Fig. 5: Schematic of the applied ABM framework for informing ecological management decisions. The workflow starts with defining objectives, followed by data collection, parameterisation, experimentation, analysis, and validation. The final step, evaluation and feedback, translates model insights into policy recommendations, research directions, and improved monitoring methods, creating a continuous improvement loop between simulations and real-world management practices.
Discussion The findings of this study demonstrate the potential of agent-based models (ABMs) to cross-over from the field of theoretical ecology and become more readily used in the field of applied ecology, particularly in the scope of management and policy for conservation and conflict resolution. By integrating high-resolution spatial and behavioural data, the presented framework enabled simulations that reflected real-world dynamics with unprecedented accuracy. For example, the case study revealed how clearfell forestry operations influence badger movement patterns, generating actionable insights into the management of bovine tuberculosis risk across a national landscape and measurable reporters for policy makers to assess and implement as required. These results underscore the utility of ABMs as tools for exploring complex ecological interactions, predicting outcomes under different management scenarios, and identifying gaps in field data which can maximise the return on investment for field based research.
Recent advancements in computational power, big data availability, and remote sensing technologies have expanded the scope of ABMs in applied ecology (Hampton et al. 2013; Skidmore et al. 2021). High-resolution telemetry datasets, such as those used in this study, now allow for the parameterization of agents with data-driven behaviours rather than relying on rule-based or arbitrary processes. Furthermore, the availability of global spatial data repositories such as MOVEBANK and the EUROMAMMALS initiative have made this data even more accessible to researchers who lack this information in their jurisdictions (Ferrari et al. 2022; Kays et al. 2022). Similarly, dynamic environmental data, including climate models and habitat indices, can transform static landscapes into systems that respond to temporal and spatial variability, with powerful new technology such as the Google Earth Engine and the Copernicus Initiative offering open-access spatial data (Tamiminia et al. 2020; Buhler et al. 2021). These advancements position ABMs as essential tools for tackling challenges in wildlife conservation and human-wildlife conflict, providing nuanced, system-specific insights that traditional methods often lack.
In applied contexts, ABMs offer a unique advantage: the ability to synthesize complex, multi-dimensional data into cohesive simulations that capture emergent dynamics. This capability enables researchers and managers to test interventions, such as altering forestry practices or adjusting hunting quotas, in a virtual environment before implementing them in the field. By simulating the potential outcomes of various scenarios, ABMs reduce the risks and costs associated with experimental management while ensuring that decisions are informed by robust, data-driven evidence (Murphy et al. 2020). As demonstrated here, ABMs are no longer theoretical tools but are increasingly integral to solving real-world ecological challenges. Field-based management experiments, while invaluable, often come with prohibitive costs, significant risks, and long timelines (Silvy, 2020). Poorly executed actions can lead to unintended ecological consequences, economic losses, and even conflicts among stakeholders (Silvy, 2020). For example, decisions such as whether to cull or vaccinate badgers as a means of controlling bovine tuberculosis often polarise stakeholders, with little consensus on the most effective course of action due to gaps in evidence or differing values (Martin et al. 2020). Moreover, the outcomes of management actions can take years to manifest, delaying the ability to evaluate their efficacy and adapt strategies accordingly.
ABMs driven by high-resolution real-world data and tailored to specific jurisdictions, provide a risk-free platform to explore these management scenarios. These models allow decision-makers to simulate the impacts of various strategies, such as altering forestry practices or implementing targeted interventions, without the ecological or economic risks of testing them in the field. By capturing the complexity of species-environment interactions and the emergent dynamics of ecosystems, ABMs offer a level of precision that can significantly enhance management planning. For instance, running simulations on the effects of vaccination versus culling in specific landscapes can reveal potential outcomes, such as reductions in disease transmission or unintended population shifts, offering a clear comparison of strategies. Importantly, this virtual experimentation fosters collaboration by providing a neutral, evidence-based tool to evaluate contentious decisions. ABMs can help stakeholders and policymakers converge on optimal solutions by demonstrating the potential consequences of various actions, thereby reducing uncertainty and facilitating consensus. By eliminating the risks associated with field trials and compressing the timeline for understanding management outcomes, ABMs empower stakeholders to make informed decisions with greater confidence.
The ability to visualise and analyse these simulated outcomes feeds directly into the need for clear and digestible evidence for stakeholders, ensuring that complex ecological dynamics are translated into actionable insights that resonate with diverse technical and lay audiences. When models are developed within platforms like NetLogo, which offer a unique advantage in their ability to communicate complex ecological processes through accessible graphical user interfaces (GUIs) and parameterised with high-resolution field data, these models provide stakeholders with a recognisable representation of their landscape, complete with agents (e.g., wildlife, humans) moving and behaving based on empirical GPS data collected within their jurisdiction. This recognisability bridges the gap between abstract modeling and real-world systems, enhancing transparency and fostering trust in the model’s outputs (Murphy et al. 2020).
Moreover, this framework allows stakeholders to see not only how agents interact with their environment but also why certain behaviours emerge (McClane et al. 2013). By demonstrating that these behaviours are grounded in real-world data, stakeholders can develop confidence in the model’s validity and relevance to their concerns. For example, stakeholders can directly observe how local wildlife responds to management interventions, such as altered forestry practices or the creation of new habitat corridors. This interactive and visual approach demystifies the modeling process, making it more accessible to non-expert audiences. Crucially, ABMs also offer the ability to run management experiments in real-time during stakeholder discussions. For instance, decision-makers can observe the immediate effects of a proposed intervention, such as a cull or habitat restoration, on key metrics like population density, resource use, or disease transmission. This interactive capability not only strengthens the rationale for proposed actions but also serves as a communication tool to illustrate why specific management strategies are being considered for field trials. By providing an evidence-based, visual platform for exploring ecological scenarios, ABMs facilitate more informed, collaborative, and transparent decision-making processes.
For policymakers and stakeholders—often the primary funders of research—this framework ensures a maximised return on investment by continuously linking field data collection and simulation modeling. The methodological framework presented in this paper establishes a circular data-economy, creating a closed-loop system that continuously improves the quality of field data collection and simulation modeling in an iterative process. In this cycle, high-quality data collected in the field is used to parameterise and validate the model, while the model generates outputs that highlight gaps and priorities for further field data collection. This iterative process ensures that both the simulation and real-world monitoring efforts evolve in tandem, continually improving the understanding and management of ecological systems. By embedding this feedback loop into the workflow, the framework not only optimises the value of data but also promotes good data practices, such as standardized collection, centralization, and accessibility. This approach prevents the loss of investment value through long-tail science and dark data, ensuring that datasets remain active contributors to scientific and management advancements (Hampton et al. 2013). The case study demonstrates the unique capacity of ABMs to uncover unexpected trends that challenge conventional thinking, providing actionable insights for management and policy. Two notable patterns emerged from the case study model, offering fresh perspectives on disease transmission risk and wildlife management strategies in Ireland (Murphy et al. 2024).
First, the spatial configuration of dairy farms emerged as a previously unrecognised driver of disease risk. Traditionally, research and policy efforts have focused on the health, longevity, and husbandry of dairy cattle as possible reasons for increased transmission risk (Broughan et al. 2016). However, the ABM revealed that the spatial arrangement and size of dairy farms within the landscape significantly influence wildlife-host interactions, thereby increasing the risk of bovine tuberculosis (bTB) exposure. This finding shifts the focus toward understanding the role of landscape structure in disease dynamics, opening new avenues for policy interventions, such as spatial planning or buffer zones, and targeted research programs aimed at mitigating risk through landscape-level management. Second, the model highlighted an unexpected relationship between farm density, habitat heterogeneity, and disease risk. The case study model found the opposite: areas with lower farm density posed a higher risk of disease transmission. This counterintuitive result suggests that wildlife hosts, such as badgers, may exhibit greater movement and contact with farms in less fragmented landscapes, where natural habitats are more contiguous. This finding underscores the need to monitor wildlife activity in low farm density areas, challenging traditional assumptions and guiding new research and monitoring programs to address these overlooked interfaces. These emergent trends demonstrate the power of ABMs to go beyond static analyses and identify nuanced, system-specific patterns. By highlighting these novel dynamics, the model provides an evidence base to inform targeted interventions, improve monitoring strategies, and refine existing policies.
ABMs for applied ecological scenarios offer immense potential, but their implementation comes with significant challenges that must be carefully considered to ensure their success. One of the foremost challenges is the intensive data requirements. High-quality, high-resolution data is crucial for parameterising agents and their environments, as this framework relies on empirical evidence to achieve realism. However, such data is not always readily available, particularly in regions or systems where monitoring efforts have been minimal. This issue contrasts with traditional ABMs, where questions can often be explored without requiring fieldwork (Murphy et al. 2020). For applied scenarios, the absence of adequate data can be a significant barrier but also presents an opportunity. Identifying these data gaps provides a compelling argument for establishing long-term monitoring programs (Ariño et al. 2016). In this sense, discovering a lack of suitable data for model parameterisation can itself be considered a valuable result, as it highlights areas where investment is needed to inform evidence-based management and policy decisions.
Another major consideration is the computational intensity required to run realistic, large-scale ABMs. These models often involve thousands of agents interacting across millions of patches in dynamic environments, which can demand significant computational resources (Murphy et al. 2020). For example, while NetLogo, a commonly used ABM platform, theoretically has no fixed model size limits, memory usage can quickly escalate. For larger models, such as those involving millions of patches or tens of thousands of agents, up to 50 GB of RAM or more may be required to achieve smooth performance. This poses a barrier for researchers with limited access to high-performance computing facilities. While NetLogo offers configuration options to optimise RAM allocation, running large-scale simulations or BehaviourSpace experiments (which multiply memory usage by running parallel simulations) can still push the limits of standard computational setups. As ABMs become more complex and data-driven, access to powerful computers and appropriate configuration of computational environments will become increasingly critical.
Another challenge is the time and expertise required to build and validate models. Constructing an ABM involves not only technical proficiency in coding and data integration but also a deep understanding of the ecological system being simulated. Collaborating with domain experts and stakeholders adds value but also increases the complexity of project coordination. Similarly, external validation requires robust datasets that are entirely independent of the input data—a requirement that can be difficult to meet, particularly for novel systems or species. Lastly, stakeholder engagement can be both a strength and a challenge (Silvy, 2020). While this framework promotes transparency and collaboration, stakeholders may have differing priorities, levels of trust in the model, or expectations regarding its outputs. For example, stakeholders may favour management actions (e.g., culling vs. vaccination) that the model does not support, creating potential conflicts. Clear communication, education about the model’s capabilities, and iterative collaboration are essential to mitigate such challenges and foster trust in the model’s outputs. Despite these hurdles, the challenges are not insurmountable. In fact, many of them—such as identifying data gaps or engaging with stakeholders—can lead to positive outcomes, such as improved monitoring programs or better-aligned management priorities. Addressing these challenges head-on ensures that ABMs remain a powerful tool for applied ecological research and decision-making.