Environmental persistence of influenza H5N1 is driven by temperature and salinity: insights from a Bayesian meta-analysis

Abstract

Environmental persistence of zoonotic pathogens is a key trait that influences the probability of zoonotic spillover. Pathogen survival outside of the host determines the window available for contact with the new recipient host species and the dose of pathogen available to that host. The longer a pathogen survives in the environment, the more disconnected the reservoir and recipient hosts can be in space and time, and the more likely that an infective dose will be available to recipient hosts. Therefore, environmental persistence is a key parameter for mechanistic models needed to predict pathogen spillover. Avian influenza can be transmitted from wildlife to poultry and people in part due to its ability to persist in the environment. Considerable work has been done to quantify trends in avian influenza persistence across environmental conditions, often published in separate studies with separate datasets. In this paper, we quantify the trends and variability of avian influenza viral persistence across environmental conditions by collating disjoint experimental data on viral particle persistence in water across many studies and a range of environmental conditions. The collated data represent 120 estimates from three different studies of the decay rates of highly pathogenic avian influenza H5N1 (90 estimates from Asia and 30 from Europe) in response to temperature, pH, and salinity. We analyzed these data with a Bayesian model to control for biases with random effects and used experimental replicates and R2 estimates of the publication's regression procedures as statistical weights. We found temperature significantly decreases persistence of H5N1 virus in water, and this effect is stronger than that of salinity alone. Salinity interacts with temperature and probably drives the most contrasting persistence scenarios between cold-saline and warm-saline water bodies, where highest and lowest persistence times could occur respectively. Our work provides needed parameters for models that examine the risk of spillover of avian influenza viruses.