IAP-25-136

From Forests to Fevers: Network Models of Disease Emergence in Human-Modified Landscapes

Zoonotic diseases impose substantial burdens on tropical communities, accounting for 26% of disability-adjusted life years lost to infectious disease in lower-middle income countries. These health impacts frequently arise from human modification and use of ecosystems, yet the socio-ecological mechanisms linking environmental change to disease emergence remain poorly understood. Predicting disease risk and designing effective interventions requires disentangling how ecosystem degradation alters pathogen transmission dynamics across heterogeneous landscapes. The complex transmission cycles of zoonotic pathogens—involving vertebrate reservoirs and human hosts—create spatial and temporal heterogeneity in disease risk that reflects both community ecology and patterns of human movement and ecosystem use. Current epidemiological models inadequately capture how pathogen flow and persistence emerge from interactions between ecological and social networks in modified landscapes. Advancing our capacity to predict disease emergence and evaluate alternative land-use policies demands integrative frameworks that explicitly link these coupled natural-human systems.

This PhD project addresses this challenge by developing a mathematical modelling framework that integrates multi-species ecological networks with human mobility and land-use patterns. The models will be parameterised using data* from an endemic region in Central or West Africa where Monkeypox (Mpox), a viral zoonosis, circulates across human-wildlife interfaces in modified forest ecosystems. Mpox provides a tractable system for this work: while human-to-human spread is a feature, spillover from its presumed rodent reservoir (e.g., rope squirrels, Gambian giant rats) is fundamentally linked to human encroachment and wildlife-human contact, often through hunting or forest-dependent activities. Crucially, the analysis will focus on how environmental change (e.g., deforestation, climate variability) alters the ecology of reservoir hosts and the risk of spillover, providing a key link between the natural environment and human health. This approach will leverage available data from public health surveillance, ecological literature, and remote sensing datasets to inform the ecological and social components of the transmission cycle.

While parameterised for a complex zoonotic pathogen, the modelling framework developed here will have broad applicability to zoonotic diseases affecting rural populations dependent on modified ecosystems. The work will advance fundamental theory on disease emergence under environmental and socio-economic change, with direct relevance to conservation ecology through its focus on biodiversity-disease relationships, and to environment and health through its examination of how landscape modification mediates human pathogen exposure and subsequent human-to-human spread.

* The data we will use to parameterise the models in this project will come from
1. World Health Organisation Case (https://www.who.int/data/).
2. Africa CDC event dashboards (https://dashboards.africacdc.org) and their Epidemic Intelligence Weekly Report (africa-cdc-epidemic-intelligence-weekly-report) which provides real-time updates on priority events (not just mpox).

The research will proceed through three complementary modelling phases, each building analytical capacity whilst generating publishable outputs.

Phase 1: Transmission dynamics across heterogeneous landscapes. The student will extend traditional (basic reproduction number) frameworks to characterise transmission processes operating across modified forest-agriculture mosaics. Supervisors at the UKCEH developed techniques for tick-borne diseases which the student will learn. This analysis will identify key mammalian reservoir hosts (e.g., rodents, primates) and the primary zoonotic and human-to-human transmission pathways in different habitat types and seasons, thereby pinpointing potential intervention targets. Parameters will be extracted from published literature on Mpox ecology and human epidemiology, and from available surveillance data, including empirically derived estimates of reservoir host abundance, pathogen prevalence, and human contact rates with wildlife across forest fragments and agricultural margins. Spatially explicit landscape representations will be developed using remote sensing products, incorporating habitat classification, vegetation structure, and land-use history under relevant deforestation and land-use change scenarios in the endemic region. Seasonal variation in zoonotic transmission risk will be modelled by linking reservoir host population dynamics and movement patterns to precipitation and vegetation indices, drawing on environmental data from the study region. Sensitivity analyses will identify parameters exerting greatest influence on system-level, guiding priorities for data acquisition and intervention design.

Phase 2: Network models linking ecological and social systems. Building from the transmission analysis, the student will develop multi-layered, spatially explicit network models that couple ecological communities with human mobility and habitat-use networks. The ecological layer will represent host-pathogen dynamics across landscape patches, incorporating realistic dispersal kernels for key vertebrate reservoirs and their spatial distribution. The social layer will capture human movement patterns among residential areas, agricultural plots, forest sites, and other socio-economic proxies for community vulnerability / adaptive capacity. The data will come from human census data such as poverty indicators, proximity to health centres, occupational indicators and potentially also indicators of variability in capacity of health systems, road networks, land-use data.

Pathogen transmission between these coupled networks will be modelled through human exposure during activities in high-risk habitat types (zoonotic spillover), with infection risk determined by local reservoir host density and pathogen prevalence, and time spent in each location. Crucially, this phase will also explicitly model human-to-human transmission dynamics within the social network layer, and how zoonotic spillover events initiate or sustain community-level outbreaks. This framework will enable exploration of how landscape configuration and human behaviour jointly determine disease risk across spatial scales from individual forest fragments to regional mosaics.

Phase 3: Intervention scenario analysis. The final phase will employ the coupled network models to evaluate alternative intervention strategies under realistic landscape change scenarios. Interventions to be tested include vaccination programmes (targeting human populations, including ring-vaccination strategies), behavioural modifications (reducing contact with wildlife/bushmeat, promoting hygiene), risk communication strategies, and environmental management policies (e.g., forest protection, sustainable land use to reduce encroachment). Landscape scenarios will incorporate ongoing forest loss, agricultural expansion, and restoration initiatives based on regional land-use trajectories. Model outputs will be designed to inform management decisions, including spatial prioritisation of intervention efforts, cost-effectiveness analyses comparing alternative approaches, and assessment of intervention robustness under uncertainty in key parameters. Remote engagement with relevant public health and environmental management authorities in an endemic African country will ensure scenarios and outputs address operational priorities, particularly those related to environmental management for disease prevention. This collaboration will be facilitated through existing networks or established via video conferencing and email exchanges.

Year 1

Months 1-4: Literature review covering zoonotic disease ecology, mathematical epidemiology, and network theory with a specific focus on orthopox viruses and the environmental drivers of Mpox spillover; training in computer programming (most likely python via LinkedIn Learning)

Months 5-12: Development of non-spatial models incorporating wildlife-human-human transmission; epidemiological and ecological modelling (at UKCEH); extraction and synthesis of transmission parameters from literature and surveillance data; sensitivity analysis and hypothesis testing; training in spatial and statistical modelling (e.g., at UKCEH where the student will learn about the techniques developed there for tick-borne diseases and adapt them to the current problem); development of spatially and seasonally explicit models.

Year 2

Months 1-6: Design and parameterisation of multi-layered network models; integration of ecological (wildlife reservoir dynamics) and social network components (human movement and contact); validation against empirical prevalence/case data.

Months 7-12: Testing and refinement of coupled network models; presentation of transmission modelling results at an international epidemiology or One Health conference; submission of first manuscript and associated code repository.

Year 3

Months 1-3: Submission of network modelling framework manuscript; training in research impact pathways and stakeholder engagement.

Months 4-12: Implementation and analysis of intervention scenarios using network models, with a focus on environmentally-driven control measures; preparation of scenario analysis manuscript.

Year 3.5

Months 1-3: Presentation at international One Health or environmental health conference; submission of final research manuscript and code; thesis writing.

Months 4-6: Thesis completion and submission.

Postgraduate research students in Stirling are automatically enrolled in the Institute for Advanced Studies. The institute runs a range of training and skills development courses and workshops designed around the Vitae Researcher Development Framework and organises postgraduate skills workshops, research seminars and support events that run throughout the year.
Students taking a PhD in mathematics subject at the University of Stirling may also attend advanced mathematical courses run by the Scottish Mathematical Science Training Centre (SMSTC) in their first year of study. These are free for registered PhD students at the University of Stirling.

The Computing Science and Mathematics division at the university of Stirling have a condor computing cluster that the student can use to run their models and will get training from their supervisors on how to use the cluster.

The student will also have access to LinkedIn Learning while registered as a student at the University of Stirling. They will have access to several programming courses (likely to be either python and/or R) on this platform if required.

In year 1 and 3 the student will spend up to 1 month at the Centre for Ecology & Hydrology who offer internal courses in Bayesian statistics for ecology, R programming, other soft skills such as paper writing.