🧬 biology

Beyond Entry: Evaluating the Internal Spread Risk of highly pathogenic Avian Influenza (HPAI) H5N1 in Australia

This study utilizes an agent-based model to demonstrate that the risk and trajectory of a potential HPAI H5N1 outbreak in Australia are heavily dependent on the incursion region, highlighting substantial threats to both poultry and wildlife populations to inform targeted national preparedness and One Health strategies.

Original authors: Catarina do Carmo Norte dos Santos

Published 2026-08-11

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Catarina do Carmo Norte dos Santos

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine a world where tiny, invisible viruses are like mischievous travelers, hopping from bird to bird, carrying a secret that could turn a harmless visit into a deadly storm. This story lives in the realm of epidemiology, the science of tracking how diseases move through populations, and it relies on a clever tool called an "agent-based model." Think of this model not as a crystal ball, but as a giant, digital sandbox. Instead of just guessing, scientists build a virtual Australia filled with thousands of tiny, computer-controlled birds. These birds aren't just static pictures; they have "energy" like a battery. They get tired when they fly, they get hungry, and they get excited when they find a juicy wetland. They make their own choices about where to go next, just like real birds do. Why does this matter? Because a super-strong flu virus, known as H5N1, is currently causing chaos around the globe. While Australia has been lucky so far, the fear is that if this virus sneaks in, it could wipe out our unique wildlife and our chicken farms. Scientists need to know: if the virus lands in one spot, where will it run next, and how fast?

This paper, titled "Beyond Entry," dives into that exact question using a digital simulation of Australia. The researcher, Catarina do Carmo Norte dos Santos, built a virtual version of the continent where they could drop a single infected bird into different starting zones and watch what happens over 100 days. They didn't just watch; they programmed the birds to act like real nomads. Some birds, like the gulls, stick close to the coast and city trash cans, while others, like the waterfowl, wander far and wide across the dry interior, chasing rain and food. The model also included farms, treating them like potential traps that birds might accidentally visit.

The results of this digital experiment suggest that where the virus first lands makes a huge difference in how the story plays out. It's not a one-size-fits-all disaster. When the simulation started in the Northern Agricultural Region of Western Australia, the virus spread widely, peaking at a prevalence of 33.809% (with a range of 6.668–60.949%) after 75 days. In this scenario, the virus stayed mostly in Western Australia and managed to breach (or "visit") farms there. However, if the virus started in the Rangelands Region, the spread was much smaller, with a peak of only 3.130% in Western Australia right at the start, and almost no spread to other states.

The most dramatic scenario, however, happened when the virus was introduced to the Limestone Coast Region. In this simulation, the virus spread significantly to specific states, reaching the Australian Capital Territory (ACT), Tasmania, South Australia, Victoria, and New South Wales, while showing zero prevalence in Western Australia, the Northern Territory, and Queensland. By day 99, the Australian Capital Territory (ACT) saw a massive infection peak of 58.531% (ranging from 44.822–72.241%), and Tasmania was hit hard with 23.833% (ranging from 18.972–28.695). Most alarmingly, this specific starting point led to a near-total breach of farms in Tasmania, with 97.945% (ranging from 97.382–98.508) of them visited by infected birds. In contrast, starting the virus in Southwest Queensland resulted in a tiny, almost non-existent spread, peaking at just 0.067% in Queensland after one day, suggesting that the high ground and distance acted like a shield.

The author suggests that these differences happen because of the birds' energy and the landscape. Birds need to be healthy and energetic to fly long distances. If they get sick, they lose energy and might not make it to the next big wetland or farm before they die or stop moving. The Limestone Coast is close to rich, productive areas where birds gather in huge numbers, so the virus spreads effectively to those specific regions. But in places like Southwest Queensland, the high elevation and distance might tire the birds out before they can carry the virus to new, crowded areas.

The study also highlights that while Australia's biosecurity is strong, the risk isn't just about the virus getting in; it's about what happens once it's inside. The simulation suggests that once H5N1 is in the wild bird population, it could jump to poultry farms and even threaten unique wildlife like sea lions and fur seals, which eat infected seabirds. The author notes that their model is a simulation, not a prediction of the future, but it serves as a powerful warning system. It suggests that if we ever see the virus, we need to know exactly where it started, because that single piece of information tells us whether we are facing a small, contained spark or a continent-wide wildfire. The model acts as a map for the future, showing us that the path of the virus depends entirely on the terrain and the tired, hungry, wandering birds carrying it.

Technical Summary: Evaluating the Internal Spread Risk of HPAI H5N1 in Australia

Problem Statement
While Oceania remains largely unaffected by the global spread of Highly Pathogenic Avian Influenza (HPAI) H5N1 (clade 2.3.4.4b), the recent detection of the virus in wildlife on sub-Antarctic Heard Island signals an increasing risk of incursion into mainland Australia. Unlike historical Low Pathogenic Avian Influenza (LPAI) strains that required mutation within poultry to become highly pathogenic, the current H5N1 clade can transmit directly as HPAI. This poses a dual threat: severe economic losses to the poultry industry (requiring mass culling) and catastrophic ecological impacts on Australia's unique biodiversity, including endangered bird species and marine mammals like the long-nosed fur seal. The primary challenge is predicting how the virus would spread internally via nomadic wild birds and identifying which regions and production systems are most vulnerable to spillover.

Methodology
The study employs an Agent-Based Model (ABM) developed in the NetLogo v6.0.3 platform, adapted from the Bronze Age Collapse (BACO) model. The simulation focuses on spatial spread driven by wild bird movements and their interactions with poultry farms.

Agents and Environment: The model simulates two distinct behavioral groups of nomadic wild birds: waterfowl (e.g., straw-necked ibis, Pacific black duck) and gulls. These agents move across a map of Australia defined by state boundaries, elevation data, and habitat productivity (wetlands and coastlines). Movement is driven by energy dynamics; birds consume energy through movement and infection and replenish it by foraging in productive patches.
Transmission Dynamics: Transmission occurs probabilistically. A "breach" is defined as an infected bird crossing a farm boundary. Transmission to farms is modeled as a discrete exposure event with a defined probability ( $\alpha = 1.51$ ). Transmission between birds occurs via flocking interactions.
Parameters: The model incorporates epidemiological parameters including a latent period ( $2 \pm 1$ days) and infectious period ( $7 \pm 4$ days). Species-specific mortality rates are applied (70% for gulls, 10% for waterfowl). A hygiene parameter is included to simulate the dampening effect of biosecurity measures on virus replication.
Experimental Design: Simulations were run for 100 days (100 ticks) with five replicates for each of four predefined incursion regions: Northern Agricultural Region (WA), Rangelands Region, Southwest Queensland, and Limestone Coast (SA). A single 365-day run was conducted to assess long-term epidemic trajectories. Initial infections were seeded randomly within the chosen incursion region.

Key Results
The simulation results indicate that the location of the initial incursion is a critical determinant of outbreak severity, duration, and geographic spread.

Incursion Region Variability:
- Limestone Coast (SA): This region produced the most severe outcomes. It resulted in the highest prevalence of infection, peaking at 58.5% in the Australian Capital Territory (ACT) and 23.8% in Tasmania at day 99. Crucially, this scenario led to a massive number of breached farms, with Tasmania experiencing a breach prevalence of 97.9% and the ACT 60%.
- Northern Agricultural Region (WA): This scenario resulted in the highest prevalence within Western Australia (33.8% at day 75) but showed no spread to other states or farm breaches outside WA.
- Rangelands Region: Produced low prevalence (max 3.1% in WA) and minimal farm breaches.
- Southwest Queensland: Resulted in negligible spread, with infection confined to Queensland and peaking at 0.067% on day 1.
Spatial Dynamics: The model demonstrated that bird movements are heavily influenced by environmental productivity and elevation. Birds tended to aggregate in Victoria and South Australia due to high-productivity wetlands. The "Central Channel Country" (flowing toward Lake Eyre) acted as a significant convergence zone.
Long-term Trajectories: Over a 365-day period, the Limestone Coast scenario showed a clear increasing trend that plateaued, suggesting sustained transmission. In contrast, the Northern Agricultural Region scenario maintained a stable, low-level infection, while Southwest Queensland and South Australia Arid Lands saw infections fade to near zero.
Farm Breaches: Breaches were strongly correlated with geographic proximity to high-density bird aggregations and productive habitats. The model suggests that while biosecurity is robust, the sheer density of birds in high-productivity wetlands near farms creates unavoidable exposure risks.

Key Contributions

Spatial Risk Assessment: The study provides a granular, state-level assessment of H5N1 spread risks, demonstrating that not all incursion points carry equal risk. It identifies the Limestone Coast and the ACT/Tasmania corridor as critical high-risk zones for both wildlife and poultry.
Mechanistic Insight: By linking bird movement to energy budgets and habitat productivity, the model offers a mechanistic explanation for why certain regions (like the Limestone Coast) facilitate rapid, widespread transmission while others (like Southwest Queensland) act as dead-ends due to elevation and distance from high-density habitats.
One Health Framework: The model explicitly connects poultry farm breaches with wild bird movements, highlighting the risk of spillover to marine mammals and endangered species in ecologically sensitive breeding areas.

Significance and Claims
The author claims that this ABM serves as a flexible tool for informing national preparedness and resource allocation. The study emphasizes that:

Targeted Surveillance is Essential: Surveillance efforts should be prioritized west of the Great Dividing Range, specifically in South Australia and around Kati Thanda–Lake Eyre, where the model predicts high bird densities and transmission potential.
Incursion Location Matters: The severity of an outbreak is not uniform; it is heavily dependent on where the virus first enters the continent.
Wildlife Vulnerability: The risk extends beyond poultry to Australia's unique biodiversity, particularly in areas where infected birds interact with susceptible marine mammals (e.g., seals) and endangered species.
Policy Utility: The model can be used to simulate specific incursion scenarios to guide biosecurity tightening and wildlife risk management strategies.

The paper maintains a modest tone regarding its limitations, acknowledging that the model simplifies complex factors such as wind and weather patterns, and that "farm breaches" in the model represent relative risk proxies rather than confirmed infection events. The author concludes that while strong biosecurity measures are in place, the unique nomadic behavior of Australian waterbirds and the high pathogenicity of the current H5N1 clade necessitate continued vigilance and adaptive surveillance strategies.

Technical Summary: Evaluating the Internal Spread Risk of HPAI H5N1 in Australia

More like this