1 Introduction

As the extent and magnitude of human activity continues to expand (IPBES 2019), the urgency to understand how wildlife populations respond to anthropogenic change is accelerating (Larson et al. 2016; Venteret al. 2016). This information is crucial for effective management and conservation policies (Pimm et al. 2014). Ecologists have long tried to understand and predict the impacts of human disturbance on wildlife populations and communities through the use of quantitative modelling approaches (Beissinger & Westphal 1998; Getz & Haight 1989). However, stressors rarely occur in isolation, with animal populations often exposed to multiple direct (e.g., harvesting; Kays et al. 2017) and indirect stressors (e.g., habitat fragmentation, Smith et al. 2019). Although it is relatively straightforward to determine how individual stressors impact populations, the complex ways that multiple stressors interact make it challenging to predict their combined effect on a population (Paniwet al. 2021). As a result, there has been an increasing focus on understanding the population-level effects, such as changes in population dynamics, geographical distribution, and/or population persistence, that can result not only from individual indirect (non-lethal) stressors but from exposure to multiple stressors (e.g., Gosselin et al. 2015; Daversa et al. 2025; Pirottaet al. 2019; Galic et al. 2018). Oftentimes, disturbances have indirect effects on populations, such as through changes in food intake, which may lead to changes in energy balance and/or body condition (e.g., Parker et al. 2009), or through exposure to pathogens or pollutants, which may result in changes in immune status (Charbonnel et al. 2008). These indirect effects can compound, leading to impacts on vital rates (e.g., survival, reproduction) that then shape population dynamics (Fig. 1, Box 1). Explicit consideration of these indirect effects has been formalised for marine mammal risk assessments (NAS 2017). Other efforts have advocated for the inclusion of indirect effects within a standardised mechanistic framework (Johnston et al. 2019). However, these are conceptual frameworks and, as such, lack information on specific approaches used to model population-level impacts of disturbance. These frameworks also fall short when considering community dynamics or management actions (but see Urban et al. 2022), which can play key roles in predicting population persistence (Fig. 1, Box 1). Interactions among stressors and discrepancies in spatial and temporal scales of impact mean that correlative approaches are often too limited to inform policy or management. Quantitative modelling approaches thus play an important role in understanding and managing ecosystems by clarifying the key mechanisms that might explain the behaviour of ecological systems (Schmolke et al. 2010). The main advantage of quantitative approaches to management decision making is their ability to predict the magnitude of effect of alternative scenarios based on underlying processes, which is rarely possible through empirical studies alone (Skogen et al. 2024). Many quantitative approaches are available for modelling the impacts of human activity on animal populations, each of which have their own assumptions, caveats, and advantages. The approaches differ in which processes are represented (and how), their spatio-temporal scales, data requirements, and in which questions they are best poised to answer. Inappropriate choices in modelling approach or structure could compromise our ability to make reliable predictions (Gerber 2006), creating greater uncertainty when deciding on management strategies. Despite the vast array of modelling approaches available, several key challenges remain when trying to predict the population-level impacts of human activity on wildlife: 1) how to get the most out of disparate data streams collected at different spatio-temporal scales; 2) how to provide scientifically informed management advice for populations when limited empirical data are available, as is typically the case; 3) how to manage uncertainty when uncertainty is ubiquitous; and 4) knowing which approach(es) to use given the available data and the question at hand. Further challenges arise when considering community dynamics, which is often necessary to accurately predict the implications of human disturbance and potential management strategies on ecological communities (Buckley and Han 2014; Zavaleta et al. 2001). While challenges 1-3 have received considerable attention (e.g., Nichols 2021; Simmonds et al. 2024; Zipkin et al. 2019; Fletcher Jr.et al. 2019), guidance on modelling choices for assessing and predicting population-level impacts of human disturbance has received less attention (but see Accolla et al. 2021; Hunter-Ayad et al. 2020; Thompson et al. 2021; Briscoe et al. 2019). Here, we provide an overview of key approaches available to model human impacts on animal populations. In doing so, we aim to provide resources for new studies to identify suitable methods and help overcome taxonomic or domain biases in model development. As part of this effort, we highlight important considerations when deciding on a modelling approach and model structure to model the direct and indirect impacts of human disturbance on animal populations. We also use two case-studies on red fox (Vulpes vulpes ) and European mink (Mustela lutreola ) to illustrate some of the considerations made during model development and the strategies used to overcome the limitations of different modelling approaches, including model integration and energetic modelling. Further, we extend existing conceptual frameworks for understanding the impacts of human disturbance by incorporating community-level responses to multiple stressors as well as the pathways by which management actions can influence a population (Fig. 1, Box 1). The information presented here can be used to identify appropriate model configurations for different research and management purposes, while also suggesting key priorities for future model development and integration.