A Critical Review of the Sun-Azimuth Hypothesis

A Critical Review of the Sun-Azimuth Hypothesis

WOLFGANG BRAEMER

Max-Planck-Institut fiir Verhaltensphysiologie, Seewiesen, Germany

I. THE PROBLEM

The ability to find compass directions with the help of the sun, and to keep the same compass direction despite the (apparent) movement of the sun during the day, has been known since 1950 [1, 2, 3]. During the last ten years this sun compass has been found to exist in many birds [3-7], several arthropods [8-14], in fishes [15, 16] of three families, and very recently in reptiles [17, 18]. In analyzing this behavior we encounter several interesting problems of modern biology. (Orien- tation, daily rhythm, seasonal periodicity, evolution.)

An essential prerequisite for this capacity is, besides sight of the sun~a clock furnishing the exact local time. Therefore, direction-finding has become a very useful method for studying this clock [19 25, 16]. Like the small hand of a mechanical clock, the compass animal changes its angle to the sun, or to an artificial light which it takes for the sun, in a 24-hour rhythm. This angle can be measured at any time and is an expression of the animal's condition at that time. The course of these periodical processes can be followed through a series of successive measurements.

The local time contributed by the clock is, however, only one of the conditions necessary to exact direction- finding. Besides the sight of the sun and the local time, the animal needs knowledge of the direction and angular velocity of the sun's movement. If one wants to investi- gate the role of the clock in this complex behavior, the other parameters must be known or at least kept constant. Direction-finding involves the cooperation of several elements, of which the clock is only one; therefore it is profitable to study, at the same time, other activities in which the same clock may be perhaps involved.

The first question which arises in sun-orientation is: what possible information given by the sun is really used by the compass animal? Generally it has been assumed that only the azimuth of the sun's position, that is, the projection of the sun's position onto the horizontal, is used. The sun's azimuth, however, changes during the day with varying speed. In order to maintain one compass direction accurately, the correction of the azimuth movement has to vary with the same rhythm. The azimuth movement varies not only during the course of the day, but also with the seasons, and in addition they are different at different latitudes. An animal which uses only the sun's azimuth for keeping a compass direction has therefore to

413

compensate different azimuth movements in equal time intervals at different times of day, at different seasons, and at different latitudes if it migrates. This is only possible if it possesses a quanti ty of information such as we can derive by complicated calculations from a nautical almanac. Now I will t ry to illustrate these problems, which arise necessarily from the sun-azimuth hypothesis.

Let us first look at the azimuth curves for different latitudes (Fig. 1). The simplest situation is at the time of the equinoxes. The extremes are the poles and the equator. At latitudes near the poles, the sun moves 15~ clockwise (North Pole) or counterclockwise (South Pole) along the horizon. An animal which uses the sun's azimuth for keeping compass directions near the pole has therefore only to compensate 15~ At the equator till noon the azimuth is an east azimuth. At noon the sun culminates in the zenith. From 12:00 to 6:00 P.M. the descending sun has a west azimuth. Be- tween 6:00 A.M. and noon an animal which is using the azimuth for keeping its compass directions has to maintain a constant angle to the sun's azimuth, and between noon and 6:00 P.M., again a constant angle, but 180 ~ opposite to the first one. At latitudes between the poles and the equator, the azimuth of the sun changes in the morning and in the evening slower than 15~ during the noon hours considerably faster. If the animal tries to maintain its compass directions with the help of the sun's azimuth, then it has to compensate slower in the morning and in the evening, and corre- spondingly faster at noon. The azimuth curve for the night is the reversed curve for the time 6 months later. For the equinox, therefore, it is simply the mirror- image of the day curve. During the course of 24 hours the azimuth changes 360 ~ corresponding to the revolution of the earth. I t is often said, therefore, that the animal has simply to calculate 15~ then it will be approximately able to keep its compass direction. This may be correct for latitudes near the pole latitudes; for lower latitudes, however, such a calculating mechanism would lead to considerable errors. Figure 1 shows the differences between an assumed steady compensation of 15~ and the real azimuth curves, to demonstrate how large the errors of the animals' directional choices would be, depending on the local time. Looking at these curves shows one that such an incomplete compass would be worse than no compass at all. There- fore, we have to postulate that at different latitudes the

414 BRAEMER

N ,o.

W,o-

I 1 ,SUN AZIMUTHS ON THE NORTHERN HEMISPHERE

ON MARCH 2.1 AND ,.SEPTEMBER 23 j (~=o)

I I

LATITU DE J

O* EQUATOR } 5" BOGOTA 20 ~ MEXICO CITY 40" PHILADELPHIA,MADRID go* NORTH POLE

S Oe

"i" k- D

N < Z D

E 270o

N ,,0- j I LOCAL

O h 3 6 9 Night. Morning

.i i - 90 = - J

I I

T,M , J I i 12 15 18 21 24

Noon Evening Night

FIGURE 1. Sun azimuth as a function of local time, at the equinox, at five different latitudes. Small figure at lower right: difference between the azimuth position of the sun, at five latitudes (see above), and the straight line corresponding to an azimuth movement of 15~ These curves show the errors to be expected in an animal which uses the sun azimuth to maintain a compass direction and can compensate only at a constant rate of 15~

azimuth movement of the sun is compensated corre- spondingly.

In addition, the azimuth curves vary differently during the seasons at different latitudes. If the compass is to function throughout the year, these differences have to be equalized (Fig. 2).

The second problem is the direction of the sun's movement. In the Northern Hemisphere, between latitude 23.5~ and the North Pole, the sun always moves clockwise from east through south to west. The compass animal must therefore compensate the sun's movement counterclockwise. At the corresponding latitudes in the Southern Hemisphere the sun moves from east through north to west. Here the compass animal must compensate clockwise. A little more complicated is the situation in the tropics, because the sun goes through the zenith and changes its direction twice a year. At the equator, for instance, the sun

culminates at the equinox exactly in the zenith; during the summer months the sun culminates in the north, thus running counterclockwise, and during the winter in the south, thus moving clockwise. Animals which live within the tropics, or which during their migrations cross the equator or the tropics with the help of their sun compass, must have the ability to change the direction of their sun computing mechanism.

In the following sections I would like to discuss the known methods with which sun-compass animals master these difficulties or with which they could master them. In the first section I describe experiments which show that the sun compass functions for 24 hours, despite the fact that in middle and lower latitudes the day is always shorter than 24 hours.

In the second section I summarize experiments in which animals were raised under such conditions tha t they could not see the sun or its movement. Despite

CRITICAL REVIEW--SUN-AZIMUTH HYPOTHESIS 415

N,,.

W,o~

S oO

_

E 2,o.

- / S R : S u n r i s e

N ~o, O h 3 6 Night Morning

SUN AZIMUTHS ON THE NORTHERN HEMISPHERE , J / / ON MARCH 21, SEPT. 23 ((~=0) I I / f / / 3 U N E 21 ( ( ~ = + 2 3 , 5 ) I I ] 15"/hLine ~ /

/ / / /

I / / 11

I

SR

LOCAL TI M E I I I 9 12 15 18 21 24

Noon Evening Night

Difference Angle between the 15~ Line and the Actual Sun Azimuth

FIGCRE 2. Upper figure: Sun azimuth as a function of local time at latitude 50~ I at the equinoxes, I I and I I I at the solstices. Lower figure: Difference between the above curves and a line representing a constant compensation of 15~ Details see Fig. 1 and text.

this, birds and fish are able to compensate the daily movement of the sun [26, 27]. In the third section I briefly discuss experiments which show that the clock which is involved in direction-finding can be changed in a predictable way through phase shifts of the light-dark regime.

In the last section I discuss the influence of daylength on the sun computing mechanism in arthropods and fishes.

II . THE 24-HOUR SUN COMPASS

I will begin by describing some experiments of my own with fishes. Most of these were carried out in

Madison, Wisconsin, during the summers of 1957 and 1958, and the rest during the last year in Seewiesen. I have used the same escape reaction which Schwassmann later used.

The main piece of experimental apparatus is a shal- low, round, rotatable tank, filled with water. I t was mounted on a pier, some distance from the shore of a lake, so that the fish could not see surrounding objects. Inside the tank were 16 small compartments, arranged in a circle with their openings to the outside, and covered with a flat plate so they could not be seen by the fish, which was released in the middle of the tank.

416 BRAEMER

B ' I I

FIGURE 3 A,B. Experimental apparatus: A in plane, B in cross-section. Explanation in text.

During the training period, 15 of the compartments were closed by means of a ring. The 16th compartment remained open. The fish was put in the center of the tank, in a small wire cage whieh prevented its escape. When the cage was lifted, the fish had to learn to seek cover in the one open compartment. The experimenter changed his position constantly, so that the fish could not use him as a landmark. For the same reason, the tank was rotated between runs during the training period (Fig. 3).

A similar arrangement was also set up in a darkroom. A 1000-watt incandescent lamp substituted for the sun. The artificial sun remained in a fixed position and at a constant altitude of 50 ~ (This method has now been considerably improved by H. Schwassmann. By means of a lever outside the tank, the small confining cage is caused to vanish into the bottom of the tank, thus releasing the fish without any stimuli which might affect its directional choice. Furthermore, the fish can now be observed by means of periscopes; the observer sits underneath the apparatus.)

The experimental animals were immature animals, mainly from the sunfish family (Centrarchidae) : Lepo- mis cyanellus (Rafin.), Lepomis gibbosus (Linn.), and Lepomis machrochirus machrochirus (Rafin.). Cichlids (Aequidens portalegrensis [Hensel]) and some salmon (Oncorhynchus kisutch) were also trained. The fish were kept in outdoor aquaria with running lake water.

They were thus exposed to the natural daylength, and could also see the sun.

The first fish trained, a bluegill (L. machrochirus), lived day and night during the whole experiment in the experimental tank outdoors. Only the north-facing compartment was left open. The idea was that the fish would train itself to seek cover in this box. Occasionally the fish was given additional training. I t was put into the center of the tank, released, and chased into the north compartment. I t was rewarded by being left in peace in this compartment. In a few days the fish had learned to swim directly to the north-facing compartment on release. Then the critical tests could begin, in which all 16 boxes were left open. Experiments were made between 8 and 9 A.M., and between 3 and 4 P.M. The results show that the fish actually had learned to flee toward the north (Fig. 4). When the sky was completely overcast, the fish swam at random in every direction. The next experiments were done indoors, between 8 and 9 A.M. and between 3 and 4 P.M. At these times the artificial sun had nearly the same altitude as the natural sun. In these experiments the fish made the same angle to the fixed artificial light as it would have made outdoors at the same time of day to the real sun. I t is thus clearly demonstrated that the sun, real or artificial, was the external reference potin used by the fish for maintaining its compass direction. The fish is able to compensate for the daily movement of the sun, and reacts to the artificial light as if it were the sun.

If it is possible to train fish to swim in a certain compass direction outdoors with the help of the sun, and then to substitute an artificial light for the sun, the reverse should also be possible. The next fish were trained indoors at a certain time of day. Since the artificial sun is fixed, it is important to train the fish exactly at the same time of day. At different times, one would have to train the fish to swim at different angles. The second fish was trained for 10 days to swim, at 1:00 P.M., toward the artificial sun. Then it was tested, early in the morning, with the natural sun outdoors. This fish swam in the direction in which the real sun would be at the training time, thus demon- strating the same ability as the first fish to keep a compass direction with the help of the sun (Fig. 5).

Next, I trained a group of five green sunfish to swim toward the artificial light at 10:00 A.M. These fish were tested in the early afternoon with the real sun; all of them took the compass direction in which the sun would be at 10:00 A.M. (Fig. 6). The same training technique was then further used for several individuals from three families. The fish were always taught to swim toward the artificial light; only the time at which they were trained was varied. At the end of ten days these fish had learned to swim toward the sun at the training time. As a result, fish which have been trained at 6:00 A.M. swim toward the east, fish trained at local noon toward the south, those trained at 6:00 P.M.

C R I T I C A L R E V I E W - - S U N - A Z I M U T H HYPOTHESIS 417

N N

ENOON

N

OVERCA �9 = SCORES OF FiSH TRAINED TO NORTH

AND TESTED iN THE FORENOON o : SCORES OF SAME FiSH TESTED iN

THE AFTERNOON

FIGURE 4. Direction choices of a fish trained to swim north. Each point represents a choice. Upper row: experiments with the natural sun, morning and afternoon. Lower left: experiments with overcast sky. The fish is not oriented. Lower right: Experiments with an artificial sun in a fixed position, morning and afternoon. The fish compensates

N

T

/ / FIGURE 5. Direction choices of a fish trained, at 1:00 P.M., to swim toward the artificial light. Upper: experiments with the artificial sun at 1:00 P.M. Lower left: Experiments with the natural sun, morning. Lower right: with overcast sky. The fish is trained to swim SSW. The arrow shows the direction of the sun at the training time.

toward the west. Fish which have been trained at midnight to swim toward the artificial sun swim in dayt ime tests to the north. They complete the revolution of the sun (Fig. 7).

N

FIGURE 6. Direction choices of 5 green sunfish with the natural sun in the afternoon. The fish were trained at 10:00 A.M. to swim toward an artificial light. All the animals compensate the movement of the sun.

I t is not impor tant tha t the fish are trained to swim toward the artificial light. In several other experiments we trained fish to swim for instance at 6:00 A.M. at an angle of 90 ~ to the left or right of the light. In tests they then swim to the north or south instead of to the east. Not all fish, however, which have been trained outdoors, or indoors at a fixed time, compensate for the movement of the sun. Some of them behave as if the sun were a simple landmark. For instance, of 15 fish trained at night, 10 kept a compass direction and the other 5 swam at a constant angle to the sun. The same behavior is observed in birds and bees. We do not

418 BRAEMER

9d E

O~

~d w

O'S

- - 0

/ X

x x e x x x- sun/ish

01 cich).id

,xx

, I , ) 12.00 18.00 24.o0 06.oo IZpO

Trainin~ Time

FIGURE 7. Fish swim in the direction shown on the ordinate, when they have been trained to swim toward an artificial sun at the time shown on the abscissa. At the times shown, individual fish were trained for 10 consecu- tive days to swim toward the artificial sun. The compass direction they then took under the natural sun was de- termined. Each symbol represents the average of 10 to 15 individual choices. The experiments with each animal were made at a different time of day, thus with different sun positions, and never at the training time. The straight line drawn through the observed points illus- trates a sun movement of 15~ The figure shows that fish can maintain a compass direction both in the daytime and at night. For them, the sun makes a complete circle. X Lepomis cyanellus or Lepomis gibbosus (Cen- trarchidae). �9 Oncorhynchus kisutch (Sahnonidae). O A equidens portalegr ensis (Cichlidae).

know what the reasons are. At first we thought that the incompleteness of the artificial sun might be involved, but the behavior also occurs sometimes in fish trained outdoors and also tested outdoors at other times of day. I t is easy to eliminate this behavior by training the fish at several times. Up to now we have preferred the use of a single training time, following the classical experiments of Kramer. If the animal then shows that it has a compass, it demonstrates at the same time that the computing mechanism for compensating the movement of the sun functions without any special conditioning of the animal by the regular movement of the sun.

The following conclusions can be drawn from the experiments described. Fish have a computing mechanism which tells them that during 24 hours the sun

makes one revolution around the horizon. The same has been shown for the honeybee [28, 29] and for starlings [30]. Indirectly it suggests that all these animals have a very accurate clock. The results suggest that the course of this clock is nearly uniform. For the water skater (Velia currens) and the beach fleas (Talitrus saltator and Talorchestia deshayesi) the sun-eompensa- tion reverses its direction after sunset and runs back- wards throughout the night [13, 14].

III. EXPERIMENTS WITH ANIMALS RAISED UNDER SUNLESS CONDITIONS

Before discussing the modifying influence of external conditions on direction-finding, we have to know which of the essential prerequisites for this capacity enter the organism by way of the genes. Experiments have been made with birds, fish, bees, and ants. Further experiments having to do with this question will also be reported in these meetings. I will first discuss experiments which demonstrate that an animal can compensate the movement of a sun which it has never in its life seen.

A starling, raised at Wilhelmshaven (51 ~ without seeing the sun, was trained and tested with a fixed artificial light. This bird compensated a clockwise movement it had never experienced. Although the length of the artificial day, in which the bird was kept, was adjusted to the season (fall) and latitude, the bird compensated a sun azimuth movement as it occurs in June. Other young starlings, treated in the same way, did not compensate for movement of the sun [26]. These findings are indirectly supported by experiments done with fish raised under sunless conditions. We raised North American green sunfish from the egg in continuous light. Twelve individuals were given a single 15-minute period of intense training (30 trials) and were then tested several hours later on the same day with the real sun. We trained them only once so as to avoid giving any periodic external clue [27]. Nine of these fish compensated a clockwise movement of the sun. Three individuals were disoriented or chose directions with a constant angle to the sun. Those fish which compensated a sun movement did so with a mean rate of 15.8~ Over 24 hours this would give a compensation of 380 ~ corresponding to a shortening of the day by 5.8 per cent. This is in good agreement with Asehoff's rule, according to which day-active animals shorten their subjective day in continuous light [34].

Other green sunfish were raised from the egg in an artificial day-night cycle, whose length always corre- sponded to the season and latitude. These fish were trained at local noon for 7-10 days to swim toward the artificial sun and then tested outdoors in the morning and in the afternoon. All of these fish compensated the sun-azimuth movement correctly according to season and latitude. We found an interesting difference between North American green sunfish and the cichlids,


which come from South America. The latter were kept, trained, and tested in exactly the same way. These fish also compensate the sun's movement approximately according to the season and to latitude 43 ~ but simul- taneously both clockwise and counterclockwise (the lat ter computation would be appropriate for latitude 43~ I t should be added that this species was first imported to the United States in 1930, and repeatedly since then. They have been reared in Florida in outdoor ponds as aquarium fish. Our adult animals came from Florida. This species occurs in South America between 20 ~ and 33 ~ S. I have also found this double compensation in another tropical cichlid, Cichlasoma biocellatum.

In contrast to this, individuals of A. portalegrensis which had always seen the sun in Madison showed a counterclockwise compensation just like that of green sunfish. They had learned not to use the clockwise one. We may therefore suppose that such a change of compensation direction may occur several times during the life of an individual. The fact that both tropical and temperate species of fish can compensate a sun movement of a higher latitude (43~ although the double direction of compensation is appropriate to the tropical distribution, lends further support to the idea, discussed further on, that external factors, such as daylength and sun altitude, can also influence the compensation process.

In the honeybee, according to Kalmus [31], a single direction of compensation, either clockwise or counterclockwise, is inherited. Lindauer has shown, however, that a learning process is involved in the establishment

of the appropriate direction of compensation for a locality. After about 500 training flights, the bees learn to compensate the sun movement in the appropriate direction. From the experiments published to date, i t is not clcar whether bees, which have never had a chance to observe the sun, are disoriented on first seeing the sun, as Lindauer claims, or whether they are only un- certain whether the sun is moving clockwise or counterclockwise, thus simulating disorientation [32]. The ant (Formica tufa L.) compensates the movement of the sun in summer experiments, after the animal could have seen the daily movement of the sun. However, in spring experiments this species did not compensate the movement of the sun. The author assumes, therefore, in accordance with Lindauer, that ants may have the ability to learn individually that the sun is moving [33].

IV. R E S E T T I N G OF THE PHYSIOLOGICAL CLOCK BY S H I F T I N G THE PHASE OF

THE L IG H T -D A RK R E G I M E

In starlings, homing pigeons, and several arthropods, the clock involved in direction-finding could be reset by shifting the phase of the light-dark cycle [19-25]. If, for instance, the artificial day, in which trained birds are kept, begins and ends 6 hours later than before, after some days the birds look for food at a point 90 ~ to the right of the direction in which they were trained. If the day begins 6 hours earlier, they search 90 ~ to the left of the original direction. After completion of the resetting, they maintain the new direction just as accurately as they do the original direction in a normal day.

N

t

A ! �9

o o �9 �9 o �9 eeo

�9 o o

FIGURE 8. Resetting experiment. Upper row: last experiment before resetting. Fish A is trained to the north, Fish B to the south, C to the west--thin black arrow. Lower row: 7th to 13th days after the resetting. The solid black arrow indicates the compass direction in which the fish must swim, if the sun azimuth changed 15~ and the fish compensated 15~ in the other direction. Each point is a critical choice.

420 BRAEMER

I have made similar experiments with fish. Just as in the birds and arthropods, it is possible to reset the clock in fish in any arbitrary way by appropriately shifting the phase of the light-dark cycle. As measured by the direction-finding, the resetting is completed within four days following sudden 6-hour or 12-hour phase shifts. In these experiments, however, a new phenomenon repeatedly cropped up, which I would now like to describe briefly. Let me single out the first group of fish in which I observed it. The group contained 3 sunfish. The first was trained to go north, the second to go south, and the third to swim to the west. After training, the fish were kept for 14 days in a darkroom in an artificial day, displaced in time 14 hours, 10 minutes with respect to the natural day. The daylength was correct for the latitude and season. Twilight was not

considered. From the seventh day on, the fish were tested once or twice daily at different times, with good sight of the sun, and put back into the artificial day. Sometimes such tests were made during the fish's subiective night. In these cases the fish were allowed to adapt for some minutes to an artificial light before being exposed to the bright sunlight (Fig. 8).

To simplify a little, let us assume that the sun's azimuth changes at a uniform rate of 15~ The fish has to change its angle to the sun in the other direction to maintain its compass bearing. Accordingly, if the fish's clock has been shifted 14 hr. 10 rain. by the change in LD phase, it should select a compass direction deviating 213 ~ clockwise from the original training direction�9

Figure 8 summarizes all the experiments from the

I D i

,

/ " , ! , , / 'y

* + ) 7 : 2 0 + ,3:

eo

ee&

9

FIGURE 9. The experiments summarized in the lower row of Fig. 8, shown individually. Irrespective of compass direction, the training direction is in each case placed toward the top (D). No. 1-9 sequence of the experiments. Solid arrow: see Fig. 8. Shaded arrow: direction in which fish swim if they observe the actual azimuth and compensate 15~ Open arrow: direction in which an animal must swim if it correctly compensates the deviations of the azimuth movement from 15~ Because of the resetting this must lead to an error. �9 = Fish trained to the north; + = fish trained to the south; A = fish trained to the west.


seventh to the thirteenth day, with the exception of Ex- periment 10 (Fig. 10). I t is clear from the results that the fish swam in the direction to be expected if the clock had been shifted. In the next figure, we see the experiments in the order in which they were carried out. Note that Experiment 2 was done at 5 P.M.; the subjective time of the fish is then 3 A.M. (Fig. 9). In Experiment 4 it is for the fish a little after 2 A.M., and Experiment 6 is exactly at the fish's midnight. In all of these night experiments the fish swim approximately in the expected compass direction. They behave as if for them the sun continues to move clockwise during the night. This also follows from the results with fish trained during their night to swim toward an artificial light. In contrast to these previous experiments, however, these three fish have never seen either the real or the artificial sun at night.

On the 13th day I again tested these fish, at local noon. Then the sun culminates. For the shifted fish it is 10:00 P.M. This time the three fish behaved completely differently than in the experiments just described. Normally they swim without hesitation in the compass direction to which they have been trained, but here they balked and behaved as if they were suffering from equilibrium disturbances. They tilted around their long axis and swam in small circles, lying on their sides. Finally they did make directional choices (Fig. 10). They seemed to be in a conflict between two directions: the original training direction on the one hand, and the direction imposed by the phase shift on the other. At noon, when the sun reaches its greatest altitude, the fish seem to "know" that it is noon, although their inner clock in this case said 10:00 P.M. This noon effect was also observed in two other groups. In the first one the clock was shifted 12 hours, so the subjective time was midnight. In the second group the subjective time was 6:00 A.M. We know from the behavior of the original shifted group that when subjective midnight falls at another time than noon, the fish choose approximately the direction to be expected on the basis of the shift (Exp. 6, Fig. 9). Thus it seems clear that this

T I 0 ~,s ~ , ~ + �9

FIGURE 10. Experiment with reset animals at local noon. The internal clock of the fish says 10:00 P.M. Thin arrow: original training direction. Solid arrow and individual symbols : see Fig. 8 and 9. The fish are in a situation of conflict between two directions.

abnormal behavior is induced by factors external to the animals and not, for instance, by their physiological state at a particular subjective time.

We began with the simplifying assumption that the sun's azimuth changes 15~ so that the fish must compensate 15~ in the opposite direction. Thus the shift of 14 hr. 10 rain. in the internal clock should result in directional choices which are shifted 213 ~ clockwise from the original training direction (black arrow).

In the experiments just described, deviations from this expected direction appear. They are alike in all three animals. They are especially well marked in the experiment at local noon, but see also Experiment 6, Fig. 9. They are made still more striking at local noon by the accompanying unusual behavior. To explain these "errors" we must first consider a problem which necessarily arises from the sun-azimuth hypothesis. If fish really find their compass directions solely with the help of the sun's azimuth, there appear at first sight to be two possible methods of dealing with the difficulties which arise from the non-uniform movement of the sun's azimuth:

I) The animal compensates at a uniform rate of 15~ Then it must make demonstrable errors in the course of the day, since the azimuth does not change at this constant rate. I t would deviate to the left of its intended bearing in the morning, at 9:00, for instance, and in the afternoon at 3:00 P.M. it would deviate the same amount to the right (Fig. l l ) . The same is true of phase-shifted animals.

II) The second possibility is that the animal is able to compensate exactly the daily and seasonal deviations of the azimuth movement from the 15~ line. I t would then be able to maintain an accurate compass bearing at any time of day. This possibility actually seems to be realized in the arthropods Talitrus saltator and Velia currens [8, 9, 14, 21].

I t must be possible to determine, from the "errors" made by phase-shifted fish, which of these methods the fish employ. If they compensate at a constant rate of 15~ then a shifted animal must behave just the same as one in normal conditions. After shifting their clocks, they must deviate to the left from the intended bearing at 9:00 A.M. and to the right at 3:00 P.M., irrespective of how much their clocks have been shifted.

If, however, the fish is able to compute the azimuth movement with corrections for time of day, then shifting the phase of its clock will also shift the phase of the compensation curve, which is now dependent on subjective time, relative to the curve of the actually occurring non-uniform azimuth movement. In un- shifted animals, the superposition of these two curves results in an exact compass bearing, since they cancel each other out. In animals shifted 14 hr. 10 rain., the superposition should lead to the errors shown in Fig.

422 BRAEMER

- 4 0 ~ l o

- o1+ .++ . .

o / . . . .

_1.200[ "" - ~ . ~ . , , , I

+ 4 0 * b

/ I I I I I I I I I I I I I I I I I I I I I I I 0 2 4 6 8 I0 12 14 16 18 20 22 24

LOCAL TIME

FIGVm~ 11. Curve a shows the difference in degrees (ordinate) between the sun's azimuth on July 21, 1958, and the straight line representing 15~ as a function of the time of day (abscissa). If a fish compensates an azimuth mov- ment of 15~ it must make the errors given by curve a. The possibility exists that the fish compensates these irregu- larities by a process of opposite sign coupled with his physiological clock (curve b) and can thus maintain a constant bearing all the time (thin straight line).

t z 0

l - L ) b J n~

IE

2 +4

4 A

6 "{" �9 :

7 "~:§ i

8

S

IC

I

. _

13 I I I I I I I §162 I I I I I 1 I I I I I I T I I 14 + .~ A

0 2 4 6 8 I 0 12 14 16 18 2 0 2 2 2 4 LOCAL TIME

I I I I I I I I I I I I I I I I I I I I I I I I I 10 12 14 16 18 20 22 24 02 0 4 06 08 10

FISH TIME

FIGURE 12. Resetting experiment. Abscissa local time, underneath it fish time. Fish time and local time are out of phase following the resetting. On the ordinate the 16 boxes are indicated = 360 ~ Dotted line through 1: original training direction. Solid line between 10 and 11: shifted direction (corresponds to the solid arrow in Fig. 9). The thin solid curve is the curve a from Fig. 11. The dashed curve is the curve b, displaced according to fish time. The thick solid curve is the result of adding the two. Each point is a critical choice. These points have no clear relationship to either of the solid curves. Symbols as in Fig. 9.

12. As one can see, however, there is no obvious relationship between the deviations actually recorded and one of these two solid lines.

From these considerations, it will be seen that it is not possible to explain the "errors" of shifted fish on the basis of the varying speed of the sun's azimuth

movement or of its compensation by the fish. We must, accordingly, look for another explanation.

The position of the sun is characterized not only by its azimuth but also by its altitude. Because of the refraction of light rays at the water surface, a water animal's view out of the water is compressed into an


angle of 97.6 ~ . Objects near the horizon appear, verti- cally compressed, at the edge of this circle of light. A rising sun, lying for a land animal on the horizon, appears to the fish accordingly just above 41.2 ~ For the fish, the sun always appears higher than for a land animal, except at the zenith. This simple relationship must be taken into account in all considerations in- volving the altitude of the sun (Fig. 13).

Assuming the fish expects to see the sun at an altitude

r 7~

f ~ - 3 o " "

I o

IO ~ 30" ..50" 70" 90* An#e c~ Elev~t lon

\

FIGURE 13. Refraction of light rays entering the water surface. Because of the refraction, the view out of the water is compressed for a water animal into a visual angle of 97.6 ~ . The apparent sun altitude is always higher than the actual.

appropriate to the time of day, this hypothetical expectation curve will be displaced by any phase- shifting performed, in Experiment 1 by 14 hr. and 10 rain. At 7:00 A.M. the two curves intersect, the actual sun rising while the expected one sinks. The intersection point in the local evening is unimportant for our discussion. I t is not known what sun altitude fishes might be expecting at night. I t seemed simplest to draw in provisionally an altitude which would be 0 ~ for a land animal and thus 41.2 ~ for a fish. The actual altitude curve is the given external clue, the expectation curve is shifted by 14 hr. 10 rain. I t is thus possible to con- front the fish with sun altitudes which deviate from its expectation (Fig. 14).

In the next figure (Fig. 15), the abscissa shows how much higher than expectation the sun was presented to the experimental animals. 0 ~ is the point of intersection of the two curves at 7:00 A.M., and 32 ~ the previously described Experiment 1 at local noon. The compass dial has been cut open and transferred to the ordinate. Each number represents one compartment (22.5~ The dashed line through the center of compartment 1 indicates the direction in which the fish were trained to swim. The solid line between compartments 10 and 11 indicates the bearing expected following the phase shift. The points represent the single choices from Figs. 9 and 10. The more the deviation from the expected direction increases, the greater the difference between the actual and the expected altitude of the sun. At local noon (32 ~ ) the fish swim in the original training direction as well as in the shifted one. In addition, they show the unusual conflict behavior.

Thus far it cannot be said with certainty that the sun's altitude is the external factor causing these deviations. I t could also be another, not yet known factor, correlated with the altitude of the sun. I therefore prefer to speak of a sun altitude-correlated factor.

All the experiments described so far are in good agreement with the assumption that the sun's altitude, or an altitude-correlated factor, gives the fish information about the time of day. This information is un- equivocal for the culminating sun at noon. The first experiment at local noon was done in July (Fig. 16). In

z 80 ~--~

�9 �9 o s ~

41 I �9 t , I l I , I �9 I . I �9 I * I . I �9 I , I . I �9 I .

4 S S IO IZ 14 IS IS I 0 E l 14 0 ~' 0 4 06 O$ LOCAL TWL

�9 ,~, ' ~ ' , ~ ' ~ o ' ~ z ~ , ' & ' ~ ' & ' & " i~) ' ,~,' ' ,:, ' ,~ ' ,~ FISH TiME

FIGURE 14. Resetting experiment. Sun altitude and time of day for a fish during the tests. Solid curve: actual altitude curve. For the night an altitude of 41.2 ~ is shown, corresponding to 0 ~ for a land animal. Dashed curve: sun-path calculated for the artificial day. If the fish expects this sun-path, it is possible to offer him sun altitudes strongly deviating from his expectation. Difference = A h.

424 BRAEMER

16

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ _ _

2

3

4

5

6

7

8

9

IG

I

13

14

0

&

I . . . . ]§ , i , I , , , , I ,

5 o I 0 o 15 o 2 0 ~ , , , I , , k , I , , , ,

2 5 ~ 3 0 "

FIGURE 15. Dependence of the direct ion choices on A h (see Fig. 14). Dashed line: original t r a in ing direction. Solid l ine: shif ted direction. Under these condi t ions A h = 0 ~ at 7:00 A.M.; A h = 32 ~ at 12:00 noon. The points are the indiv idual choices f rom Figs. 9 and I0. See these for explanat ion of symbols.

~ ~ ~ Fi= FISH TIME LT = LOCAL TIME TRAINING DIRECTION

FIGURE 16. Direct ion choices of three groups of fish wi th the cuhn ina t ing midday sun (middle row from top to bo t tom) . In each case the t r a in ing direct ion is to the top. After complet ion of the t ra in ing the fish were reset by shif t ing the phase of the l igh t -dark cycle in the di rect ion of the arrows in the diagram. Fish t ime and local t ime are out of phase. Left row: last exper iment before the noon exper iment . Middle row: exper iment at local noon. R igh t row: exper iments following the noon experiment . At local noon these fish swim bo th in the original t r a in ing direct ion and in the direct ion to be expected if they used only the i r phase-shi f ted clock.

these fish the clock was shif ted 14.10 hrs. A t local noon the sun a l t i tude was 73 ~ . The difference be tween the a l t i tude expected b y the fish and t h a t of the real sun was 32 ~ . The second exper iment was done in August . A t

local noon the sun al t i tude was 64 ~ . The difference between the expected and ac tual a l t i tudes of the sun was 23 ~ . The th i rd exper iment was done in September. At local noon the sun a l t i tude was 61 ~ The difference


40* N

40 ~ S

Mar Jun Sept Dec Mar Jun

FIGURE 17. Day length as a function of latitude and season. The curves connect points of the same daylength. The sine curve indicates the time when the sun passes through the zenith (modified after Baker 1938 [35]).

between the expected and the actual altitudes was 20 ~ . The results of these experiments suggest that the fish knew in each case that it was local noon, because in all 3 experiments the fish showed a conflict between the time of day indicated by their internal clocks (10:00 P.M., midnight, and 6:00 A.M. respectively) and that indicated by the sun altitude, i.e., noon. This conflict was manifested in the abnormal behavior and in the two compass directions taken by the fish.

If the sun altitude is really the factor causing this effect, it would mean that the seasonal variations in altitude are also known to the fish.

At all times of day other than noon, the information from the sun's altitude is ambiguous, because of the symmetrical shape of the altitude curve. Each altitude occurs at two times of day, for instance, at 9:00 A.M. and 3:00 P.M. Then the fish behave as if they select that information from the sun altitude which conflicts the least with that coming from their internal clock.

V. INFLUENCE OF DAYLENGTH

The seasonal changes of the sun's azimuth movement are a function of the changing declination of the sun's arc. This causes a corresponding sequence of changes in daylength, which are different at different latitudes. The resulting photoperiod could affect the quantitative compensation of the sun's azimuth movement. In Fig. 17, we see the variation of the photoperiod or daylength for different latitudes depending on the season. Each curve connects points of equal daylength. These curves show that any given daylength is ambiguous. For instance, a 12:12 day occurs on the equator throughout the year, and at all other latitudes at the equinoxes.

From the study of these curves it is already clear that the length of the day cannot alone be the adequate cause for the modification of the compensation mecha-

nism necessary at different latitudes. At the equinoxes we find a 12:12 day at all latitudes, but very different azimuth curves. A bird which migrates at this time from the North Pole to the equator, or beyond, needs additional external factors to enable its compass to function at different latitudes, even if it has an internal seasonal clock.

Up to the present there are no experiments with a sun-orienting species in which the orientation has been studied at all the different daylengths possible, begin- ning with a 24-hour day (constant light) and going on to an L / D ratio of 18:6, 12:12, and so on. There is also no analysis of the sun-compass behavior of animals kept during the whole year under constant external conditions, a 12:12 day for instance. I t is thus not known whether an internal annual periodicity manifests itself in sun-orientation. For these reasons, it is very difficult to interpret the experiments which have so far been made with different photoperiods.

Among the factors which might influence the compu- ration of the sun-azimuth movement, we should not forget twilight. At every season we have a gradient, with a long twilight near the poles and a considerably shorter one on the equator. The importance of twilight for several biological functions is clear, for example, for the waking-up time of birds [36]. I t is not yet known whether the duration of the twilight affects the computing mechanism for direction-finding.

In the introduction we had to postulate that the sun's azimuth movements, which vary according to season and latitude, are correctly compensated. The results described in Section I I I , with fish raised under sunless conditions, give us an indirect hint that the photoperiod influences the quantitative computing of the sun's azimuth movement in fish. There animals were kept in an artificial LD with the same L : D ratio which we had at the same season at latitude 43~ After being trained with the artificial sun, they were tested with the natural sun, whose altitude was correct according to the time of day, season, and latitude. Both South American ciehlids and North American green sunfish compensated in this experiment a sun- azimuth movement corresponding approximately to latitude 43 ~ . They kept the compass direction to which they had been trained, though the cichlids compensated both clockwise and counterclockwise. How did the fish know they were on latitude 43~ The artificial day paid no attention to twilight. The experiments therefore suggest that the fish got the information about the latitude and the season from the daylength and/or the sun altitude.

Further experiments were made with green sunfish kept in different photoperiods, but with the sun's altitude kept constant in both the training and the test situation. One group of fish was kept for 4 weeks in a summer day; another group in a winter day for latitude 43 ~ The fish kept in the shorter winter day compen-

426 BRAEMER

sated, in the same span of time, a significantly smaller sun-azimuth movement than those kept in the summer day [37]. Similar experiments have been done previously with Velia currens [21]. Kept in the summer day of latitude 48 ~ this insect compensates exactly the sun- azimuth movements of this latitude and season, and kept in a winter day i t compensates the winter azimuth curve for the same latitude. All these experiments with Velia were made with an artificial sun, and i t could be shown that the altitude of the sun was unimportant.

VI. CONCLUSIONS

The ability to maintain compass directions with the help of the sun, widely distributed in the animal king- dom, has in recent years been investigated in the most various animals. In all closely investigated species, the sun compass functions throughout the 24 hours. I t requires that the movement of the sun be calculated. Two different types of compensation have so far been found. In Type I, the animals compensate as if the sun moves in a clockwise direction throughout the 24 hours. In Type II , the animals compensate a clockwise movement during the daytime; between sunset and sunrise, they behave as if the sun moved counterclockwise.

Experiments with animals reared under sunless conditions show that the capacity to calculate the sun's movement is innate in the starling and in fish. In experiments with North American sunfish and South American ciehlids, differences in the innate direction of compensation were found which are appropriate to the differing distributions.

The quantitative computation of the sun movement is affected by the length of the day in which the animals live. Experiments with fish further indicate that the sun's altitude, or a factor correlated with it, also affects the computation. (In the author's opinion it is not impossible that the interaction of daylength and sun altitude so affects the computation quantitatively that it is correct for all latitudes and times of year.)

By shifting the phase of the light-dark cycle, corresponding to a displacement along the same degree of latitude, it is possible in all species so far examined to reset the clock involved in direction-finding in a predictable way.

ACKNOWLEDGEMENTS

These studies of the sun compass of fish in Madison were supported from National Science Foundation Grant No. 3339 to Professor A. D. Hasler, University of Wisconsin. The author wishes to express his warm appreciation to Professor Hasler for the invitation to work in his laboratory and for his unfailing material and moral support.

To Horst Schwassmann, John Neess, and Warren Wisby I am grateful for valuable discussions both in person and by ]etter. They have made an important contribution to the success of the fish orientation experiments. I would also like to thank John Burchard

for careful reading of the manuscript and for his assist- ance in the translation.

REFERENCES 1. FRISCH, K. YON. 1950. Die Sonne als Kompass im

Leben der Bienen. Exper., VI/6: 210-221. 2. KRAMER, G. 1950. Orientierte Zugaktivit~tt gek~-

figter SingvSgel. Naturwissenschaften, 37: 188. 3. KRAMER, G., and U. v. SAINT PAUL. 1950a. Stare

(Sturnus vulgaris L.) lassen sich auf Himmels- richtungen dressieren. Naturwissenschaften, 37: 526-527.

4. - - . 1950b. Ein wesentlicher Bestandteil der Orien- tierung der Reisetaube: Die Richtungsdressur. Z. Tierpsych., 7: 620-632.

5. KRAMER, G. 1952. Experiments on bird orientation. Ibis, 94: 265-285.

6. .1957. Experiments on bird orientation and their interpretation. Ibis., 99: 196-227.

7. SAINT PAUL, U. v. 1953. Nachweis der Sonnenorien- tierung bei n~tchtlich ziehenden VSgeln. Behavior, 6: 1-7.

8. PARDI, L., and F. PAPI. 1953. Ricerche sull 'orientamento di Talitrus saltator (Montagu) (Crustacea-- Amphipoda). Z. vergl. Phys., 35: 459-489.

9. PAPI, F., and L. PARDI. 1953. Ricerche sull 'orientamento di Talitrus saltator (Montagu) (Crustacea-- Amphipoda). Z. vergl. Phys., 35: 490-518.

10. PARDI, L. 1953/54. L'orientamento diurno di Tylos latrellei (Aud. & S a v . ) (Crustacea-Isopoda ter- restria) Bolletino dcll ' Ist i tuto e Museo di Zoologia dell 'universita di Torino, 4: 1-32.

11. PAPI, F., and L. SERRETI. 1955. Sull'existenza di un senso nel tempo in Arctosa perita (Latr.) (Araneae Lycosidae). Memorie LXII--Serie B: 98-104.

12. PAPI, F. 1955. Orientamento astronomico in alcuni Carabidi. Memorie LXII--Serie B: 82-97.

13. PARDI, L. 1957. L'orientamento astronomico degli animali risultanti e problemi attuali. Bollettino di Zoologia, 24: 473-523.

14. BiRu~ow, G. 1956. Lichtkompassorientierung beim Wasserli~ufer Velia currens F. (Heteroptera) am Tage und zur Nachtzeit. Z. f. Tierpsych., 13: 463-484.

15. t~IASLER, A. D., n. M. IIORRALL, W. J. WISBY, and W. BRA~MeR. 1958. Sun-orientation and homing in fishes. Limnology and Oceanography, 8: Nr. 4: 353-361.

16. BRAEMER, W. 1959. Versuche zu der im Richtungs- finden der Fische enthaltenen Zeitsch~tzung. Verh. d. deutschen Zool. Ges. Zool. Anz. 23. Sup- plementband: 276-288.

17. GOULD, E. 1957. Orientation in box turtles. Terra- pene c. Carolina. (Linnaeus). Biological Bull. 112, Nr. 3: 336-348.

18. FISCHER, K., and G. BIRUKOW. 1960. Dressur von Smaragdeidechsen auf Kompassrichtungen. Na- turwiss., 47: 93-94.

19. HOFFMANN, K. 1953. Experimentelle Nnderung des Richtungsfindens beini Star durch Beeinflussung der "inneren Uhr". Naturwiss., 40: 608-609.

20. - - . 1954. Versuche zu der im Richtungsfinden der V6gel enthaltenen Zeitsch54zung. Z. f. Tierpsych. 11: 453-475.

21. BIRUKOW, G., and E. BuscH. 1957. Lichtkompass- orientierung beim Wasserliiufer Velia currens F. (Heteroptera). Z. f. Tierpsych., 14: 184203.

22. PARDI, L., and IV[. GRASSI. 1955. Experimental modification of direction-finding in Talitrus saltator (Montagu) and Talorchestia deshayesei (Aud.)


(Crustaeea--Amphipoda). Experientia, XI: 202- 211.

23. PAPI, F. 1955. Experiments on the sense of time in Talitrus saltator (Montagu) (Crustacea--Amphi- poda). Experientia XI/5: 1-5.

24. SCHMIDT-KOENIG, K. 1958. Experimentelle Ein- flussnahme auf die 24-Stunden-Periodik bei Brief- tauben und deren Auswirkungen unter besonderer Berficksichtigung des Heimfindevermhgens. Z. f. Tierpsych., I5: 301-331.

25. - - . 1960. The sun azimuth compass: one factor in the orientation of homing pigeons. Science, 13I: 826-828.

26. HOFFMANN, g . 1953. Die Einrechnung der Sonnen- wanderung bei der Richtungsweisung des sonnenlos aufgezogenen Stares. Naturwiss. 40: 148.

27. BttAEMER, W., and H. SCHWASSMANN. Der Sonnen- kompass bei sonnenlos aufgezogenen Fischen. (In preparation.)

28. LINDAUER, M. 1954. Dauert~nze im Bienenstock und ihre Beziehung zur Sonnenbahn. Naturwiss., 41: 506.

29. - - . 1957. Sonnenorientierung der Bienen unter der Nquatorsonne und zur Nachtzeit. Naturwiss., 44: 1-6.

30. HOFFMANN, K. 1958. Die Richtungsorientierung von Staren unter der Mitternachtssonne. Z. f. vergl. Phys., 4I: 471-480.

31. KALMUS, H. 1956. Sun navigation of Apis mellifica in the southern hemisphere. J. exp. Biol., 88/3: 554-565.

32. LINDAUER, M. 1959. Angeborene und erlernte Kom- ponenten in der Sonnenorientierung der Bienen. Bemerkungen und Versuche zu einer Mitteilung yon Kalmus. Z. f. vergl. Physiol., 42: 43-62.

33. JANDER, R. 1957. Die optische Riehtungsorientierung der roten Waldameise (Formica rufa L.) Z. f. vergl. Physiol., 40: 162-238.

34. ASCaOFF, J. 1958. Tierische Periodik unter dem Ein- fluss von Zeitgebern. Z. f. Tierpsyeh., 15: 1-30.

35. BAKER, J. R. 1938. Latitude and egg-season~ in old world birds. Proc. Zool. Soc. London Ser. A, 108: 557-582.

36. FRANZ, J. 1949. Jahres- und Tagesrhythmus einiger V6gel in Nordfinnland. Z. f. Tierpsych., 6: 309-329.

37. SCHWASSMANN, H., and W. BRAEMER. Photoperi- odic response in the sun orientation rhythm of fish. (In preparation.)

DISCUSSION SCIIMIDT-KOENIG: Referring to my remark after Professor Birukow's paper, I should like to emphasize the following details: following the tests during daytime in which the two birds allow for the sun's angular velocity by roughly 15~ the birds were subjected to a time shift of 12 hours (shift to reversed day). When they were examined during their personal night, which was actually the natural day outdoors under the natural sun, the birds, one better than the other, compensated continuously for the sun as running counterclockwise. At the birds' personal midnight and only then, however, the direction to be expected agrees with the training direction of the bird. This is to be predicted entirely on the basis of compensation for the sun's movement in azimuth. BR.&EMER: When I first observed the noon-response I thought that this might be due to the physiological

situation of the fish at 10:00 P.M. However, as I said, if the internal clocks were reset by another amount, it was found that this effect always occurred at local noon independent of the internal situation of the fish. If it is certain in your experiment that the clockwise compensation in the subjective night is not a response to the unusual first test at night, we must consider that this type of behavior, which so far has only been observed in arthropods, also occurs in vertebrates. I do not see, however, how you predict this on the basis of compensation for the sun's movement in azimuth on higher latitudes. Horst Schwassmann told me that he had some observations indicating that some fish show the same clockwise compensation which Dr. Schmidt-Koenig just described. However, in considering the sun-azimuth hypothesis, I tried to explain in my paper that a compensation of roughly 15~ is not accurate enough to keep compass directions. People who believe that the azimuth of the sun is the only external reference point for keeping the same compass direction at different times should offer experimental evidence. This has only been done for arthropods! [8, 9, 14, 21.] ENRIGHT: An additional aspect of the sun-azimuth hypothesis that has puzzled me is this: if only the direction of the horizontal component of the sun's position is significant, and the magnitude of this component is relatively small during the midday hours, one might expect that the probability of error would be greater during midday. However, none of the experimental data on any organism that I can recall indicates any greater random scatter, or general loss of precision in the orientation, during the midday hours, relative to early morning or late afternoon. Dr. Braemer's data shows anomalous behavior at local noon, but this is a directional anomaly rather than a matter of random errors. I would therefore like to ask Dr. Braemer whether this apparently unvarying precision seems a relevant objection to the sun-azimuth hypothesis. BRAEMER: As far as I know, no data have been published which describe the scatter in relation to local time. This is a good question that somebody should answer with experimental evidence. SCttWASSMANN: Braemer's experiments with phase- shifted fish have been repeated at Madison. The effect of the midday sun could be reconfirmed. However, the separation of scores into the two directions was not distinct and seemed to depend on the amount of shift and the experimental conditions. Thus, it was possible to produce the "noon response" in two fish not at local noon but a little before 12 hrs. by shifting the phase first and training the fish to an artificial sun in the shifted position. The behavior which was displayed by the fish, when tested under the high sun, was the same as described by Braemer. I assume that the changed behavior can have been caused by the high altitude of the sun which deviated so much from what the fish should have "expected" in their shifted condition.

A Critical Review of the Sun-Azimuth Hypothesis

Documents

Transcript of A Critical Review of the Sun-Azimuth Hypothesis