Binoculars to bird flu: using the citizen science database 'eBird' to inform avian influenza surveillance on wild waterfowl in the Fraser Valley, British Columbia, Canada

Wild waterfowl are the natural reservoir for avian influenza (AI), a serious viral pathogen in domestic poultry. Wild waterfowl are presumed to transmit AI onto farms during seasonal migrations, making them a major focus of AI... [ view full abstract ]

Wild waterfowl are the natural reservoir for avian influenza (AI), a serious viral pathogen in domestic poultry. Wild waterfowl are presumed to transmit AI onto farms during seasonal migrations, making them a major focus of AI surveillance. However, effective AI surveillance is limited by the paucity of data on local wild bird populations in many jurisdictions, including the Fraser Valley region (British Columbia), which is a high-density region of poultry farming that has experienced multiple AI outbreaks. During the 2014/2015 outbreak it was recognized that 'eBird', an expert-moderated online citizen scientist database containing wild bird observations, may surmount this knowledge gap.
To determine if eBird data are effective at characterizing local waterfowl populations and can inform regional wild bird AI surveillance by identifying (a) optimal timing for surveillance, (b) specific waterfowl species that could be associated with AI outbreaks in poultry, and (c) ideal wetlands for a novel environmental surveillance tool. Species abundance and observer effort were collected from eBird for 29 local species of waterfowl across the Fraser Valley between Mar 2004 and Feb 2015. A general additive model (GAM) was constructed to determine predicted seasonal and inter-annual trends in species abundance. eBird also identified 15 wetlands within the outbreak area via waterfowl population density and diversity (sediment collected from these wetlands during the 2014/2015 outbreak underwent targeted re-sequencing for AI detection and characterization).
EBird was able to capture seasonal trends in waterfowl abundance and preliminary analysis suggests mid-Sep to early Dec as the optimal surveillance time-frame. However, there was no consistent trend in species’ abundance in relation to timing of past AI outbreaks. The H5N2 outbreak virus was detected in 9/15 wetlands identified by eBird.
EBird can be used to determine regionally-appropriate surveillance time-frames and site selections for use in future AI surveillance.