Flight Speeds among Bird SpeciesAllometric and Phylogenetic Effects

A tracking radar dish sweeps the night sky over southern Sweden. On the scope, blips move through the dark — dozens of them, at different altitudes and different speeds. An observer with a telescope identifies each one by species as it passes through the beam. The radar logs range, elevation, and bearing every ten seconds. And behind all of this data is a simple aerodynamic prediction, one that every single one of those blips was supposed to obey. Most of them didn't. That is the setup for a study by Thomas Alerstam and colleagues, published in 2007, that tracked the flapping flight speeds of one hundred thirty-eight bird species ranging from ten grams to ten kilograms — three orders of magnitude in body mass — and tested one of the cleanest predictions in animal locomotion science. The prediction comes straight from physics. Lift, the force that keeps a bird in the air, depends on air density, wing area, the wing's lift-producing efficiency, and the square of flight speed. In steady horizontal flight, lift must balance weight. So a bird with a small wing area relative to its weight — what aerodynamicists call high wing loading, defined simply as weight divided by wing area — has to move faster to generate enough lift. Because lift scales with the square of speed, cruising speed should scale with the square root of wing loading. That's an exponent of zero point five. Add a geometric assumption — that wing area scales predictably with body mass across species — and you get a second prediction: cruising speed should scale with body mass to the one-sixth power, or about zero point one six seven. These aren't arbitrary numbers. They follow directly from the physics of flight under the assumption that birds are, to a first approximation, geometrically and dynamically similar to one another. To test those predictions, Alerstam and colleagues needed clean airspeed data across a wide range of species. They got it by combining two large radar datasets. Their own data from Sweden and the Arctic, collected between 1979 and 1999 at five sites in southern Sweden plus two icebreaker expeditions, yielded one thousand three hundred ninety-nine radar tracks of one hundred two visually identified species, with tracks averaging about six minutes each and altitudes reaching up to three thousand six hundred meters. Wind at each bird's altitude was measured by releasing balloons carrying radar reflectors, and those wind vectors were subtracted from the radar-derived ground speeds to get true airspeeds. Then the team took one more step: they converted true airspeeds to equivalent airspeeds — Ue — by adjusting for the lower air density at altitude. Since lift depends on air density, a bird flying fast at altitude might be working no harder than a bird flying slower at sea level. Equivalent airspeed corrects for that, putting all species on a common footing. A second major dataset from Bruno Bruderer's group, collected across Switzerland, Germany, Israel, and Spain using the same radar methods, added sixty-four more species. Combined, the two datasets gave one hundred thirty-eight species. For the twenty-eight species that appeared in both, the mean equivalent airspeeds matched closely — a reassuring sign that the measurement approach was consistent. Now for the result. Across all one hundred thirty-eight species, equivalent airspeed ranged between about eight and twenty-three meters per second. That's a roughly threefold range in speed across a one thousand fold range in body mass. Already that should feel compressed. And when Alerstam and colleagues ran the regressions, the numbers confirmed it. Equivalent airspeed scaled with body mass to the power zero point one three, not the predicted zero point one six seven. It scaled with wing loading to the power zero point three one, not the predicted zero point five zero. These aren't trivially different. The wing loading exponent is barely more than half of what aerodynamic theory demands. Phylogenetically corrected analyses — using independent contrasts to remove the statistical dependence among related species — gave essentially the same answer: exponents of about zero point one two for mass and zero point three two for wing loading. The compression of the speed range is real, and it doesn't go away when you control for shared ancestry. Think about what that means concretely. Wing loading across the species in this dataset spans roughly a tenfold range. With an exponent of zero point five, as theory predicts, that should produce a three point two fold range in cruising speed. With the observed exponent of zero point three one, the actual speed range is only about twofold. Small, low-wing-loading birds fly faster than they should; large, high-wing-loading birds fly slower. Something is squeezing the distribution toward the middle. Part of that squeeze has a geometric explanation. Alerstam and colleagues found that aspect ratio — wingspan squared divided by wing area, a measure of how slender the wings are — increases significantly with both body mass and wing loading. Larger, more heavily loaded birds tend to have more slender wings. Slender wings are aerodynamically more efficient: they generate lift with less drag, which means a bird can sustain flight at somewhat lower speeds than its wing loading alone would demand. So the geometry is doing some of the work. The team tested this directly by including aspect ratio in their statistical models. But here is the key finding: aspect ratio only partially explains the compression. When aspect ratio was added to models that already included wing loading or phylogenetic group, it did not improve model fit by the Akaike information criterion. The wing loading exponent remained far below zero point five even after accounting for wing shape. Something else is going on. That something else turns out, in large part, to be evolutionary history. This is where the study's findings become genuinely striking. After accounting for mass and wing loading, the single most powerful predictor of cruising speed was which broad lineage a bird belonged to. Alerstam and colleagues grouped species into six major monophyletic groups — evolutionarily coherent lineages — and found that phylogenetic group alone explained fifty-five percent of the variation in equivalent airspeed, an adjusted R-squared of zero point five five. A model combining wing loading with phylogenetic group was the best-fitting model overall, reaching an adjusted R-squared of zero point six four. Body mass alone, by contrast, explained only about twelve percent of the variation. Wing loading explained close to half. But neither alone, nor together, came close to the explanatory power that phylogeny added. The pattern shows up as consistent group-level deviations from the overall scaling lines. Birds of prey and herons fly more slowly than their mass and wing loading would predict. Songbirds and many shorebirds fly faster. These aren't random scatter — they're coherent signals tied to lineage. The authors note that flight mode likely plays a role: different groups use different patterns of continuous flapping, intermittent flapping with short glides, or bounding flight, and these modes have different relationships between speed and power. Muscle physiology, wing kinematics, and vortex wake structure — all of these differ systematically among lineages and could all contribute to the phylogenetic signal. But Alerstam and colleagues are careful not to pin it on any single mechanism. Phylogeny, as they frame it, is an empirical summary of the correlated functional adaptations and constraints that evolution has built into each lineage. It predicts speed well without yet telling us exactly why. Step back and the picture that emerges is this. Birds are not passive aerodynamic objects that obey scaling laws the way aircraft do. Evolution has actively shaped their cruising speeds, and the result is a compressed distribution — what the authors call a speed corridor. Fly too slowly, and you're vulnerable to wind, predators, and inefficient muscle operation. Fly too fast, and landing becomes difficult, maneuvering suffers, and energetic costs climb. Natural selection appears to have pushed against both extremes, producing a speed range that is far narrower than aerodynamic scaling alone would allow. Aspect ratio differences account for part of that compression geometrically. Phylogenetic history accounts for much of the rest empirically. But the full mechanistic explanation — why the wing loading exponent is zero point three one and not zero point five — remains open. Kinematic differences, muscle power-speed relationships, and aerodynamic wake structure are all candidates, and the radar data, for all its scale and precision, can tell us what the speeds are but not yet why evolution settled on them. That is the next question this study hands to the field: one hundred thirty-eight species tracked through the dark, their speeds measured to a fraction of a meter per second, and a pattern that the simplest physics cannot fully explain. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.