Knowledge about species’ distributions is the foundation of biodiversity conservation. While distributions are typically visualized as static range maps, many species have strongly dynamic distribution patterns . Here we describe our approach at visualizing the dynamic patterns of migratory bird populations to more accurately convey distributional information and to study the links between populations and environments across the full annual cycle. We use a variety of data visualization approaches to (1) convey the dynamic nature of hundreds of migratory bird species distributions across the Western Hemisphere, (2) display detailed weekly abundance patterns for each species, (3) present the amount of time and area a species occurs across seasons, and (4) indicate the changes in habitat use across seasons. Methods. We train an ensemble model that learns the associations between observations gathered by the citizen science project eBird and environmental features using NASA earth imagery. Next we predict weekly relative abundance at a 3.3 × 3.3 km spatial resolution within the northern portion of the Western Hemisphere (>3 million predictions per week). Models are run on the Azure cloud computing service using R and Hadoop workflows. A team of statisticians, ornithologists, and designers then analyze model output and develop data visualizations. Results. From the model output we generate weekly maps of distribution and display them as animations. In contrast to typical static, uniform, and coarsely defined distributions represented by geographic range maps, our results clearly show a species’ annual distributional dynamics. Next, we estimate species’ spatial patterns of duration across the annual life cycle based on the weekly estimates of relative abundance. When displayed on a map, these patterns show striking differences in occupancy and duration across a species’ full annual distribution. Finally, we quantify associations between species’ occurrences and land cover diversity and composition. We then estimate how these associations vary through an entire year and across multiple regions. We then use ‘cake plot’ visualizations to demonstrate seasonal shifts in land-cover associations for migrating birds. Conclusions. Our approach is the first to visualize the distributional dynamics of entire populations of migratory bird species across the full annual cycle, and highlight the strong spatiotemporal variation that has been obscured by traditional range and distribution visualizations. By visualizing annual distributional dynamics, we provide a robust and objective ecological baseline for assessing the implications of climate and land-use change for migratory bird species and the development of comprehensive conservation strategies.
