There is strong evidence that migratory birds inherit an endogenous directional programme, steering inexperienced migrants in a certain direction for a certain period of time [1], [2]. This programme alone does not enable migrants to navigate toward their unknown species-specific wintering grounds and thus, it does not allow birds to compensate for a displacement [3]. With experience, this programme develops into a goal area navigation programme allowing the birds to accurately pin-point at least their breeding and winter grounds. Why this is so remains poorly understood [4]. Being able to compensate for displacements would presumably lead to evolutionary advantages but these have apparently not led to the evolution of such a capability in young birds, possibly because of evolutionary or mechanistic constraints but also because juveniles may benefit from a more open strategy.

The presumed change from a simple bearing-based orientation to true navigation is largely unstudied. It is generally assumed that the development of the navigational system in adult birds, which enables accurate goal navigation, relies on experience during the first migration [5], [6]. In general, juvenile birds making their first migration do not show signs of adult navigation after release in experimental displacements where the compensation requires some form of true navigation [7], [8]. Despite this, there are no theoretical objections to juvenile migrants being able to correct for displacements: Juveniles could correct by using experienced-based navigating toward a site already visited on migration, as birds already on the way on their first migration could be acquiring information necessary for later navigation.

The way the initial programme interacts with and is influenced by external factors such as wind, topography, habitat etc. is still largely unknown. For example, winds can exert a strong influence on migrating birds, because wind speeds can easily reach the endogenous self-propelled flight speed of most birds. A few studies have indicated an ability at the individual level to compensate for wind displacements [9], [10], whereas others have not [11]. In juvenile birds with no prior knowledge of the migratory route or wintering grounds, the behavioural responses to such displacements are crucial for the successful arrival at the species-specific wintering grounds.

Here, we study naturally (ie. likely by wind) and experimentally displaced juvenile birds of nocturnally and solitarily migrating species at the Faroe Islands in the northeastern Atlantic Ocean far off the normal migration route of European land birds. So far responses to displacements and especially to external factors have proven very difficult to study, mainly because of problems in following free-flying migratory birds [12]. Thus, we applied advanced radio telemetry methods to efficiently track the movements of departing birds after the displacement to the Faroe Islands and combined this with the more traditional orientation cage studies.

Juvenile birds of three long-distance migratory species, experimentally displaced 1100 km from Denmark to the Faroe Islands, shifted their orientation similarly and in accordance with compensation for the displacement. Before displacement, juvenile birds tested in Emlen funnels in Denmark were oriented toward southwest (α = 236°, r = 0.544, N = 29, P<0.001, Rayleigh test; Figure 1b) similar to the normal migration direction as found from ring recoveries (P>0.7, Watson-Williams test). After displacement, the orientation in funnels on the Faroes shifted counter-clockwise toward south-southeast (α = 168°, r = 0.516, N = 25, P<0.001; Figure 1c) significantly different from the orientation in Denmark before displacement (N = 29/25, F_1,52 = 12.88, P<0.001, WW test). Radio tracking of the displaced birds later released on the Faroes revealed departure directions that were significantly different from the orientation before displacement (N = 29/13, F_1,40 = 40.18, P<0.001, WW test) and in even more counter-clockwise shifted directions toward southeast and east (α = 102°, r = 0.711, N = 13, P<0.001; Figure 1d) than in the orientation cages (N = 25/13, F_1,36 = 10.01, P = 0.003, WW test; though the difference was not significant at level of the individual: α = 2°, r = 0.415, N = 11, P>0.05, Confidence Interval test). This difference between funnel orientation and departure directions could have been caused by winds, with birds on average departing in westerly tailwinds and with the average tailwind vector significantly different from random only when experimentally displaced birds departed (Figure 2). Furthermore, the overall pattern of headings was similar to that of vanishing bearings and headings were not more constant. Testing changes in orientation at the individual level also resulted in very similar results with the orientation after displacement differing significantly from the orientation in Denmark both for funnel tests and vanishing bearings (P<0.05 and 0.01, respectively, Confidence Interval test; Figure S1). Because of the relatively large scatter, it was not possible to determine whether orientation back toward the capture site or toward the winter grounds fitted the data best (Figure S2).

Similarly, birds naturally displaced to the Faroe Islands, presumably by wind drift, and followed with radio telemetry, apparently compensated for the natural displacement by departing in directions toward south-southeast (α = 155°, r = 0.734, N = 8, P = 0.009, Rayleigh test; Figure 1e). The direction differed significantly from that found in birds caught and tested in cages within the normal migration route in Denmark (N = 8/29, F_1,35 = 8.46, P = 0.006, WW test) and also from the departure directions of experimentally displaced birds (N = 8/13, F_1,19 = 5.75, P = 0.027, WW test) but not from the orientation of displaced birds in funnels.

In general, no differences among species were apparent. Only if combining naturally and experimentally displaced birds, the vanishing bearings from the Faroe Islands differed slightly among species (N = 11/3/7, F_2,18 = 4.11, P = 0.034, WW test). Garden warblers Sylvia borin took off in a more southerly direction (α = 181°, r = 0.939, N = 3, P = 0.058) than both willow warblers (α = 93°, r = 0.797, N = 7, P = 0.007) and blackcaps (α = 124°, r = 0.663, N = 11, P = 0.005). The differences in funnel orientation and headings among species were smaller and not significant.

The compensatory reaction in drifted juvenile birds could be caused by reverse path integration (dead-reckoning) to register the passive displacement or an even simpler non-specific reaction to westward displacement by winds during a migratory flight as in corrective early morning flights [13], [14]. Directional compensation does not need to be a very precise mechanism but could be a non-specific reaction to any westward displacement. Despite the expectations from previous releases of displaced birds, the additional compensation by experimentally displaced birds indicates surprisingly, that in some circumstances, juveniles may be able to correct for displacements from the normal route. The juvenile passerines observed here might well have used a simple form of true navigation.

The navigation of birds on the Faroes could have been based on an experience-based system that was initiated earlier than is generally assumed in these or other species. The tracks taken by juvenile Eleonora's falcons suggest the possibility of goal-area navigation toward the winter grounds independent of the adults [15] and such restricted migration has also been shown in several other species [16].

Previously, only a few studies of bird migrants have hinted at the existence of some form of navigational responses in juveniles [8], [17], in contrast to a number of studies showing no compensation [2], [19], [20]. However, in juvenile turtles changes in orientation as a response to experimental changes of the magnetic field may indicate the use of an inborn navigational system based on magnetic ‘sign post’ cues [21]–[23] similarly to the apparently inborn responses to magnetic field changes reported in birds where fat deposition needed for crossing an ecological barrier is apparently partly controlled by magnetic cues [24], [25].

Our results indicate that there could be minor differences in reactions among species. The reasons for such differences might relate to different motivational states and differences in migration routes and distances, but warrant further in-depth studies. Blackcaps tested on the Faroes by Rabøl [26] were oriented in the normal migration direction of the Norwegian population in contrast to the pattern seen here and such a difference could easily be the result of even small differences in conditions on arrival between years. The expected evolutionary advantages associated with being able to compensate for displacements are not fully obvious because juveniles may benefit from dispersing longer than adults as well as from a bet-hedging strategy with regard to migratory directions [27]. In this study, the perhaps most pronounced long-distance migrant, the garden warbler, showed the least reaction to the displacement. Though no firm conclusions can be drawn due to small sample sizes of this species, this could be because for garden warbler the direction to a far winter quarter changes less than in the shorter-distance migrants or be because immediate compensation for displacement is not necessarily optimal when making optimal use of winds to reach a goal far away [28].

Magnetic cues seems to be the most obvious candidates for forming the basis of the navigational responses to the displacement [29], [30] with relatively large magnetic field changes for the distances involved in this study (change in declination: 9°; dip: 4°; magnetic intensity: 1500 nT). Inexperienced migrants have been shown to increase fattening and change their innate directional preference in response to changes in the magnetic field [25], [31] but the involvement of the magnetic field in migratory navigation remains unresolved. Experiments with silvereyes and northern wheatears have shown responses indicating that the magnetic field can be used for navigation [32], [33]. Other cues such as celestial [18] or olfactory cues [34] may also be involved.

Given the restricted range of the tracking methods used here, the ultimate fate of the displaced birds is not known. To be able to find out whether birds are able to find their normal winter grounds would require being able to track the full migrations of these birds. We believe that the possibility of further experimentation with free-flying birds will result in much improved possibilities for investigating the complex relationship with the environment and that such experimentation is likely to enable us to understand the fascinating navigational mechanisms in long-distance migrating birds. With the development of smaller tracking devices this appears possible in a near future.