Publication: Using museum specimen records to show migration in Painted Buntings
Migration is closely associated with birds in the popular imagination, from the writings of Aristotle and Homer to modern documentary films. Which is why it may be surprising to learn tracking migratory birds remains a significant scientific problem. While radio telemetry and GPS geologgers can both reveal the movements of birds across the landscape, they are limited to larger species that can tolerate their weight, and to small sample sizes due to their expense. Small samples sizes, never ideal in science, are a particular problem here because of the inherent stochasticity of migration: individuals may veer off course, and are frequently subject to predation. Because of this, it’s hard to know whether those individuals researchers do successfully track are representative of their population or species.
An alternative to this “direct” approach of inferring migration patterns is to use population level data and determine whether densities of a particular species in a particular region change over time, possibly indicating many individuals of that species are going somewhere else (e.g., migrating). For example, the citizen science initiative eBird records bird observations from across the globe and archives them in a freely accessible online database. These records can be then used to visualize patterns that would otherwise escape us, as my labmate CJ did by taking records of Rufous Hummingbirds from western North America and plotting them across the months to show their circular migratory pattern around the Rockies (see the captivating .gif here).
However, though eBird’s international records are improving, its data remains biased toward the United States and Western Europe. Aggregated museum specimen records offer relatively large datasets with better geographical and historical representation, and are a natural extension of this indirect (or metanalytical) approach. These data do, however, come with their own biases, particularly regarding the lack of any standardized information around collecting effort, or how much time a given scientist spent sampling birds in a given region. Without this information, it’s hard to know whether a large number of individual specimens represent high population densities, or simply the fact that the collector spent a long time in that area.
To overcome this problem, Sievert Rohwer (emeritus curator of ornithology at the Burke) developed a formula known as an abundance index, or AI for short.* AIs correct raw specimen counts of a particular species at a particular time and place for collecting effort by transforming them into a ratio over the total number of birds of other species at the same time and place expected to be collected in a similar way. What this means is that using AI values, species are assumed to be abundant only if their specimens make up a high proportion of the total load of individual birds in that region and interval. Sievert and colleagues have already used abundance indices to illustrate migratory double breeding and faunal change in NW Mexico, and the method seems likely to continue to provide opportunities to use natural history museum collections in new ways.
Which brings us back to migration. Recently, I worked with Sievert and colleagues at University of Oklahoma and Universidad Nacional Autónoma de México to apply AIs to the question of where Painted Buntings (Passerina ciris) went after arriving in Mexico for the winter from their breeding grounds in the southern US. Both Sievert and the team at Oklahoma suspected the western population of Painted Buntings (our focus) first arrived in Sinaloa in the fall before moving down the coast later in the winter, eventually traveling back to their breeding grounds up the gulf coast. In other words, we thought the birds were tracing a circle around Mexico’s central mountains.
Using a comprehensive database of all Mexican birds in museums worldwide assembled by Adolfo G. Navarro-Sigüenza, we plotted AI values across the months to visualize any potential patterns in space and time. We then tested for the statistical significance in the change in bunting densities among regions (in case our eyes were tricking us), and using remote sensing data, tested to see if these changes we observed were correlated to changes in live green vegetation (known as the “green wave” hypothesis). The result? Our hypothesis of “circular” movement was supported, and bunting specimens were nonrandomly distributed on areas with the greatest amount of live green vegetation across the winter.
There’s a lot more you could do with AIs and museum specimen records, which seem like they might be particularly useful for determining crude population trends in difficult-to-survey regions. In the mean time, you can read our paper here.
*Interestingly, a group of Japenese botanists developed the abundance index method independently, which they used to assess changes in medicinal plant population sizes with herbarium records.