Networks on the wing

Devices called proximity loggers are helping researchers understand how a barn swallow interacts with its flight-mates—and what connectedness could mean for an individual’s fitness.

By Laura Booth. Photographs by Matt Wilkins and Iris Levin.

In the shabby but well-loved barn where I spent most of my childhood, a coterie of domestic fowl was ever-present. My mother, who dotes on her hens (and ducks, and guineas, and, intermittently, quail), anthropomorphizes her broods in a way that would make any good scientist cringe. But none of these tame birds captured her imagination like the pair of barn swallows who took up residence with us when I was a teenager.

Thinking of the barn swallows brings me back to July, to the sultry summer evenings after the horses had been turned out for the night and my parents and I would sit on the barn aisle, watching as the male and female (or, “husband and wife,” so my mother insisted) swallows leapt and dove in intricate flight over the lush pasture, punctuating their dance by slicing scythe-like over each other in midair. For my family and I, it felt natural to associate their swift, elegant gestures as a conversation between partners. Watching them sit on the paddock gate, a comfortable distance apart, brought to mind a contented couple enjoying the hard-won peace of familiar silence.

But as Dr. Iris Levin, a post-doctoral researcher working in Dr. Rebecca Safran’s lab at the University of Colorado, Boulder, can tell you, interpreting the meaning of social interactions between individual barn swallows is far from intuitive.

Levin first became interested in barn swallow sociality after reading the results of a study by Dr. Safran and her colleagues; they found that sex hormone concentrations and body mass in male swallows were altered after researchers experimentally darkened their breast feathers. Safran’s earlier work had demonstrated that the hue of a male’s feathers determines his luck in mating: females consistently prefer darker to paler males.

The findings of the manipulation experiment indicate that, for a male barn swallow, “there’s some feedback from his color, from the way he looks, into his physiology,” says Levin. “And it’s undoubtedly coming from how he’s being treated by other birds.” She was intrigued: “When I saw that, I thought, social behavior has to be mediating this interaction between physiology and phenotype,” she says. And she immediately wanted to know: “Can we actually measure that?”

This was no rhetorical question; counting and describing swallow interactions is tricky because of the birds’ small size and swift flight. In long-term avian studies, researchers attach colored bands to birds’ legs to keep track of individuals in a population. It’s possible to visually associate which individuals spend time with which other individuals—a technique for observations of social behavior that biologists call “the gambit of the group,” first introduced in 1999—using these color bands, but only during birds’ slower interactions.

Levin explains, “Let’s say you have a feeder in your backyard and you have a color-banded population of chickadees. Every morning, for five minutes, you document the birds who are together at the feeder, and you see that red-black is always with green-yellow. There are ways, statistically, to quantify social networks from these group associations…in a fairly robust manner.”

But Levin wanted to do better than the gambit of the group, which is limited in one important way: its measurement of a social interaction is subjective. Using this method, a biologist is the one who decides how close a bird must be to another bird for those individuals to be ‘interacting.’

“Everyone comes up with what they argue is a biologically relevant proximity,” Levin says, citing examples from past studies: lizards who share the same den, fish who swim in a single shoal, beetles within a certain distance on a log. “But we’re still all guessing, because none of us are our own study organisms.” By using the gambit of the group alone, Levin would not have been able to distinguish between the close, midair interactions that my parents and I once found so thrilling and instances of individuals loosely associating while kicked back on a telephone wire. To do that, she needed a way to count which individuals were interacting and how far apart they were from each other.

Eventually, Levin turned to a technological solution in a miniature package: automated tags that produce information about which birds are interacting, how close together they are, and how long the interaction lasts.

These tags, which birds wear in the form of tiny backpacks, are formally called ‘digital radio transceiver proximity loggers,’ and represent a sophisticated adaptation to clunkier technologies developed in the recent past. Previous proximity devices were limited to big collars by the weight of heavy batteries; they were intended to measure the movements of large mammals, such as dairy cows, so that people didn’t have to spend hours recording cow interactions.

But researchers quickly noticed that such devices could do more than make studying behavior of large animals more convenient: they offered a window into studying small animals in a way that wasn’t possible before. For such study systems, Levin says, “Proximity tags are going to make a massive difference.”  

Levin first deployed proximity loggers on her birds during the summer 2014 breeding season, when parent swallows were tending to their second clutch of chicks for the year. When two tagged birds came together, their proximity loggers detected each other and began recording data, which was later rerouted to Levin’s laptop for analysis.

After her 2014 field season ended, Levin headed back to the lab to build a social network based on her proximity logger-generated barn swallow interaction data. She was shocked to find just how breathtakingly close swallows came when rapidly sweeping and diving over one another in flight. “I literally have encounters that last for one pulse only, which means the encounter is anywhere up to 20 seconds long,” she says “Based on the radio signal strength between the tags, these birds are on top of each other.”

With this information in hand, speculating about how a social interaction for a barn swallow looks is no longer the question that keeps Levin up at night; rather, her mind is now occupied with the issue of what these interactions can tell us about barn swallow biology.

Levin wants to know how social connectedness—whether a swallow is popular or unpopular— corresponds to an individual’s reactivity to stress. Her preliminary results indicate that, among other factors, the answer depends on whether she analyzes loose interactions between birds (within five meters) or very close interactions (within ten centimeters). At close proximity, the birds with the most social interactions, regardless of their sex, also produce the highest levels of stress hormones in response to a stressor. Does this mean the most interactive, ‘popular’ birds are also the most stressed out?

That remains an open question. “My pet hypothesis is that it’s reflecting something about social status,” Levin says. “It seems like captive studies show that subordinates are more stressed out, and that wild studies show that the dominant individuals are more stressed out.” This isn’t surprising, Levin explains: “If you’re a subordinate in a cage with a bunch of animals and you can’t cope or get away, maybe you’re going to suffer more stress.” In the wild, conversely, “Maintaining dominance is actually kind of stressful.”

For now, Levin is most excited simply to be able to ask such fine-scale questions about social networks—questions that would have been limited in the past by a lack of technological ability to answer them.

According to Levin, the advantage of proximity loggers over a gambit of the group approach is a reduction in the bias introduced when scientists decide how close ‘interacting’ individuals must be to each other. “There’s so much bias when you do behavioral observation,” she says. “It’s biased because we’re not perfect observers, it’s biased because we push the animals around, it’s biased because some animals don’t mind that you get close to them and others do.” By using proximity loggers that record interactions along a spectrum of distances, Levin says, “I think we can really eliminate a lot of that bias.”

Nevertheless, Levin remains circumspect about the role technology can play in helping us answer biological questions. “My first field experience was spent watching hours and hours of savanna sparrows finding nests, watching behavior, documenting behavior, and figuring out where birds territories were. And that was so important,” she says. Rather than “strapping backpacks on your birds and walking away,” she argues, researchers can most effectively use automated technologies as a complement to prior research that has established a natural history framework in which to interpret the data. “I think that if we didn’t know anything about barn swallows, we’d be really bad at figuring out what this stuff means,” she says.

Levin’s reservation on the subject of technology hints at what is, in my opinion, one of the subtlest beauties of science as a practice: it is built on past human endeavors, especially ones that failed or didn’t turn out as the researcher expected (an idea that is more fully considered in Columbia neuroscience professor, Dr. Stuart Firestein’s, book “Ignorance: How it Drives Science”). Only with the context established by Dr. Safran’s work did the question of connections between social interactions and physiology in barn swallows arise. Proximity tag technology provides a means for accessing robust data for this analysis, but it hasn't tidily answered Levin’s questions. Rather, it has brought a suite of new questions to the fore, all of which have yet to be explained.

In this way, the role of technology is integral but secondary to the spark of human curiosity that drives us to figure out, for example, whether the most socially connected swallows really do mount different stress responses from their less-connected counterparts. And now that Levin has constructed her first social networks using data obtained from the proximity loggers, Levin’s questions have evolved: her next project is to repeat Dr. Safran’s phenotype manipulation experiment to see whether altering an individual bird’s phenotype changes its position in the social network, as well as whether stress responses track with those changes in a predictable way. “I don’t think any of the work that I’ve done yet…is all that revealing,” she says modestly. “It’s nice that it confirms what we already knew. But what we’re really excited about is going out and messing with it.”