How Sea Turtles Find Their Way Home

Two numbers are enough to find one beach

Junior level — plain language, no maths

A female sea turtle hatches at night, scrambles down a beach into the surf, and vanishes. She spends the next twenty or thirty years crossing entire oceans. Then, as an adult, she comes back - often to within a few kilometres of the beach she was born on, having never once returned in the meantime. There are no signposts in the open sea and nothing to see. So what is she steering by?

Here is the trick, and it is worth taking slowly. The Earth is a magnet, and a compass needle does not just point north - if you let it tip freely, it also dives into the ground at an angle. Near the equator the field runs almost flat. Near the poles it plunges almost straight down. So the steepness of that dive tells you how far north or south you are. That is one number.

But one number is not enough, and the simulation above shows you exactly why. Match only the dive angle and a whole line lights up across the map - you could be anywhere along it, hundreds of kilometres from home. You need a second, independent reading, and the field provides one: how strong it is. Strength also changes across the map, but along a slightly different direction. So it draws a second line - and two lines that are not parallel cross in exactly one place.

That crossing point is a magnetic address, and it is what a turtle uses. It seems she memorises the address of her home beach in the first hours of her life, while crawling down the sand to the water. Decades later, out in the open Atlantic, she compares what she feels now with what she remembers: too steep means go south, too weak means head for the coast. She follows the mismatch until both readings match at once - and that is the beach.

Inclination, intensity and a bicoordinate fix

Student level — the core equations

Navigation to a goal from unfamiliar territory needs two separate things. A compass tells you which way you are pointing; a map tells you where you are. Plenty of animals have a magnetic compass. Sea turtles are among the very few known to have a magnetic map, and it works because the geomagnetic field carries two independent, position-dependent quantities.

The first is inclination, the angle at which field lines dip below horizontal. For a dipole field the relation is exact and beautifully simple: \(\tan I = 2\tan\lambda_m\), where \(\lambda_m\) is magnetic latitude. Inclination is 0° at the magnetic equator and 90° at the poles, so it is essentially a latitude readout. Its contours - isoclinics - run roughly east-west.

The second is total intensity \(F\), which ranges from about 25 to 65 µT over the globe. It also varies with latitude, but the real field is not a perfect centred dipole: crustal and core anomalies tilt its contours - isodynamics - away from the isoclinics. That non-parallelism is the entire point. Two families of contours that cross at an angle define a unique intersection, exactly as latitude and longitude do. Where they run parallel, the fix degenerates into a line and the system fails.

The evidence is direct. Kenneth and Catherine Lohmann placed hatchling loggerheads in coil systems reproducing the field of remote points on the North Atlantic gyre. The turtles oriented along the heading that would keep them inside the gyre from that location - a place-specific response to a field they had never physically visited, and adaptive in each case, since the wrong heading means being swept into lethally cold water. In the simulation, drag her off station and watch the two lines separate from the remembered pair; the navigation rule is simply to reduce both mismatches at once.

What a map sense must compute, and the receptor nobody has found

Scholar level — full mathematical depth

01Compass, map, and the difference between them

True navigation - reaching a goal from a site never previously visited, without following a known route - requires positional information. A compass is insufficient in a strict logical sense: displaced to an unknown location, a compass-only animal can hold any heading but cannot determine which heading to hold. It must first answer where am I, and that demands at least two scalar fields whose values vary independently with position. The geomagnetic field is the only global scalar field an animal can plausibly sense that offers two.

02Why the two coordinates work, and where they fail

For a geocentric axial dipole, inclination is a pure function of magnetic latitude via \(\tan I = 2\tan\lambda_m\), and intensity varies as \(F = F_0\sqrt{1+3\sin^2\lambda_m}\) - so for a perfect dipole both are functions of latitude alone, the contours are parallel, and the pair specifies a line, not a point. The map only works because the real field is not a centred dipole: non-dipole terms from the core and crust rotate the isodynamics relative to the isoclinics by an angle that varies regionally. The quality of the fix is set by that angle, and along stretches of coast where it approaches zero the system is intrinsically ambiguous. This is a genuine constraint, not a quibble, and it predicts where natal homing should be least precise.

03The experimental case

The Lohmanns' coil-array work remains the strongest evidence for a bicoordinate magnetic map in any animal. Tethered hatchling loggerheads exposed to fields replicating widely separated points on the North Atlantic gyre oriented along regionally appropriate headings - not a single fixed direction, but a different, adaptive direction for each simulated location. Later work showed juveniles could be conditioned to distinguish magnetic signatures of feeding sites, demonstrating learned magnetic positional discrimination rather than only innate responses.

04The resolution problem

Natal homing precision of order tens of kilometres implies intensity discrimination well under 100 nT against a background near 45,000 nT - a part in several hundred - and inclination discrimination of a small fraction of a degree. No known receptor mechanism comfortably delivers that. Three families compete: magnetite-based transduction, where biogenic single-domain \(\text{Fe}_3\text{O}_4\) torques against the field; radical-pair chemistry in cryptochromes, whose singlet-triplet yield depends on field orientation and so gives an inclination compass but no obvious intensity magnitude sense; and electromagnetic induction, which requires the highly conductive ampullary organs of elasmobranchs and does not apply here. No magnetoreceptor cell has been unambiguously identified in any vertebrate. The behaviour is far better characterised than the organ.

05Secular variation as a natural experiment

Because the geomagnetic field drifts, following a signature and following a place make divergent predictions, and the planet supplies the manipulation free of charge. Brothers and Lohmann analysed nineteen years of loggerhead nesting along the Florida coast against modelled isoline motion: nesting density rose where isolines converged and fell where they diverged. Natal homing in this species is therefore best described as homing to a magnetic signature which, on average, happens to sit near the natal beach - an important distinction, since the two come apart over decades.

06What imprinting has to be

If the signature is learned rather than inherited, the learning window is extraordinarily brief: the crawl from nest to surf, plus perhaps the first hours offshore. That imposes hard requirements - the value must be acquired in minutes, stored for two to three decades without rehearsal, and remain retrievable with enough fidelity to steer a final approach. No other vertebrate memory system is known to hold an analogue scalar that long to that precision. It is also testable in principle: displacing eggs between beaches should shift adult homing to the incubation or emergence site, and the partial evidence from translocated clutches is consistent with that, though the experiment takes thirty years to run.