Identifying migratory birds’ population bottlenecks in time and space

The annual disappearance and reappearance of billions of migratory animals from seasonal environments across the planet is a stunning natural phenomenon. Naturalists have marveled for more than 2,000 y at these migrations, at least since Aristotle pondered birds’ seasonal disappearance. Birds are the preeminent seasonal commuters because of their physiological capacity for sustained long-distance flight, extraordinary navigational adaptations, and diversity of movement patterns; and since Aristotle’s time we have learned much about how and why birds migrate (1). What’s been tougher to decipher is how and where migrants’ populations are controlled. Sampling and tracking the millions of individuals in many species is challenging because of far-flung seasonal movements that can span entire continents and shuttle between them, the baffling array of ways to migrate, and the potential for population limitation in summer, winter, spring or fall transits, or some combination thereof. Overcoming many of these challenges, Kramer et al. (2) in PNAS contribute significantly to understanding migratory birds’ population bottlenecks by linking different distribution patterns in two closely related Vermivora wood warblers (Parulidae) to ecological conditions precipitating decline in a geographically distinctive population of one of the two species.

Proliferation of technological and conceptual advances is powering a renaissance in migratory birds’ population biology. For example, widespread populations sometimes show a strong migratory connectivity pattern, defined by a tendency for individuals that breed near each other also to winter near each other (3) (see Fig. 1 for a hypothetical example). Strong migratory connectivity is now documented in a variety of species using stable isotope ratios (e.g., ref. 4), genetic population structure (5), and light-level geolocators (e.g., refs. 2 and 6) and more spatially accurate GPS tags (7). Small migratory birds can carry these devices without ill effect, but birds must be recaptured near the deployment site, typically a year later, to retrieve the device and download the archived data, making representative sample sizes labor-intensive. Another important technological advance is full annual cycle tracking. For example, one can now study ecological circumstances in one season that carry over into a subsequent season (3) using a variety of methods, including intrinsic stable isotopic signatures of birds occupying different quality wintering habitats, that subsequently determine timing of northward migration with consequences for breeding fitness (e.g., refs. 8 and 9). Comprehensive citizen science, long-term continental-scale population databases—including the Christmas Bird Count, Breeding Bird Survey (BBS), and e-Bird—can be linked to complete annual-cycle and range-wide ecological conditions, including remotely sensed data. For example, El Niño–La Niña rainfall fluctuations, manifested as satellite-detected vegetation greenness where American redstarts (Setophaga ruticilla) winter, accounted for BBS population fluctuations in eastern but not western North American breeding populations, presumably by influencing winter–spring survival (10).

Fig. 1.

Hypothetical migratory species distribution incorporating strong migratory connectivity. Two breeding subpopulations (different colors) undergo chain migration (northern breeding population migrates to northern wintering area, and southern breeders winter further south). Higher fitness natal dispersal movements within subpopulations are indicated with solid black lines; higher fitness migration routes with solid blue lines. Lower fitness natal dispersal and migration indicated with dashed lines. Weak migratory connectivity (not illustrated) would be indicated by individuals from each breeding population migrating to either wintering region with no fitness penalty.

Kramer et al. (2) advance understanding in two important ways. First, they combine multiple technologies to distinguish among all of the possible ecological circumstances that could be driving population decline. Using light-level geolocators strapped to the backs of 41 recaptured individuals deployed across their breeding range, Kramer et al. show that golden-winged warblers (Vermivora chrysoptera) exhibit strong migratory connectivity. Moreover, the more southern Appalachian Mountain breeding population has declined precipitously, corresponding with relatively severe conversion of native forest to other land uses in this population’s northern Colombian wintering areas. As controls, the Great Lakes population of golden-winged warblers and the entire low-connectivity blue-winged warbler (Vermivora cyanoptera) population (n = 25 geolocators) were relatively steady, corresponding with less deforestation in Central American wintering ranges. This study thus integrates geolocators, long-term (since 1966) BBS population trends, and historical forest land-cover data from appropriate winter regions. The two Vermivora species breed, choose habitat, and feed similarly, making it less likely that the breeding grounds or en route migration locations are contributing to the population dynamics. Thus, a strong case is made for habitat restoration in the South American wintering areas of the Appalachian breeding population of golden-winged warblers. Kramer et al. also provide corroborative data from other declining migratory birds that winter in the same South American regions.

Second, Kramer et al. (2) integrate ecological and evolutionary approaches to help understand population changes in these two Vermivora species. Biologists have been studying ecological and evolutionary aspects of migrant birds for a long time (e.g., ref. 11), but rarely integrate these different approaches in the same study. Kramer et al. (2) suggest that forest loss in northern South America, where the Appalachian population of V. chrysoptera winters, has created an ecological trap that threatens continued decline and possible extirpation of this population if it cannot quickly adapt to its wintering area. This study also exemplifies a comparative (evolutionary) approach involving two closely related species, which helps control for many different factors that could plausibly cause population change. Even more importantly, Kramer et al. use this comparative approach to focus on the question of why some migratory species have evolved strong connectivity in the first place. This is an evolutionary question that will require additional emerging technologies to address. Specifically, it is useful to know what genetic differences have become fixed in the two subspecies of golden-winged warblers, and what traits are thereby coded, by way of keeping the two breeding populations from mixing more (i.e., the pattern suggested by the strong connectivity in this species).

Understanding the evolution of migratory connectivity patterns is an important research frontier. The kinds of structured populations in strong-connectivity species imply not just ecological but also evolutionary population independence. To see how this might arise evolutionarily, consider American redstarts (S. ruticilla), a widely distributed species with strong north–south connectivity (12), as illustrated hypothetically in Fig. 1. Redstarts also undergo chain migration, in which northern breeding populations migrate to northern wintering sites and southern breeders to southern wintering sites. In parallel, redstarts also have a north–south gradient in nesting behaviors, including clutch size, timing of breeding, and nest size and location in the vegetation (13), generally heritable traits in birds. Plausible selection gradients include predator types and climate. Such patterns suggest that individual redstarts dispersing too far to the north or south of their natal population would try to reproduce under inappropriate conditions, with reduced success. This should select evolutionarily for reduced natal dispersal distance, thus evolutionarily reinforcing the population isolation implicit in strong connectivity. The separation of wintering populations could be maintained by migration timing. For example, birds wintering in regions in which food supply favors earlier migration (the southern hypothetical population in Fig. 1) should have higher fitness migrating to the southern breeding range, where conditions become favorable for breeding earlier in the spring.

Strong migratory connectivity allows for, and probably results from, birds becoming locally adapted both in summer and winter.

Kramer et al. in PNAS contribute significantly to understanding migratory birds’ population bottlenecks by linking different distribution patterns in two closely related Vermivora wood warblers (Parulidae) to ecological conditions precipitating decline in a geographically distinctive population of one of the two species.

Identifying the ecological factors underlying such adaptations is an important research frontier, as is understanding how connectivity is related to differences among species in natal dispersal. Such information can help structure range-wide population models, including geographic patterns of isolation, and the implicit resistance to gene flow. A variety of ecological population models are available for wide-ranging migratory birds (14), including spatially structured network models (e.g., ref. 15), but these models have yet to incorporate genetically structured populations to address the kinds of evolutionary questions motivated by species differences in migratory connectivity, and the kinds of conservation problems addressed with golden-winged warblers (2).

Rapid advances in genomic technologies now allow studying how evolution has shaped migrant bird distributions and ecologies. For example, a scan of the Wilson’s warbler (Cardellina pusilla) genome using SNPs has facilitated a comprehensive map (genoscape) of its migratory connectivity (5). More recently Bay et al. (16) show with another parulid, the yellow warbler (Setophaga petechia), how to integrate genome-wide information with ecological population dynamic approaches range-wide so as to better understand current and future impacts of climate change on local populations, some of which are declining in response to changing precipitation patterns, particularly in Rocky Mountain breeding sites. This study has also linked population changes to particular genes, associated with migratory and exploratory behavior.

What all these studies show in aggregate is that the tools are now available not only to help identify the population bottlenecks in threatened migratory bird populations, particularly those with strong migratory connectivity, but also to identify the ecological drivers and evolutionary processes involved in local population differentiation and adaptation. This genetically based geographic variation arises even in continuously distributed migratory birds, which should help address such fundamental questions as speciation, insofar as combined ecological and evolutionary technologies are revealing hitherto underappreciated within-species geographic variation on which speciation can operate.