A nocturnal mammal, the greater mouse-eared bat, calibrates a magnetic compass by the sun Richard A. Hollanda,1, Ivailo Borissovb,c, and Bj��rn M. Siemersc aDepartment of Migration and Immuno-Ecology, Max Planck Institute for Ornithology, 78315 Radolfzell, Germany bBulgarian Bat Research and Protection Group, National Museum of Natural History, 1000 Sofia, Bulgaria and cSensory Ecology Group, Max Planck Institute for Ornithology, 82319 Seewiesen, Germany Edited by Thomas Kunz, Center for Ecology and Conservation Biology, Boston, MA, and accepted by the Editorial Board March 9, 2010 (received for review October 28, 2009) Recent evidence suggests that bats can detect the geomagnetic field, but the way in which this is used by them for navigation to a home roost remains unresolved. The geomagnetic field may be used by animals both to indicate direction and to locate position. In birds, directional information appears to be derived from an interaction of the magnetic field with either the sun or the stars, with some evidence suggesting that sunset/sunrise provides the primary directional reference by which a magnetic compass is calibrated daily. We demonstrate that homing greater mouse- eared bats (Myotis myotis) calibrate a magnetic compass with sun- set cues by testing their homing response after exposure to an altered magnetic field at and after sunset. Magnetic manipulation at sunset resulted in a counterclockwise shift in orientation com- pared with controls, consistent with sunset calibration of the mag- netic field, whereas magnetic manipulation after sunset resulted in no change in orientation. Unlike in birds, however, the pattern of polarization was not necessary for the calibration. For animals that occupy ecological niches where the sunset is rarely observed, this is a surprising finding. Yet it may indicate the primacy of the sun as an absolute geographical reference not only for birds but also within other vertebrate taxa. navigation | orientation | sun compass | Chiroptera | sensory ecology Sneticthe ince discovery that migrating birds can use the geomag- field to designate direction, research on the manner in which different animals use this cue for orientation and navi- gation has flourished (1���4). As a general notion, animals could potentially use the geomagnetic field for obtaining directional information, the ���compass sense��� (5), and/or to locate position, the ���map sense��� (6). In birds, a number of seemingly contra- dictory findings suggested that the geomagnetic field interacts with either the sun or the stars to provide directional information (7). Calibration by such celestial cues may be important in mi- grating animals to correct for declination error [the difference between geographical and magnetic north (8)] or to avoid in- creased paths through inaccuracies (9). Recent evidence suggests that polarized light cues at sunset and sunrise may provide the primary directional reference that calibrates a magnetic compass in migrating birds (8���11). Surprisingly, only recently has it been shown that bats are able to detect the geomagnetic field (12���14), although so far evidence exists only for two species, one in the United States (12, 13) and one in Asia (14). Using the technique of Cochran et al. (8), in which the animal is exposed to an altered magnetic field at sunset and then released into the natural magnetic field, the results of ref. 12 suggested that, like birds, bats used the geomagnetic field to determine direction following calibration by sunset cues. However, this interpretation remains uncertain on two counts. First, the experimental design in ref. 12 involved transport in a variable magnetic field, making it uncertain which component of the bat���s navigation system (map, compass, or compass calibra- tion) was critically affected. Second, the technique of rotating the magnetic field at sunset, followed by release into the natural magnetic field, has been criticized in being unable to distinguish between the hypothesis of sunset calibration and an ���over- compensation��� response to the calibration of a celestial compass by the magnetic field (15, 16). These authors (15, 16) argue that this is a similar response to that shown by homing pigeons kept in an altered magnetic field for long periods. These birds showed a counterclockwise deflection on their second release following displacement, after having shown a clockwise deflection upon their first release (17). Most insectivorous bats occupy an (almost) exclusively noc- turnal niche (18), and thus the way in which celestial and mag- netic cues could interact to provide directional information could potentially differ from birds. This is especially true for the species chosen in this study, Myotis myotis, the greater mouse- eared bat. This bat emerges from its day roost to forage after the sun���s disk has passed below the horizon (19, 20), although the postsunset glow is still visible. Thus, this species represents an excellent cross-taxon comparison of the generality of the hypothesis that the sun provides the primary geographical ref- erence by which all other directional cues are calibrated (10) or whether the cues used for orientation depend on the eco- logical niche of the species. Thus, the aim of our study was to test the hypothesis that sunset cues calibrate the magnetic field by comparing the homing orientation of displaced bats exposed to a rotated magnetic field. In the first of two experiments, bats were exposed to a magnetic field rotated east during sunset. If bats were using a magnetic compass calibrated with sunset cues, we would expect their ori- entation to be rotated counterclockwise from controls. If they calibrated a star compass with the magnetic field, we would predict them to be rotated clockwise with respect to controls (cf figure S1 in ref. 12). In the second experiment, bats were exposed to a magnetic field rotated east after all sunset cues had dis- appeared. This comparison has the potential to distinguish between sunset calibration and alternative hypotheses. If the experimental bats in this treatment were oriented counter- clockwise with respect to controls, this would indicate that observation of the sun���s position at sunset is not responsible for the effect. Results Experiment 1: Exposure to a Rotated Magnetic Field at Sunset. Both control and experimental groups had orientations significantly different from random (Rayleigh test: controls: r = 0.71, P = 0.012, n = 8 experimentals: r = 0.65, P = 0.028, n = 8). Fig. 1 shows the orientation of the two groups. There was a significant difference between the mean angular directions of the two groups (Watson���Williams test, F1,14 = 7.69, P = 0.014), with the Author contributions: R.A.H. and B.M.S. designed research R.A.H., I.B., and B.M.S. per- formed research R.A.H. and B.M.S. analyzed data and R.A.H and B.M.S. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. T.K. is a guest editor invited by the Editorial Board. 1 To whom correspondence should be addressed. E-mail: rholland@orn.mpg.de. www.pnas.org/cgi/doi/10.1073/pnas.0912477107 PNAS | April 13, 2010 | vol. 107 | no. 15 | 6941���6945 ECOLOGY

experimental group being shifted 77.56�� counterclockwise from the controls. The 95% confidence interval did not include the home direction in either group (confidence interval test, home direction = 224�� controls: 95% confidence interval = 223.4��, P 0.05 experimentals: 99% confidence interval = 157.3��, P 0.01). Experimental (median 7.5 min) and control bats (median 8 min) vanished quickly from the release site, and there was no significant difference between groups (Mann���Whitney U test: U = 25.5, P = 0.491). There was also no significant difference in homing times between the two groups (Mann���Whitney U test: U = 16, P = 0.749), although very few bats homed on the first night of release (Fig. 2). Experiment 2: Exposure to a Rotated Magnetic Field After Sunset. Both groups were significantly oriented (Rayleigh test: controls: r = 0.62, P = 0.039, n = 8 experimentals: r = 0.758, P = 0.006, n = 8). Fig. 3 shows the orientation of the two groups, and again the 95% confidence interval did not include the home direction in either group (confidence interval test, home direction = 224�� controls: 95% confidence interval = 223.7��, P 0.05 exper- imentals: 95% confidence interval = 212.9��, P 0.05). There was no significant difference between the two groups, either in mean angular direction (Watson���Williams test: F1,14 = 0.037, P = 0.85) or in distribution (Watson���s U2 test: U2 = 0.113, 0.5 P 0.2). Again there was no significant difference in time to vanish [Mann���Whitney U test: U = 23, P = 0.340 medians 6.5 min (experimental bats) and 8 min (controls)] or homing times (Mann���Whitney U test: U = 19, P = 0.483), with few bats homing the same night as release (Fig. 3). There was no sig- nificant angular difference between control groups from experi- ments 1 and 2, either in the mean angular difference (Watson��� Williams test: F1,14 = 0.003, P = 0.954) or the distribution (Watson���s U2 test: U2 = 0.085, 0.5 P 0.2). There was also no difference between the controls from experiment 1 and the magnetic-treated group from experiment 2, either in the angular difference (Watson���Williams test: F1,14 = 0.08, P = 0.781) or the distribution (Watson���s U2 test: U2 = 0.043, P 0.5). However, there was a significant difference between the magnetic-treated groups from experiments 1 and 2 (Watson���Williams test: F1,14 = 7.253, P = 0.017). Twenty-six of 32 released bats homed while an automated receiver that we had deployed at the site of capture was in place. There was no difference in homing times between any of the four groups (Kruskal���Wallis test: ��2 = 5.361, P = 0.147). However, we note that it appears that the bats from night 1, both control and experimental, took longer to home than bats on any other night (Fig. 2). The most likely explanation for this difference are 2-day rain storms that followed the release on night 1 (Materials and Methods), which may have forced the bats to take extra stopover nights before homing to the site of capture. Discussion The results of the first experiment in which the magnetic field was rotated at sunset are consistent with the hypothesis that observ- ing some aspect of sunset calibrates the magnetic field, which is then used to provide bats with directional information. However, this does not conclusively demonstrate that it is the observation of the sunset that is responsible for this calibration, and therefore cannot be distinguished from the overcompensation hypothesis (15). The second experiment, in which the same treatment was used but after all traces of sunset had disappeared, thus provided the crucial test for the hypothesis of sunset calibration of the magnetic field. The lack of an effect on the experimental group in this experiment clearly indicates that it is the exposure to some aspect of the setting sun that is responsible for the shifted ori- entation. As far as bats are concerned, it provides no support for Fig. 1. Vanishing bearings of control (black triangles) and experimental bats (white triangles) exposed to an east-shifted magnetic field at sunset. Mean bearing and vectors of the two groups are shown. The arrow (H) at the edge of the circle indicates the direction to the cave where the bats had been captured. Fig. 2. Box plots of homing times for the two experiments. Medians are plotted with 25th and 75th percentiles at the lower and upper boundaries of the box means are given as hatched lines. Results are plotted by night of release. Experimental groups had the magnetic field shifted 90�� east during the treatment, with bats released on nights 1 and 2 having a view of the sunset and those on nights 3 and 4 only stars. For statistics, see text. Fig. 3. Vanishing bearings of control (black triangles) and experimental bats (white triangles) exposed to an east-shifted magnetic field 1 h and 45 min after sunset. Details as in Fig. 1. 6942 | www.pnas.org/cgi/doi/10.1073/pnas.0912477107 Holland et al.