When you are learning a new subject, one important first step is to learn the terms used in that field of study. That’s why if you took Intro Biology, for example, you probably used flashcards to learn the definition of terms like “resting metabolic rate,” “ectotherm,” and “keystone species.” When researching a new idea, scientists sometimes have to create new terms with agreed upon definitions- that’s not always easy, even if they are all humans and speak the same language. Now imagine that you are trying to define the terms of another species’ “language” that you do not speak. Consider the Black-capped Chickadee. Is “chicka-dee-dee” equivalent to “chicka-dee-dee-dee-dee”? What about “fee-bee” and “fee-bee-bee”? By “fee-bee-bee” do you mean the “cheeseburger” call, or the one my colleague calls the “hey, sweetie” call? Confusion ensues.

In a study published in the journal Bioacoustics at the end of July, IBP Acoustic Monitoring Specialist Katelyn “Keke” Ray, along with co-authors from Western Michigan University, Syracuse University, and the City University of New York, used an analytical method called “fuzzy clustering” to quantify and describe alarm calls in Red-winged Blackbirds. This work was part of Keke’s master’s thesis. Fuzzy clustering probably has the most approachable and fun name of any statistical analysis we’ve heard of, but it's a powerful, and relatively new, way to define vocalizations more quantitatively, especially “graded” vocalizations in which the same call type can confer different information depending on its length or some other variation (e.g., the “chicka-dee-dee…” call in chickadees).

Keke and colleagues recorded the vocalizations of 74 male Red-winged Blackbirds around Kalamazoo, Michigan to create a library of over 13,000 vocalizations. They then selected over 1,500 high quality recordings for analysis and included 50 different parameters, like duration or minimum frequency, per alarm call in the analysis. The researchers used a specific type of fuzzy clustering called “k-means” to identify the number of call types and the extent to which the calls shared acoustic features, and to define thresholds categorizing discrete, intermediate, and graded call structures. They found that male blackbirds used 6 different types of alarm calls: check, cluck, zeer, seee, se-er and ti-er. These call types shared acoustic features to varying extents. The researchers found that the “cluck” calls were discrete, meaning that there was little variation within that call type. The “zeer”, “see”, “se-er” and “ti-er” were graded calls, meaning that within each call type there was variation in one or more acoustic features (e.g., length, frequency, or number of syllables) and that there is more overlap between these different graded call types than between discrete call types. “Check” calls were intermediate, less discrete than “cluck” calls but less graded than the 4 graded calls. Previous studies of Red-winged Blackbird vocalizations defined calls qualitatively (subjectively) and some of these call types were lumped into a single type when the fuzzy clustering method was used. On the other hand, the fuzzy clustering method defined additional call types that qualitative methods lumped into a single call type: “cheer.”

We talked to Keke about the study and the interesting ideas that come up when studying animal communication.