Migratory blackcaps can use their magnetic compass at 5 degrees inclination, but are completely random at 0 degrees inclination

It is known that night-migratory songbirds use a magnetic compass measuring the magnetic inclination angle, i.e. the angle between the Earth’s surface and the magnetic field lines, but how do such birds orient at the magnetic equator? A previous study reported that birds are completely randomly oriented in a horizontal north-south magnetic field with 0° inclination angle. This seems counter-intuitive, because birds using an inclination compass should be able to separate the north-south axis from the east-west axis, so that bimodal orientation might be expected in a horizontal field. Furthermore, little is known about how shallow inclination angles migratory birds can still use for orientation. In this study, we tested the magnetic compass orientation of night-migratory Eurasian blackcaps (Sylvia atricapilla) in magnetic fields with 5° and 0° inclination. At 5° inclination, the birds oriented as well as they did in the normal 67° inclined field in Oldenburg. In contrast, they were completely randomly oriented in the horizontal field, showing no sign of bimodality. Our results indicate that the inclination limit for the magnetic compass of the blackcap is below 5° and that these birds indeed seem completely unable to use their magnetic compass for orientation in a horizontal magnetic field.

In spring 2015, birds were tested in the control conditions (67°NMF and 67 °CMF) and in the 0° inclination conditions (0°NMF and 0°CMF) to investigate further, whether the blackcaps were bimodal or randomly oriented in the field with a flat inclination. The birds oriented significantly in their appropriate migratory direction when they were tested in a magnetic field pointing towards magnetic north with a magnetic inclination of 67° (spring 2015 67°NMF: group mean orientation = 38° ± 28°, r = 0.66, p < 0.01, N = 13, Fig. 4a). When the horizontal component of the 67° inclined field was rotated − 120°, the birds turned their orientation accordingly (spring 2015 Figure 1. An illustration indicating how the Earth's magnetic field at the magnetic equator may appear to a bird, which "sees" the magnetic field. Hypothetical signal modulation patterns using the assumptions also used in Ritz et al. 26 and Solov'yov et al. 32 for a bird changing its viewing direction clockwise in 90° increments in a magnetic field of 0° inclination. The four circles represent a full 360° sweep, showing all cardinal directions, from north (left circle) to west (right circle). Each "view" covers 180°.

Discussion
We could independently confirm the results of previous studies in European robins, Erithacus rubecula 9 , and garden warblers, Sylvia borin 47 : blackcaps completely fail to orient at 0° inclination, and we saw no robust signs of bimodal orientation. This is surprising because their magnetosensory system should have been able to separate the north-south axis from the east-west axis (see Fig. 1). The fact that they do not show this presumed capability in their orientation behaviour in an Emlen funnel can have many reasons.
One possibility is that the birds' brains not only consider the absolute sensory information from various potentially navigation relevant cues, but that they also evaluate the quality of the sensory information arriving from each sense 49 . This could take place in one or more candidate brain regions outlined in Mouritsen et al. 49 . In that case, the north-south ambiguity, even though it might be in principle detectable, may lead the birds to ignore the magnetic cues altogether.
Another open question is how precisely birds are able to detect magnetic declination angles, and at which minimum inclination angle, the birds are still able to use their magnetic inclination compass. For vertical inclination angles we know that blackcaps are able to use a field with 85° inclination, but not one with 88° inclination 50 . Savannah sparrows tested near the actual magnetic north pole even seemed to be able to orient in a field with 88.6° inclination 51 . In the present study, we could show that blackcaps can orient as well in a magnetic field with a shallow 5° inclination as they can in the normal geomagnetic field found around Oldenburg (67° inclination). Thus, for shallow inclination angles, the limit is certainly also better than 5° as it is for steep inclination angles. Because the Earth's magnetic field changes ca. 0.009° per km, our results determines the upper limit for the extent of the magnetic compass blind zone around the equator to (2 × 5°)/(0.009°/km) = 1110 km. If the inclination angle detection limit would turn out to be ca. 2-3° as seems to be the case for the steep inclinations, the magnetic compass blind zone would be ca. 440-660 km wide.
A few studies have suggested that exposure to 0° magnetic inclination can trigger a reversal of birds' orientation direction from "equatorward" to "poleward" and vice versa 44,46,52 but see ref. 45. We avoided to expose our birds to 0° inclination in autumn for this reason and to prevent that this field could function as a migratory stop signal in blackcaps, which do not cross the magnetic equator. Because our birds were exposed to the various magnetic inclination conditions in a semi-random pattern, we can state that the blackcaps in our study, which had no stellar cues available, did not reverse their orientation in the control condition, after being exposed to 0° inclination for 1-2 hours several times during spring. Had such exposures to 0° inclination led to orientation direction reversal in spring, the birds tested in our control condition should have become random or bimodal, Figure 4. Blackcaps fail to orient in a 0° inclination magnetic field. In the spring migratory season 2015, the blackcaps were well oriented in their natural spring migratory direction in the geomagnetic field of Oldenburg (a) and turned their orientation accordingly when the magnetic field was turned 120° counter-clockwise (c). In contrast, the birds were randomly oriented when the inclination angle was set to 0° (0°NMF (b) and 0°CMF (d). For a description of the circular diagrams, see legend to Fig. 2. because the control tests and 0° inclination tests were intermixed throughout spring. However, in the studies observing reorientation 44,46,52 , the birds were tested in autumn and they were exposed to the 0° inclination for two full days and nights.
In conclusion, blackcaps seem unable to orient in a completely horizontal magnetic field, and their angular determination capabilities related to shallow inclination angles is better than 5°. Considering our current knowledge about the sensitivity of the birds' magnetic inclination compass, at least two important questions remain: (1) What orientation mechanism(s) do birds use to successfully cross the less than 1110 km broad area around the magnetic equator, where the birds seem unable to use their magnetic inclination compass? Celestial cues would be the most likely solution 2,3 . (2) How and at what point are these alternative cues triggered? The point at which the magnetic inclination becomes too shallow to resolve for the birds' magnetic inclination compass would be one option.

Magnetic fields.
A double-wound, three-axial, Merritt four-coil system (2 × 2 × 2 m) was used to create the experimental magnetic fields 11,28,48 . The coils were run by three constant current power supplies (KEPCO BOP 50-4M, Kepco Inc., Flushing, NY, USA), one for each axis. The experiments were performed within the center of the coils, where the homogeneity of the field applied by the coils was 99% or better.
First, we pre-tested the blackcaps to make sure they were in migratory mood. Pre-testing was performed in two different field conditions: in the natural geomagnetic field (67°NMF) of Oldenburg (field strength = 48,600 ± 240 nT [standard deviation], inclination = 67.3° ± 0.4°; horizontal direction 360° ± 0.1°), and in a changed magnetic field, in which magnetic North was turned 120° counter clockwise (67°CMF: field strength = 48,600 ± 250 nT; inclination = 67.4° ± 0.3°; horizontal direction = − 120° ± 2°). In the NMF condition, the same amount of current that was needed to create the CMF condition was sent through the two subsets of coil windings but in opposite (antiparallel) directions, so that the effective changes in the natural geomagnetic field strength were less than 10 nT.
During the critical experiments, we re-tested the birds under the control conditions (67°NMF and 67°CMF with 67° inclination), and in conditions where the inclination was changed to 5° (5°NMF and 5°CMF performed in spring 2014 and autumn 2014) or 0° (0°NMF and 0°CMF performed in spring 2014 and spring 2015) while the total field intensity remained the same as in the control conditions ( Table 1). The magnetic field conditions present inside the funnels were measured daily before the experiments started and remained the same for two consecutive nights per bird.
All magnetic field conditions were set and controlled (Table 1)  Inc., Natick, MA, USA) to confirm the consistency of the magnetic field conditions during the entire experimental period. Due to the inevitable heterogeneities created by the coils, minor deviations from the desired fields could not be avoided, but these were measured and are reported in Table 1. Behavioural tests. In spring 2014, the group of tested birds was unusually well oriented 50 . Thus, fewer tests per condition were needed until the group of birds showed a significant orientation in the control condition compared to most other years (compare data given in ref. 50 with data obtained from the same huts in refs 28,29,33,34 and 53).
All behavioural tests took place in wooden huts covered with aluminum plates on the inside, which were grounded and therefore acted as Faraday cages that shielded the inside from time-dependent electromagnetic fields in the frequency range up to ~20 MHz by approximately two orders of magnitude 54 . All electrical equipment (power supplies, computers etc.) was placed outside the experimental room in aluminum-shielded shelves to prevent electromagnetic interferences caused by the equipment to affect the birds. During testing, the room was illuminated with dim, diffused light (2.5 ± 0.25 mW/m 2 ) produced by light bulbs (see spectrum given in ref. 28). Hence, the static magnetic field was the only available cue for orientation. All behavioural tests were conducted in the following way: one hour before the start of the experiments, the birds were placed outdoors in wooden transport boxes fitted with 7 cm diameter mesh-covered peepholes to enable them to see twilight cues and parts of the evening sky. This gives them the possibility to calibrate their magnetic compass [2][3]55 . At sunset (± 10 min), the birds were placed in modified aluminum 'Emlen funnels' (35 cm diameter, 15 cm high, walls 45° inclined 56 ). The walls of the funnels were lined with scratch-sensitive paper (Blumberg GmbH, Ratingen, Germany) on which the birds' migratory restlessness became visible as scratches 57 .
Nine birds were tested simultaneously twice each night. The second testing started approximately 1.5 h (± 10 min) after the first test started. In the second test round, the birds were switched to another hut or a different funnel position so that we could exclude that the birds had transferred any possible non-magnetic cues from one test to the next.
Orientation data analyses. After the end of the experiments, all scratch-sensitive papers were evaluated independently by two researchers relative to the overlap point to estimate the mean direction of the scratches. The evaluators did not know in which cardinal direction (N, S, E or W) the overlap point had been fixed. The cardinal direction of the overlap point was chosen randomly and varied in-between test rounds and nights. If the scratch-sensitive paper fulfilled one of the following three criteria it was excluded from further analysis: (1) If the two independently estimated mean directions differed by more than 30°, a third observer was consulted. In the case that no agreement between the three was achievable, the paper was regarded as random (10% of all papers). (2) Papers with less than 30 scratches were classified as inactive, because blackcaps typically left fewer than 30 escape scratches when removed immediately after placing them in Emlen funnels 9,29,50 (25% of all papers) (3) If the distribution of the scratches were bimodal (0.3% of all papers). In all other instances, subsequently, the bird's mean direction was determined as the mean direction of the scratches corrected for the direction of the overlap point. As a result, the number of active and directed tests per bird in the different conditions differed slightly. The average mean heading of an individual bird in a given experimental condition was calculated by addition of unit vectors in each of the mean directions of the individual tests. The group mean vectors were calculated by vector addition of these individual mean directions followed by division by the number of birds tested in the given condition. The significance of the group mean vector was tested using the Rayleigh-test 58 . Differences in group mean orientations between birds tested in different magnetic field conditions were tested by the Mardia-Watson-Wheeler test (MWW; see ref. 58). To test for bimodal orientation, each angle was doubled and the group mean vector of the doubled angles was tested for significance with Rayleigh-test 58 .