CASE STUDY
ScenarioClearfell forestry operations are a common land-use practice in Ireland,
creating significant disturbances in agroforestry mosaics. These
disturbances can alter the behaviour and movement patterns of wildlife,
particularly European badgers, which are a known wildlife host for
bovine tuberculosis (bTB). Badger movement into farms increases the risk
of bTB transmission to cattle, posing challenges for disease management.
Understanding how forestry disturbances influence badger behaviour is
crucial for developing evidence-based strategies to mitigate disease
risk. This case study applies the ABM framework to simulate badger
responses to clearfell forestry and identify management interventions
that could reduce bTB transmission. See Fig. 5 for an infographic of the
framework. Full details of the study are available in Murphy et al.
(2024).
ObjectiveThe aim of this study was to simulate the behavioural response of
European badgers to clearfell forestry and assess its role in
influencing bovine tuberculosis (bTB) transmission to cattle. By
understanding how forestry disturbances alter badger movements and
interactions with farms, the model sought to inform evidence-based
management strategies to mitigate disease risk in agroforestry mosaics
across Ireland.
Data CollectionHigh-resolution data were gathered from multiple sources to construct a
realistic and comprehensive model environment. GPS telemetry data from
202 badgers tracked over seven years provided detailed movement
information, while sett survey data contributed fine-scale information
on real-world nesting sites and population estimates. Environmental
layers, such as CORINE land-use classifications and clearfell records,
detailed the spatial and temporal dynamics of habitat and human-induced
disturbances. Administrative data, including farm boundaries and
forestry management zones, delineated areas of human-wildlife
interaction. These diverse datasets ensured that the parameterisation
process was informed by real-world observations, enhancing the model’s
ability to reflect actual badger behaviours in response to clearfell
forestry.
Parameter DatabasesTelemetry data were analysed to derive movement metrics, including step
lengths, turning angles, and net squared displacement, which captured
the variability in badger movement patterns. Kernel Density Estimation
(KDE) was used to calculate home ranges, defining the spatial extent of
badger territories. Resource Selection Functions (RSFs) quantified
habitat preferences, standardizing GPS fixes by habitat availability to
identify favoured land-use types across different seasons. These
parameter databases informed agent behaviour within the model, ensuring
that simulated badger movements closely mirrored observed patterns.
ParameterisationAgents were parameterised with realistic movement and behavioural rules
derived from the telemetry and RSF analyses. Sex- and age-specific
parameters allowed for differences in movement dynamics, while habitat
preferences informed decision-making during foraging and dispersal. The
environment was parameterised using spatial layers that integrated farm
boundaries, forestry operations, and habitat patches. Clearfell sites
were dynamically updated in the model to simulate post-harvest
regeneration, influencing badger resource availability and movement
patterns. This parameterisation ensured that agents adapted their
behaviours based on real-world changes in the landscape.
Experiment & AnalysisThe ABM was implemented across 100 clearfell sites selected randomly
across Ireland, thus allowing the model to be used across a national
landscape. Simulations ran for one year post-clearfell, with 25
iterations per site, capturing spatial and temporal dynamics of
badger-farm interactions. Output data included detailed records of
“alien encounters” (badgers entering farms outside their home range as
a proxy for increased disease transmission risk), habitat use, and
clearfell attributes. The model revealed that clearfell size, habitat
diversity, and badger density were significant predictors of alien
encounters. Statistical analysis employed Generalized Additive Mixed
Models (GAMMs) to evaluate the influence of these predictors, capturing
non-linear relationships and providing nuanced insights into the drivers
of badger behaviour. The GAMMs achieved 94% deviance explained,
demonstrating the model’s effectiveness in replicating real-world
dynamics.
ValidationIndependent datasets were used to validate model outputs, ensuring their
applicability to real-world scenarios. Farm-level bTB breakdown data
from 12,000 farms over a four-year period provided an external benchmark
for model predictions. A Pearson correlation between predicted encounter
rates and observed bTB incidence confirmed the model’s accuracy in
identifying risk hotspots, thereby demonstrating its reliability for
informing management decisions. Validation highlighted both the
strengths and limitations of the model, identifying areas where further
refinement was needed.
Evaluation & FeedbackThe results were reviewed collaboratively with stakeholders and
policymakers to ensure their practical relevance. The model confirmed
expected trends, such as the influence of badger density and herd type
on bTB risk, but it also uncovered previously overlooked factors. For
example, farm density and landscape heterogeneity were found to play a
critical role in facilitating badger movement, insights that had not
been fully considered by policymakers. These unexpected findings
highlighted gaps in field data, such as limited sett survey coverage and
a shortage of telemetry data outside of two specific regions. By
identifying these data gaps, the model provided a roadmap for future
monitoring efforts, guiding targeted field surveys to improve data
coverage and refine future simulations. This feedback loop between site
and simulation enhanced the model’s accuracy and ensured its continued
relevance in informing policy decisions.