Estimating epidemiological parameters of highly pathogenic avian influenza in common terns using exact Bayesian inference.

Abstract

Highly pathogenic avian influenza (HPAI) is a contagious viral disease that has led to the culling of huge amounts of poultry as well as the mortality of countless wild birds. The recent panzootic that began in 2021 has been particularly notable for its devastating effect on seabird populations around the globe. Whilst transmission of HPAI within poultry has been relatively well studied, the recency of the current panzootic, combined with data collection challenges, means that much less is known about key epidemiological parameters, such as reproduction numbers, $R_0$ , of HPAI in wild populations. We develop methodology to carry out exact Bayesian parameter inference using reversible jump Markov chain Monte Carlo applied to mortality data in the form of daily carcass counts over the duration of subsequent outbreaks in a colony of common terns, Sterna hirundo, in 2022 and 2023. We estimate $R_0$ to be 3.7 (95% CI $2.3;7.2$ ) in 2022, and 3.2 (95% CI $1.7;7.0$ ) in 2023. The probability of mortality for an infected bird was estimated to drop from 0.26 (95% CI $0.24;0.28$ ) in 2022 to 0.14 (95% CI $0.11;0.20$ ) in 2023. Our findings furthermore suggest direct bird-to-bird transmission to be the predominant driver of infection within the colony, with environmental transmission playing a negligible role. We interpret our results to suggest that daily carcass removal may have kept environmental transmission at bay and that increased immunity and/or a change of the strain of HPAI may have caused the drop in mortality, but that facilitating 'social distancing', for example by providing more suitable breeding habitat, such that breeding densities can be reduced, will be key to reduce disease transmission in colony-breeding seabirds such as the terns.