fields and are nondirectional in their re- sponse. The above results indicate that the vertical response is mediated by a movement-detecting mechanism with the characteristics of the optomotor re- sponse. On the other hand, the lateral response, which is the basis of the cen- tering response as discussed in [2], is mediated by a movement-detecting system with substantially different properties. This confirms our earlier contention [2] that there are at least two distinct movement-detecting systems in insects: one mediating the optomotor response, and the other the centering response. Given that the two responses serve quite different roles for the flying bee, it is not surprising that they should be mediated by different movement-de- tecting mechanisms. The optomotor mechanism integrates motion over large areas of the visual field, as would be expected from a system organized for course control. It does not measure image velocity reliably: the response at a given velocity is strongly dependent on image structure [3]. Such a system would probably suffice, however, for stabilizing yaw, pitch, roll, or altitude: all that is needed is a signal that con- veys the direction of movement of the image reliably. Our observation that the vertical response is manifest at a temporal frequency of 50 Hz, but not at 100 Hz is consistent with the data of Kunze [7], who found that the opto- motor response of the bee is negligible

at frequencies at or beyond 100 Hz. The mechanism mediating the centering response, on the other hand, measures image speed reliably, and indepen- dently of image structure, as we have demonstrated in [2]. This property is essential for a system that is to measure the ranges of surfaces independently of their texture. The small receptive fields used by this mechanism would poten- tially enable range to be measured with good angular resolution. It is intriguing that this mechanism measures the speed, but not the direction of image motion. This may well be a computa- tional shortcut adopted by the nervous system, given that the images of objects always move backwards on the eye when the bee flies forward. A nondirec- tional, speed-measuring mechanism has the additional advantage that it would not produce the large errors that a directional velocity detector would when an edge moves in a direction nearly perpendicular to the detector's preferred null axis. This "obliquity problem" is discussed in detail in [8]. Further work is needed, however, to unravel the significance of nondirec- tional motion detection. It is note- worthy that a nondirectional mechanism has also been implicated in the process by which a peering locust estimates the range of an object [9]. Nondirectional, motion-sensitive neu- rons, with receptive fields 2 °-10 o wide, have recently been found in the locust medulla [10]. Nondirectional

movement-detecting mechanisms have also been implicated in visual course control in flying Drosophila [11, 12].

We thank M. Egelhaaf, M. Ibbotson, T. Maddess, and D. Osorio for helpful comments on the manuscript and T. Goodsell for assistance with the il- lustrations.

Received September 21, 1992

1. Kirchner, W. H., Srinivasan, M. V.: Naturwissenschaften 76, 281 (1989)

2. Srinivasan, M. V., Lehrer, M., Kirchner, W. H., Zhang, S. W.: Vis. Neurosci. 6, 519 (1991)

3. Reichardt, W., in: Processing of Optical Data by Organisms and by Machines, p. 465 (W. Reichardt, ed.). New York: Academic Press 1969

4. Egelhaaf, M., Borst, A.: Naturwissen- schaften 79, 221 (1992)

5. Goetz, K. G. : J. Comp. Physiol. 99, 187 (1975)

6. Hausen, K., Egelhaaf, M., in: Facets of Vision, p. 391 (D. G. Stavenga, R. C. Hardie, eds.). Berlin: Springer 1989

7. Kunze, E: Z. vergl. Physiol. 44, 656 (1961)

8. Srinivasan, M. V., in: Nonlinear Vision, p. 353 (R. B. Pinter, B. Nabet, eds.). Boca Raton: CRC Press

9. Sobel, E. C. : J. Comp. Physiol. A 167, 579 (1990)

10. Osorio, D. : Vis. Neurosei. 7, 345 (1991) 11. Wolf, R., Heisenberg, M. : Nature 323,

154 (1986) 12. Heisenberg, M., Wolf, R.: J. Comp.

Physiol. A 163, 373 (1988)

Naturwissenschaften 80, 41-43 (1993) ©Springer-Verlag 1993

Magnetic Orientation of Honeybees in the Laboratory D. E. Schmitt and H. E. Esch

Department of Biological Sciences, University of Notre Dame, Notre Dame, Indiana 46556, USA

Animals respond in various ways to earth-strength magnetic fields. With certain bacteria there is unambiguous evidence for magnetic orientation [1]. In the case of honeybees (Apis melli- fera) there are reports of the elimina- tion of the systematic variations (Mi6-

weisung) in the bees' dances by com- pensation of the earth's field [2]; disap- pearance of the Mil3weisung when the dances align with the projection of the earth's magnetic vector onto the honey- comb [3]; orientation of horizontal dances with respect to cardinal and

intercardinal compass directions [3]; alignment of comb-building with the earth's field [3, 4]; the ability to dis- tinguish between goals based on the magnetic conditions [5] and the effect of magnetic fields on the bees' sense of time [6]. We have demonstrated direct and manipulable orientation of hon- eybees to the earth's magnetic field in the laboratory. The experimental pro- tocol enables us to answer questions re- garding the nature of the magnetic re- ceptor. Other uses include the study of the role of a magnetic sense in the dance language as well as neu- roethological and ecological questions. Bees were collected individually as they flew from the opening of an eastward

Naturwissenschaften 80 (1993) ©Springer-Verlag 1993 41

facing hive. Each bee was carried in a light-tight container to a darkened (dim, red light invisible to bees) labora- tory two floors below and transferred to a freshly cleaned, 9 cm in diameter test arena with eight radial channels of equal dimension. The arenas were constructed from plastic Petri dishes and the wedge-shaped partitions be- tween channels were made of beeswax. The arena with bee was inserted be- tween horizontal, circular arrays of in- flared emitter-detector pairs (915-nm emission). Time spent by a bee at the end of each channel was recorded. Single bees were monitored for 10-min periods in total darkness, Channels and detectors were rotated between trials to randomize nonmagnetic cues. Pre- ferred direction vectors (Fig. 1) were

a 9 0 °

2 7 0 o

18G 0 o

Z / O °

Fig. 1. Circular distributions of preferred direction vectors from bees collected as they exited their eastward facing hive and then monitored in the test arena, a) Twenty-five bees were exposed to the ambi- ent field, b) 29 bees were exposed to a 90 °- shifted field. Magnetic east is indicated for both treatments. The ellipses indicate the axis of the circular coils used to shift the field. Under both treatments the bees oriented towards magnetic east (ambient field: P ~ 0.001, Hodges and Ajnes' test,

= 19 °; shifted field: P = 0.030, ~p = 79 o)

42

computed trigonometrically [7]. These vectors represent the tendency of the animal to orient in a given direction and their utility was verified using the phototactic response of bees. The Hodges and Ajne test [8] was employed to test whether the hive-exiting bees were oriented in the hypothetical direc- tions of magnetic east (geographic east in the ambient field or geographic north in a shifted field). Observed mean directions, ap, for the treatment samples are indicated by the ar- rowheads inside the unit circles in Fig. 1. Twenty-five bees were exposed to the ambient field while monitored. Twenty-nine other bees were exposed to an artificial field in which the horizontal component of the field was shifted 90 ° counterclockwise. This was accom- plished with two 31.5-cm circular coils separated by 19.5 cm. The average intensity at the center of the artificial field was within 5 % of the ambient field with a 17 % variation in intensity across the test area. Figure 1 a shows that all 25 bees in the ambient field continued to orient east- ward (0 °) as they did when exiting the hive ( P < 0.001). The mean direction for the sample was 19 ° . Presumably, no cue other than the earth's magnetic field was available to the bees by which this could be accomplished. That the bees use the magnetic field to determine direction was confirmed by their re- sponse to the shifted field (Fig. 1 b). In this case the bees oriented about the new hypothetical direction 90 ° to the first ( P = 0.030). The new mean direc- tion for this sample was 79 °. There is no difference between ambient and artificial field treatment groups in the average amounts of time registered at the ends of the channels (208 +_+_ 52 vs. 217 ___ 60 s, respectively; n.s., t-test). The lengths of the preferred direction vectors are significantly different between the ambient- and the artificial- field treatments (0.18 _+ 0.08 vs. 0.27 + 0.17, respectively; P < 0.02, t-test). The differences in vector lengths may be due to the 17 % inhomogeneity of the artificial field. I t is clear that bees can use the earth's magnetic field to obtain directional in- formation. An important next step is to identify the properties of the receptor. Three categories of properties help to describe a magnetoreceptor. These cat- egories involve the receptor's ability to

Naturwissenschaften 80 (1993)

provide information about the mag- netic vector of the external field: (1) symmetry (unipolar vs. axial re- ceptors), ( 2 ) p r o j e c t i o n (intensity-de- tecting vs. directional receptors), and (3) planes and volumes of sampling (in- definitely many but there are the biologically relevant frontal, sagittal, and transverse planes). Sagittal and transverse sampling planes are both ex- amples of dorsoventral planes. Com- binations of receptor sampling planes can lead to volumes of sampling de- pending upon how the animal's nervous system integrates the informa- tion. Based on the data presented in this paper the likelihood of the bees possessing an axial, directional, frontal receptor can be considered small. Otherwise, the bees would have been unable to select the direction of mag- netic east exclusively and could have only, at best, oriented along the mag- netic east-west line. The results pre- sented here do not, alone, exclude the possibility of an axial, directional, dor- soventral receptor, or, dip compass. Dip compasses require another, inde- pendent reference axis provided by gravity, the horizon, etc. [9]. The an- gle that the earth's field makes with the horizontal (or gravity) is used to identify magnetic north or the population- specific preferred magnetic direction. However, if these results are consid- ered along with the observation of the disappearance of the MiBweisung when the dances align with the projection of the magnetic vector into the comb (BTP in Fig. 2, see also [3]), then it appears unlikely that bees are employing a dip compass. For bees on the vertical comb there is no constant reference frame other than the plane of the comb. Even still, the bees would be unable to de- termine the direction BTPbased on the "dip" of the magnetic vector into the comb (angle ~ in Fig. 2). This is be- cause [~ is in no way salient (neither the largest nor smallest angle) and there- fore does not identify the direction of BTP. Regardless of the orientation of the comb in the vertical plane, the angle [3 would never be the largest dip angle measured with respect to the comb, c~ would be (see Fig. 2), except in the spe- cial case where a = [3 = 0 °. Neither would [3 ever be the smallest angle, the vertical component of the field, gener- ally speaking, lies in the plane of the comb and would therefore nearly al-

©Springer-Verlag 1993

a

i I

, i

T I

(x/-r ! - - " Y.)-/B,'--.. . - '" D

b v

Fig. 2. a) Illustration of the relationship of the magnetic elements of the earth's field with a honeycomb oriented at some arbitrary angle c~ to the field (cf. b). Vectors BTP and B~P are components of the total vector, BT, and its horizontal component, BH, BTp and B~P lie in the plane of the comb. The vertical component, By, already lies in the plane of the comb. BTand BH make angles [3 and c~, respectively, with the plane of the comb. The fron- tal plane of the bee (b) coincides with the plane of the comb when the bee is dancing or following a dance. If a bee were using a dip compass referenced to the plane of the comb (there are no other likely references), it would have to detect the direction of Ba~ to account for the disappearance of the Migweisung in the directions of B ~ . However, the angle [3 is in no way special and therefore does not permit detection based on dip. The angle a is the largest angle that the magnetic field makes with the comb, and By obviously makes the smallest angle (0 °) with the comb. See text for a discussion of receptor mecha- nisms consistent with these facts and observed behaviors

ways make an angle of 0 ° with the comb. There are a multitude of possible compass mechanisms: three candidates for magnetoreceptors are consistent with the above discussion. The first re- ceptor model is a directional compass that samples in at least two planes: the dorsoventral and the frontal planes. A

"spherical" compass that sampled in three mutually perpendicular planes or more would suffice and could be an ax- ial compass. The second possibility for a compass mechanism is a polar, frontal, directional compass. Two ex- perimental treatments involving the ex- posure of bees to reversal of the

ver t ica l component of the earth's mag- netic field (horizontal polarity re- maining the same) and reversal of both the horizontal and vertical components (dip angle remaining the same) will cat- egorically resolve the question of wheth- er the bee's compass is polar or axial. The third receptor model is an intensity- detecting receptor which could be restricted to sampling in the bee's frontal plane or along a single body axis. Unpublished results show that bees do detect gradients in magnetic fields. This supports a compass model that re- sponds to intensity as well as direction.

Received September 7, 1992

1. Blakemore, R. E: Science 190, 377 (1975)

2. Lindauer, M., Martin, H.: Z. vergl. Physiol. 60, 219 (1968)

3. Lindauer, M., Martin, H., in: Animal Orientation and Navigation, p. 559 (S.R. Galler et al., eds.). Washington, D. C. : NASA 1972

4. DeJong, D.: J. Comp. Physiol. 147, 495 (1982)

5. Walker, M. M., Bitterman, M. E. : J. Comp. Physiol. A 157, 67 (1985); J. Exp. Biol. 141, 447 (1989); 145, 489 (1989); Kirschvink, J. L., Kobayashi Kirschvink, A.: Am. Zool. 31, 169 (1991)

6. Lindauer, M.: Proc. XV Int. Congr. Entomol., p. 450 (1977)

7. Batschelet, E.: Circular Statistics in Biology. London: Academic Press 1981

8. [7], p. 61 9. Wiltschko, W, Wiltschko, R. : Science

176, 62 (1972); Kiepenheuer, J. : Behav. Ecol. Sociobiol. 14, 81 (1984)

Naturwissenschaften 80, 43-46 (1993) @Springer-Verlag 1993

Electric Signaling and Impedance Matching in a Variable Environment The Electric Organ of Mormyrid Fish Actively Adapts to Changes in Water Conductivity

B. Kramer and B. Kuhn

Zoologisches Institut der Universitfit, W-8400 Regensburg, FRG

The electric organs of elephant nose fish, the Mormyridae, are adaptations for active electrolocation ([1]; recent review, [2]) and electrocommunication (reviews, [3-5]). As an any battery, the

current and voltage that the electric organ generates depend on the resis- tance of the load [6, 7]. The comparison of marine and freshwater strongly electric fish revealed anatomical and

physiological adaptations matching the electric organ to the great difference in impedance between the two environ- ments [6]. Weakly electric fish live in tropical freshwaters of low and seasonally variable conductivity (about 5-150 gS/cm, or a resistivity of 200 to 7 kE2.cm). It is unknown whether weakly electric fish are able to adapt the biophysical properties of their electric organ to this great impedance varia- tion. The inability to do so would greatly affect the EOD (electric organ discharge) waveform, as shown in the mormyrids Pollimyrus isidori and Pe- trocephalus bovei [8], and probably re- duce the usefulness of the EOD as a

Naturwissenschaften 80 (1993) ©Springer-Verlag 1993 43