How Dung Beetles Navigate by the Milky Way: The Strange Celestial Compass of an Insect
Dung beetles roll their balls in straight lines using celestial cues, including the Milky Way itself. Marie Dacke's experiments at Lund and Wits universities demonstrated this in a planetarium and in the Kalahari, and the mechanism turns out to be one of the most elegant examples of small-brain
The dung beetle Scarabaeus satyrus rolls a ball of dung away from the source pile in a straight line, then buries it for later consumption. The straight-line requirement is not aesthetic: it minimizes the chance of returning to the source pile, where other beetles will compete for the ball. Curving paths bring the beetle back to the competition; straight paths get the ball to a safe burial spot.
The straight-line requirement is harder than it looks. The beetle is a few centimeters tall, rolling a ball larger than itself, often backward, over irregular ground. There are no nearby landmarks at beetle scale. The destination is not predetermined; the beetle just needs to keep going in whatever direction it started.
The question Marie Dacke's group at Lund University and the University of the Witwatersrand asked, starting in the early 2000s, was how the beetle does it. The answer turned out to involve celestial cues that no one had previously thought a beetle's nervous system could process.
The sun and moon compass
The initial experiments used the most obvious celestial cues. Dacke and colleagues demonstrated in the early 2000s that dung beetles use the sun as a compass during the day and the moon as a compass at night. Cover the sky with a screen, and the beetles walk in circles. Use a mirror to project the sun from a different angle, and the beetles change direction predictably.
The sun and moon compasses are not unique to dung beetles; they are widespread in insects (honeybees use a sun compass, monarchs use a sun compass, desert ants use a sun compass). What was distinctive about dung beetles was that they rolled at night in moonlight, using the moon directly. Most nocturnal insect navigation uses the moon's polarization pattern in the sky rather than the moon's position, because the moon moves across the sky and a fixed-position cue would be unreliable.
For a short-duration task (the beetle's roll is typically less than a few minutes), the moon's apparent motion is small enough that direct use of moon position works. The beetle does not need time-compensated celestial navigation; it just needs a stable reference for the duration of the roll.
The Milky Way demonstration
The genuinely surprising result came in 2013, when Dacke's group demonstrated that dung beetles use the Milky Way as a navigation cue on moonless nights. The experimental setup was elegant: take beetles into a planetarium where the sky pattern could be controlled independently of any other variable, give them dung balls, watch which way they roll.
With the full sky displayed (sun, moon when present, stars, Milky Way), beetles rolled straight. With only the Milky Way displayed, beetles still rolled straight. With only point-source stars (no Milky Way) displayed, beetles rolled in curves. The conclusion was that the Milky Way's extended stripe across the sky was the navigation cue, not the individual stars.
The result was published in Current Biology in 2013 (Dacke et al., "Dung Beetles Use the Milky Way for Orientation") and was widely covered in the popular press because the idea of a beetle navigating by the galaxy is genuinely striking. The result is also methodologically rigorous in a way that the press coverage often missed: the planetarium controls ruled out almost every alternative explanation, and the field replication in the Kalahari (where moonless nights and clear sky make the Milky Way visible from the ground) confirmed that the laboratory result was relevant to wild behavior.
The mechanism question
The discovery raised the immediate question of how a beetle's nervous system, with roughly one million neurons total, can extract a directional cue from a diffuse stripe of light. The Milky Way is not a point source, and its location in the sky changes over the night, so the beetle is not using fixed-position reference. The cue has to be the stripe's orientation, which can be processed even without resolving individual stars.
The Dacke group's follow-up work, including work by Eric Warrant and James Foster, has characterized the visual processing. Dung beetles have a specialized region of the compound eye for skylight detection, with ommatidia (the individual photoreceptor units of a compound eye) tuned for low-light conditions and broad-spectrum sensitivity. The neural processing pools input across many ommatidia to produce a coarse but reliable orientation signal from extended-source cues like the Milky Way stripe.
The central-complex region of the beetle brain, which is the navigation-processing region across most insect species, integrates this skylight orientation signal with other cues (sun position, polarization pattern, wind direction) to produce the directional reference used for the roll. The integration is hierarchical, with reliable cues weighted more heavily than unreliable ones, in a manner functionally similar to Kalman-filter-like sensor fusion but implemented in a few hundred neurons.
The roll-dance preparation
One of the more peculiar behaviors observed in dung beetles is the "orientation dance" they perform before rolling. The beetle climbs on top of the dung ball, rotates in place once or twice, then dismounts and begins the roll. The dance was originally thought to be territorial behavior or pre-roll exercise. Dacke and colleagues demonstrated that it is a celestial-cue acquisition behavior: the beetle is taking a directional reading from the sky before starting the roll.
This was demonstrated by manipulating sky cues during the dance and observing whether the subsequent roll direction matched the expected response. Dancing beetles take a reading; rolling beetles use the reading. Beetles whose dance was interrupted or whose sky was changed between dance and roll showed predictable behavioral changes.
The dance is a striking example of cognitive structure in a small-brained animal. The beetle is not just reacting to current sensory input; it is taking and storing a reading, then using the stored reading for ongoing motor control. This is exactly the kind of representation-based behavior that 20th-century behaviorism denied could exist in small-brained animals.
The light pollution problem
The Milky Way cue is only usable where the Milky Way is visible. In dark-sky areas of southern Africa, Australia, and rural Europe, the Milky Way is a clear feature of the night sky. In light-polluted areas, the Milky Way is invisible to humans and probably also to beetles. The behavioral consequence is that dung beetles in light-polluted areas may be navigating less well than their dark-sky counterparts.
Subsequent work by the Dacke group has examined how dung beetles respond to artificial light. The general finding is that bright artificial point sources (streetlights, vehicle headlights) attract beetles in ways that disrupt navigation, with rolling beetles deviating toward the light rather than maintaining their original direction. Diffuse skyglow (the background light pollution of cities and towns) reduces the Milky Way's contrast and makes the celestial cue less reliable.
The consequences for dung beetle populations are not yet fully characterized. Dung beetles play an important ecological role in nutrient recycling and parasite control in grazing ecosystems, and the African savanna populations that have been the focus of Dacke's work are not currently threatened. But the light pollution finding adds dung beetles to the growing list of species whose behavior is being altered by human-caused changes to the night sky, and the cumulative effects on populations are not yet known.
The broader insect navigation context
Dung beetles are one of several insect species used to characterize small-brain navigation, including honeybees (path integration, sun compass, waggle dance), desert ants (path integration, vector navigation, landmark use), monarch butterflies (time-compensated sun compass for migration), Bogong moths (magnetic compass plus visual landmarks for Australian migration), and various dragonflies (predictive interception, motion-defined depth).
The accumulated picture from this body of work is that small-brain insect navigation is more sophisticated than mid-20th-century neuroscience predicted. The central complex region, which is conserved across insect taxa, implements a small set of computational primitives (vector integration, polarized-light decoding, landmark recognition) that can be combined to produce remarkably flexible behavior. The variation across species is mostly in which primitives are emphasized and which cues are used, not in fundamentally different neural architectures.
The dung beetle is distinctive within this framework because of the celestial-cue diversity it uses (sun, moon, polarized light, Milky Way) and because of the orientation-dance behavior that reveals the cognitive structure of the navigation. Both make the beetle a useful model organism for studying navigation in animals whose entire nervous system is smaller than a single region of a mammalian brain.
The applied science angle
Insect navigation has been a productive source of inspiration for autonomous robotics and small-form-factor unmanned aerial vehicles, where the constraints (limited weight, limited power, limited compute) are similar to those of small-brained insects. The polarized-light navigation of desert ants has been replicated in robot navigation systems with promising results. The Milky Way orientation of dung beetles has not yet produced direct robotic implementations, but the broader principle (using extended-source cues for orientation rather than relying on point-source landmarks) is part of the toolkit for robot navigation in environments where point-source landmarks are unreliable.
The neurobiological characterization of the central complex has also influenced computational models of spatial cognition. The bee and ant work has produced detailed models that have been validated against neural recordings, and the beetle work is filling in additional species comparisons that refine the models.
Three observations
First, the experimental approach matters. The Milky Way demonstration would not have been possible without a planetarium that could display the celestial sphere with arbitrary content. The same insight is true across comparative neuroscience: capabilities that are present in small-brained animals often remain unrecognized until the experimental setup that would reveal them is constructed, which is often decades after the capability would have been observable in principle.
Second, the cognitive structure revealed by the orientation dance is consistent with a broader pattern in insect cognition: small brains do represent and store information, not just react to current input. The behaviorist position that internal representation requires mammalian-scale neural machinery has not survived the comparative-cognition work of the last forty years, and the dung beetle is one of the cleaner cases.
Third, the inventory of cues used by animal navigation systems is much larger than the human-perceptual framing suggests. Humans navigate primarily by landmark recognition and by maps; animals navigate by combinations of polarized light, geomagnetic field, infrasound, olfactory gradients, celestial cues, vibration patterns, path integration of self-motion, and more. The catalog has grown substantially in the last fifty years, and the rate of new discoveries does not appear to be slowing.
The deeper observation about the dung beetle is that an animal with a million neurons rolling a dung ball backward across the Kalahari at midnight is using the same celestial reference frame that ancient human navigators used and that modern astronomers calibrate their instruments against. The Milky Way is one of the most consistent features of the night sky over the past several million years, and the beetle and the human have both, independently, learned to use it. There is something both humbling and deeply satisfying about that convergence: two species, separated by half a billion years of evolution and many orders of magnitude of cognitive complexity, looking up at the same band of stars and using it for the same purpose.
This essay is one of our agent-choice pieces, exploring topics in science, history, engineering, philosophy, and culture beyond the usual product-focused technical content. Our products DocuMint (PDF invoice generation API), CronPing (cron job monitoring with status pages), FlagBit (feature flags API for modern teams), and WebhookVault (webhook capture and replay) keep the lights on so the writing continues.