Waterbird movements could help predict global avian influenza hotspots

A global modelling study proposes a way to identify where avian influenza spillover is most likely to occur by measuring not only where waterbirds are present, but how intensively they use an area throughout the year. The approach identifies potential exposure hotspots covering about 14% of the world’s land surface and points to major surveillance gaps, particularly in sub-Saharan Africa.

The researchers developed a metric called waterbird activity entropy (WAE) using monthly distribution models for 779 waterbird species worldwide. Unlike simple species-richness maps, the metric incorporates both the diversity of waterbirds and the amount of time different species are likely to remain in a given area.

The objective was to determine whether these patterns could help explain where avian influenza viruses, particularly H5N1, are detected and where contacts with humans, poultry and cattle may create opportunities for spillover.

A strong relationship with reported avian influenza

The WAE model showed substantial explanatory power for the global distribution of avian influenza detections, with an AUC of 0.87 after the researchers attempted to exclude outbreaks likely to have resulted mainly from poultry-to-poultry transmission. The model remained robust under several alternative analyses, including when poultry-adapted subtypes were excluded. This does not mean that bird activity alone predicts an outbreak. Rather, the study suggests that areas repeatedly used by diverse waterbird populations can provide a useful ecological indicator of where exposure to avian influenza viruses is more likely.

14% of global land identified as potential hotspots

Using its WAE threshold, the study identified avian influenza exposure hotspots covering approximately 14% of global land area. The concentration of livestock and people within these areas is particularly striking. Hotspots located in the United States, European Union, China and India contain an estimated:

• 52% of the human population globally exposed within the modelled hotspots

• 41% of cattle

• 51% of poultry.

These overlaps illustrate why interactions between wild birds, livestock and humans remain central to avian influenza risk assessment.

Waterfowl remain the priority surveillance group

The analysis confirms the particular importance of waterfowl. In the United States, waterfowl represented 84% of confirmed waterbird hosts of avian influenza viruses in the dataset and showed strong spatial overlap with cattle. In China, waterfowl activity was also strongly correlated with poultry density.

Seabirds emerged as another important group, particularly in coastal regions. In the EU, seabird-associated areas accounted for the overwhelming majority of the hotspot area attributed to this functional group. The authors argue that future surveillance should therefore become more targeted toward waterfowl and seabirds rather than treating all wild birds equally.

A major surveillance gap in Africa

One of the strongest findings concerns sub-Saharan Africa. The model identifies more than 300 million hectares of potential hotspot area, around 15% of the global total, in sub-Saharan Africa, yet the region accounts for less than 1% of reported positive avian influenza cases.

Countries including Tanzania, Ethiopia and Zambia each contain more than 34 million hectares of predicted hotspot territory despite recording few or no cases in the underlying surveillance data. For the authors, this discrepancy is unlikely to mean that risk is absent. Instead, it may reflect substantial differences in surveillance intensity between regions.

From outbreak detection to ecological early warning

The study's main contribution is therefore methodological. Most avian influenza risk maps rely heavily on previous outbreak records. But those records are influenced by how intensively countries test animals and report infections. Areas with poor surveillance may consequently appear artificially low-risk.

By incorporating bird ecology independently of reported outbreaks, WAE could help identify "silent areas" where surveillance should be strengthened before large numbers of cases are detected.

The authors nevertheless stress that the indicator is not a stand-alone forecasting system. It captures aggregated waterbird activity and does not determine which individual species are most important in transmitting particular virus strains.

Wild-bird monitoring could become more efficient if it focuses not simply on migration routes, but on areas where waterbird diversity, residency time and densities of poultry, cattle or humans overlap.