Naturwissenschaften

, Volume 94, Issue 10, pp 813–819

Homing pigeons (Columba livia f. domestica) can use magnetic cues for locating food

Authors

  • Peter Thalau
    • Fachbereich BiowissenschaftenJ.W.Goethe-Universität Frankfurt
  • Elke Holtkamp-Rötzler
    • Fachbereich BiowissenschaftenJ.W.Goethe-Universität Frankfurt
  • Gerta Fleissner
    • Fachbereich BiowissenschaftenJ.W.Goethe-Universität Frankfurt
    • Fachbereich BiowissenschaftenJ.W.Goethe-Universität Frankfurt
Original Paper

DOI: 10.1007/s00114-007-0259-6

Cite this article as:
Thalau, P., Holtkamp-Rötzler, E., Fleissner, G. et al. Naturwissenschaften (2007) 94: 813. doi:10.1007/s00114-007-0259-6

Abstract

An experimental group of homing pigeons (Columba livia f. domestica) learned to associate food with a magnetic anomaly produced by bar magnets that were fixed to the bowl in which they received their daily food ration in their home loft; the control group lacked this experience. Both groups were trained to search for two hidden food depots in a rectangular sand-filled arena without obvious visual cues; for the experimental birds, these depots were also marked with three 1.15 × 106 μT bar magnets. During the tests, there were two food depots, one marked with the magnets, the other unmarked; their position within the arena was changed from test to test. The experimental birds searched within 10 cm of the magnetically marked depot in 49% of the test sessions, whereas the control birds searched there in only 11% of the sessions. Both groups searched near the control depot in 11 and 13% of the sessions, respectively. The significant preference of the magnetically marked food depot by the experimental birds shows that homing pigeons cannot only detect a magnetic anomaly but can also use it as a cue for locating hidden food in an open arena.

Keywords

PigeonMagnetic fieldConditioning

Introduction

The use of magnetic parameters in migration and homing has been demonstrated for a number of avian species (see Wiltschko and Wiltschko 1995, 2005 for review). Conditioning experiments with magnetic stimuli were less successful, with the negative results by far outnumbering the few positive ones (Wiltschko and Wiltschko 1995, 1996). One of the few successful experiments was reported by Reille (1968) who used cardiac conditioning, i.e., stimulus-induced acceleration of heart pulse rate, and obtained significant answers to magnetic stimuli. Bookman (1977) successfully conditioned homing pigeons, Columba livia f. domestica, in a flight tunnel to discriminate between an inhomogeneous magnetic field of about 50 μT and a weak field of 2 μT, and Mora et al. (2004) recently trained homing pigeons to detect the presence or absence of a magnetic anomaly of about 189 μT. In both cases, the magnetic stimulus indicated the correct alternative of a conditioned two-way choice to obtain a food reward.

These two successful operant studies used differences in magnetic intensity as stimuli. Attempts to condition birds to magnetic directions also met with problems and produced many negative results (see Wiltschko and Wiltschko 1995, 1996). Only recently, Freire et al. (2005) successfully trained young domestic chickens, Gallus gallus, to search in specific magnetic compass directions for a social reward. First positive results indicating that pigeons can also be conditioned to magnetic compass directions have been found by Wilzeck et al. (unpublished data) who tested their birds in an operant chamber with a food reward.

In the present experiments with homing pigeons, we again used a magnetic anomaly as stimulus, but the task was different from that of the earlier studies: the anomaly served as a ‘magnetic landmark’ to indicate the position of a hidden food depot in an open arena.

Materials and methods

Test birds

The subjects were adult homing pigeons between 5 and 9 years old at the time of the experiments. Age and sex of the test birds are included in Table 2 and in “Results,” with, e.g., 5-8022-f indicating that pigeon 8022 was a 5-year-old female. All pigeons, except ‘bbwf,’ had a background of numerous homing flights: In their first year of life, they had participated in a standard program of flock toss flights up to 40 km in the cardinal compass directions. Before the present experiments, they had homed singly from experimental releases up to 100 km from the home loft, with the number of these flights varying between individuals. ‘bbwf’ was of unknown origin; it had joined the University flock on its own more than half a year before the experiments began.

The pigeons were divided in two subgroups, designated experimental birds and controls. The two groups lived in adjacent pigeon pens with aviaries. They had access to food and water ad libitum. However, maize kernels, their favorite food, was removed from their normal food ration and served as a reward to motivate them in the tests (see below).

Preparing the pigeons for the experiment

The designated experimental pigeons (groups M) learned to associate a magnetic cue with their food. For 3 to 4 weeks, which included training (see below), they received their daily food ration in a bowl that was equipped with the same type of bar magnets used in training and later in the tests. Feeding them in a magnetically marked bowl continued during testing to reinforce the association between food and a magnetic anomaly. The other pigeons were fed from an identical bowl, but without the bar magnets; they lacked the magnetic experience and served as controls (group C).

At the same time, both groups of pigeons had to learn to dig for hidden food. For each training session, only one pigeon was released into a rectangular wooden arena (2.80 × 2.20 m and 2.80 m high), which was the same where later the tests took place. It was homogeneously illuminated, and the floor was completely covered with a layer of sand, 15 cm thick. In its center, we placed two food depots, about 30 cm apart, each consisting of three cardboard tubes (diameter 5 cm, length 10 cm), which formed a row of about 15 cm.

For the control birds, the two food depots were not magnetically marked, whereas for the experimental birds, in one food depot, each of the three tubes contained a bar magnet with a remanence of 1.15 × 106 μT (AlNiCo500, 6-mm diameter, 30 mm long) that stuck vertically into the sand layer with its south pole pointing upwards (see Fig. 1). The anomaly used in our experiments was designed to be rather irregular with changes in intensity and direction, just to mark the food depot.
https://static-content.springer.com/image/art%3A10.1007%2Fs00114-007-0259-6/MediaObjects/114_2007_259_Fig1_HTML.jpg
Fig. 1

A magnetically marked food depot before it was covered with sand, with the maize kernels and the bar magnets visible

When the pigeons approached the magnetic depot from a northern direction, they experienced a decrease in both intensity and inclination angle. In contrast, when approaching from south, the intensity also decreased, but the inclination increased (Table 1). At a distance of about 15 cm, the inclination was 90° and magnetic north turned from geographic north to geographic south because the local field was now dominated by that of the bar magnets. At 5 cm from the bar magnets, the intensity increased above 200 μT and, thus, exceeded the range of our magnetometer (FL3-BT Triaxial Fluxgate-Magnetometer, Stefan Mayer Instruments, Germany).
Table 1

Changes in magnetic intensity and inclination angle relative at various distances to the magnetic anomaly

Distance from anomaly (cm)

South of the anomaly

North of the anomaly

Intensity (μT)

Inclination (°)

Intensity (μT)

Inclination (°)

80

46.9

66

47.1

66

70

46.6

66

47.1

66

60

46.5

65

47.0

66

50

46.4

65

46.7

66

40

45.9

64

46.2

66

30

44.9

63

45.1

68

20

42.4

58

41.8

75

10

31.5

35

23.1

80a

The local geomagnetic field was 47.0 μT, with 66° inclination.

aAbout 15 cm from the anomaly, magnetic North shifted to geographic South.

At the beginning of the training to dig for the hidden food in the sand, the maize kernels were plainly visible on top of the sand. During the following training sessions, they were successively covered by sand, until finally, at the end of the training, they were completely covered by a 1 cm layer of sand.

A pigeon was considered to be successfully trained when it showed pronounced digging for more than 10 min, with digging defined as a movement of the beak resulting in a visible displacement of sand. The training period lasted about 3 weeks. Of a total of 30 pigeons, 24 successfully learned the task to dig for hidden food, 16 of the experimental group and 8 control birds. These birds were then tested in the experiments reported here. Of the 16 experimental pigeons, 8 had previously participated in pilot experiments with an identical setup of food depot (group M1), while the other 8 and the control birds did not have any previous experience in searching for food depots (groups M2 and C).

Test performance

For the tests, we used two food depots, one marked with three magnets as described above, while the other, the control depot, was unmarked. The two food depots, with all three tubes containing three maize kernels, were hidden in the sand and covered with a layer of sand 1 cm thick. As pigeons may use room features like shape, size, and the distance of a depot to the walls as a cue for locating the hidden food (see Gray et al. 2004), the position of the two food depots was varied pseudorandomly from session to session to render these cues useless. However, the food depots were never placed at a previously used position. The distance between the control and the magnetically marked depot was at least 150 cm. Before each test session, the sand of the arena was completely cleaned and smoothened to avoid visual cues.

A test session lasted 15 min. The pigeons were tested once a day, twice per week, which added up to a total of 7 to 10 test sessions per bird. They were observed with the help of a small video camera mounted at the ceiling of the experimental arena. The camera was connected to a video recorder and a TV screen in an adjacent room.

A test was scored as ‘food obtained,’ when the pigeon found and ate at least one maize kernel from one of the food depots. A searching attempt was scored to be ‘near a food depot’ when the pigeons had concentrated more than 50% of their searching efforts within a distance less than 10 cm from a food depot; otherwise, it was considered a searching ‘elsewhere.’

Data analysis and statistics

The analysis is based on the scores of each pigeon. Bird 8158 (in two sessions) and bird 127 (in one session) found both food depots; in these cases, only the depot found first was considered. Three pigeons from group M1, one from M2, and one from the control group were inactive in more than half of the test sessions; their data are not included in the analysis.

A one-way analysis of variance (ANOVA) was used to detect differences within the data, with Bartlett’s test for equal variances used to test for differences within each group and Dunnett’s multiple comparison test to test for differences between groups. For analyzing the responses to the two depots, the two groups M1 and M2 were pooled. The pooled responses of the experimentals and the controls were compared using the Mann–Whitney U test, while the search efforts of each group at the two food depots were compared with the sign test.

Results

We had planned to perform 10 test sessions per pigeon, but the pigeons from group M1 ceased to show searching activity after 7 and the birds from the control group C after 8 test sessions.

Behavioral observations

When pigeons were released into the test arena, they mostly began immediately to search for food by digging in the sand with their beak. In the first test sessions, they began searching in the center of the arena, where the food had been during the training sessions. When they had been successful in finding a food depot during a previous session, they normally started their search at about the point in the arena where that food depot had been. In later test sessions, the pigeons tended to continue this ‘win–stay strategy,’ at least until they found another food depot. This was the only consistent pattern we observed. In some test sessions, birds seemed to follow distinct routes with continuous pecking and digging, whereas in others, they ran and flutter around seemingly randomly, digging and pecking in all sectors of the arena.

Altogether, the searching behavior of our pigeons showed a considerable inter- and intra-individual variance with respect to the time needed to locate food depots, which varied from 1 to 15  min. In numerous sessions, they did not find one at all. Pigeons that were successful in finding a food depot normally kept eagerly digging at this location until the end of the test session, even if there were no more hidden maize kernels. Occasionally, however, they moved to other parts of the arena but always returned to the place where they had discovered the food.

Searching success

The distribution of the searching behavior of the individual pigeons are documented in Table 2; Table 3 compares the attempts of the experimental birds that associated food with the magnetic anomaly and the magnetically inexperienced control groups.
Table 2

Location of searching efforts and food finding of the individual pigeons

Bird

Tests

Magnetically marked food depot

Control food depot

Searched elsewhere

Searched nearby

Found food

Total

Searched nearby

Found food

Total

Group M1

        

6-737-f

7

1 (14%)

3 (43%)

4 (57%)

0

1 (14%)

1(14%)

2 (27%)

5-8022-f

7 (6)

1 (17%)

2 (33%)

3 (50%)

0

0

0

3 (50%)

5-8031-m

7

2 (29%)

3 (43%)

5 (71%)

0

0

0

2 (29%)

5-8111-m

7

3 (43%)

2 (29%)

5 (71%)

0

1 (14%)

1 (14%)

1 (14%)

?-bbwf-?-*

7

0

1 (14%)

1 (14%)

1 (14%)

1 (14%)

2 (29%)

4 (43%)

Group M2

        

9-1255-f

10 (8)

2 (25%)

2 (25%)

4 (50%)

0

0

0

4 (50%)

8-2161-m

10 (8)

1 (13%)

3 (38%)

4 (50%)

0

0

0

4 (50%)

7-127-m

10 (6)

3 (50%)

0

3 (50%)

0

1 (17%)

1 (17%)

2 (33%)

7-261-f

10 (7)

1 (14%)

1 (14%)

2 (29%)

0

0

0

5 (71%)

7-325-m

10 (8)

2 (25%)

3 (38%)

5 (63%)

0

0

0

3 (38%)

5-8006-?

10 (8)

2 (25%)

2 (25%)

4 (50%)

1 (13%)

0

1 (13%)

3 (38%)

5-8158-?

10

0

3 (38%)

3 (30%)

0

4 (40%)

4 (40%)

3 (30%)

Control group C

        

5-9006-f

8

0

0

0

0

1 (13%)

1 (13%)

7 (88%

5-9009-f

8 (7)

0

0

0

0

1 (14%)

1 (14%)

6 (86%)

5-9010-m

8 (7)

0

1 (14%)

1 (14%)

0

2 (29%)

2 (29%)

4 (43%)

5-9011-m

8 (6)

0

1 (17%)

1 (17%)

0

0

0

5 (83%)

5-9016-f

8 (5)

0

1 (20%)

1 (20%)

0

0

0

4 (80%)

5-9020-f

8 (6)

1 (17%)

1 (17%)

2 (33%)

0

0

0

4 (67%)

5-9101-f

8

0

0

0

0

2 (25%)

2 (25%)

6 (75%)

In the bird number, the first digit indicates the age of the bird, m and f at the end indicate whether the birds was male or female, respectively; question mark denotes that the sex is unknown. Tests column gives the number of test and, in parentheses, the number of active tests, if not identical.

M1 magnetically experienced pigeons also involved in a pilot study; M2 magnetically experienced pigeons without previous test experience; C control birds without magnetic experience

Table 3

Comparing food finding and location of searching efforts between pigeons that associated food with a magnetic anomaly and magnetically naive control birds

 

12 Pigeons with magnetic training (89 test sessions)

7 Control pigeons (47 test sessions)

U

Difference significant?

Food found

    

Food with magnets

25 (28%)

4 (9%)

12.5

p < 0.01

Control food depot

8 (9%)

6 (13%)

36

n.s.

Ratio M:0:C, sign test

9:1:2 p < 0.05

3:0:4 n.s.

  

Searching attempts

    

Near food with magnets

18 (20%)

1 (2%)

11.5

p < 0.01

Near control food depot

2 (2%)

0

35

n.s.

Ratio M:0:C, sign test

10:1:1 p < 0.01

1:6:0

  

Elsewhere

36 (40%)

36 (77%)

5.5

p < 0.001

The column U gives the test statistic of the Mann–Whitney U test; the ratio M:0:C indicates which depot was found more often, with 0 indicating that both were found equally often. The ratio of search attempts of the controls near the control depot cannot be tested because of the ties that decrease the sample size.

M magnetically marked depot; C control depot; n.s. not significant

Statistical analysis indicated differences within the data set (one-way ANOVA: F2,16 = 12.39, p = 0.0006). Within each group, the pigeons showed a consistent searching behavior, i.e., there were no ‘lucky birds’ that affected the results of their groups more than others (Bartlett’s test for equal variances, 2.854; p = 0.24). The data of the two experimental groups M1 and M2 are significantly different from those of the control group (Dunnet’s multiple comparison test: q = 4.415 and q = 4.063, respectively, p < 0.01 in both cases). The two experimental groups do not differ from each other (Dunnet’s multiple comparison test: p > 0.05) so that they were pooled for the analysis with the sign test and the Mann–Whitney U test (see Table 3).

The experimental pigeons that had learned to associate the magnetic anomaly with food searched near the magnetically marked depot and found the food in this depot more often than the control depot, whereas there was no such difference in finding the two depots in the control group (see Table 3, columns). Also, the experimental birds searched at the magnetically marked depot with and without success more often than at the control depot, while there was no such difference between the two depots in the control birds. The control birds, on the other hand, searched significantly more often ‘elsewhere’ in the arena (Table 3, lines). The rate of finding the control depot in the experimental birds is very similar to the rates of finding either depot in the control birds, indicating that it might reflect the chance level.

Discussion

Our results demonstrate that pigeons can be successfully trained to associate a local distortion of the magnetic field with food and are able to use this magnetic anomaly to locate hidden food.

A ‘magnetic landmark’

The pigeons used a ‘win–stay strategy,’ beginning their search where they had previously found food. We cannot define what types of cues or landmarks they used for relocating the point of a previous success in the arena, but findings of Tommasi et al. (1997; Tommasi and Vallortigara 2000, 2004) with chickens and Gray et al. (2004) with pigeons suggest that they may have used the shape of the room and/or the distance between the former food depot and the wall as a cue. Not finding food there caused them to continue searching. When the experimental birds came near the magnetically marked food depot, and probably had detected the anomaly, they started to dig vigorously at this site and very often found the food. The observation that they did not show such a response near the control depot and that the control birds did not show a similar response at all indicates that they could use the anomaly as a ‘magnetic landmark’ as an indicator of the hidden food.

Yet a ‘magnetic landmark,’ in contrast to a visual landmark, cannot be perceived from a distance—the birds must approach it rather closely. Electrophysiological recordings showed that pigeons responded to changes in magnetic intensity of 0.2 μT (Semm and Beason 1990; Beason and Semm 1996), which would mean that our pigeons may have recognized changes of the magnetic field already at a distance of 50 to 80 cm from the magnetically marked depot, depending on the direction of approach. It is unclear, however, how large a change in intensity must be to alert the pigeons and motivate them to start searching. Our setup did not allow us to determine the critical distance. In any case, the area where the pigeons could have perceived the ‘magnetic landmark’ was probably considerably larger than the hidden food depot itself—searching for the magnetic anomaly would thus have enhanced the experimental pigeons’ chances to find food. Yet these pigeons moved about in the test arena pecking and digging with their beak in the sand and throwing it around, behaving just like the controls. This seems to suggest that they also primarily searched for the food itself, using the anomaly as an indicator only when they, by chance, came across it without actively searching for it.

The controls, in contrast, not having any cue that would indicate the presence of a nearby food depot—the anomaly did not mean anything to them—continued to search in the entire arena until they, by chance, found food. This is reflected by their significantly higher portion of searching away from the food depots.

Detecting changes in magnetic intensity

The present study, like the earlier work of Bookman (1977) and Mora et al. (2004), documents behavioral responses to changes in magnetic intensity. In particular, the stimulus Mora et al. (2004) used, a ‘waveshaped’ anomaly produced by a coil system, with an intensity varying from 44 to 189 μT, that is, in a similar order of magnitude as the stimulus used here. The task in our study, however, was considerably more complex than the conditioned two-way choices in the earlier studies.

Mora et al. (2004) reported that a small magnet of 2,500 μT attached to the cere disrupted the pigeons’ response to the anomaly stimulus, as did local anesthesia of the nasal skin and bilateral dissection of the ophthalmic branch of the trigeminal nerve. This indicates that the test anomaly was perceived by receptors innervated by the trigeminal system, the system where the electrophysiological responses to changes in magnetic intensity has been recorded (Semm and Beason 1990; Beason and Semm 1996). The same is to be assumed for our present study. The ophthalmic branch of the trigeminal nerve innervates region of the head where iron-rich material has been identified in passerines and pigeons, namely, the ethmoid region near the nasal cavity (e.g., Beason and Nichols 1984; Beason and Brennon 1986; Williams and Wild 2001) and the upper beak (Hanzlik et al. 2001; Winklhofer et al. 2001; Fleissner et al. 2003). The treatments by Mora et al. (2004) would have affected receptors at both sites.

The idea that animals, in particular birds, use biogenic magnetic material for magnetoreception has been discussed since more than 25 years. Receptors based on magnetite (Fe3O4) were first suggested by Yorke (1979) and Kirschvink and Gould (1981); in the meantime, several models on how such receptors may mediate magnetic information have been forwarded (e.g., Kirschvink and Walker 1986; Edmonds 1996; Shcherbakov and Winklhofer 1999; Davila et al. 2003). Based on detailed anatomical and physical studies that had revealed complex receptor structures consisting of two different iron minerals, maghemite and magnetite (Fleissner et al. 2007), Stahl et al. (2006) developed a new model for a magnetite-based receptor system. This system would act as a biological ‘magnetometer’ detecting the local magnetic intensity.

Receptor systems based on iron minerals in birds are usually discussed as providing a magnetic component of the avian navigational ‘map’ system involved in determining its position with respect to a goal (e.g., Beason et al. 1997; Munro et al. 1997; Wiltschko and Wiltschko 2005; Fleissner et al. 2007). The present study shows that pigeons can use information from a local distortion of the local magnetic field in a rather flexible way: They cannot only be conditioned to use such stimuli in two-choice tests to discriminate between alternatives (Bookman 1977; Mora et al. 2004) but also as signal when actively foraging, focusing their digging activity on a magnetically marked food depot in an open arena. We demonstrated this in a small-scale arena of a little more than 6 m2, but our findings invite speculation that pigeons may also use this ability for navigation, incorporating the position of magnetic anomalies as ‘magnetic landmarks’ into their navigational ‘map.’

Acknowledgements

Our work was supported by the Deutsche Forschungsgemeinschaft (grant to W.W.). We thank Jörg Oehlmann for help with the statistical analysis, Helmut Prior and Christiane Wilzeck for their critical comments and four anonymous reviewers for their helpful suggestions. The experiments were performed in accordance with the rules and regulation of animal welfare in Germany.

Copyright information

© Springer-Verlag 2007