Mycobacterium bovis Molecular Typing and Surveillance 69
the landscape over time. It remains plausible
that the mechanism of spread is not adequately
captured in official records. The number of
badger- derived isolates examined was again too
low to draw robust conclusions about the direc-
tion of transmission between cattle and badgers
within the home range of MLVA010.
This extended study illustrates the potential
and some of the limitations inherent in this
phylodynamics approach. The exceptionally
slow mutation rate imposes some limitations on
what can be achieved, even from bacterial WGS
data. For example, it may not be possible to
determine the underlying transmission tree, the
‘who infected whom?’ question, as similarly con-
cluded in human TB studies. This study illus-
trated the increased precision and discrimination
provided by bacterial WGS and was able to detect
some MLVA switching events that would other-
wise have been undetected. More widespread
application of bacterial WGS to M. bovis genomic
epidemiology should provide novel and impor-
tant insights into the dynamics of M. bovis
persistence and spread. The most relevant con-
trol questions may be better addressed using
approaches that integrate more directly bacte-
rial WGS with additional epidemiological data.
However, if sampling of hosts has been suffi-
ciently dense and over sufficient time, it may be
possible to identify ‘transitions’ between hosts
and to provide some estimate of the extent to
which local prevalence is cattle-driven versus
wildlife-driven.
In Michigan, USA, M. bovis from experi-
mentally infected white tail deer populations has
been subjected to pathogen WGS, revealing
within-host evolution and diversification of the
pathogen in different tissues (Thacker et al.,
2015). More recently, genomic epidemiological
approaches have revealed potential airborne
transmission of M. bovis between human hosts
in Nebraska (Buss et al., 2016) and introduction
of infection into Minnesota with possible links to
Mexico and south-western states of the USA
(Glaser et al., 2016). In the latter case, phyloge-
netic techniques were used to confirm the break-
down was a result of a very recent, single
introduction (Glaser et al., 2016).
Bacterial WGS looks set to revolutionize the
way we do veterinary bacteriology, much as it is
doing for human medical microbiology (Loman
et al., 2012; Loman and Pallen, 2015; Arnold,
2016). It is likely that pathogen WGS will replace
many traditional methods for detection, identifi-
cation, antimicrobial resistance (AMR) testing
and epidemiological typing where the resolution
of WGS can be scaled to provide the desired reso-
lution, including the highest possible resolution
of disease transmission dynamics. When inte-
grated with dense epidemiological, ecological
and population genetics data and advanced
mathematical modelling, including Bayesian
inferences (Kao et al., 2014, 2016; Biek et al.,
2015; Trewby et al., 2016), this powerful
approach, once the sole preserve of studies on
fast evolving RNA viruses (du Plessis and Stadler,
2015), is providing unprecedented insight
into disease maintenance and spread in the
landscape in epi-systems involving complex
host, pathogen and environment interactions
( Blanchong et al., 2016). An elegant recent
approach of genomic epidemiology to study
Brucella abortus illustrates the power of this new
approach. Using animal location and movement
data, mathematical modelling and Bayesian
inferences they detected transmission events
(transitions) between domestic and wildlife host
species (Kamath et al., 2016).
These phylodynamic approaches have their
limitations with pathogens that exhibit low
mutation rates such as M. bovis (Biek et al.,
2015). However, the unprecedented level of
resolution that genomic data provides, especially
when applied over densely sampled bacterial
populations from multiple hosts and collected
over long time frames, promises to be a game
changer (Sintchenko and Holmes, 2015). It is
consequently a time of upheaval in bacterial
molecular epidemiology as researchers adapt to
new tools that are opening up new opportunities
(Gaiarsa et al., 2015). A major challenge in the
near future will be how to better implement bac-
terial WGS for disease forensics. It will become
necessary for laboratories to agree a nomencla-
ture scheme for naming M. bovis genomes that
facilitates inter-laboratory comparisons and the
inclusion of extant whole-genome sequences
that have been deposited in open-access data-
bases. It is already feasible to generate a spoligo-
type pattern from raw sequence reads (Coll et al.,
2012). Provided sequencing costs continue
to fall and issues around incorporation of
existing typing data, standardization of nomen-
clature and methodologies can be resolved,