Using remote sensing to examine wintering waterfowl distributions and ecology in the United States to create a novel avian influenza virus biosecurity tool for the poultry industry
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
The ongoing global HPAI outbreak in commercial poultry, first reported in the USA in February 2022 has emphasized the need for improved food biosecurity and novel approaches pertaining to nationwide wild bird surveillance. The Mid-Atlantic (MA) and Central Valley of California (CVC) are critical sites for waterfowl species in the winter, which promotes the interface between wild and domestic birds and increases the risk of an HPAI outbreak. Improving the understanding between predictors of wintering waterfowl distributions on both the fine and broad scale will allow facility managers to make more informed biosecurity decisions. Using the network of US NEXRAD’s and fine scale GPS-tracking of known HPAI reservoirs; Greater Snow Geese (Anser caerulescens atlanticus) (N = 59) and Canada Geese (Branta canadensis) (N = 9), my objectives were to 1) model winter waterfowl distributions at the broad scale in the MA and CVC as a function of weather, temporal, and environmental characteristics using boosted regression tree modelling and 9 years of NEXRAD data (2014–2023); 2) use high frequency GPS-tracking of geese for four winters (2019–2023) to demonstrate how a percentile risk ranking system can be used to gauge exposure to wintering waterfowl prior to an HPAI outbreak and discuss a risk management analysis framework for facility managers; and 3) use high frequency GPS-tracking of geese for four winters (2019–2023) to model waterfowl exposure in goose points per hour at individual poultry facilities as a function of temporal, geographic, and landcover predictors. ☐ I was able to capture the variability in effect size of 28 different covariates across space and time within two geographic regions which are critical to nationwide poultry and egg production. In general, environmental and geographic predictors had the strongest relative effect on predicting wintering waterfowl distributions in both regions, while land cover composition had regional and temporal specific effects. Daily mean temperature was a major predictor of increasing waterfowl distributions in both regions throughout the winter, while precipitation had differing effects, increasing waterfowl densities in the MA, while decreasing in general within the CVC. Increasing waterfowl densities in California are strongly tied to the flooding of the landscape and rice availability, whereas waterfowl in the MA, where water is less limiting, are generally governed by waste grain availability and emergent wetland on the landscape. Waterfowl distributions in the MA generally increased when distances to the Atlantic coast and lakes decreased, while increasing waterfowl distributions in the CVC were more strongly tied to decreasing distances to lakes. ☐ I demonstrated how a percentile ranking system based on goose points per hour at individual poultry facilities may be successfully used as a metric to predict AI outbreaks at poultry facilities within the Delmarva Peninsula in the Mid-Atlantic region of the US. Seven facilities that had an H5N1 outbreak during a 34-day window were within the 78th – 99th percentile of risk based on this system, with one facility even being ranked within the ‘top-25’ of 6,021 facilities. I was also able to document the first ever fine scale movement of wintering waterfowl in proximity to poultry facilities, with the minimum distance for Canada Geese (CANG) and Greater Snow Geese (GSGO) being documented at 3.9 m and 1.7 m respectively to poultry facilities. Exactly one-third of all Delmarva Peninsula poultry facilities had waterfowl present across winters, which is an alarming rate given our small sample size relative to the entire populations of both species. ☐ I was able to capture the spatial and temporal variability in mean exposure rate of geese in proximity to poultry facilities within a region of high poultry farm production across four winters. Time of the winter was the best predictor of goose exposure for both species, with GSGO generally greater during mid-late winter and CANG fairly consistent across the entire winter. Colder winter temperatures (particularly below 0°C) exponentially increases the rate of GSGO exposure at farms as they feed on waste grain to offset energy deficits. Goose exposure was similarly high for both species across the winter during both the day and night and emphasizes the need for 24 hr surveillance. Facilities that were closer to sewage treatment facilities, national wildlife refuges, waterbodies, or the coast had higher relative goose exposure, highlighting the importance of these sites at various times through the winter for geese. Many inland sites also have relatively high rates of goose exposure and geographic features such reservoirs, lakes, quarries, and gravel/sand pits may play a disproportionate role in supporting large numbers of geese.
Description
Keywords
Avian influenza virus, Biologging, Waterfowl distributions, Geese, Poultry