Environmental transmission is likely a primary driver of transmission for influenza A viruses (IAVs) in waterfowl hosts due to their natural ecology (e.g., dabbling behavior) and the ability of IAVs to survive for long periods in cold water. In order to test the relative importance of direct (bird-to-bird) transmission and environmental (via a shared water resource) transmission routes, we conducted a series of transmission experiments in which eight groups of four contact ducks were sequentially exposed to a single infected focal mallard. Each contact group was only allowed exposure to a focal duck for a time period long enough to allow infection of contacts but not shedding. After exposure to the focal duck, contact groups were removed and then housed individually until we could ascertain their individual infection statuses. We repeated this procedure for six focal mallards and their associated contact groups.
We used Bayesian heirarchical logistic regression to quantify the transmission probability to contact ducks as a function of the viral shedding curves in focal ducks (direct) and the concentration of virus in shared water (environmental). Models were implemented in WinBUGS software and compared using DIC.
Results/Conclusions
Our models show that contaminated water is the primary driver of transmission compared to direct contact. Water transmission was included in the top four models tested and there were only minor differences in AIC values for the water only model compared to models that included water and other sample types. The model with water viral load as the only predictor performed well with an AUC of 0.92. This model shows that mallards can become infected at very low concentrations of virus in the water, with contact ducks having a 50% probability of infection at 215 In addition, the experimental results showed high levels of individual heterogeneity in the transmission probability associated with a given viral load. This variability suggests other, unmodeled factors, such as behavioral differences between individuals, likely influence the probability of infection.