Machine Learning and Duck Butts with Holly Ganz and Ladan Doroud

I spoke to Holly Ganz and Ladan Doroud about their recent paper in mSystems.

Like Comment

Community-Level Differences in the Microbiome of Healthy Wild Mallards and Those Infected by Influenza A Viruses

Holly H. GanzLadan DoroudAlana J. FirlSarah M. HirdJonathan A. EisenWalter M. Boyce



Waterfowl, especially ducks and geese, are primary reservoirs for influenza A viruses (IAVs) that evolve and emerge as important pathogens in domestic animals and humans. In contrast to humans, where IAVs infect the respiratory tract and cause significant morbidity and mortality, IAVs infect the gastrointestinal tract of waterfowl and cause little or no pathology and are spread by fecal-oral transmission. For this reason, we examined whether IAV infection is associated with differences in the cloacal microbiome of mallards (Anas platyrhyncos), an important host of IAVs in North America and Eurasia. We characterized bacterial community composition by sequencing the V4 region of 16S rRNA genes. IAV-positive mallards had lower species diversity, richness, and evenness than IAV-negative mallards. Operational taxonomic unit (OTU) cooccurrence patterns were also distinct depending on infection status. Network analysis showed that IAV-positive mallards had fewer significant cooccurring OTUs and exhibited fewer coassociation patterns among those OTUs than IAV-negative mallards. These results suggest that healthy mallards have a more robust and complex cloacal microbiome. By combining analytical approaches, we identified 41 bacterial OTUs, primarily representatives of Streptococcus spp., Veillonella dispar, and Rothia mucilaginosa, contributing to the observed differences. This study found that IAV-infected wild mallards exhibited strong differences in microbiome composition relative to noninfected mallards and identified a concise set of putative biomarker OTUs. Using Random Forest, a supervised machine learning method, we verified that these 41 bacterial OTUs are highly predictive of infection status.


Seasonal influenza causes 3 to 5 million severe illnesses and 250,000 to 500,000 human deaths each year. While pandemic influenza viruses emerge only periodically, they can be devastating—for example, the 1918 H1N1 pandemic virus killed more than 20 million people. IAVs infect the respiratory tract and cause significant morbidity and mortality in humans. In contrast, IAVs infect the gastrointestinal tract of waterfowl, producing little pathology. Recent studies indicated that viruses can alter the microbiome at the respiratory and gastrointestinal mucosa, but there are no reports of how the microbiota of the natural host of influenza is affected by infection. Here we find that the mallard microbiome is altered during IAV infection. Our results suggest that detailed examination of humans and animals infected with IAVs may reveal individualized microbiome profiles that correspond to health and disease. Moreover, future studies should explore whether the altered microbiome facilitates maintenance and transmission of IAVs in waterfowl populations.

Ben Libberton

Science Communicator, Freelance

I'm a freelance science communicator, formerly a Postdoc in the biofilm field. I'm interested in how bacteria cause disease and look to technology to produce novel tools to study and ultimately prevent infection.