Kilian J Murphy1
1Laboratory of Wildlife Ecology and Behaviour, SBES,
University College Dublin, Ireland
*Corresponding Author:
kilian.murphy@murphyecology.com
Abstract1. Agent-based models (ABMs) are powerful tools for exploring ecological
systems but have historically been applied predominantly to theoretical
research rather than practical management and policy contexts.
Traditionally, ABMs rely on simplified agents and environments, often
lacking integration with real-world data, limiting their utility in
addressing complex, jurisdiction-specific ecological challenges.
2. This research presents a novel framework for building applied ABMs
that integrate high-resolution spatial, behavioural, and environmental
datasets. By leveraging telemetry data, remote sensing, and
site-specific monitoring, the framework creates realistic, data-driven
emulations of ecological systems. These models enable researchers and
policymakers to test management and policy interventions virtually,
offering insights into system-level dynamics and emergent trends without
the risks, costs, or delays of field-based experiments.
3. We highlight this framework’s ability to significantly enhance the
realism of ABMs, increasing their robustness, transparency, and
trustworthiness through rigorous data input and validation. By
uncovering critical data gaps and fostering a circular feedback loop
between simulation outputs and field data collection, the framework
ensures continuous refinement of models and monitoring programs. We
demonstrate the efficacy of this approach using a published workflow
examining the role of badger movement in a disturbance landscape and its
implications for disease transmission, showcasing how the framework
supports evidence-based management and policy decisions.
4. This research establishes a repeatable protocol for integrating ABMs
into applied ecology, enhancing their capacity to inform evidence-based
decisions. By promoting transparency, collaboration, and trust among
stakeholders, it positions ABMs as an essential tool within applied
ecology for addressing pressing conservation challenges and
human-wildlife conflicts in a cost-effective, adaptable, and
scientifically rigorous manner.
Introduction In the Anthropocene, ecological systems are shaped by human activity,
whether that is through positive forces such as conservation and
management of ecosystems or negative impacts such as habitat
destruction, climate change, overexploitation of natural resources,
pollution, and the introduction of invasive species (Young et al. 2016;
Von Essen et al. 2023). These impacts have heightened pressures on
biodiversity and ecosystem services, often placing wildlife in direct
conflict with human interests (Simkin et al. 2022). This tension is
particularly evident in shared production landscapes (e.g. agricultural
landscapes, forestry and aquaculture), where wildlife species interact
frequently with human interests, leading to a wide variety
human-wildlife conflicts across the globe (König et al. 2020;
Munguia-Carrara, 2020; Göttert & Stark, 2022). Addressing these
challenges demands applied ecological approaches that integrate diverse
data sources, including monitoring from remote sensing technologies,
systematic field surveys, and citizen science initiatives that are
transformed into actionable evidence through state-of-the-art modeling
techniques, such as agent-based and predictive ecological models, which
capture the complexity of interactions between species, environments,
and human activity (Silvy, 2020; Murphy et al. 2020; Morera-Pujol et al.
2023). Crucially, applied ecology extends beyond data analysis; it
necessitates transdisciplinary collaboration among scientists,
policymakers, and stakeholders to co-design conservation and management
strategies (Silvy, 2020; Fuller et al. 2020; Murphy et al. 2022a).
The field of applied ecology aims to foster win-win scenarios where
ecological integrity is preserved while supporting human livelihoods and
societal needs (Hopkins et al. 2021; Hegwood et al. 2022). Applied
ecology is a cornerstone of evidence-based management and policy, but it
remains a time-, resource, and cost-intensive endeavour (Manfredo et al.
2021). Shortages in personnel, data availability, or even appetite for
evidence-based decision-making can significantly limit its broader
implementation, posing challenges to its role in informing wildlife
conservation and management (Mafredo et al. 2021). While applied ecology
has tested frameworks for addressing conservation challenges (e.g.
adaptive wildlife management), its practical implementation faces
significant challenges (Dressel et al. 2018; Mansson et al. 2023). One
of the foremost challenges lies in the inherent complexity and
variability of ecological systems, where universal problems change often
require system-specific solutions. Developing effective solutions
requires longitudinal data to evaluate the long-term impacts of
interventions, which are costly to establish and maintain long-term
(Caughlan & Oakley, 2001; Lindenmayer et al. 2022). Without this
iterative feedback process, conservation efforts risk
oversimplification, leading to plans that fail to capture the nuance of
real-world systems and deliver sustainable outcomes.
Another critical challenge lies in the patchy availability and
inconsistent quality of ecological data, which can significantly limit
applied ecology’s effectiveness. Monitoring efforts are often unevenly
distributed, with some regions or taxa receiving extensive attention
while others remain poorly understood (Stephenson et al. 2015). This
imbalance frequently forces researchers to rely on proxies or indices as
substitutes for high-quality data, which may oversimplify ecological
complexity and reduce the accuracy of conclusions (Murphy et al. 2022b).
Additionally, many ecological datasets are difficult to access or
entirely lost—a phenomenon known as ”dark data.” These datasets, often
produced by small-scale, investigator-led studies, are rarely curated or
shared due to limited funding, lack of incentives, and the complexity of
managing heterogeneous data types (Hampton et al. 2013). The issue is
compounded by a historical reliance on “long-tail science,” where
small projects generate large amounts of data without sufficient
resources for data management or sharing (Hampton et al. 2013). As a
result, much of the information remains siloed, inaccessible for broader
synthesis or reuse. Poor data infrastructure and insufficient
collaboration further hinder the development of comprehensive datasets
necessary for informing conservation and management decisions. Despite
these challenges, applied ecology has made significant strides by
integrating cutting-edge technologies and fostering collaboration across
disciplines. Addressing these pitfalls is essential to furthering its
impact and ensuring that conservation and management efforts are both
effective and equitable.
Agent-based modelling (ABMs) is a method that holds immense potential to
address many of the pitfalls inherent in conducting research in applied
ecology, offering a framework to integrate diverse data sources and
explore the complexity of ecological systems in unprecedented detail
(Murphy et al. 2020). By simulating individual agents (e.g., animals,
plants, humans) and their interactions with the environment, ABMs can
capture the emergent dynamics of ecosystems, allowing researchers to
predict the outcomes of conservation interventions under various
scenarios (McClane et al. 2011). These models are particularly
well-suited to tackle system-specific challenges, such as
species-specific responses to habitat management, or to assess the
cascading impacts of management actions across landscapes (McClane et
al. 2011; Murphy et al. 2020). However, while their use is increasing in
applied ecology (Recio et al. 2020; Gritter et al. 2024; Murphy et al.
2024; Thompson et al. 2024), the application of ABMs to real-world
applied ecological management and decision-making has historically been
limited relative to the potential of the method.
Traditionally, ABMs have been used predominantly in theoretical or
”blue-sky” ecological research, focusing on advancing fundamental
ecological theory rather than providing actionable insights for applied
contexts (Chivers et al. 2014; Ringelman, 2014; Kane et al. 2016;
Teckentrup et al. 2018). This stems, in part, from the perception that
ABMs lack the nuance and complexity needed to reflect real-world systems
accurately. ABMs in ecology have traditionally not been parameterised
with the level of realism required for evidence-based management.
Instead, they often use oversimplified agents and environments, aiming
to extrapolate patterns and mechanisms that can be generalised add
nuanced evidence about the ecology of the species but not necessarily
how to management them within a specific jurisdiction (Chivers et al.
2014; Carter et al. 2015). While this approach has provided valuable
insights into ecological processes, it has limited the ability of ABMs
to address system-specific management questions directly. Many ABMs
assess their efficacy based on their ability to recreate patterns
observed in the field, yet they often lack validation using independent,
external, and longitudinal datasets (Carter et al. 2015). This absence
of robust validation, which would confirm that model outputs align with
real-world outcomes over time, has been a significant barrier to their
application in high-stakes decision-making. Without this level of proof,
ABMs struggle to gain the trust needed to inform contentious management
scenarios or resolve conflicts involving multiple stakeholders, who
often have significant stakes in the outcomes and cannot afford to rely
on models that lack the realism to address their specific circumstances.
Recent advancements, however, are rapidly transforming the potential of
ABMs for applied ecology. The growth of big data repositories (e.g.
MoveBank), high-resolution remote sensing (e.g. Google Earth Engine),
and satellite imagery (e.g. Copernicus) now provides a wealth of
spatially and temporally detailed datasets that can be used to
parameterise ABMs with new levels of realism. Furthermore, the
integration of artificial intelligence, including machine learning
techniques, has enhanced the ability to develop ABMs capable of
capturing complex, non-linear dynamics and emergent behaviours in the
model. Improvements in computational power and software capabilities
have also significantly reduced the time and resource requirements for
running large-scale simulations with millions of agents and hundreds of
procedures. Crucially, the community-driven, continual development of
popular modeling environments such as R (R Core Team, 2024) and NetLogo
(Wilensky, 1999) has played a pivotal role in advancing the capabilities
of ABMs. These platforms now enable the seamless integration of diverse
datasets and sub-models through extensions (e.g. gis, csv) ,
facilitating the creation of hyper-realistic agents and environments.
This progress has allowed ABMs to scale up their applications, testing
applied ecological systems in ways that were previously unfeasible.
Combined with improvements in computational power and software
efficiency, these advancements have made large-scale simulations
increasingly accessible and practical for addressing real-world
ecological challenges.
In this paper, we present a methodological framework for building
realistic agent-based models tailored to informing applied ecological
challenges, hereafter referred to as “applied ABMs.” We demonstrate
how to effectively parameterise agents and their environments using
diverse input datasets and outline methods for generating data to feed
into multivariable modelling and external validation protocols. This
approach aims to deliver evidence-based insights for management and
policy while integrating feedback loops that inform field data
collection systems through applied ABMs. By fostering an adaptive data
economy, this framework addresses issues such as dark data and long-tail
science, promoting iterative improvement in applied ecological
management. To illustrate the framework’s practical application, we
refer to its use in a previously published case study (Murphy et al.
2024), which investigates the role of wildlife hosts in the transmission
of a zoonotic disease within a disturbance-driven landscape. This
example highlights how applied ABMs can address pressing ecological and
management questions with enhanced realism and practical utility.