Fig. 1: Schematic representation of key parameters used to
parameterise badger agents in an agent-based model (ABM). Central to the
model is telemetry data from GPS-collared badgers, informing various
behavioural and ecological parameters. Moving clockwise from the top:
activity scheduling, habitat selection, step length and turning angle,
movement speed, social networks, behaviour, migration and dispersal, and
home range. These parameters enable realistic, data-driven agent
behaviours in simulations.
To ensure the accuracy of these analyses, data must be carefully
screened. This involves removing erroneous fixes, such as points with
unrealistically large step lengths, which can distort movement metrics.
Tools such as the dplyr package (Wickham et al., 2014) can
facilitate filtering and cleaning of datasets. Stakeholder engagement is
essential during this process, as their local expertise often aids in
identifying and correcting errors. Additionally, for species exhibiting
temporal activity patterns, such as badgers, transforming movement data
into bursts (e.g., nightly tracks) allows for accurate computation of
daily movement metrics. This burst-level analysis is crucial for
parameterising time-dependent behaviours and ensuring high-resolution
temporal alignment with the model’s objectives.
Spatial and environmental data analysis provides critical site-specific
and dynamic inputs that enhance the resolution and realism of the
modelled environment (Fig. 2). Land-use maps, habitat quality indices,
and environmental datasets (e.g., NDVI, temperature, precipitation) can
be analysed using the sf package (Pebesma, 2018) for spatial data
handling. For dynamic environmental parameters, raster data such as NDVI
or climate models can be processed using the raster (Hijmans,
2015) or terra (Hijmans et al., 2022) packages. Temporal changes,
such as seasonal shifts in resource availability or extreme weather
impacts, can then be incorporated into the model to represent dynamic
landscapes. For example, combining flood risk maps with weather data can
simulate habitat changes due to flooding events, affecting agent
movement, resource distribution, and habitat suitability.