Module 5: Magnetoreception

In 1976, Klaus Schulten proposed that a light-induced electron transfer between two molecules could produce a pair of radicals whose spin states evolve coherently under the combined influence of internal hyperfine couplings and the external geomagnetic field. Because the chemical reactivity of the radicals depends on whether their total electron spin is in a singlet (anti-parallel) or triplet (parallel) configuration, a small anisotropic magnetic field can bias the product yield along a specific direction. This was developed into a detailed biological hypothesis by Thorsten Ritz, Peter Hore, and colleagues (Biophys. J., 2000; Hore & Mouritsen, Annu. Rev. Biophys., 2016).

Spin Hamiltonian

For a radical pair with electron spins \(\mathbf{S}_1,\mathbf{S}_2\) and nuclear spins\(\mathbf{I}_i\) each coupled to \(\mathbf{S}_1\) by an anisotropic hyperfine tensor\(A_i\), the standard Hamiltonian is

Because the hyperfine tensor is molecule-fixed while the Zeeman term is lab-fixed, rotating the molecule relative to the field changes the eigenstate structure and therefore the singlet yield. The classic result is that the yield depends on the angle \(\theta\) between\(\mathbf{B}\) and the hyperfine axis—giving an intrinsic inclination compass without any need for polarity discrimination.

Why Earth's field works at all

The Earth's field is only 25–65 µT, giving an electron Larmor frequency of about 1.4 MHz. For the radical-pair compass to function, two conditions must be met. First, the hyperfine coupling must be comparable to the Zeeman coupling—otherwise the field is too weak to appreciably perturb the spin evolution. This matches measured cryptochrome hyperfine couplings of a few MHz. Second, the coherence time must be long enough for the field-dependent evolution to complete before radical recombination or decoherence wipes out the signal. Timmel et al. (Mol. Phys., 1998) showed that the optimum is \(\tau\sim 1\;\mu\)s — remarkably, the coherence times measured in isolated cryptochromes (Xu 2021) sit in exactly this range.

Cryptochromes (CRY) are blue-light photoreceptors found throughout the animal and plant kingdoms. They are structurally homologous to photolyases but carry out photoreception and circadian regulation rather than DNA repair. Upon absorbing a blue photon, the bound FAD cofactor is excited; a sequence of electron transfers through three or four tryptophan residues produces a transient FAD∙⁻–Trp∙+radical pair. This pair survives on the ~µs timescale needed for field-dependent spin dynamics and is the leading candidate for the primary magnetosensor.

CRY4 and the European robin

The landmark paper by Xu et al. (Nature, 2021) purified CRY4 from European robin (Erithacus rubecula) and performed direct in vitro spectroscopy under applied magnetic fields. The result: robin CRY4 exhibited a clear magnetic field effect (MFE) on its spin dynamics, with the sensitivity peaking near Earth's field strength. Critically, CRY4 from the non-migratory chicken (Gallus gallus) showed a much weaker MFE, and pigeon CRY4 (partial migrant) showed an intermediate effect—strongly suggesting that natural selection has tuned the protein's spin dynamics in migratory species. This was the first direct biochemical demonstration that avian cryptochrome magnetosensitivity is evolutionarily adaptive.

Inclination vs polarity compass

A seminal set of behavioural experiments by Wolfgang and Roswitha Wiltschko (Science, 1972;J. Exp. Biol., 1995) showed that European robins orient using the inclination of the Earth's field (the angle between the field vector and gravity) rather than its polarity. When the vertical component of the lab field is reversed, the birds reverse orientation; when only the horizontal component is reversed, they do not. This maps directly onto the radical-pair prediction: the compass is intrinsically inclination-based because it reads out the angle between \(\mathbf{B}\) and the hyperfine tensor axis, ignoring overall sign. Magnetite compasses, by contrast, are inherently polar.

The alternative—or complementary—magnetoreception mechanism is based on biologically-synthesised magnetite (Fe₃O₄). Blakemore (Science, 1975) first documented magnetotactic bacteria (Magnetospirillum magnetotacticum) that contain chains of ~20 nanometre-sized magnetite crystals, precisely tuned to the single-magnetic-domain size range (Moskowitz et al., Earth Planet. Sci. Lett., 1993). These bacteria use passive magnetic alignment to swim downward along Earth's field lines into oxygen-poor sediments.

The Fleissner group (Naturwissenschaften, 2003) described iron-containing structures in the upper beak of homing pigeons (Columba livia) that looked like magnetite sensors, and subsequent behavioural work suggested that pulse remagnetisation could disrupt pigeon homing. However, Treiber et al. (Nature, 2012) later showed that the cells originally identified as magnetite-containing sensors were, in fact, macrophages that had engulfed environmental iron particles. The status of avian magnetite sensors is therefore currently unresolved. Trout (Oncorhynchus mykiss) nasal tissue has magnetite-containing cells with clearer electrophysiology (Eder et al., 2012), and magnetite is also present in honeybee abdomens (Gould, 1978) and in the dermal tissue of various teleost fish.

The key experimental tool that distinguishes the two mechanisms is the pulse remagnetisation test: a brief, intense (\(>100\) mT) magnetic pulse permanently reorients single-domain magnetite particles but has no effect on radical-pair dynamics. Birds subjected to such pulses show altered orientation for several weeks, consistent with a magnetite-based “map” sense in addition to the radical-pair compass.

Loggerhead sea turtles (Caretta caretta) and green sea turtles (Chelonia mydas) provide the most rigorous evidence for magnetic map sense in vertebrates. Lohmann and colleagues (Nature, 2001; Curr. Biol., 2012) placed hatchling loggerheads in a large magnetic coil and exposed them to field parameters corresponding to different points along their natural Atlantic migration route. The turtles consistently oriented in the direction that would have kept them inside the North Atlantic gyre at each simulated location. They used two independent magnetic parameters—inclination and intensity—to form a 2D geomagnetic coordinate system.

Brothers & Lohmann (Curr. Biol., 2015) then analysed thirty years of nesting-beach return data along the east coast of Florida. As the geomagnetic field drifts (secular variation), the magnetic signature of individual beaches shifts over time; Brothers showed that female loggerhead distribution tracks these shifts exactly, providing the first population-level evidence that sea turtles imprint on the geomagnetic signature of their natal beach and return 30+ years later using that signature as a coordinate target.

The underlying receptor is believed to involve magnetite rather than radical pairs, because the turtles must integrate intensity information over multiple cues and operate at depth, without light. But this is not yet confirmed at the cellular level.

The monarch butterfly (Danaus plexippus) undertakes a 4000 km autumn migration from eastern North America to a handful of overwintering colonies in central Mexico. Reppert and colleagues (PLoS ONE, 2008; Nature Comm., 2014) showed that monarchs use a time-compensated sun compass based on their antennal circadian clock, and that this sun compass is complemented by a cryptochrome-based inclination magnetic compass. Mutation of CRY2 abolishes the magnetic compass without affecting the sun compass. Monarchs are thus currently the cleanest genetic demonstration of the radical-pair hypothesis.

Pacific salmon (Oncorhynchus) return to their natal rivers with uncanny precision after years at sea. Putman et al. (Curr. Biol., 2013) demonstrated that sockeye salmon use geomagnetic intensity and inclination together to re-locate their river entry point, exactly analogous to the sea-turtle case. Salmon have magnetite-containing cells in the olfactory epithelium (Walker et al., 1997) that provide the likely sensor.

Honeybees contain abdominal magnetite (Gould, 1978) and can be trained to associate a food reward with a specific compass direction; their waggle dance also becomes disoriented under artificially reversed fields. Whether the bee sensor is radical-pair or magnetite-based is unresolved.

Several controversial mammalian cases have been reported: resting cattle and deer align along the N–S magnetic axis (Begall et al., PNAS, 2008); dogs preferentially defecate aligned N–S (Hart et al., Frontiers in Zoology, 2013); foxes pouncing through snow preferentially attack toward magnetic north (¸ervený et al., 2011). These studies all survive statistical scrutiny but lack a confirmed sensor or neural pathway. Wang et al. (eNeuro, 2019) reported a human alpha-band EEG response to rotating magnetic fields, but replication remains ongoing.

Studying magnetoreception requires specialised methods that can unambiguously distinguish magnetic effects from other environmental cues. The main tools:

• Emlen funnels: paper-lined cones in which migratory birds hop during their nocturnal migration restlessness; footprints on the paper record preferred orientation. Developed by Stephen Emlen (1966).
• Helmholtz-coil cages: large-volume uniform fields that allow simulation of arbitrary field intensity, inclination, and declination.
• Conditioned-place preference: train the animal to expect reward in one field direction; test with the field rotated.
• Pulse remagnetisation: distinguishes magnetite from radical-pair mechanisms (see Section 3).
• Radio-frequency disruption: oscillating RF fields at 1–10 MHz disrupt radical-pair dynamics but leave magnetite unchanged. Ritz et al. (Nature, 2004) used this to confirm the radical-pair mechanism in European robins.
• Genetic knockouts: CRY1/CRY2 knockout flies (Gegear et al., 2008) lose the magnetic compass; same with CRY2 knockdown monarchs.

Earth field variation

The Earth's field intensity varies from about 25 µT near the magnetic equator to 65 µT near the poles; the inclination angle runs from 0° at the equator (field parallel to the surface) to ±90° at the poles (field vertical). Declination (the angle between magnetic and geographic north) varies both geographically and temporally, by up to tens of degrees over centuries. Any animal using the magnetic field as a positional cue must either tolerate this drift or update its calibration from alternative cues such as the sun, stars, or polarisation.

No migratory animal studied so far uses the magnetic field in isolation. The picture that has emerged over fifty years of behavioural experiments is of a hierarchy of cues: the sun, star patterns, polarisation of skylight, magnetic field, and visual landmarks are each weighted differently depending on availability, time of day, age of the animal, and developmental experience. Juvenile pied flycatchers (Ficedula hypoleuca), for instance, calibrate their magnetic compass against the rotational axis of stars during a sensitive pre-migratory window; birds raised under rotating star patterns orient toward the false stellar pole even when placed outdoors later.

Cue-integration models—often Bayesian or optimal-combination frameworks (Åkesson & Bensch, 2014)—provide a quantitative language for this hierarchy. Module 8 of this course develops the cross-modal neural substrate, including the avian Wulst and the mammalian head-direction system.

• Is the radical pair actually in vivo responsible for behavioural orientation, or are the Xu 2021 results a necessary-but-not-sufficient condition? Genetic rescue experiments are needed.
• What is the neural readout pathway? The retinal origin of avian cryptochrome signals is well-established (Zapka et al. 2009), but the convergence onto cluster N of the forebrain is still being mapped.
• Are there additional cryptochrome paralogs (CRY1a, CRY4) with specialised magnetic functions?
• Do mammals with demonstrated alignment behaviours (cattle, dogs, foxes) actually use their magnetic sense for navigation, or is alignment a byproduct of another function?
• Can the radical-pair mechanism be miniaturised into a quantum sensor for human technology?

Where in the brain is magnetic information processed? For the radical-pair compass, the primary signal originates in retinal photoreceptor–like cells containing cryptochrome. Heyers et al. (PLoS ONE, 2007) traced magnetoreception-responsive regions in night- migrating garden warblers (Sylvia borin) using immediate-early-gene (ZENK) mapping and identified a discrete region of the forebrain called Cluster N, which is active only during nocturnal magnetic orientation. Cluster N receives input from the retinal magnetosensitive ganglion cells via the thalamofugal visual pathway, suggesting that the bird literally “sees” the magnetic field as a pattern overlaid on its visual scene.

For the magnetite-based map sense, the likely pathway runs through the ophthalmic branch of the trigeminal nerve (V1). Mora et al. (2004) showed that zinc-sulphate anaesthesia of V1 disrupts map-based orientation in homing pigeons without affecting the compass. Heyers et al. (2010) followed up with detailed neuroanatomical tracing showing that V1 projects to the principal sensory trigeminal nucleus (PrV), whose neurons show graded responses to slowly varying magnetic field intensity.

The dual pathway is consistent with Wiltschko's two-receptor hypothesis: one compass (radical-pair, retinal, light-dependent, inclination-only) and one map sense (magnetite, trigeminal, light-independent, polarity-sensitive). The interaction between the two systems during long-distance navigation is still being worked out.

One of the most striking aspects of the radical-pair mechanism is that it requires quantum spin coherence on microsecond timescales at body temperature, in a noisy aqueous environment. Naively, one would expect decoherence to destroy any quantum effect within nanoseconds. How does nature achieve \(\mu\)s coherence?

The answer lies in the choice of “spin bath”. Radical-pair electrons couple strongly to surrounding nuclear spins (protons, nitrogen-14) but only very weakly to the fluctuating electronic environment, because the orbital degrees of freedom of a radical are largely quenched. The effective decoherence times are dominated by spin relaxation via nuclear dipole–dipole coupling and paramagnetic O₂ in solution, both of which can be suppressed by the protein scaffold. Kattnig et al. (2016) showed that the cryptochrome fold provides a low-decoherence pocket around the FAD cofactor, analogous to the way photosynthetic reaction centres shield their excitonic states.

Current consensus: the radical-pair compass represents an example of functional quantum biology — not a macroscopic coherent wavefunction, but a coupled pair of spins whose short-time coherent evolution produces a measurable macroscopic output (product yield) that depends on external magnetic fields. It is analogous to the Stern- Gerlach experiment but running continuously on every photo-activated cryptochrome molecule in the bird's retina.

Cai, Guerreschi & Briegel (Phys. Rev. Lett., 2010) raised the possibility that entanglement between the two radical electrons might extend the useful coherence time beyond naive estimates. This is still debated, but the existence of a functioning compass at Earth-field strength seems to require the spin dynamics to be at least in the “coherent regime” in the technical sense of reference (Kominis, 2015).

The field has advanced rapidly in the past five years. The Xu 2021 demonstration of cryptochrome magnetosensitivity in vitro finally bridged the long gap between theoretical predictions and empirical biochemistry. Subsequent work by the same Oldenburg group (Oxford collaboration) extended the measurements to additional cryptochrome paralogs and other avian species, consistently finding that migratory species have enhanced field effects compared to sedentary relatives.

On the magnetite side, the Keays group (Treiber et al. 2012 and subsequent work) has largely ruled out magnetite-containing sensors in the upper beak of pigeons, but confirmed functional magnetite sensors in trout olfactory epithelium (Eder et al. 2012) and several teleost species. The sensor appears to be a single-domain magnetite particle physically coupled to a mechanosensitive ion channel, though the exact identity of the channel is still unknown.

Behavioural experiments continue to discover new magnetoreceptive taxa. Magnetic alignment has been confirmed in domestic chickens choosing roosting direction (Wiltschko et al. 2020), in salmon parr orienting along stream flow under geomagnetic cues (Putman et al. 2020), and in various insects including fruit flies (Drosophila), where a CRY-dependent magnetic preference can be genetically dissected at the single-cell level (Bae et al. 2016; Bradlaugh et al. 2023).

Looking ahead, the combination of CRISPR-Cas9 genetic manipulation in model organisms (mouse, zebrafish, fly) with improved magnetometer-equipped behavioural rigs is expected to produce the first single-cell electrophysiology of a confirmed vertebrate magnetosensor within the decade. The decades-old mystery of magnetoreception may finally be close to a complete mechanistic picture.

The extraordinary sensitivity of biological compasses has attracted significant attention from quantum-sensing researchers. If a protein can detect a ~50 µT field using a few radical pairs at room temperature, perhaps a similar device could serve as a low-cost magnetometer for medical, geophysical, or navigational applications.

Maeda et al. (Nature, 2008) demonstrated the first artificial chemical compass: a synthetic carotenoid–porphyrin–fullerene triad whose triplet yield shifts by ~1% under an applied 50 µT field, with sensitivity comparable to the biological system. Their device works on exactly the same principles: photo-induced electron transfer creates a radical pair, hyperfine coupling mixes the spin states, recombination is spin-selective, and the product yield depends on the angle of the external field.

More recent work has moved toward nitrogen-vacancy (NV) centre magnetometry in nanodiamond, which exploits a completely different mechanism (optically detected magnetic resonance of deep-trap electron spins) but has achieved biological-scale sensitivity at the single-molecule level. The 2020s may finally yield practical radical-pair magnetometers competitive with SQUID technology at room temperature.

The Earth's magnetic field is not static. On human timescales, the North magnetic pole drifts several tens of kilometres per year, and the total intensity at a given point can change by several percent per century. Geomagnetic reversals, in which the field polarity flips, have occurred roughly every 200–300 kyr on average. These facts pose a serious challenge for any animal that relies on the field as a stable positional reference.

The evidence summarised in Section 4 shows that sea turtles, at least, cope by tracking secular variation: the genetic structure of loggerhead populations matches the time-evolving magnetic signature of natal beaches (Brothers & Lohmann, 2015). How this is achieved mechanistically is unclear. Possibilities include continual recalibration from astronomical cues, genetically imprinted “attractor” ranges rather than exact values, or coincidentally matched temporal scales between secular variation and individual lifetime.

Over longer timescales, polarity reversals must occasionally have created catastrophic navigational failure for species relying on polarity compasses. Ecological modelling (Hruschka & Lohmann, 2022, in prep.) suggests that inclination compasses are robust to polarity reversals (they work equally well in either polarity) whereas magnetite-based polarity compasses would fail. This may provide an evolutionary rationale for the predominance of inclination compasses in long-lived vertebrates.

Do humans have a magnetic sense? The evidence remains contested. Wang, Connolly et al. (eNeuro, 2019) reported that human participants placed in a magnetically shielded room with a rotating weak magnetic field produced a robust drop in alpha-band EEG power within tens of seconds of field rotation, with correct rotational specificity. Subsequent attempts at replication have had mixed results. If the effect is real, it would imply a previously unrecognised neural receptor, because human behaviour shows no obvious overt compass ability.

One intriguing line of investigation is whether the human CRY protein retains magnetoreceptive capability: Foley et al. (Nature Comm., 2011) showed that human CRY2 expressed in Drosophila can rescue the magnetic-orientation phenotype in CRY-knockout flies, suggesting that at least the basic quantum machinery is preserved in the human homolog. Whether that machinery is wired to a behavioural output in humans remains unknown.

From a cultural perspective, magnetoreception has deep resonance with traditional navigational knowledge. Pacific Islander wayfinders used a multimodal cue hierarchy strikingly similar to that of a migratory bird: dominant stars, swell patterns, cloud banks, bird behaviour, and (possibly) geomagnetic cues via observed compass effects in iron-rich volcanic samples. Reading the ocean was an explicit multisensory calculus long before science formalised the notion of sensor fusion. The parallel between an albatross and a Polynesian navigator reading cues over thousand-kilometre ranges is one of the most beautiful convergences in the study of biological and cultural cognition.

Magnetoreception exhibits two complementary mechanisms that coexist in nature: the light-dependent radical-pair compass (cryptochrome-based, inclination-only, short-range orientation) and the light-independent magnetite map sense (iron-oxide-based, polarity- sensitive, long-range positioning). The combined evidence from the Xu 2021 biochemistry, the Wiltschko behavioural studies, the Lohmann sea turtle imprinting, and the Reppert monarch CRY2 genetics now provides a reasonably complete picture of the radical-pair side. The magnetite pathway is still under active investigation.

The central unresolved challenge remains: how does a quantum spin-dependent chemical reaction in a single cryptochrome molecule scale up to reliable behavioural orientation in the whole animal? The answer probably involves population averaging over millions of cryptochrome-containing photoreceptors, integrated with predictable noise-filtering in downstream neural circuitry. In this respect, the magnetic compass is analogous to the olfactory discrimination of single molecules: a rare noisy event at one receptor amplified by ensemble statistics and circuit integration into a macroscopic behavioural output.

Looking ahead, the module reinforces the recurring motif of this course: sensory biology often pushes against fundamental physical limits (thermodynamic, quantum, shot-noise) and solves these limits by clever biological engineering at the receptor and circuit level. Magnetoreception is perhaps the most extreme example, with the sensor operating at the edge of quantum coherence and the physical signal weaker than in any other natural modality.