Locating and collecting badgers
Members of the public, local authorities and countryside organisations in Wales reported the locations of found dead badgers to the Welsh Government or APHA Field Services who recorded the map reference details. The collection by APHA staff of badger carcases that met pre-defined criteria took place between 1st September 2014 and 31st December 2016. There were instances where the collection of reported carcases was not attempted. Reasons for not attempting collection included safety concerns arising from the specific location of the carcass or the non-availability of staff (or other necessary resource) to undertake the collection (supplementary table online ST1). In some further instances collection was attempted but was unsuccessful because either the reported carcase could not be found or the condition of the carcase made it unsuitable for further investigation (for example viscera were herniated externally through wounds, there was severe myiasis (flystrike), the carcase was distended with gas or it was flattened).
Depending on where they were found, the carcases were delivered to the APHA Veterinary Investigation Centres in Carmarthen and Shrewsbury and to the Wales Veterinary Science Centre in Aberystwyth (from 28/01/2016) where they were stored at between 2 and 8 °C for no more than four days before post-mortem examination.
Post-mortem examination and sampling
Of the 1863 dead badgers reported, 841 were collected and 681 (37% of reported carcasses) were suitable for post-mortem examination. The prevalence of bTB was calculated for suitable carcasses only. The sampling protocol was adapted from the standard and detailed protocols described in a comparison by Crawshaw and others24, so that fewer overall samples were taken than the detailed protocol but that the samples chosen were the ones most likely to yield M. bovis if present.
The initial external examination comprised the following: measuring the length from nose tip to tail base (cm), assessing and recording tooth wear, recording the sex of the animal and whether female animals were lactating, andrecording any evidence of vaccination: To temporarily identify vaccinated badgers, the guard hairs of the dorsal back (usually caudal) were trimmed and a coloured marker applied at the same site at the time of vaccination. During the external examination of the badger carcases, to attempt to detect recent vaccination, the skin and hair of the back was visually examined for guard hair trimming and coloured marker. Furthermore the examination recorded any evidence of trap injury, or of illegal interference with the animal and the presence and location of bite wounds.
A detailed examination of the lungs, pericardial sac, liver and kidneys was conducted at post-mortem examination. The lungs were examined by making multiple longitudinal incisions approximately one centimetre apart. At least four slices were made in the liver and three slices in the kidney24.
Each lymph node of all suitable badgers was incised at least once and a pool of lymph nodes (pool 1) consisting of the retropharyngeal, bronchial, mediastinal and hepatic lymph nodes (or as many as were detectable) was collected for subsequent bacteriological culture. If any gross internal lesions suggestive of tuberculosis were observed or if bite wounds were detected, the lesioned tissue and/or excised bite wounds were added to a separate container (pool 2). The pooled samples for bacteriological culture were preserved in 15 ml of 1% aqueous cetylpyridinium chloride. Samples were sent to the Animal Plant & Health Agency Laboratory inStarcross, Devon on the day of examination, for next day receipt. After taking samples, two or three incisions were made in the muscles of the anterior thigh of both hind legs to look for any potential adverse reactions to BCG vaccination24.
Culture and molecular typing
On receipt at APHA Starcross, the samples were washed in sterile 0.85% saline solution, homogenized by standard methods, inoculated onto six modified Middlebrook 7H11 agar slopes26 and incubated at 37 °C. Pool 1, and Pool 2 (if collected), were cultured separately. The slopes were examined weekly from the end of week 2 for a maximum of 12 weeks. M. bovis isolates were harvested when growth was sufficient for genotyping and sent to APHA Weybridge.
Genotyping was performed using spoligotyping27 and VNTR typing (Exact Tandem Repeat Loci A to F)9,28. Spoligotyping confirmed that the isolates were M. bovis. Genotypes of M. bovis were labelled according to the current APHA convention, using numbers to represent spoligotypes and lower case letters to represent the VNTR pattern within each spoligotype. The same genotyping methods were applied to the cultures of M. bovis from cattle slaughtered as part of the national bTB control programme. The cattle genotype home ranges were determined for 2015 (Data Systems Group, Dept of Epidemiological Sciences, Animal & Plant Health Agency Weybridge); for inclusion a genotype had to have been present for at least three years, on at least two different 5 km × 5 km grid squares, in the last 5 years (with a 10 km buffer applied).
Disease status in cattle herds
The disease status in cattle herds from the five TB regions of Wales has been calculated for all Spatial Units (for definition please refer to introduction) with at least 10 badger submissions and compared with prevalence estimates for the badger populations in these areas. The metric used to describe disease status in cattle is average annual herd prevalence during most of the AWBFD sampling period (2015–2016, number of herds under restriction at any point of year/registered live herds in region). Herd size is a known risk factor for bovine TB29; therefore, direct standardisation was used to adjust for varying herd sizes in the Spatial Units3,30,31. Briefly, annual stratum-specific prevalence was calculated for each Spatial Unit across four strata (reduced from six in the cited studies due to the small herd denominator in Wales Spatial Units) of herd size and then applied to the reference population, which comprised the sum of cattle populations across all Spatial Units. The standardised population was then used for herd-level disease measures, resulting in a standardised herd prevalence (Fig. 4).
Badger prevalence mapping
For the Spatial Unit badger prevalence map each Spatial Unit was labelled with the number of submissions, the number of positive results and the resulting prevalence estimate. When added to ArcMap (Esri ArcGIS 10.2.2) the Spatial Unit layer was symbolised using the prevalence value and six pre-defined range values or classes were applied and colour ramped. Spatial Units with less than 10 AWBFD submissions (“insufficient data”) were not colour ramped. All Spatial Units were labelled with the number of positives/number of submissions.
The prevalence of bTB in badgers was estimated among the sampled badgers as in previous studies3,8 with the underlying assumption that the carcasses collected were representative of the overall population.
Analysis was performed to test the null hypotheses (Ho) that:
There are no differences in badger bTB prevalence between the five TB Areas in Wales.
There was no change in overall badger bTB prevalence in Wales between the surveys in 2005–2006 and in 2014–2016.
There was no change in badger bTB prevalence within the five TB Areas between the surveys in 2005–2006 and in 2014–2016.
There was no change in badger bTB prevalence between the surveys in 2005–2006 and in 2014–2016 within the Intensive Action Area (IAA), site of a 2012–2015 badger vaccination trial.
There is no correlation between cattle herd prevalence and estimated bTB prevalence in dead badgers in different geographic regions.
There is no difference in cattle herd prevalence between areas with infected badgers and those with no evidence of bTB in badgers.
For statistical purposes, for the comparisons of prevalence estimates between the TB Areas and over time, both Intermediate TB Areas were combined. All data were tested for normality with the Kolmogorov–Smirnov test (SPSS Version 21 for Windows). Since the dependent variable in these analyses, prevalence, is a rate and fulfilled the criteria for normality, the z-test for comparisons of population proportions was used to test for the statistical significance of differences between prevalence estimates between the TB Areas and over time. A condition of the z-test is that each sample contains at least 10 observations in each category of the dependent variable and for comparisons between samples with less than 10 submissions in at least one category, Fishers Exact test was used32. To explore the correlation between prevalence in badgers and cattle, linear regression analysis was undertaken to calculate Pearson’s coefficient of correlation (SPPS 22 for Windows). Spatial Units with > 10 badger submissions with one or more positives were compared with those which had none, using the two sample t test with unequal variances. In order to compare genotypes between badgers and cattle, as in the previous Wales badger survey3, the authors calculated the associations between the frequency distribution of the genotypes in badgers and the frequency distribution in cattle for each TB Area. In order to prevent ties and to account for the small number of positive badger submissions, frequencies were adjusted, by replacing them with their deviations from expected values which were calculated as (TB Area subtotal) × (Wales genotype subtotal)/(Wales grand total). A Spearman rank correlation between the ranks of genotypes in badgers and their ranks in cattle was then calculated for each TB Area.