Zoonotic bacteria of the Borrelia burgdorferi sensu lato complex are transmitted to humans by ticks and cause Lyme disease, which has infection foci through North America, Europe, and Asia. Human risk is partly shaped by variation in the relative abundance of competent reservoirs, host species than can transmit bacteria to uninfected larval ticks. While life history strategy is implicated as a driver of competence among mammal reservoirs, competence appears more idiosyncratic among avian reservoirs. Yet because many birds have relatively high tick burdens, can transport infected larvae over long distances, and often occupy suburban environments, resolving the interspecific drivers of competence could have implications for human risks. Here, we systematically collated data on reservoir competence (i.e., the ability of birds to infect larvae or the presence of infected larvae on wild birds) for 160 avian species. To learn the trait profile of a competent bird and to guide future studies of bird–pathogen interactions, we applied boosted regression trees to binary competence data. We used a suite of species traits that could affect tick exposure and host competence, weighted all trees by the number of studies per bird species, and used a 70:30 data split to train and then test the model.
Results/Conclusions
Of the 160 bird species included in our database, 79 showed evidence of reservoir competence for bacteria in the B. burgdorferi sensu lato complex. B. garinii was the most common species (32%), followed by B. burgdorferi sensu strico (18%), B. valaisiana (18%), and B. afzelii (16%). Our boosted regression model characterized competent avian species with 78% accuracy. Traits most important for describing competent birds included geographic range size, incubation time, population trends, body mass, fledging age, maximum lifespan, clutch size, and diet, suggesting that competent avian reservoirs are geographically widespread, fast-lived, non-threatened, and less-insectivorous species. We applied our model to trait data for 11,000 birds to identify 22 unsampled species with a high probability of being competent (95th percentile) on the basis of trait similarity with known competent birds. These include the northern mockingbird, Pallas’s bunting, the sedge warbler, the green-tailed towhee, the yellow-legged buttonquail, the peaceful dove, the savannah sparrow, the eastern meadowlark, the American goldfinch, the African blue tit, and the white-eyed vireo. These results highlight potential bird species and geographic regions as high-priority targets for Lyme disease surveillance