4 Discussion
In this study, we describe an outbreak of tuberculosis caused byM. bovis in a Brazilian safari park that lasted for at least 15 years. This outbreak was a result of multiple pathogen introductions and culminated with the death of many animal specimens. There are two main reasons that can explain why this outbreak was so severe. First, starting in 2005 the park passed through management changes and financial hurdles that led to animal overcrowding and malnutrition. Many animals of different species shared a small enclosure and feeding practices were not adequate to serve them all. Overcrowding and malnutrition are known risk factors of bTB (Pollock & Neill, 2002). Second, the park owners often acquired and introduced animals of unknown bTB status (i.e. without diagnostic testing) into the premises, particularly cattle and buffalo. Collectively, these factors may have facilitated pathogen introduction and spread in the property as well as the development of clinical disease by the wild animals.
Our results suggest two possible timeline scenarios for the dynamics of this bTB outbreak. One scenario involves at least three differentM. bovis introductions occurring in the time period between the park inauguration in 1977 up until around 2003. The founding M. bovis subsequently evolved into many different strains over time, fueled by persistent animal to animal transmission. These strains were likely circulating at low levels or asymptomatically until the above-described conditions of overcrowding and malnutrition became conducive of a deadly outbreak. As M. bovis is endemic in the region (5.2% of infected herd prevalence (Queiroz et al., 2016)), a second scenario involves the introduction of many different strains over the years, possibly through cattle, contributing to the high M. bovis diversity observed in the property. Although cattle and wildlife were separated by a fence, escape and mixing of animals in both enclosures were reportedly common events. Alternatively, pathogen introduction may have occurred with the acquisition of specimens of infected wild animals, since TB cases occur in wild animals in captivity in Brazil (Zimpel et al., 2017; Rocha et al., 2011; Murakami et al., 2018; Ikuta et al., 2018; Murakami et al., 2012).
As a wild animal reservoir of bTB has not been identified in Brazil so far, in its inception the PNCEBT was designed to cover cattle and buffalo only. Hence notifications of TB cases in wildlife are not mandatory, diagnostics are not standardized, and there are no official guidelines for outbreak resolution or prevention in wildlife. These factors likely influenced the course of the outbreak described herein. Thus, we believe the inclusion of wildlife in the PNCEBT can bring great benefits in terms of animal welfare, while increasing biosafety in zoos and parks. In addition, regulation of the diagnosis of tuberculosis in wild animals would have an impact on reducing outbreaks that may occur through sale or trade of infected animals. The many outbreaks seen in Brazilian captive populations over the years (Zimpel et al., 2017; Rocha et al., 2011; Murakami et al., 2018; Ikuta et al., 2018; Murakami et al., 2012), along with the one reported herein, should form the rationale necessary to develop this nationwide plan to control and report TB in wildlife, as also suggested previously (Valvassoura & Ferreira Neto, 2014). Notwithstanding, for these policies to become effective, it is necessary to integrate federal and state surveillance services and environmental regulatory agencies.
Our study adds to the importance of evaluating potential paths of transmission to understand bTB outbreaks. Using phylogenetic analysis and M. tuberculosis -based cut-offs for the number of SNPs, many different strains of M. bovis were found circulating in the park, with four recent transmission events involving five deer. The low SNP distance between these five genomes suggests that the same M. bovis strain was transmitted from one animal to another, and the amount of accumulated genetic changes is just a reflection of microevolution. In addition, one of the animals was co-infected with multiple strains, as it presented many heterogeneous SNPs. This mixed infection occurs when the individual is exposed to a single infection event carrying multiple strains, or to several infection events with different strains throughout its life, resulting in a superinfection. Such condition is normally seen in conditions of high disease endemicity, as described in this park.
Using the SNP-cutoff approach, recent transmission events between the surveyed cattle farms and the safari park were not observed in this study. It is possible that the sampling of cattle farms was not comprehensive enough to capture the true M. bovis diversity of Rio Grande do Sul state. The origin of the cattle maintained at the safari park was unknown to the authors of this study; thus, it is also possible that the region from which they originated was not covered in the cattle sampling. Alternatively, the M. bovis strains infecting deer and llamas have been introduced a long time ago in the park (as suggested by the dating analysis) and evolved in geographic isolation. Nevertheless, the phylogenetic relatedness among strains indicates that M. bovis was introduced in the park from somewhere in the region at some point in time.
Although commonly used as a measure of M. tuberculosis orM. bovis transmission between individuals, the SNP cutoff approach was never standardized for M. bovis , is sensitive to the pipeline used to detect SNPs (Guimaraes & Zimpel, 2020), and also a simplistic approach to follow the complex evolution of tuberculous mycobacteria (Menardo et al., 2019). The BEAST analysis performed herein rejected a strict clock evolution model of M. bovis , corroborating previous studies on M. bovis (Zimpel et al., 2020; Crispell et al., 2019; Salvador, 2019) and M. tuberculosis(Menardo et al., 2019; Rutaihwa et al., 2019). We are also unaware if evolutionary rates of M. bovis are influenced by the disease form (e.g. latent versus clinical disease) or host species. Out of the 32 possible transmission links, we observed discordant spoligotype patterns in five, from which three were due to loss of spacers, but two either gained or lost spacers (Appendix Figure 5). Therefore, results reported regarding recent transmission events should be interpreted with caution; and further studies should be conducted to address how we can best use genomics to infer transmission of M. bovis strains and how these can be compared to traditional genotyping techniques such as spoligotyping and MIRU-VNTR.
This study has limitations. Unfortunately, M. bovis isolates from all other animals that died during the outbreak were not obtained, precluding our ability to evaluate the true genetic diversity of the pathogen in the park. We were also unable to control sample collection at the slaughterhouse where deer were culled, thus preventing a better description of the lesions found, number of affected animals, and host species. And finally, we did not have access to records of animal introductions that occurred in the park, impeding a more comprehensive analysis of pathogen introduction.