Geomagnetic imprinting: A unifying hypothesis of long-distance natal homing in salmon and sea turtles.
] constitutes a remarkable behavioral phenomenon. It requires the learning of locale-specific information capable of guiding long-distance homeward movements, after perhaps years of elapsed time and based on very limited experience. Spatial familiarity acquired through exploratory movements and outbound migration may be an important component of natal homing in many species [
]. Nevertheless, imprinting by neonates prior to first migration on locale-specific values of a wide-ranging gradient cue could in principle provide a natal-site signature targetable over very long distances without requiring the exploratory experience needed to build some kind of navigational “map.” Suitable cues could be provided by components of the Earth’s magnetic field, such as magnetic inclination (the angle between the Earth’s magnetic field and the Earth’s surface) or magnetic intensity (the overall strength of the Earth’s magnetic field).
Here, we used available ringing data from the British Trust for Ornithology to investigate the role of magnetic cues in the natal philopatry of Manx shearwaters (Puffinus puffinus), a small but long-lived pelagic Procellariform seabird with a trans-equatorial, trans-Atlantic migration. Manx shearwaters breed on islands in the East Atlantic across a latitudinally diverse but longitudinally constrained range (∼−5°E to −10°E, ∼45°N to 65°N). From ringing recoveries  and geolocators [
Migration and stopover in a small pelagic seabird, the Manx shearwater Puffinus puffinus: insights from machine learning.
], both adult and first-year birds are known to spend the boreal winter on the Patagonian shelf, Argentina. Immatures return to European colonies for the first time from around 3 years post-fledging, with breeding commencing several years later [,
]. Since 1954, some 2,996 Manx shearwaters from colonies in Britain, Ireland, and the Isle of Man have been ringed as chicks and later recovered as returners (>1 year later) at or close to a breeding colony, with around 4% of these birds recovered at non-natal colonies, indicating a high degree of natal philopatry.
Secular variation, changes in the Earth’s magnetic field caused by the movement of magnetic material in the core, causes inclination and intensity at a given latitude to increase or decrease year-on-year. This is associated with a corresponding northward or southward movement in the latitude at which specific parameter values occur (see Figure 1 for magnetic shifts over the sampled period). Thus, if Manx shearwaters make use of magnetic parameters during natal homing, such changes might be reflected in the errors or biases associated with where they return to breed, allowing for the following qualitative prediction: with decreased magnetic intensity and/or inclination, birds will be more likely to recruit to the north of their natal colony, while increases will lead to a greater chance of recruiting to the south (see Figure 1). Further, if inclination and/or intensity is being used to infer latitude, then we can make two further quantitative predictions. First, we predict that the intercept of the linear regression of latitude against inclination and/or intensity should be zero (i.e., when there is no change in inclination or intensity there is also no change in latitude). Second, if inclination is used to infer latitude, the gradient of the inclination versus latitude linear model should be approximately 1.34, the ratio of degrees inclination to degrees latitude in the UK. By comparing the magnetic field at the point when a fledgling leaves its colony and the field at the point it returns (3 years later [,
]), we can calculate a change in the magnetic field parameters of interest. We can then assess how this change predicts the change in latitude at recruitment.
To test whether changes in magnetic inclination predicted changes in recruitment latitude, we fitted a linear model between the changes in the inclination angle and magnetic intensity and the actual changes in latitude between ringing and recruitment. Because ringing effort at colonies of different latitudes has not remained constant, changes in the relative recovery effort in the north and south of the species range could result in apparent mean latitudinal changes, even if birds were to disperse at random with respect to latitude. As such, we modeled out the bias in average latitudinal change caused by sampling latitude over time by first mechanistically modeling the expected latitudinal bias owing to sampling and then including it as a predictor in our linear model (see STAR Methods for details). Additionally, we also tested for an effect of change in sea surface temperature (SST) on the change in latitude between bird fledging and recruitment, as this variable is often assumed to play a key role in distribution changes in oceanic organisms [
Evidence for geomagnetic imprinting as a homing mechanism in Pacific salmon.
We found that shifts in recruitment latitude are significantly predicted by shifts in inclination (LM; F = 34.7935, p = 4.1 × 10−10, Figure 2) and the change in the sampling latitude bias (see STAR Methods for more information; LM; F = 71.3764, p = 2.2 × 10−16; total model r2 = 2.8%). In order to ensure that the statistically significant effect of inclination observed was not the product of zero-inflated distributions, we also tested for statistical significance using a rank order randomization (see STAR Methods). This too suggested a significant effect of inclination shift on recruitment latitude (p a priori prediction. We also noted that the linear model intercept was close to the expected 0 value (−0.013 ± 0.020 [SE]) and that the estimated gradient of inclination change versus recruitment latitude change was, as with the intercept, close to our expected value of 1.34 (1.34 ± 0.23 degrees latitude per degree inclination). We found that intensity change is not a significant predictor of latitudinal change during recruitment when considered either as part of a linear regression model (LM; F = 0.001, p = 0.99) or in a rank-order randomization (p > 0.05). Similarly, the model was not changed significantly by the inclusion of changes in SST as a predictor (LM; F = 2.44; p = 0.118).
Within our ringing data, we also have 1,207 records of adult shearwaters (aged 3 years or greater) that were recaptured a second time. Of these between-year adult recoveries, we find only 6 individuals that change colony. Given that in this time 109 fledgling shearwaters changed colony, we suggest that, unlike with naive returners, non-magnetic cues may contribute to orientation in adult shearwaters or, alternatively, that adult shearwaters update their magnetic inclination target yearly and are thus more resilient to fluctuations in the Earth’s magnetic inclination.
Our finding that changes in latitude between fledging and recruitment are significantly predicted by changes in inclination over the same period is consistent with magnetic inclination being used directly by shearwaters as a measure of latitude when returning to colonies for the first time. This is supported not only by the qualitative expectation that northward and southward shifts in inclination between the year of hatching and the probable year of recruitment predict northward and southward recruitment of birds but also by the quantitative predictions of an inclination-based measure of latitude. The linear model intercept value of 0 suggests that when inclination is constant, there is no change in recruitment latitude. Furthermore, the gradient of the effect is remarkably close to the predicted 1.34° change in latitude per degree change in inclination. Conversely, despite being correlated with inclination (LM; F = 77.708, p = 2.2 × 10−16), total magnetic intensity did not predict latitude at recruitment. Given that this is a correlative study, it is possible that the results detailed above arise due to the confound of magnetic inclination with another variable. However, we found no effect of changes in sea surface temperature on recruitment latitude, which would probably be the most plausible non-navigational driver of post-natal dispersal in a pelagic seabird. As such, given how precisely the data fit the magnitude and direction of our a priori predictions, we believe that the most parsimonious explanation is the direct use of magnetic inclination in seabird natal philopatry.
Nocturnal life of young songbirds well before migration.
] or information gathered on the first outward migratory journey. The mechanisms involved have, however, been seldom investigated. Unlike the learning of a detailed familiar area map, or a larger scale extrapolated navigational map, around the home site, the specific task of first returning to the natal site long after fledging (and a potentially rapid first autumn migration like that seen in shearwaters ) may favor a targeting mechanism based on some form of imprinting that requires no exploratory experience. Support for a “geomagnetic imprinting hypothesis” [
Geomagnetic imprinting: A unifying hypothesis of long-distance natal homing in salmon and sea turtles.
] has been found in sea turtles (Chelonioidae) and several teleost fish species (notably the Salmoniformes), where natural variation in the Earth’s magnetic field and contrasts induced by local topography suggest the use of magnetic cues specific to the natal site when navigating [
An inherited magnetic map guides ocean navigation in juvenile Pacific salmon.
Our results suggest that, as in sea turtles and teleost fish taxa, shearwaters might imprint onto the magnetic inclination (but apparently not magnetic intensity) of their natal colony prior to or around fledging to provide a potentially very long-distance targeting mechanism for natal homing three or more years later. Most likely, such a mechanism would provide only approximate position (latitude), with other cues focusing specific colony choice (whether similarly learnt location cues, or social and habitat attractors). Hence, and given the geographical spacing of active breeding colonies, small shifts in magnetic inclination at a given latitude should lead only to shifts in the probability of recruiting at the natal versus another colony (see Figure 2). Consequently, we see significant latitudinal shift with inclination shift on the population level, but with comparatively few individuals (109) exhibiting the precise expected shift.
The role of the earth’s magnetic field in avian navigation has had a controversial history. Considerable evidence exists for the use of an inclination compass in night-migrating passerines [
]. Interestingly, in procellariforms, the taxon studied here, there is also as yet no empirical evidence for the use of magnetic cues when navigating either after experimental displacement or during natural foraging excursions [
Olfaction and topography, but not magnetic cues, control navigation in a pelagic seabird: displacements with shearwaters in the Mediterranean Sea.
]. The results presented here, then, provide the first evidence at least that young seabirds must have a sensitivity to magnetic inclination. Our finding that unlike fledglings, adult birds are not sensitive to inclination shifts suggests that inclination is not the only indicator of latitudinal position in experienced returners, which, in turn, is parsimonious with previous studies suggesting that spatial position in adult procellariforms is unlikely to be primarily ascertained using magnetic cues.
Additionally, even in fledglings, inclination cannot be the only indicator of geographic position since it only provides a guide to latitude. Other cues, magnetic or otherwise, must be required to provide fledglings with longitudinal information. Further, inclination cannot be used to determine latitude if acted upon using a simple monotonic response, since inclination forms an approximate mirror-image pattern across the equator. Consequently, trans-equatorial movement requires at least a reversal of response either side of the magnetic equator, a problem shared with the magnetic inclination compass [
While our results constitute the first evidence for magnetoreception in a seabird as well as a mechanistic explanation of natal philopatry following first time migration, they also demonstrate the power of re-analyzing historic ringing data when attempting to understand the mechanisms underpinning the migrations of millions of birds worldwide. While ultimately, as with any correlative study, experimental verification will be needed in future, these results nonetheless represent the first attempt to explain the sensory basis of avian natal philopatry, one of the great enigmas of animal behavior.