds with milk licenses than herds without. We posit that this reflects animal management practices, as beef herds are likely to operate by purchasing larger volumes of animals, retaining these animals in the herd for only a short period before sending cattle onwards or to slaughter. Beef herds have been linked to the presence of multiple reactors [26] arising from the purchase of animals with undetected infection, possibly from many different geographical locations. Indeed, we identified the inwards movement of animals as a negative influence on whether RTA badgers and cattle share MLVA types. We hypothesise that high cattle turnover in these herds means that the local MLVA types are less likely to become established in the immediate environment, limiting the opportunity for a co-localisation of M. bovis genotypes between badgers and cattle. The findings may also suggest that inter-species transmission is perhaps not particularly efficient [9] given that a shared M. bovis epidemic is less likely to be observed in herds associated with a high animal turnover. Conversely, animals in dairy settings may be more likely to be exposed to local M. bovis genotypes, resulting in the repeated emergence of M. bovis genotypes via introduction from local sources, possibly exacerbated by within-herd amplification. The diagnostic single intradermal comparative cervical tuberculin test (SICCT) may also be less effective in dairy herds [47] and infected dairy cattle may be less likely to exhibit visible lesions post mortem [48]. This is highly suggestive of the fact that dairy herds may be at elevated risk of within-herd recrudescence of the same M. bovis genotypes compared to beef herds.

Limitations

As alluded to, the main limitation of this study is the nature of the RTA badger survey. The general limitations of these data are well acknowledged [16] but the bias in collection localities was presumed to have the most impact on these results. Nevertheless, this does not limit the utility of the dataset in making some inferences about the spatial relationships in MLVA types in, and between, the badger and cattle populations, with the caveat that a more representative RTA badger dataset would enable more robust inferences to be drawn. Ideally, a more comprehensive sampling of M. bovis genotypes in the badger population in NI would be undertaken which could permit the pathogen population structure to be derived in badger hosts. Furthermore, it is understood that cattle farms in NI are highly fragmented, and can consist of multiple, distal land parcels. The use of land parcel locations may provide a more accurate and precise indication of true herd locations and quantify the opportunity for herds to interact spatially with neighboring herds. This could shed more light on true spatial association of M. bovis genotypes in cattle herds. While co-localisation of molecular types between hosts, in various settings and at differing genetic and geographical scales, have been reported previously, our analyses extend our understanding of associated risk factors. Whilst molecular epidemiology using MLVA is being superseded by phylodynamics using whole-genome sequencing and modelling to investigate transmission dynamics [7], we argue for broad spatial scales, and for national monitoring there is still considerable value in using MLVA for bTB surveillance and epidemiological analysis.

4. Materials and Methods

4.1. Study Area

The area of NI is approximately 13,500 km2. The bTB programme is administered across 10 Divisional Veterinary Offices (DVOs) and 123 administrative patches. There are approximately 1.6 million cattle in NI, distributed throughout approximately 20,000 herds. This includes some 2500 dairy herds (313,549 cattle) and 14,000 beef herds (247,009 cattle), amongst others [49].

4.2. Study Data

4.2.1. M. bovis Molecular Typing Data

Whilst herd-level MLVA surveillance (MLVA typing on the first reactor) has been ongoing since 2003, from 2008, animal-level M. bovis MLVA typing is carried out on every SICCT reactor and lesioned animal identified at routine slaughter (LRS). MLVA analysis was carried out using established high resolution methods [18,19,50]. The eight M. bovis variable numbers of tandem repeats (VNTR) loci genotyped were MV2163B/QUB11B, MV4052/QUB26A, MV2461/ETRB, MV1955/Mtub21, MV1895/QUB1895, MV2165/ETRA, MV2163/QUB11A and MV3232/QUB3232. VNTR results were concatenated into a Multi-Locus VNTR Analysis (MLVA) string which constituted the molecular type of the isolate; this string was further simplified in a local laboratory nomenclature which reflected the previously assessed herd-level prevalence of MLVA types.

4.2.2. Cattle Data

Animal-level MLVA profiles of isolates were associated with anonymised breakdown-level data made available from the Animal and Plant Health Information System database (APHIS) [51], administered by the Department of Agriculture, Environment and Rural Affairs (DAERA). This enabled the determination of the number of M. bovis MLVA types present in each confirmed bTB breakdown; this dataset has been described in full elsewhere [25]. Additional relevant epidemiological variables included in these data were the breakdown start and end dates, presence of a milk license (dairy herds) or no milk license (non-dairy herds), herd size at the time of bTB breakdown, the number of inwards and outwards cattle movements in the year before breakdown, and the herd DVO. The geo-referencing of registered homestead locations (here referred to as herd locations) were available in the form of the first four digits of the six figure Irish grid reference, which provides reasonable estimations of cattle herd density and distribution. These data are available in an anonymised format in Supplementary Data 1. Actual herd numbers have been removed, and herd locations were “jittered” by between −999 m and +999 m to protect herds from being identified.

4.2.3. Badger Data

The RTA badger dataset has been described previously [16,20,52]. Briefly, from 1998, the carcasses of badgers suspected to have died from accidental causes (e.g., road traffic accidents, RTAs) were collected by a wildlife officer from the Department of Agriculture, Environment and Rural Affairs (DAERA). Badger carcasses were checked for the presence of visible lesions consistent with tuberculosis, and defined tissues and bodily fluid samples were also collected for bacterial culture. Culture-confirmed M. bovis underwent further MLVA analysis by the Agri-Food and Biosciences Institute (AFBI) to determine the bTB spoligotype and MLVA genotype of the isolate. The geo-referenced collection location for RTA badgers was recorded to within 100 m of the actual location. This study included only badgers collected after 2008 to align temporally with the cattle data. These data are available in an anonymised format in Supplementary Data 2. Badger collection localities were “jittered” by between −999 m and +999 m.

4.3. Assessment of MLVA Clustering within the Cattle Population

To visualise the extent of spatial clustering of MLVA types, the distribution of the twelve most common MLVA types were plotted in geographic space. The presence of spatial clustering in MLVA in cattle herds was confirmed by a permutation analysis. Firstly, the median Euclidean distance between herds infected with a given MLVA type was derived. Next, a random sample of herds was selected from the cattle herd population, with the sample size equal to the number of herds infected with the MLVA type of interest. The median Euclidean distance between the herds in this sample was then calculated. This process was repeated 999 times to generate a sampling distribution of distances. Finally, the actual median distance between herds infected with each MLVA type was compared to those derived from the sampling distribution using Wilcoxon signed rank tests.
Next, a pairwise Euclidean distance matrix between all herds infected with the same MVLA type was generated. The cumulative frequency of these distances was used to investigate the spatial spread of herds within clusters; if the majority of pairwise distances lay below the distance represented by the 50th percentile, this indicated spatial clustering. However, if the pairwise distances were broadly distributed (i.e., 50% of pairwise distances equal to or greater than the 50th percentile), this suggested instead that infected herds are widely distributed across space. This process was also conducted separately for herds with and without milk licenses.

4.4. Intra- and Interspecific Nearest Neighbour Analysis

Following similar methods to Trewby (2016) [20] and Abernethy et al. (2011) [15], the Euclidian Nearest Neighbour (NN) distance was calculated between each RTA badger to the closest RTA badger with the same M. bovis MLVA type, and to the closest RTA badger with a different M. bovis MLVA type. Only badgers within 7 km of each other were included, as this represents the approximate 95 percentile of the dispersal movement kernel of badgers in the ROI [53]. Furthermore, the RTA collection dates had to fall within two years of each other; it is reasonable to assume that a two year window will adequately capture co-localisation of infection without introducing ambiguity from associating entities across longer temporal windows. Similar NN distance measures were calculated for RTA badger-cattle herd pairs, repeated separately for only those herds with a milk license, and those without a milk license. Again, only badger and cattle hosts within 7 km of each other were included, and the RTA collection date and herd breakdown period had to be within two years of each other. Finally, NN distances were calculated between cattle-cattle pairs, under the same criteria. The 7 km limit was considered acceptable for cattle-cattle associations in this instance, as we do not presently have information on the Euclidian distances associated with between-herd cattle movements. The null hypothesis was that distances between hosts which share M. bovis MLVA types were not significantly different to distances between hosts that do not. This was tested using paired Wilcoxon signed rank tests, and p was established at ≤0.05. To interrogate any limitations in the resolution of spatial data (i.e., the RTA badger and cattle herd co-ordinates were each subject to a 100 m error), a sensitivity analysis was conducted whereby the analyses were re-run 100 times, with values between 1m and 99 m added to, or subtracted from, each of the cattle and badger latitude and longitude co-ordinates.

4.5. Distance Based Similarlity Analysis

Using the approach established by Goodchild et al (2012) [27], the Odds Ratio (OR) of an M. bovis MLVA type match was calculated as a function of distance between hosts that yielded MLVA-typed M. bovis, up to a distance of 7 km. The two-year temporal association was again applied to this analysis. These data were used in the construction of logistic Generalised Linear Mixed Models (GLMMs), with a binary outcome indicating whether or not the M. bovis MLVA types matched. The single explanatory fixed-variable was distance, and herd DVO was allowed to vary with a random intercept.

4.6. Determining the Factors Associated with RTA Badgers and Nearby Cattle Herds Sharing MLVA Types

We modelled factors associated with RTA badgers and nearby cattle sharing the same MLVA type. The outcome of interest was ‘whether any cow in a herd within a 7 km radius of an RTA badger shared the badger MLVA type’, was entered as a binary variable (1 = yes, 0 = no), and was modelled via a binomial GLMM. This analysis was again limited to breakdowns occurring within two years before or after an RTA collection date. Explanatory variables were: the presence or absence of a milk license, the distance between an RTA badger and registered cattle homestead, herd size, the number of inwards and outwards cattle movements, and the number of MLVA types isolated from a bTB breakdown. Initial univariable analysis involved visual assessments of each predictor, including Cleveland dot plots and boxplots, and fitting loess curves to assess linearity in the logit. Covariates were then assessed for co-linearity using both multi-panel scatterplots and correlation values; variables with a correlation coefficient greater than 0.5 or less than −0.5 were considered for removal, with the aim of retaining the most biologically relevant predictor(s). DVO was allowed to vary with a random intercept. The final model was arrived at via a backwards stepwise routine [54], and the impact of variable removal at each stage was assessed by comparing model AIC values, examining changes in model coefficients and assessing confounding. Where potential confounding was identified, it was investigated by running separate analyses on the suspected confounders. The influence of outliers and influential points was assessed by re-running models with potential influential points removed, and comparing the model coefficients.
All data processing and analyses were carried out using Microsoft Excel and R version 3.4.4 (R Core Team 2014). The packages rgdal [55], rgeos [56] and ggplot2 [57] were used to create maps and figures, models were built using lme4 [58], and dplyr [59] was used for data handling.

Supplementary Materials

The following are available online at https://www.mdpi.com/2076-0817/9/7/592/s1; Supplementary Data 1 and Supplementary Data 2; Supplementary Material 1, Figure S1. The spatial distribution of MLVA type 001, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S2. The spatial distribution of MLVA type 003, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S3. The spatial distribution of MLVA type 004, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S4. The spatial distribution of MLVA type 005, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S5. The spatial distribution of MLVA type 006, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S6. The spatial distribution of MLVA type 007, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S7. The spatial distribution of MLVA type 009, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S8. The spatial distribution of MLVA type 010, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S9. The spatial distribution of MLVA type 027, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S10. The spatial distribution of MLVA type 117, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once; Figure S11. The spatial distribution of MLVA type 122, shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once. Supplementary Material 2: Figure S12. The Cumulative Density Function (CDF) of MLVA type 001. Figure S13. The Cumulative Density Function (CDF) of MLVA type 002. Figure S14. The Cumulative Density Function (CDF) of MLVA type 003. Figure S15. The Cumulative Density Function (CDF) of MLVA type 004. Figure S16. The Cumulative Density Function (CDF) of MLVA type 005. Figure S17. The Cumulative Density Function (CDF) of MLVA type 006. Figure S18. The Cumulative Density Function (CDF) of MLVA type 007. Figure S19. The Cumulative Density Function (CDF) of MLVA type 009. Figure S20. The Cumulative Density Function (CDF) of MLVA type 010. Figure S21. The Cumulative Density Function (CDF) of MLVA type 027. Figure S22. The Cumulative Density Function (CDF) of MLVA type 117. Figure S23. The Cumulative Density Function (CDF) of MLVA type 122.

Author Contributions

Conceptualization, A.A., R.S. and A.W.B.; Data curation, J.G.; Formal analysis, G.M.; Funding acquisition, R.S. and A.W.B.; Methodology, G.M., A.A., C.M., E.P., R.S. and A.W.B.; Project administration, R.S. and A.W.B.; Resources, R.K., C.M. and E.P.; Software, G.M. and J.G.; Supervision, A.A., R.S. and A.W.B.; Writing—original draft, G.M., A.A., R.S. and A.W.B.; Writing—review & editing, G.M., A.A., R.S. and A.W.B. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Department of Agriculture, Environment and Rural Affairs for project 16-3-06.

Acknowledgments

The authors wish to extend particular gratitude towards the following individuals: Hannah Trewby, whose PhD research inspired some of the methodology, Emily Courcier and Maria O’Hagan who were associated with managing the DAERA badger RTA data, and Darrell Abernethy who previously established methods for investigating RTA badger-cattle relationships in NI. The authors also wish to extend thanks and acknowledgement to all of the AFBI staff involved in bovine tuberculosis histology, pathology, bacteriology, molecular microbiology, case confirmation and MLVA typing, along with the Epidemiology, Molecular biology and Immunology cluster (https://twitter.com/AFBI_EMbI) for technical support and discussion. We are also thankful to the DAERA Veterinary Service and abattoir staff who deliver the NI bTB eradication programme, those staff who maintain and manage the DAERA APHIS database, and all staff involved the DAERA badger RTA scheme.

Conflicts of Interest

The senior author, Dr Andrew W. Byrne, is a member of the Pathogens journal editorial board. Aside from this, the authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References

  1. Gortázar, C.; Ferroglio, E.; Höfle, U.; Frölich, K.; Vicente, J. Diseases shared between wildlife and livestock: A European perspective. Eur. J. Wildl. Res. 2007, 53, 241–256. [Google Scholar] [CrossRef]
  2. De La Rua-Domenech, R.; Goodchild, A.; Vordermeier, H.; Hewinson, R.; Christiansen, K.; Clifton-Hadley, R. Ante mortem diagnosis of tuberculosis in cattle: A review of the tuberculin tests, γ-interferon assay and other ancillary diagnostic techniques. Res. Vet. Sci. 2006, 81, 190–210. [Google Scholar] [CrossRef] [PubMed]
  3. Morris, R.; Pfeiffer, D.U.; Jackson, R. The epidemiology of Mycobacterium bovis infections. Vet. Microbiol. 1994, 40, 153–177. [Google Scholar] [CrossRef]
  4. Mathews, F.; Macdonald, D.W.; Taylor, G.M.; Gelling, M.; Norman, R.; Honess, P.E.; Foster, R.; Gower, C.M.; Varley, S.; Harris, A.; et al. Bovine tuberculosis (Mycobacterium bovis) in British farmland wildlife: The importance to agriculture. Proc. R. Soc. B Boil. Sci. 2005, 273, 357–365. [Google Scholar] [CrossRef] [Green Version]
  5. Gallagher, J.; Clifton-Hadley, R. Tuberculosis in badgers; a review of the disease and its significance for other animals. Res. Vet. Sci. 2000, 69, 203–217. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  6. Byrne, A.; Kenny, K.; Fogarty, U.; O’Keeffe, J.; More, S.J.; McGrath, G.; Teeling, M.; Martin, S.; Dohoo, I. Spatial and temporal analyses of metrics of tuberculosis infection in badgers (Meles meles) from the Republic of Ireland: Trends in apparent prevalence. Prev. Vet. Med. 2015, 122, 345–354. [Google Scholar] [CrossRef]
  7. Crispell, J.; Benton, C.H.; Balaz, D.; De Maio, N.; Ahkmetova, A.; Allen, A.; Biek, R.; Presho, E.L.; Dale, J.; Hewinson, G.; et al. Combining genomics and epidemiology to analyse bi-directional transmission of Mycobacterium bovis in a multi-host system. eLife 2019, 8, 45833. [Google Scholar] [CrossRef]
  8. Gortázar, C.; Delahay, R.J.; McDonald, R.A.; Boadella, M.; Wilson, G.J.; Gavier-Widen, D.; Acevedo, P. The status of tuberculosis in European wild mammals. Mammal Rev. 2011, 42, 193–206. [Google Scholar] [CrossRef]
  9. Böhm, M.; Hutchings, M.; White, P.C.L. Contact networks in a wildlife-livestock host community: Identifying high-risk individuals in the transmission of bovine TB among badgers and cattle. PLoS ONE 2009, 4, e5016. [Google Scholar] [CrossRef]
  10. Campbell, E.L.; Byrne, A.; Menzies, F.D.; McBride, K.R.; McCormick, C.M.; Scantlebury, M.; Reid, N. Interspecific visitation of cattle and badgers to fomites: A transmission risk for bovine tuberculosis? Ecol. Evol. 2019, 9, 8479–8489. [Google Scholar] [CrossRef]
  11. Barasona, J.; Vicente, J.; Díez-Delgado, I.; Aznar, J.; Gortázar, C.; Torres, M. Environmental presence of mycobacterium tuberculosis complex in aggregation points at the wildlife/livestock interface. Transbound. Emerg. Dis. 2016, 64, 1148–1158. [Google Scholar] [CrossRef] [PubMed]
  12. NIAO. The control of Bovine Tuberculosis in Northern Ireland; Northern Ireland Audit Office: Belfast, Northern Ireland, 2018.
  13. Allen, R.A.; Skuce, R.A.; Byrne, A.W. Bovine tuberculosis in Britain and Ireland a perfect storm? The confluence of potential ecological and epidemiological impediments to controlling a chronic infectious disease. Front. Vet. Sci. 2018, 5, 109. [Google Scholar] [CrossRef] [PubMed]
  14. Reid, N.; Etherington, T.R.; Wilson, G.J.; Montgomery, W.I.; McDonald, R.A. Monitoring and population estimation of the European badger Meles meles in Northern Ireland. Wildl. Biol. 2012, 18, 46–57. [Google Scholar] [CrossRef] [Green Version]
  15. Abernethy, D.A.; Walton, E.; Menzies, F.; Courtier, E.; Robinson, P. Mycobacterium bovis surveillance in European badgers (Meles meles) killed by vehicles in Northern Ireland: An epidemiological evaluation. Épidémiologie et Santé Animale 2011, 216–218. Available online: https://pdfs.semanticscholar.org/7313/a7149c10fed40e4432c8fb9d8c0653445e25.pdf (accessed on 30 June 2020).
  16. Courcier, E.A.; Menzies, F.D.; Strain, S.A.J.; Skuce, R.A.; Robinson, P.A.; Patterson, I.A.P.; McBride, K.R.; McCormick, C.M.; Walton, E.; McDowell, S.W.J.; et al. Monitoring Mycobacterium bovis in Eurasian badgers (Meles meles) killed by vehicles in Northern Ireland between 1998 and 2011. Vet. Rec. 2017, 182, 259. [Google Scholar] [CrossRef]
  17. Kamerbeek, J.; Schouls, L.; Kolk, A.; Van Agterveld, M.; Van Soolingen, D.; Kuijper, S.; Bunschoten, A.; Molhuizen, H.; Shaw, R.; Goyal, M.; et al. Simultaneous detection and strain differentiation of Mycobacterium tuberculosis for diagnosis and epidemiology. J. Clin. Microbiol. 1997, 35, 907–914. [Google Scholar] [CrossRef] [Green Version]
  18. Skuce, R.; Mallon, T.R.; McBride, S.H.; Couzens, C.; Gordon, A.W.; McDowell, S.; Bsc; McCormick, C.M.; Hnc; Pgdip(Comp), H.B.; et al. Mycobacterium bovis genotypes in Northern Ireland: Herd-level surveillance (2003 to 2008). Vet. Rec. 2010, 167, 684–689. [Google Scholar] [CrossRef]
  19. Skuce, R.; Breadon, E.; Allen, A.; Milne, G.; McCormick, C.; Hughes, C.; Rutherford, D.; Smith, G.; Thompson, S.; Graham, J.; et al. Longitudinal dynamics of herd-level Mycobacterium bovis MLVA type surveillance in cattle in Northern Ireland 2003–2016. Infect. Genet. Evol. 2020, 79, 104131. [Google Scholar] [CrossRef]
  20. Trewby, H. The genetic and Spatial Epidemiology of Bovine Tuberculosis in the UK: From Molecular Typing to Bacterial Whole Genome Sequencing 2016; University of Glasgow: Glasgow, UK, 2016. [Google Scholar]
  21. Trewby, H.; Wright, D.; Breadon, E.L.; Lycett, S.J.; Mallon, T.R.; McCormick, C.; Johnson, P.; Orton, R.J.; Allen, A.R.; Galbraith, J.; et al. Use of bacterial whole-genome sequencing to investigate local persistence and spread in bovine tuberculosis. Epidemics 2016, 14, 26–35. [Google Scholar] [CrossRef] [Green Version]
  22. Fielding, H.R.; McKinley, T.J.; Silk, M.J.; Delahay, R.J.; McDonald, R.A. Contact chains of cattle farms in Great Britain. R. Soc. Open Sci. 2019, 6, 180719. [Google Scholar] [CrossRef] [Green Version]
  23. Doyle, L.; Courcier, E.; Gordon, A.; O’Hagan, M.; Menzies, F. Bovine tuberculosis in Northern Ireland: Risk factors associated with duration and recurrence of chronic herd breakdowns. Prev. Vet. Med. 2016, 131, 1–7. [Google Scholar] [CrossRef] [PubMed]
  24. Menzies, D.F.; Neill, S.D. Cattle-to-cattle transmission of bovine tuberculosis. Vet. J. 2000, 160, 92–106. [Google Scholar] [CrossRef]
  25. Milne, G.; Graham, J.; Allen, A.; McCormick, C.; Presho, E.; Skuce, R.; Byrne, A. Variation in Mycobacterium bovis genetic richness suggests that inwards cattle movements are a more important source of infection in beef herds than in dairy herds. BMC Microbiol. 2019, 19, 154. [Google Scholar] [CrossRef] [PubMed]
  26. O’Hagan, M.; Matthews, D.; Laird, C.; McDowell, S. Herd-level risk factors for bovine tuberculosis and adoption of related biosecurity measures in Northern Ireland: A case-control study. Vet. J. 2016, 213, 26–32. [Google Scholar] [CrossRef] [Green Version]
  27. Goodchild, A.V.; Watkins, G.H.; Sayers, A.R.; Jones, J.R.; Clifton-Hadley, R.S. Geographical association between the genotype of bovine tuberculosis in found dead badgers and in cattle herds. Vet. Rec. 2012, 170, 259. [Google Scholar] [CrossRef]
  28. Woodroffe, R.; Donnelly, C.A.; Johnston, W.T.; Bourne, F.J.; Cheeseman, C.L.; Clifton-Hadley, R.S.; Cox, D.R.; Gettinby, G.; Hewinson, R.G.; Le Fevre, A.M.; et al. Spatial association of Mycobacterium bovis infection in cattle and badgers Meles meles. J. Appl. Ecol. 2005, 42, 852–862. [Google Scholar] [CrossRef]
  29. Olea-Popelka, F.; Flynn, O.; Costello, E.; McGrath, G.; Collins, J.; O’Keeffe, J.; Kelton, D.; Berke, O.; Martin, S. Spatial relationship between Mycobacterium bovis strains in cattle and badgers in four areas in Ireland. Prev. Vet. Med. 2005, 71, 57–70. [Google Scholar] [CrossRef]
  30. Rodríguez, L.D.; Romero, B.; Bezos, J.; De Juan, L.; Álvarez, J.; Castellanos, E.; Moya, N.; Lozano, F.; Gonzalez, S.; Sáez-Llorente, J.L.; et al. High spoligotype diversity within a Mycobacterium bovis population: Clues to understanding the demography of the pathogen in Europe. Vet. Microbiol. 2010, 141, 89–95. [Google Scholar] [CrossRef]
  31. Romero, B.; Aranaz, A.; Sandoval, A.; Álvarez, J.; De Juan, L.; Bezos, J.; Sánchez, C.; Galka, M.; Fernández, P.; Mateos, A.; et al. Persistence and molecular evolution of Mycobacterium bovis population from cattle and wildlife in Doñana National Park revealed by genotype variation. Vet. Microbiol. 2008, 132, 87–95. [Google Scholar] [CrossRef]
  32. De La Cruz, M.L.; Pérez, A.; Bezos, J.; Pages, E.; Casal, C.; Carpintero, J.; Romero, B.; Dominguez, L.; Barker, C.M.; Díaz, R.; et al. Spatial dynamics of bovine tuberculosis in the autonomous community of Madrid, Spain (2010–2012). PLoS ONE 2014, 9, e115632. [Google Scholar] [CrossRef] [Green Version]
  33. Green, D.; Kiss, I.Z.; Mitchell, A.P.; Kao, R. Estimates for local and movement-based transmission of bovine tuberculosis in British cattle. Proc. R. Soc. B Boil. Sci. 2008, 275, 1001–1005. [Google Scholar] [CrossRef] [PubMed]
  34. White, P.W.; Martin, S.W.; De Jong, M.C.; O’Keeffe, J.; More, S.J.; Frankena, K. The importance of ‘neighbourhood’ in the persistence of bovine tuberculosis in Irish cattle herds. Prev. Vet. Med. 2013, 110, 346–355. [Google Scholar] [CrossRef] [PubMed]
  35. Guta, S.; Casal, J.; Napp, S.; Saez, J.L.; Garcia-Saenz, A.; De Val, B.P.; Allepuz, A. Epidemiological investigation of bovine tuberculosis herd breakdowns in Spain 2009/2011. PLoS ONE 2014, 9, e104383. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  36. Byrne, A.W.; White, P.W.; McGrath, G.; James, O.; Martin, S.W. Risk of tuberculosis cattle herd breakdowns in Ireland: Effects of badger culling effort, density and historic large-scale interventions. Vet. Res. 2014, 45, 109. [Google Scholar] [CrossRef]
  37. Denny, G.O.; Wilesmith, J.W. Bovine tuberculosis in Northern Ireland: A case-control study of herd risk factors. Vet. Rec. 1999, 144, 305–310. [Google Scholar] [CrossRef]
  38. Vernon, M.C. Demographics of cattle movements in the UK. BMC Vet. Res. 2011, 7, 31. [Google Scholar] [CrossRef] [Green Version]
  39. Mekonnen, G.A.; Ameni, G.; Wood, J.; The ETHICOBOTS consortium; Berg, S.; Conlan, A. Network analysis of dairy cattle movement and associations with bovine tuberculosis spread and control in emerging dairy belts of Ethiopia. BMC Vet. Res. 2019, 15, 262. [Google Scholar] [CrossRef]
  40. Picasso, C.; Alvarez, J.; VanderWaal, K.L.; Fernandez, F.; Gil, A.; Wells, S.J.; Perez, A. Epidemiological investigation of bovine tuberculosis outbreaks in Uruguay (2011–2013). Prev. Vet. Med. 2017, 138, 156–161. [Google Scholar] [CrossRef] [Green Version]
  41. Gopal, R.; Goodchild, A.; Hewinson, G.; Domenech, R.D.L.R.; Clifton-Hadley, R. Introduction of bovine tuberculosis to north-east England by bought-in cattle. Vet. Rec. 2006, 159, 265–271. [Google Scholar] [CrossRef]
  42. Adkin, A.; Brouwer, A.; Downs, S.; Kelly, L. Assessing the impact of a cattle risk-based trading scheme on the movement of bovine tuberculosis infected animals in England and Wales. Prev. Vet. Med. 2016, 123, 23–31. [Google Scholar] [CrossRef] [Green Version]
  43. Palisson, A.; Courcoul, A.; Durand, B. Role of Cattle Movements in Bovine Tuberculosis Spread in France between 2005 and 2014. PLoS ONE 2016, 11, e0152578. [Google Scholar] [CrossRef]
  44. Brown, E.; Marshall, A.H.; Mitchell, H.J.; Byrne, A. Cattle movements in Northern Ireland form a robust network: Implications for disease management. Prev. Vet. Med. 2019, 170, 104740. [Google Scholar] [CrossRef]
  45. Trewby, H.; Wright, D.M.; Skuce, R.A.; McCormick, C.; Mallon, T.R.; Presho, E.L.; Kao, R.R.; Haydon, D.T.; Biek, R. Relative abundance of Mycobacterium bovis molecular types in cattle: A simulation study of potential epidemiological drivers. BMC Vet. Res. 2017, 13, 268. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  46. DAERA. The Test and Vaccinate or Remove (TVR) Wildlife Intervention Research Project. Year 4 Report 2017; DAERA: Belfast, UK, 2017.
  47. Lahuerta-Marin, A.; Milne, G.; McNair, J.; Skuce, R.; McBride, S.; Menzies, F.; McDowell, S.; Byrne, A.; Handel, I.; Bronsvoort, B.D.C. Bayesian latent class estimation of sensitivity and specificity parameters of diagnostic tests for bovine tuberculosis in chronically infected herds in Northern Ireland. Vet. J. 2018, 238, 15–21. [Google Scholar] [CrossRef] [PubMed]
  48. Byrne, A.; Graham, J.; Brown, C.; Donaghy, A.; Guelbenzu, M.; McNair, J.; Skuce, R.A.; Allen, A.R.; McDowell, S.W. Bovine tuberculosis visible lesions in cattle culled during herd breakdowns: The effects of individual characteristics, trade movement and co-infection. BMC Vet. Res. 2017, 13, 1–14. [Google Scholar] [CrossRef]
  49. DAERA. Final results of the June agricultural census 2019. In 2020, Statistics and Analytical Services Branch, Policy, Economics and Statistics Division Belfast; DAERA: Belfast, UK, 2020. [Google Scholar]
  50. Skuce, R.A.; McDowell, S.; Mallon, T.R.; Luke, B.; Breadon, E.L.; Lagan, P.L.; McCormick, C.M.; McBride, S.H.; Pollock, J.M. Discrimination of isolates of Mycobacterium bovis in Northern Ireland on the basis of variable numbers of tandem repeats (VNTRs). Vet. Rec. 2005, 157, 501–504. [Google Scholar] [CrossRef] [PubMed]
  51. Houston, R. A computerised database system for bovine traceability. Rev. Sci. Tech. Off. Int. Epizoot. 2001, 20, 652–661. [Google Scholar] [CrossRef]
  52. Abernethy, D.A.; Denny, G.; Pfeiffer, D.U.; Wrigley, K. Survey for mycobacterium bovis infection in road-traffic-accident badgers in Northern Ireland. In Proceedings of the 10th International Symposium on Veterinary Epidemiology and Economics, Vina del Mar, Chile, 17–21 November 2003. [Google Scholar]
  53. Byrne, A.; Quinn, J.L.; O’Keeffe, J.; Green, S.; Sleeman, D.P.; Martin, S.W.; Davenport, J. Large-scale movements in European badgers: Has the tail of the movement kernel been underestimated? J. Anim. Ecol. 2014, 83, 991–1001. [Google Scholar] [CrossRef] [Green Version]
  54. Zuur, A.F.; Ieno, E.N.; Walker, N.; Saveliev, A.A.; Smith, G.M. Mixed Effects Models and Extensions in Ecology with R; Springer Science and Business Media LLC: Berlin/Heidelberg, Germany, 2009. [Google Scholar]
  55. Bivand, R.; Keitt, T.; Rowlingson, B.; Pebesma, E.D.Z.E.R. rgdal: Bindings for the Geospatial Data Abstraction Library. 2010. Available online: https://www.researchgate.net/publication/247474582_rgdal_Bindings_for_the_Geospatial_Data_Abstraction_Library (accessed on 30 June 2020).
  56. Bivand, R.; Rundel, C. rgeos: Interface to Geometry Engine—Open Source (’GEOS’). 2017. Available online: https://rdrr.io/cran/rgeos/ (accessed on 30 June 2020).
  57. Wickham, H. ggplot2: Elegant Graphics for Data Analysis; Springer: New York, NY, USA, 2009. [Google Scholar]
  58. Bates, D.; Mächler, M.; Bolker, B.; Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 2015, 67, 1–48. [Google Scholar] [CrossRef]
  59. Wickham, H.; Francois, R.; Henry, L.; Müller, K. dplyr: A Grammar of Data Manipulation. 2017. Available online: https://dplyr.tidyverse.org/reference/dplyr-package.html (accessed on 30 June 2020).
Pathogens 09 00592 g001 550
Figure 1. The spatial distribution of study herds with (green), and without (purple) a milk license in NI.
Figure 1. The spatial distribution of study herds with (green), and without (purple) a milk license in NI.
Pathogens 09 00592 g001
Pathogens 09 00592 g002 550
Figure 2. The spatial distribution of the twelve most common Multiple Locus Variable Analysis (MLVA) types in (a) cattle, (b) road-traffic accident (RTA) badgers, and (c) MLVA types common to both cattle and RTA badgers, where circles with a black outline represent isolates from RTA badgers, and circles without a black outline indicate cattle herds.
Figure 2. The spatial distribution of the twelve most common Multiple Locus Variable Analysis (MLVA) types in (a) cattle, (b) road-traffic accident (RTA) badgers, and (c) MLVA types common to both cattle and RTA badgers, where circles with a black outline represent isolates from RTA badgers, and circles without a black outline indicate cattle herds.
Pathogens 09 00592 g002
Pathogens 09 00592 g003 550
Figure 3. The spatial distribution of the most common MLVA type (002), shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once, and orange dots represent herds from which the MLVA type was never isolated.
Figure 3. The spatial distribution of the most common MLVA type (002), shown in (a) all herds, (b) herds with a milk license, and (c) herds without a milk license. Green dots represent herds from which the MLVA type was isolated at least once, and orange dots represent herds from which the MLVA type was never isolated.
Pathogens 09 00592 g003
Pathogens 09 00592 g004 550
Figure 4. Visualization of the permutation analysis for all herds (black), herds with a milk license (red) and herds without a milk license (blue). The histograms show the sampling distribution of median distances between all herds infected with each of the twelve MLVA types (a–l), generated from 999 random permutations of the data. Vertical lines show the actual observed median distances between herds infected with each MLVA type. A vertical line intersecting with a histogram of the same colour indicated that the observed median distance between herds infected with the same MLVA type was not distinguishable from the median distances in a random selection of herds. The observed distances between all herds, dairy herds and non-dairy herds infected with types 001 and 006 were in very close agreement, and therefore appear as single lines on the figure.
Figure 4. Visualization of the permutation analysis for all herds (black), herds with a milk license (red) and herds without a milk license (blue). The histograms show the sampling distribution of median distances between all herds infected with each of the twelve MLVA types (a–l), generated from 999 random permutations of the data. Vertical lines show the actual observed median distances between herds infected with each MLVA type. A vertical line intersecting with a histogram of the same colour indicated that the observed median distance between herds infected with the same MLVA type was not distinguishable from the median distances in a random selection of herds. The observed distances between all herds, dairy herds and non-dairy herds infected with types 001 and 006 were in very close agreement, and therefore appear as single lines on the figure.
Pathogens 09 00592 g004
Pathogens 09 00592 g005 550
Figure 5. The relationship between probability of an MLVA match between two hosts as a function of distance in the (a) badger-badger context, (b) badger-cattle herds context, and (c), cattle-cattle herds context. Black lines show the fitted logistic regression, and the grey areas illustrate the 95% Confidence Intervals. The y-axis shows the probability that two hosts share a matching MLVA type; p(match), as the distance between the hosts increases, shown on the x-axis.
Figure 5. The relationship between probability of an MLVA match between two hosts as a function of distance in the (a) badger-badger context, (b) badger-cattle herds context, and (c), cattle-cattle herds context. Black lines show the fitted logistic regression, and the grey areas illustrate the 95% Confidence Intervals. The y-axis shows the probability that two hosts share a matching MLVA type; p(match), as the distance between the hosts increases, shown on the x-axis.
Pathogens 09 00592 g005
Table
Table 1. The distances associated with the 25th, 50th, 75th and 100th percentiles from a frequency distribution of pairwise distances between herds infected with a given MLVA type. The values of the 25th, 50th and 75th percentiles are shown as a percentage of the maximum extent (100th percentile).
Table 1. The distances associated with the 25th, 50th, 75th and 100th percentiles from a frequency distribution of pairwise distances between herds infected with a given MLVA type. The values of the 25th, 50th and 75th percentiles are shown as a percentage of the maximum extent (100th percentile).
MLVADatasetTotal Infected Herds25th Percentile (km)50th Percentile (km)75th Percentile (km)100th Percentile (km)
001All114128.7 km (19.3%)52.7 km (35.5%)77.6 km (52.2%)148.5 km
001Dairy34729.2 km (19.2%)52.8 km (34.7%)76.4 km (50.1%)152.4 km
001Non Dairy79427.8 km (16.9%)52.5 km (31.9%)78.2 km (47.5%)164.5 km
002All188530.9 km (18.5%)52 km (31.3%)76.5 km (46.0%)166.4 km
002Dairy54528.3 km (18.9%)54.1 km (36.1%)80.9 km (54.0%)149.9 km
002Non Dairy134031.1 km (18.7%)50.8 km (30.6%)74.5 km (44.8%)166.4 km
003All46014.9 km (10.0%)29 km (19.4%)51.2 km (34.2%)149.6 km
003Dairy11817.4 km (13.6%)32.2 km (25.1%)51.5 km (40.1%)128.4 km
003Non Dairy34213.5 km (9.1%)27 km (18.1%)50.7 km (33.9%)149.6 km
004All73626.1 km (16.7%)45.6 km (29.1%)64.2 km (41.0%)156.5 km
004Dairy20427.6 km (19.7%)48.1 km (34.3%)68.3 km (48.7%)140.5 km
004Non Dairy53225.4 km (16.3%)44.4 km (28.6%)62.7 km (40.4%)155.3 km
005All100123.8 km (14.3%)51.3 km (30.7%)79.3 km (47.4%)167.3 km
005Dairy26617.9 km (12.0%)41.2 km (27.4%)77.6 km (51.7%)150.1 km
005Non Dairy73526.1 km (15.6%)53.4 km (31.9%)79.8 km (47.7%)167.3 km
006All81013.9 km (9.5%)23.6 km (16.1%)40.9 km (27.9%)146.4 km
006Dairy24112.9 km (8.8%)23.2 km (15.9%)44.7 km (30.5%)146.4 km
006Non Dairy56914 km (10.3%)23.5 km (17.3%)39.6 km (29.2%)135.8 km
007All38819.8 km (13.8%)39.6 km (27.6%)65.6 km (45.6%)143.8 km
007Dairy12014.3 km (11.0%)27.6 km (21.3%)57 km (44.1%)129.4 km
007Non Dairy26822.7 km (15.8%)43.7 km (30.4%)67.9 km (47.2%)143.8 km
009All35517.1 km (10.5%)27.1 km (16.7%)44.9 km (27.7%)162.0 km
009Dairy14012 km (9.4%)20.4 km (15.9%)30.6 km (23.9%)128.1 km
009Non Dairy21519.3 km (11.9%)31.7 km (19.5%)52.5 km (32.4%)162.0 km
122All1578.5 km (5.8%)16.3 km (11.1%)32.1 km (21.7%)147.5 km
122Dairy606.6 km (8.5%)11.9 km (15.2%)22.7 km (29.1%)77.9 km
122Non Dairy9710.4 km (7.1%)19 km (12.9%)36.7 km (24.9%)147.5 km
010All14814.2 km (10.2%)39.7 km (28.6%)59.6 km (42.9%)138.9 km
010Dairy4615 km (12.0%)43.7 km (35%)59.9 km (48.1%)124.7 km
010Non Dairy10212.9 km (9.7%)37.7 km (28.3%)59.3 km (44.4%)133.5 km
027All15718.7 km (11.4%)46.9 km (28.6%)88.5 km (54.1%)163.8 km
027Dairy2627.8 km (20.3%)61.2 km (44.8%)84.1 km (61.6%)136.4 km
027Non Dairy13113.8 km (9.2%)38.9 km (25.9%)84.1 km (56.1%)150.0 km
117All11615 km (10.0%)26.2 km (17.5%)38.8 km (25.9%)149.7 km
117Dairy3712.3 km (9.0%)23.7 km (17.3%)37.4 km (27.3%)136.8 km
117Non Dairy7914.6 km (9.8%)25.2 km (17.0%)39.5 km (26.6%)148.5 km
Table
Table 2. Nearest-neighbour (NN) distances for pairs of RTA badgers and cattle herds which share MLVA types, and which do not share MLVA types.
Table 2. Nearest-neighbour (NN) distances for pairs of RTA badgers and cattle herds which share MLVA types, and which do not share MLVA types.
NN Distance for Hosts Which Share an MLVA TypeNN Distance for Hosts Which Do Not Share an MLVA Type
Median (km)Q1 (km)Q4 (km)Median (km)Q1 (km)Q4 (km)Difference in Medians (km)
Badger-Badger2.441.224.043.331.945.220.89
Badger-Cattle (all herds)0.820.501.641.490.832.250.67
Badger-Cattle (non-dairy)1.140.611.942.551.623.741.41
Badger-Cattle (dairy)1.700.852.732.881.854.031.18
Cattle-Cattle (all herds)1.120.612.320.820.511.270.29
Cattle-Cattle (dairy)1.600.852.951.390.822.020.21
Cattle-Cattle (non-dairy)1.280.672.660.930.581.500.35
Table
Table 3. Model coefficients for the factors associated with RTA badgers sharing at least one MLVA in common with nearby cattle herds. The median (Med) and Inter-Quartile Range (IQR) are reported for continuous variables, and where appropriate the maximum (Max) value is also included. The number of instances and percentages are reported for binary variables.
Table 3. Model coefficients for the factors associated with RTA badgers sharing at least one MLVA in common with nearby cattle herds. The median (Med) and Inter-Quartile Range (IQR) are reported for continuous variables, and where appropriate the maximum (Max) value is also included. The number of instances and percentages are reported for binary variables.
VariableMatch
n = 9471 (61.4%)		No match
n = 5964 (38.6%)		Est.Std. Errorz ValueOR95% CI Lower95% CI Upper
Intercept--0.190.190.991.210.831.75
Number of MLVA types (per type)Med: 1
IQR: 1–1; Max: 10		Med: 1;
IQR: 1–1; Max: 5		0.850.0420.282.342.162.54
Distance (per km)Med: 4.4 km
IQR: 2.9–5.7 km		Med: 4.9 km
IQR: 3.3–6 km		−0.120.01−11.610.890.870.91
Milk license (absent)6083 (64.2%)4261 (71.5%)−0.280.04−7.420.750.700.81
Inwards cattle movements (per 10 animals)Med: 4
IQR: 2–9		Med: 5
IQR: 2–11		−0.020.001−11.000.990.980.99

Share and Cite

MDPI and ACS Style

Milne, G.; Allen, A.; Graham, J.; Kirke, R.; McCormick, C.; Presho, E.; Skuce, R.; Byrne, A.W. Mycobacterium bovis Population Structure in Cattle and Local Badgers: Co-Localisation and Variation by Farm Type. Pathogens 2020, 9, 592. https://doi.org/10.3390/pathogens9070592

AMA Style

Milne G, Allen A, Graham J, Kirke R, McCormick C, Presho E, Skuce R, Byrne AW. Mycobacterium bovis Population Structure in Cattle and Local Badgers: Co-Localisation and Variation by Farm Type. Pathogens. 2020; 9(7):592. https://doi.org/10.3390/pathogens9070592

Chicago/Turabian Style

Milne, Georgina, Adrian Allen, Jordon Graham, Raymond Kirke, Carl McCormick, Eleanor Presho, Robin Skuce, and Andrew W. Byrne. 2020. "Mycobacterium bovis Population Structure in Cattle and Local Badgers: Co-Localisation and Variation by Farm Type" Pathogens 9, no. 7: 592. https://doi.org/10.3390/pathogens9070592

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop