Guest Post: Starling Murmuration, by Harmony Lu

Harmony, working at a field site in New Mexico

Hello! You may be surprised to see so much text below, considering the fact that I just announced a brief hiatus! Well, you are in for a treat. My good friend, Harmony Lu, who is studying Geology-Biology at Brown, has written a captivating post on the mechanics behind large-scale bird flocking. Specifically that of starlings, who are capable of choreographing their flight to an astounding degree of synchronization. When they do this, the flock resembles an enormous, mutable cloud. As if the birds were listening to music at a frequency beyond our grasp, they coordinate their movements like synchronized swimmers, and the flock assumes different shapes and patterns on time. To see starling murmuration in action, make sure to watch the video at the bottom of the post.

When Harmony first discovered starling flocking, she asked the question that everyone would — “How do they do that??” — and then set out to answer it herself. She presents her findings in the post below, which goes through the science of modeling bird flocks, theories on how starlings accomplish their astonishing feat, and the implications of understanding birds’ flocking mechanisms. Thanks, Harmony, for this illuminating piece!  – Steph

Starling Murmuration, by Harmony Lu

At the high school I attended, everyone in band class also participated in marching band. Thus, every fall for several years, I spent countless hours on the football field and in parking lots dressed sort of like a toy soldier, attempting to move in a coordinated fashion with approximately 100 other ungainly teenagers in a pseudo-militarized fashion while making noise that was supposedly music. And for all that effort, I am pretty sure that we could never achieve marching in a straight line.

Perhaps this experience has led to my fascination with bird flocking. Watching birds fly with seeming effortlessness in groups riddles me with both jealousy and curiosity.

Of the most incredible bird flocking patterns is starling murmuration: when up to tens of thousands of European starlings, Sturnus vulgaris gather and fly together in a breathtaking dance. During murmuration, a starling flock turns, twirls, expands, contracts and plays in the sky as a smokey tuft of living birds. Birds don’t crash into each other or get stranded from the flock and left behind. So how do hundreds to thousands of individual birds all navigate in one cohesive bunch?

Starling murmuration over a beautiful sunset. Similar to schools of fish, starling flocks accomplish harmonious movements across an extraordinarily large scale

Well, before getting into that, let’s start with “why bother?” Behavioral scientists have amassed a large amount of evidence that gathering and moving in large groups can make predation more difficult. Many species of insects, fish and mammals, as well as birds, use these techniques. Further studies on birds show support that associating larger groups can increase the average amount of food that a bird consumes. This is likely because each bird can spend less time on activities that end up split amongst the flock, such as looking out for danger, and can then more time foraging for food. In the case of starlings, murmuration is a nightly ritual during the winter and also serves a few other purposes. Before settling down to roost for the night, flying around in the murmuration may help warm the birds up. The large starling flock also organizes during the seeming spontaneous flight – by the time that the birds actually land to sleep, they have arranged according to a social hierarchy.

So how does this happen? Although at times in the past it was proposed that bird flocks were controlled by the gods or that they could communicate with all of the flock members through some form of telepathy, scientists now believe that this is a remarkable occurrence of “self-organized collective behavior”. Essentially, this means that each of the individual birds is interacting locally with certain others (not even considering most of the massive flock), but because every bird in the flock follows similar behavior patterns, a cohesive behavior as a flock emerges.

Based on this theory, scientists have been modeling flocks using computer simulations since the 1980’s. All birds in the flock are controlled by a certain algorithm. The first of these computer models were based on the principle that each bird responded to all other birds within a certain distance, which is called “metric” organization. Each bird was programmed to continuously maintain specific distance-based relationships with the nearest birds: get close but not crash into neighbors and travel in the same direction. These models could create cohesive flocking patterns, which suggested that this might be how flocks are actually organized.

Unfortunately, these computer-simulated flocks failed to show the same fast responses to predator signals that real ones do. Furthermore, at the time that these models were developed, there wasn’t a good way to test whether the computer-generated flocks were actually behaving like real starling flocks (which is a pretty integral step in testing a model’s accuracy). Since then, some researchers have worked on methods to extract 3D positioning information from video footage of starling flocks. A team of European researchers working under the group name STARFLAG – Starlings in Flight, collected and analyzed footage of starling flocks in Rome. They used two stereo cameras set 25m apart and could shoot up to 8 consecutive seconds of photographs at 10 pictures per second. Although they shot over 500 of these 8 second “clips”, only 10 of them were suitable to convert entirely into 3D information.

Still, when these data about true flocking structures became available in 2008, researchers quickly realized that the metric algorithms weren’t quite right. Instead, the STARFLAG team came up with an alternative algorithm from studying the 3D flocking footage. Instead of maintaining the specific relationships with starlings that are just a certain distance away, each starling in their model tracks seven other birds, regardless of distance. This has been called “topological organization”. This newer theory has several advantages: not only does it match the data better, but it explains how the flock structure can change so rapidly. For example “compactness” of the flock can change drastically without risking it disintegrating entirely. But it turns out that even topological models have struggled to reproduce flocking behavior that truly emulates groups of birds in real life, so clearly there is still work to be done.

Understanding large flocking mechanisms isn’t just cool because of a scientific curiosity (although it is pretty awesome by that alone), but engineers and economists hope to put some of these principles to use for human systems. Some uses include large swarms of aerial surveillance robots that may fly together in groups. But for now, we still don’t know quite how nature does it. Even if we do come up with an accurate algorithm for creating such responsive self-organized behavior, I think we will still be able to watch and be amazed when thousands of birds can dance through the air so easily.

A lone starling, perched on a cable.

Some resources and links about starling murmuration:

Ballerini, H et. al. 2008. Empirical investigation of starling flocks: a benchmark study in collective animal behavior. Animal Behaviour. 76: 201-215.

Beauchamp, G. 1998. The effect of group size on mean food intake rate in birds. Biological Reviews of the Cambridge Philosophical Society. 73: 449-472.

Highfield, R. 2008. Study of starling formations points way for swarming robots. The Telegraph.

Friederici, Peter. 2009. Flight Plan. Audubon Magazine:


4 thoughts on “Guest Post: Starling Murmuration, by Harmony Lu

  1. Harmony
    I really like your post. I continue to be fascinated by the flocking of birds. I am a scientist though not in this area, but emergent patterns in nature are a recurrent theme in my art. What is interesting is how differently flocks of different species behave: this week I was observing flying formations of starlings (small groups of maybe up to 500), geese and lapwings. So whatever the underlying algorithms, I guess these are species specific. Interesting that you draw the parallels with human groupings also. I hope you continue to post on this.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s