Electric Signal Reception

An electric signal is perceived as current flows through the specialized low-resistance cutaneous sensory organs belonging to the lateral line system of certain fishes. All the known examples of electroreceptors may be classed as either ampullary or tuberous, based on their anatomical structure. In marine environments, electroreception is fairly widespread, but most of the species possess ampullary electroreceptors. The Ampullae of Lorenzini, found in nearly all sharks, skates, and rays, consists of a flask-shaped ampulla lying deep beneath the surface of the skin and connected to the exterior by a long neck or canal that may be as long as one-third of the fish's body length. The sensory cells lying embedded in the wall of the ampulla are responsive to low-frequency electrical stimuli. A marine catfish, Plotosus, is known to have similarly constructed ampullae that are presumably electroreceptive, and some migratory eels, such as Anguilla rostrata, found in salt water during part of their life cycle, show behavioral responses to weak electrical stimuli but are not yet known to have electroreceptors (Rommell and McCleave, 1972).

Two principal groups of freshwater fishes with well-developed electric capabilities are the gymnotid fish of South America and the mormyriform fishes of Africa. The Gymnotoidei, which are a characoid-related suborder of Cypriniformes (Ostariophysi), consist of four families of closely related fish estimated at between sixty and eighty species. The Mormyriformes, which belong to the suborder Osteoglossomorpha, are composed of two families with an estimated two hundred or more species. All known members of both these groups possess both ampullary and tuberous electroreceptors. Ampullary organs are basically the same flask-shaped structures but with short necks filled with a jellylike substance extending to the surface; they are also known to be responsive to low-frequency electrical stimuli. Tuberous organs, consisting of a receptor cavity buried under layers of loosely packed epithelial cells with no canal to the outside, are responsive primarily to high-frequency stimuli (greater than 50 Hz). Sharing freshwater habitats with the mormyrids and the gymnotids is the very large group of electroreceptive catfish (Siluriformes, Ostariophysi), which are all thought to possess ampullary, or low-frequency, but not tuberous, electroreceptors. An occasional freshwater elasmobranch, such as the freshwater stingray (Potamotrygon circularis), also possesses ampullary electroreceptors (Szabo et al., 1972).

In all likelihood, additional species of electroreceptive fishes will be added to this list as researchers begin to look at different fishes' behavioral responses to weak electric fields. Although there is some confusion as to terminology, excellent reviews of the extensive work on the physiology and anatomy of electroreceptors may be found in Szabo (1965), Lissmann and Mullinger (1968), Bennett (1970, 1971b), Bullock (1973), Scheich and Bullock (1974), Kalmijn (1974), and Fessard (1974).

Signal Production

The production of electric currents is not as widespread as electroreception. In addition to the electrogenic gymnotids and mormyrids discussed above, there is one species of freshwater electric catfish (Malapteruridae). There are also several marine electric fishes, including the electric rays (Torpedinidae), electric skates (Rajidae), and stargazers (Uranoscopidae), as discussed in the reviews by Bennett (1971a), Lissmann (1958), Grundfest (1957, 1960), and Bullock (1973).

Electric currents are generated in specialized organs that are derived from either muscle or nerve, as shown from physiological, anatomical, and pharmacological studies of mature and developing tissue (see review in Bennett, 1971a). In muscle-derived organs, which are the most common, several long columns of multinucleated cells, called electrocytes (Bennett, 1971a), either run the length of the fish (e.g., in the gymnotids) or are localized in specific regions in the tail, head, skin, or pectoral fins (Grundfest 1957, 1960).

Fig. 1 illustrates the mode of action of a simple electric organ from the electric eel, Electrophorus electricus (family Electrophoridae, Gymnotoidei). In this species, placque-shaped cells, innervated on their posterior faces, lie within chambers of connective tissue to form a fairly accurately aligned column. When a nerve discharge excites the posterior face of the placque it generates a spike, which typically overshoots the zero potential by +50 mV. This can be demonstrated by recording the potential difference between two microelectrodes, one placed outside the posterior face of the cell and another placed inside (Fig. IB). But the anterior face of this cell has a low resistance and is not excitable, even if stimulated electrically. When the micro-electrode is advanced through the anterior face into the extracellular space, the discharge potential generated across the posterior face shows up across the entire cell as shown in the oscilloscope tracings in Fig. 1C. Because of the basic asymmetry of these electrocytes, the synchronous discharges from adjacent cells will summate to produce a relatively large voltage. Looking external to the electric eel, the discharge recorded in the water is simply a head-positive, monophasic spike, lasting several milliseconds, and attaining as much as several hundred volts.

The mechanism of activity of electrocytes in other species is usually more complicated. Biphasic discharges are produced by the gymnotid Hypopomus artedi (family Rhamphichthyidae), for example, because both the posterior and anterior faces of the electrocytes are electrically active, firing slightly out of phase with each other (Bennett, 1961). Other species exhibit other mechanisms that result in a variety of complex wave forms (see review in Bennett, 1971a).

All the electric organs known for members of the gymnotoid family Apteronotidae appear to be derived from neural tissue (de Oliveria Castro, 1955; Waxman et al., 1972). Apteronotus albifrons, for example, has an electric organ made up of enlarged loop-shaped spinal neurons, which are myelinated (Waxman et al., 1972). The ap-teronotids are interesting because they seem to be exploiting frequencies at the highest possible limit for electric signaling—1,800 pulses per second, as shown for at least one species (Bullock, 1969; Steinbach, 1970). Whereas sound signaling often occurs at much higher frequencies as small structures are set into vibration, electric signaling has an apparent upper frequency limitation imposed by neural activation and reactivation times of electric organs. There are no known examples of species that can de-couple part of their electric organs in order to achieve higher frequencies. This is probably because in so doing, the current-generating capability, which is directly related to the number of synchronized cells in the electric organ, would be significantly reduced, thereby severely affecting signal range.

Fig. 1. The mechanism of additive discharge in the electrocytes of the electric eel, Electrophorus electricus. Left: Schematic diagram of two electrocytes in the column showing the orientation within the body and the direction of current flow. Right: Oscilloscope tracings of potentials recorded differentially between two electrodes. Positive voltages on the right-hand electrode go upwards.

A.When both electrodes are external to the posterior (innervated) face of the electrocyte, no potential can be seen, save a slight artifact.

B.When the right-hand electrode is advanced through the posterior membrane into the cell, a -90 mv resting potential can be seen. During the electric organ discharge (EOD), this potential changes to show a reversal potential of about +50 mv at the peak of the spike.

C. When the electrode advances through the anterior face (uninnervated), the resting potential disappears but the spike does not. Consequently, during the EOD there is an additive potential difference across each electrocyte. (After Keynes and Martins-Ferreira, 1953.)

In spite of a great deal of diversity in the structure and physiology of electric organs, all electric fish are capable of producing their own signal energy. In this regard, electric communication is distinguished from other communication modalities, which depend on available sources of energy such as sunlight. While this ability permits signaling at night when no external sources of energy are available, it also means that the evolution of signal amplitudes will be consistent with the presence of background noise in the environment.

Signal Transmission

Several peculiar properties of current flow in water are important to our understanding of the evolution of electrical communication. Some of these signal-transmission properties are considered here.

CONDUCTION VELOCITY

Electric signals travel so rapidly in water that conduction times may be considered instantaneous for most biological systems. In this respect the electric modality resembles the visual. Small time delays due to finite conduction times are biologically meaningful in other communication modalities. With sound detection, for example, differences in phase or time of arrival at the two ears provide cues about the location of the sound source (Steven and Newman, 1934; Marler, 1959; Konishi, 1974). Even monaural localization in vertebrates with a pinna depends on non-instantaneous conduction velocities (Batteau, 1967).

Electric fish probably cannot utilize time delays for signal localization, but they may use other mechanisms. Although no definitive work has been done to test the accuracy of spatial localization, Knudsen (1974) has been able to train gymnotid fish to make a choice between a sinusoidal electrical signal coming from a dipole on either the left or the right as they are free-swimming in a nylon mesh starting area in the center of an aquarium. Spatial localization presumably depends on a comparison of signal amplitudes at different parts of the body or on comparisons of amplitudes of different locations made sequentially while swimming. Signal amplitude differences at the skin are known to play an important role in the active detection of objects using electrolocation (Hagiwara and Morita, 1963; Hagiwara, Szabo, and Enger, 1965; Heiligenberg, 1973b, 1975).

The rapid signaling possible in the electric modality contrasts with that for the chemical modality—especially in water (Wilson, 1970).

SIGNAL RANGE

The range of electric signal transmission is limited; estimates vary between several cm and several meters. Granath et al. (1967, 1968) measured the strength of the electric field at various distances from the gymnotid Apteronotus albifrons, as well as its conditioned-response threshold for perception of an electric field. They then estimated that one electric fish should be able to sense another at approximately 3 m. Using a similar approach, Knudsen (pers. comm.) estimated the threshold for signal detection in another gymnotid, Eigenmannia, to occur between 25 cm and 200 cm. The situation is undoubtedly complex. Signal range probably depends on several factors: the size of the signaler and consequent amplitude of its EOD (Brown and Coates, 1952); the size of the receiver and consequent sensitivity of its electroreceptors (Bennett, 1971b); the angular orientation of the receiver's body with respect to the signaler's, and vice versa; the conductivity of the water; and the presence of nonconductors (the bottom, the surface, nonconducting objects) near the sender and receiver that might compress or distort the field. Other factors of importance would include the sensitivity of the receiver's receptors, the nature of the signal, the presence of noise, etc.

Moller and Bauer (1973) demonstrated that there are significant negative correlations between the discharge frequencies of two individual mormyrids (Gnathonemus petersii) when they are separated by distances smaller than 30 cm. When moved further apart, the EOD frequencies of the two fish were unrelated. Similar results were observed using a different technique by Russell et al. (1974). Gnathonemus produces an "echo" response to a conspecific's EOD after a characteristic delay. This response diminishes in intensity as the distance between the two fish increases, until at 30 cm it is no longer evident. The lack of responsiveness in these unconditioned behavioral tests may not be an accurate assessment of the maximum distance of communication; however, these estimates all imply a short-range system that is consistent with the severe rate of attenuation of an electric field surrounding a dipole source. The peak-to-peak electric potential surrounding an Eigenmannia falls off according to the inverse square of distance (Fig. 2A) when measurements are conducted in a large (3 m diameter, 1 m deep) tank (Knudsen, 1975), and the electric field falls off according to the inverse third power of distance, as would be expected for a dipole source (Smyth, 1968). Measurements conducted in a small tank (1 m diameter, 16 cm deep) show the electric field falling off according to r ^-2.3 instead of r ^-3 (Knudsen, pers. comm.). In confined areas the rate of signal decrement may be reduced by the formation of electrical images by the surface and bottom, and signals may therefore extend further horizontally.

Fig. 2. The range of electric signaling is limited by the severe rate of signal attenuation with distance. Top: The peak-to-peak potential falls off according to the inverse square of distance from the null of the fish for various angles. Squares indicate negative values; circles, positive values. Bottom: The electric field strength falls off according to the inverse cube of distance for various angles. All measurements were made on an 18.6 cm-long gymnotid, Eigenmannia virescens, in a 3 m-diameter, 1 m-deep tank filled with water at a conductivity of 2.6 X 10^-2 mho/m. (Data courtesy of Erik I. Knudsen, 1975; from Hopkins, 1974a.)

DIRECTIONALITY

Electric currents lack the directionality characteristic of visual signals. At the source, an electric signal is broadcast in all directions in a typical dipole-shaped field (Hopkins, 1974). Even if the signaler bends its body in one way or another, there probably is not a significant narrowing of the beam. Electric signals are also capable of crooked-line transmission in that they flow around rocks or fallen trees in their path. Because of this, spatial localization is difficult and consequently signals involving spatial patterning are probably insignificant, except at extremely close range. Crooked-line transmission does allow signaling in dense vegetation, however; and suspended particulate matter—a common impediment to visual signaling in tropical fresh water—does not affect current flow.

FADE-OUT

An electric signal fades as soon as it is discontinued, thus necessitating that the recipient be present when the signal is emitted. In contrast to the lingering nature of an odor or visual mark, this property is something of a disadvantage; however, in combination with a rapid conduction velocity, it makes the electric modality ideal for transmitting information that is likely to fluctuate rapidly with time. Signals that allow predictions about an animal's motivation to attack or to flee, for example, need to be transmitted rapidly (Marler and Hamilton, 1966). Fast conduction and fade-out also permit the use of time-varying signals (see discussion in Wilson and Bossert, 1963), and, as shown in the next section, the information content of most electric signals depends on their temporal structure.

BACKGROUND NOISE

The ultimate determinant of the range of a communication signal is the nature and amplitude of background noise in the channel. The predominant source of electrical noise, aside from that from nearby electric fish, is the extremely low frequency (ELF) electromagnetic radiation of terrestrial origin, which appears to be caused principally by lightning (Soderberg, 1969; Watt, 1960; Liebermann, 1956a, 1956b). Because thunderstorms are extremely common in Africa and South America, where electric fish are found, and because electromagnetic waves from lightning travel long distances according to the inverse first power of distance (Watt, 1960), lightning discharges are frequently of sufficient amplitude to be detected by electric fish (Hopkins, 1973). Other inanimate sources of electrical noise may be magnetic storms, earthquakes, the movement of water through the earth's magnetic field (Kalmijn, 1974; Bullock, 1973). Biological sources, particularly other electric fish, add to this noise and create a substantial interference problem. Field observations show that electric fish are commonly found in groups, with many fish spaced only centimeters apart. Not infrequently these groups are composed of members of several species.

Functions of Electric Signals

Although our knowledge of the functions of electric signals is limited by comparison with other modalities, there appears to be a comparably rich repertoire of signals serving as designators and prescriptors (Marler, 1961): signals that dispose the interpreter to make responses appropriate for a particular species, sexual partner, individual, age class, omotivational state of signaler, or to some aspect of the environment.

SPECIES-SPECIFIC SIGNALS

If a signal evokes a response in members of only one species, then it suggests that it is species-specific and reduces the uncertainty about the species identity of the signaler. Such responses may play a crucial role in reproductive isolation or may aid in forming and maintaining social groups for protection or foraging. In searching for species-specific signals, we begin by demonstrating that the physical properties of the signal of interest are distinctive and characteristic in a given population. The resting discharges of the gymnotid fishes from Guyana show certain species-typical patterns, not only in wave form (Fig. 3) but also in frequency. Stein-bach (1970) found similar differences among gymnotids in the Rio Negro, Brazil.

The regular and continuous electrical emissions of Sternopygus macrurus and Eigenmannia virescens, for example, are clearly distinguishable from all other sympatric species on this basis. Both produce "wave" or "tone" discharges in which the impulse is long compared to the interval between impulses, and when compared to other "wave" species, their frequencies are unique (Fig. 4). Eigenmannia in nonbreeding condition respond aggressively toward a Plexiglas fish model with electrodes playing tape recordings of their own species' discharges with headbutting attacks and electrical threat displays (see below), but respond less to recordings of other sympatric species, as shown in Fig. 5. In addition, sine waves of the characteristic frequency are as effective as tape recordings in eliciting agonistic behavior from Eigenmannia, whereas frequencies outside the species range are ineffective (Hopkins, 1974b). Similarly, Sternopygus males give courtship displays in response to sinusoidal electrical stimuli mimicking the discharge frequency of a female Sternopygus, but hardly respond at all to stimuli mimicking other sympatric species (Hopkins, 1972b, 1974c).

Gymnotoid Wave Forms—Moco-moco Creek

Fig. 3. Oscilloscope tracings of the wave forms of the ten most common gymnotids from Moco-moco Creek in Guyana. The records were obtained by placing one electrode near the head and one near the tail. Head positive signals are deflected upwards. (From Hopkins, 1974a.)

But not all wave species have characteristic frequencies. Those in the family Apteronotidae, in particular, show a great deal of overlap in pulse frequency, as shown for Sternarchorhamphus macrostomus and Apteronotus albifrons (Fig. 4), found sympatric in Guyana, frequently in the same habitat. These two species have discharges that are similar with respect to frequency, wave form, and polarity. It is not known by what mechanisms species recognition is accomplished in this case, but it does not appear to involve electrical characteristics of the undisturbed resting discharges. The resting discharges of many of the "pulse" species also do not appear to be species-specific. Gymnotus carapo produces a typical "pulse" discharge in which each impulse is separated by a relatively long interval. It responds aggressively to a wide variety of electrical stimuli but shows its lowest threshold for attack toward 1,000 Hz sinusoidal stimuli—the predominant component frequency of its own individual pulses (Black-Cleworth, 1970). Gymnotus is unspecific in its attacks directed toward other species of electric fish, showing only a moderate preference for attacking other Gymnotus or other pulse species with similar frequencies, e.g., Steatogenes elegans, as shown in a series of hetero-specific aggression experiments. Further research is needed to determine if the characteristic low-frequency or long-duration pulses of Hypopomus artedi or the characteristic high-frequency pulses of Gymnorhamphichthys hypostomus effectively elicit species-specific responses.

Fig. 4. Distribution of discharge frequencies, corrected to 25°C water, of the four species of wave-discharging gymnotids from Moco-moco Creek, Guyana. (From Hopkins, 1974c.)

The shape of the electric field, the use of multiple frequencies, time relationships between pulses of both sender and receiver, frequency modulations, and other time-varying signals might also serve as species-specific signals. The first two possibilities are characteristic of the resting discharge of several species; the others involve temporal modifications of the discharge. Steatogenes elegans and Gymnorhamphichthys hypostomus are known to possess specialized accessory electric organs located on the underside of the head in the region surrounding the urogenital papilla, which fire in synchrony with the main electric organ in the tail (Bennett, 1971a). These organs alter the shape of the electric field locally, and at a range of several centimeters they might serve to identify the signaler (Hopkins, 1974). Adontosternarckus sachsii is unique in South America because it produces two frequencies at once —one using its main electric organ in the tail and the other using its accessory organ in the chin (Bennett, 1971a). Gnathonemus petersii 's echo responses to artificial electrical pulses and to pulses from conspecifics, occurring after a delay of 12 to 14 msec (Russell et al., 1974) might also serve as a species-identification signal in a way analogous to the flash-answer system known for certain fireflies (Lloyd, 1966). This hypothesis needs to be tested by comparing the "echo" responses of sympatric species. As Bell et al. (1974) point out, echo responses could be adapted for preventing coincident pulses with neighbors.

The use of time-varying signals such as frequency modulations and discharge cessations— known to occur in most species of electric fish— is potentially a diverse way of encoding species-specificity. But it has not been shown conclusively that any species relies on these types of signals.

We might expect there to be selection pressure for optimization of certain characteristics of electrical pulses used for electrolocating that might tend to result in convergent evolution on a single type of resting discharge (e.g., Scheich and Bullock, 1974). This trend would counter the intrinsic signal value of species differences in resting discharges. But when two electric fish approach each other in a stream, their electrolocating pulses are likely to be the first indicators of each other's presence. If these pulses are species-specific, rapid and efficient species identification would be possible. Also, any physiological mechanism, such as stimulus filtering, that enhances homospecific communication signals at the expense of extraneous noise produced by other species would be an obvious advantage; and if species-specificity and filtering depended on some property of the resting discharge, the signal-to-noise ratio of the electrolocating system would improve at the same time.

Fig. 5. Results of playback of tape-recorded signals to captive Eigenmannia virescens. Each bar indicates the median number of responses (per 2 minutes) recorded during playback for six fish. Responses include: R=retreats, A=approaches, T=threats, B=at-tacks, and X=discharge interruptions, directed at a Plexiglas model carrying the playback electrodes.

SEXUAL SIGNALS

As an organism comes into reproductive condition, it may respond to signals emitted by members of the opposite sex. While Sternopygus macrurus are reaching sexual maturity, for example, the discharge frequencies of males and females diverge, with males adopting low frequencies (50 to 90 Hz) and females adopting higher frequencies (100 to 150 Hz), with little overlap (Hopkins, 1972b). This frequency difference has communicative significance. As noted before, males respond to the resting electric discharge of females or to sine waves mimicking the female's frequencies by giving electrical displays thought to play a role in mate attraction or in courtship (Fig. 6). They do not respond to sine waves mimicking other males or to those mimicking other sympatric species with wave discharges.

In contrast to this example, electrical courtship displays consisting of brief interruptions in the otherwise continuous EOD can be experimentally elicited from a sexually mature male Eigenmannia during its breeding season by connecting wires to a tank containing another Eigenmannia. A male, a female, or sine waves are all equally effective in eliciting this response: there are no sex differences in the EOD frequency of Eigenmannia. However a male's courtship is reduced when he is presented with a tape recording of another courting male giving discharge interruptions, as it is for a sine wave interrupted artificially (Hopkins, 1974b). Whereas Sternopygus uses its resting discharge to evoke sexual responses, Eigenmannia depends on modifications in its discharge.

Fig. 6. Responses to sinusoidal stimuli by male Sternopygus macrurus in the field. The vertical scale represents the mean increase in the number of responses per one-minute stimulus period, minus the one-minute control period, for twenty trials at each stimulus, and for two males. The horizontal axis compares sine waves with frequencies (as shown) mimicking a male Sternopygus, a female Sternopygus, an Eigenmannia virescens, or an Apteronotus albifrons. Responses include the number of rises, or increases in frequency followed by decreases back to the resting frequency; frequency maxima, or points at which the EOD frequency goes through a maximum; and interruptions, or cessations in the discharge. (From Hopkins, 1972.)

INDIVIDUAL SIGNALS

Responses evoked by signals emitted by a particular individual, a mate, a companion, or a rival, may play an important role in maintaining cohesive social groups of electric fish. Individual recognition requires signals for which there is great diversity within the species, in contrast to species recognition, which requires similarity among members of the species and lack of variation within the individuals. Because each individual fish of a wave-discharging species utilizes an extremely narrow frequency band in its normal resting EOD (as little as 0.5 percent during 10-minute sampling periods; Bullock, 1969) compared to the range for the species, frequency could easily encode information about individual identity, at least over short time spans. Evidence suggests that a male Sternopygus may be able to recognize his mate because of the unique frequency relationship between his mate and himself. During fieldwork in Guyana in 1971, two pairs of Sternopygus were discovered just prior to spawning, and in both cases the male's frequency was exactly one octave below that of the female (Hopkins, 1974c). When a male and a female Sternopygus come together for breeding, either one may change its frequency to be in an octave relation with the other, thereby imitating the partner and facilitating mate recognition.

Even in the mormyrids, which have extremely variable frequencies, individual recognition may be accomplished by an individual's use of different preferred discharge intervals or combinations of intervals. Malcolm (1972) has shown consistent individual differences in the interval histogram patterns of several isolated Gnathonemus petersii, but it remains to be shown whether conspecifics respond differentially to these differences.

AGE-CLASS SIGNALS

In several species of gymnotids, newly hatched fish discharge differently from adults. Of course, in all species there is a good correlation between the size of a fish and the amplitude of its discharge. In Trinidad, young Gymnotus carapo between 6 and 25 mm in length discharged at 15 to 35 pulses per sec; they then gradually adopted the adult frequency of 40 to 60 pulses per sec. Hypopomus brevirostris juveniles in Guyana discharged at higher frequencies (up to 90 per sec) than did the adults (30 to 40 per sec). Monophasic discharges produced by juvenile Apteronotus albifrons gradually change into the adult biphasic discharge by the time the fish are 50 mm long, but the frequency remains in the species-typical range throughout development (Hopkins, 1972a). Because there are no data showing age-specific responsiveness to any of these age-correlated signals, we cannot be sure that they convey age-specific information to the recipients. It is possible that they merely represent developmental changes in the electric organ or control centers.

MOTIVATIONAL SIGNALS

An interpreter may respond to certain signals in a way that is consistent with the most-probable following action of the signal emitter. Such signals appear to allow the receiver to make a prediction about the subsequent behavior of the signaler once the signal has been emitted, and they are therefore important in facilitating social interactions (Marler, 1961; Hazlett and Bossert, 1965; Nelson, 1964). They are exemplified in the electrical modality by signals for threat, submission, and courtship.

Threat

Black-Cleworth analyzed events during fighting behavior in Gymnotus carapo in semi-natural aquaria and found several categories of electric signals that provided clues about the probability of forthcoming attacks from the signaler. One type, an SID display, consisting of a Sharp Increase in the EOD frequency followed by a Decrease back to the resting frequency (Fig. 6 in Black-Cleworth, 1970), usually accompanies and frequently precedes attacks or biting by dominant fish. This was also observed by Valone (1970). SIDS rarely accompany retreats and are consequently more common in the repertoire of dominant fish than of subordinates. Most likely, as a result of repeated temporal association between signal and action, the receiver comes to predict that an attack is highly likely following an SID display and responds appropriately. In an analysis of actions of recipients following SID displays—considered to be "responses," by definition—Black-Cleworth showed that recipients are unlikely to approach or attack but are likely to retreat. Furthermore, a resident Gymnotus in an aquarium is less likely to approach, and spends less time near, electrodes playing artificial SID displays than the same electrodes playing unmodulated pulses.

Analogous displays occur in other species; in fact, the SID format appears to be in widespread use by gymnotids and mormyrids with both wave and pulse discharges. During fighting behavior in Gymnorhamphichthys hypostomus, for example, aggressive individuals give frequent SID displays of varying amplitude and duration. Hypopomus brevirostris even has an exaggerated burst of high-frequency pulses called a "rasp" discharge, in which the normal pulse frequency of 50 Hz is suddenly elevated to several hundred Hz, in addition to its more typical SIDS, which function as aggressive threats (Hopkins, 1974a). While the wave discharge of Apteronotus albifrons is typically a very steady tone at about 1 kHz, an analogous "chirp" display, a sudden increase in frequency by as much as 30 percent, followed by a gradual return to the resting frequency, also serves as an aggressive threat (Black-Cleworth 1970; Hopkins, unpublished). Similarly, Sternopygus macrurus produces a brief SID when disturbed in its natural hiding places, a display that possibly functions as a heterospecific threat. Eigenmannia virescens occasionally produces brief SIDS— termed short rises—during attacks (Hopkins, 1974a).

Among the mormyrids, Möhres (1957), Szabo (in Lissmann, 1961), Bauer (1972), and Bell et al. (1974) found that smooth, rapid increases in frequency to a high level, sometimes followed by a cessation of the discharge, commonly accompanied attack, head butting, and a vigorous antiparallel fighting posture with several species, including Gnathonemus petersii (Fig. 7). The high-frequency bursts of pulses in Gnathonemus often cause a similar reciprocal discharge from a fish of equal aggressive motivation (Fig. 7A, B), but cause a cessation or reduction in frequency in a clearly subordinate fish (Bell et al., 1974). Thus, it appears that this display also serves as a threat by indicating that attack is imminent.

Several species of electric fish produce SID-like accelerations in discharge frequency while attacking prey. For electric rays, eels, and catfish, the long series of high-frequency pulses may actually stun the prey (Bauer, 1968, 1970; Belbenoit and Bauer, 1972), while in other species, such as the stargazer (Astrocopus) and Gymnotus carapo, its function is not known (Pickens and McFarland, 1964). Black-Cleworth has suggested that SIDS may have evolved in conjunction with their function in attack or biting prey, having evolved for display purposes through a process similar to the ritualization of intention movements (Daanje, 1950). If this is true, SIDS would be an example of what Darwin (1872) called "serviceable associated habits." Alternatively, SIDS could be a correlate of strong arousal and may have evolved a display function from this basis. While a sudden increase in frequency followed by a decrease may be an arbitrary representation of its designatum—which in this case is pending attack (see Hockett, 1960; Altmann, 1967)—the convergent evolution of the SID displays in gymnotids and mormyrids strongly supports the contention that the form of the display is not arbitrary but that there are selection pressures on its form that lead to some inevitable pathway of evolution. As yet, these selection pressures are not understood.

It is possible that tonic shifts in the EOD frequency could serve as a signal for threat, particularly in the mormyrids, where the discharge seems to alternate between a highly variable pulse frequency and a highly regular frequency (Moller, 1970). During dominance-determining antiparallel fights in Gnathonemus petersii, for example, the discharge sometimes alternates between one of several preferred intervals, 9 msec or 15 msec (Fig. 7), and in several instances, discharges appear to be delivered in pairs or triplets that result in discrete peaks in the interpulse interval histograms (Bell et al., 1974; Harder et al., 1964; Moller, 1970). Tonic shifts in frequency may also be related to dominance behavior in Gymnotus carapo, where Box and Westby (1970), Westby and Box (1970), and Westby (1975) found a correlation between the mean discharge frequency of an individual and the outcome of its aggressive encounters. Fishes with higher mean frequencies tended to be the more aggressive in their limited sample (Box and Westby, 1970). Tonic shifts in frequency are directly related to arousal among pulse fish, as has been shown in several studies of diurnal rhythmicity of the EOD frequency (Lissmann and Schwassmann, 1965; Schwassmann, 1971; Moller, 1970; Black-Cleworth, 1970).

Fig. 7. Interval diagrams of electric discharges from two mormyrids, Gnathonemus petersii, during fighting. The interdischarge intervals of one fish are plotted as a function of time above the center lines; those of its opponent are plotted below. Very long intervals are off-scale and are not shown in this display.

a. The fish in the upper diagram gives two attacks accompanied by high-frequency (small-interval) discharges. The second attack evokes a brief acceleration in EOD rate in the second fish.

b. Both fish are involved in antiparallel fighting; the fish in the lower diagram gives a head butt that is accompanied by a brief slowing and then acceleration. This fish also shows an interesting alternation between 9-msec and 16-msec intervals.

Convergent evolution like that for SID displays is even more striking in another class of electrical threat signals consisting of brief interruptions in the otherwise steady discharge. Described in Gymnotus (Valone, 1970; Black-Cleworth, 1970; Box and Westby, 1970), discharge "breaks" or cessations lasting 1.5 sec or less frequently occur simultaneously with attack, rarely with retreat. They often precede attacks, but are more typical of individuals who eventually lose fights (Black-Cleworth, 1970). A respondent is more likely to retreat from a fish giving breaks than to approach—a response that is consistent with the attack motivation of the signaler. Discharge breaks are common in other pulse-emitting gymnotids such as Hypopomus beebei (Black-Cleworth, 1970), Hypopomus artedi, Gymnorhamphichthys hypostomus, and Rhamphichthys rostratus (Hopkins, unpublished) and appear to serve a similar function.

Discharge interruptions are also an important display in the wave species, Eigenmannia virescens, as shown in Fig. 8. They are frequently given at the same time as attacks, approaches, or darting-threat movements. Following an interruption, the recipient is likely to withdraw in retreat or else do nothing, but is unlikely to attack or to approach (Hopkins, 1974b). The African counterpart of Eigenmannia, Gymnarchus niloticus, also uses brief cessations (breaks) in its discharge for threats (Fig. 8). They are typically delivered at territorial boundaries in laboratory situations, while attacking, or while giving open-jawed threats. Breaks cause the interpreter to retreat or else defend its territorial boundary. And because the wave form, polarity, and frequency of Eigenmannia and Gymnarchus are so similar, the convergence in this electrical display used for threat seems even more remarkable. We are, once again, reminded of the possibility that this signal, which allows the receiver to predict motivation, may not be arbitrary. Yet, it is not clear that there is any iconic relationship between the signal and its designatum; rather it appears to be a case in which the signal is neither arbitrary nor iconic (Marler, 1961; Altmann, 1967) but somehow physically adapted to its function in the animal's social behavior. This adaptation is not understood.

Submission

Some signals evoke responses that are consistent with a reduced likelihood of attack or an increased likelihood of retreat, withdrawal, or quiet resting on the part of the signaler. These are submissive signals. Responses appropriate to this situation might vary, depending on the context. If an indicator of waning of aggressiveness were to be given during a serious fight, for example, the interpreter might renew fighting with increased vigor. With dominance clearly established, however, a submissive signal might result in a cessation of attack by the respondent. In this case, the submissive signal that reduces stimuli normally eliciting attack would be called an "appeasement display" (Moynihan, 1955; Tinbergen, 1959; Dunham et al., 1968).

Fig. 8. Sound spectrograms of discharge interruptions as threat displays in the African species Gymnarchns niloticus and the South American species Eigenmannia virescens. Spectrograms were prepared using a Kay Electric 7030A Spectrum Analyzer, bandwidth = 37.5 Hz.

One of the better-studied submissive signals is known from Gymnotus carapo. A discharge arrest, or a complete cessation in the EOD for up to three minutes, fills the criterion for a display that reduces or is the antithesis of an attack-eliciting stimulus (Darwin, 1872; Tinbergen, 1959). Black-Cleworth considers any cessation longer than 1.5 sec to be an "arrest." She found that arrests were given exclusively by subordinate fish—those that had been defeated in aggressive encounters—and that they frequently were accompanied by retreat, but rarely by approach or attack. In her analysis of responses, arrests were likely to be followed by approach or by an absence of activity, but rarely by retreat; and biting attacks constituted a lower percentage of responses than expected. Because of this and because the normal EOD is typically an attacking-eliciting stimulus in Gymnotus, she concluded that arrests serve as appeasement signals.

Other pulse gymnotids such as Hypopomus artedi, Gymnorhamphichthys hypostomus, and Hypopomus beebeii produce discharge arrests, but the function is less well known for these species. Although they are given by subordinate fish during agonistic interactions, much like Gymnotus, arrests can be evoked by a wider variety of stimuli, including various frequencies of sine waves, single pulses, or even metal objects near the fish. Bennett (1968) has suggested that arrests might function in hiding; they might also provide a period of quiet listening to the environment.

A completely analogous discharge arrest occurs in the repertoire of subordinate Gymnarchus niloticus. As Szabo and Suckling (1964) and Harder and Uhlmann (1967) noted, Gymnarchus produces cessations lasting for periods up to 20 minutes at a time. Arrests in Gymnarchus, defined as cessations lasting longer than 1.5 sec, were given by subordinate individuals during agonistic interactions in which dominance was clearly established. Arrests accompanied retreats from the dominant, and once given, the dominant's attack level appeared to be reduced. Dominants usually ignored fish with an arrested discharge, only resuming their attacks when their opponent turned its discharge on again (Hopkins, unpublished).

Another class of submissive signals are encoded as frequency modulations in the resting EOD. Eigenmannia produces slight increases in its discharge frequency followed by a decrease back to the resting frequency (Fig. 9). The frequency change is typically on the order of 1 percent, and the duration can be as long as 40 sec. This display is given by subordinate fish—those who have given the least number of attacks in a standard watch and who have lost a competition for a hiding place during the daytime—and very rarely by dominants. These "long rises" are given at the same time as retreat from a dominant. Although this display reflects a clear lack of aggressiveness on the part of the signaler, the conditions under which it might serve as an appeasement display are unknown (Hopkins, 1974b). Remarkably enough, Gymnarchus also produces a submissive signal consisting of modulations in its EOD frequency. As can be seen in Fig. 9, these modulations usually consist of a decrease in the resting discharge followed by an increase back to the resting discharge. Often when a subordinate fish gives a discharge arrest, its discharge resumes and undergoes a long period of modulating frequency, as shown in the examples in Fig. 9. Frequency modulations are only given by subordinate fish, and they also appear to reduce the attack levels from the opponent.

Courtship

At certain times of the year, specialized signal exchanges between males and females appear to facilitate mating behavior in several ways: by reducing the distances between the male and the female, by overcoming aggressive tendencies of partners, by arousing sexual responsiveness, and by synchronizing spawning. Our knowledge of the reproductive behavior of this group of fishes is unfortunately so limited that we cannot be sure of all the functions.

Fig. 9. Sound spectrograms of frequency modulations in the discharge of Gymnarchus niloticus and Eigenmannia virescens. Spectrograms were prepared using the Kay Electric 7030A Spectrum Analyzer, bandwidth = 1.1 Hz.

Sternopygus macrurus in the Rupununi District of Guyana come into breeding condition during the months of April and May—one month before the start of a three-month rainy season. At this time, when the sex of an adult can be identified by the frequency of its regular wave discharges (see above), large males take up residence in hiding places in the creeks that provide excellent cover and protection: under undercut banks, inside sunken logs or stumps, or under large rocks. Although they are not known to defend these hiding places, defense would be nearly impossible to observe under natural conditions. It is not uncommon for several such males to be hiding within meters of each other and, in some cases, for an occasional female or young individual to be present. These males produce a remarkable series of modifications in their discharge, consisting of both rises in frequency and interruptions in the discharge whenever a female passes their hiding place. Examples of this "song" from two males are shown in Fig. 10. There does not seem to be any strict temporal patterning to these discharge modifications, and one individual seems to be different from the next one in the patterns of rises and interruptions that they produce. Interruptions generally lasted between 0.3 and 1.7 sec, and rises sometimes reached 85 Hz above the resting frequency. The natural stimulus eliciting this response is a passing female, but sinusoidal stimuli with frequencies in the female range were just as effective. The female's response to these signals is uncertain, but because the modulated signals are only given by males, only during the breeding season, and only in the presence of a female, most likely the signal normally elicits approach by the female to the male's hiding place so that further courtship activities can take place. The signals may also have a sexually stimulating effect. To an observer, these signals serve to identify a male who is in reproductive condition who is in possession of an appropriate hiding place.

Female Sternopygus also produce electric signals at a later stage in the sexual behavior, when the male and female appear to be paired. In two cases in which an observed male and female were ready for spawning and in which the male's and female's discharges were one octave apart, the female produced slight modulations in her frequency, as shown in Fig. 11. These interesting signals illustrate a unique aspect of the electric modality not discussed earlier.

Because of the octave relationship between the male and the female, the second harmonic of the male's discharge adds to the fundamental of the female's to produce a resultant signal that is amplitude-modulated at the difference, or "beat," frequency of the two signals. The amplitude or depth of modulation is particularly pronounced when the two signals have similar amplitudes, that is, when the two fish are of similar size and are near each other. These beat-frequency amplitude modulations show up in the spectrograms in Fig. 11. When the female changes her frequency slightly, the beat frequency changes too, but the change in the beat frequency is proportionally much larger than the absolute change in the female's frequency. Thus, the two fish act as a "frequency amplifier."

Consider: A male discharge frequency of 65 Hz. His second harmonic at 130 adds to his mate's discharge at 132 Hz to produce an amplitude-modulated signal with a beat frequency of 2 Hz. Now, when the female increases her frequency by 4 Hz, up to 136 Hz, the beat frequency changes to a new value of 6 Hz ( 136-130 = 6 Hz)—three times its earlier value. Thus, comparatively small changes in a female's discharge produce relatively large changes in the beat frequency. Frequency amplification might be exploited to great advantage by Sternopygus for purposes of intra-pair communication. Scheich (1974) has demonstrated that cells in the Torus Semicircularis of the midbrain of the closely related Eigenmannia are sensitive to low-frequency beats produced by the addition of two signals of similar frequency. These cells respond differently to different wave form envelopes and to different beat frequencies—thus the neuronal mechanism for beat-frequency detection may also exist in Sternopygus.

Fig. 10. Examples of naturally occurring sequences of rises and interruptions recorded in the field from two different male Sternopygus macrurus.

A. Recording made on May 1, 1970, Moco-moco Creek, Guyana.

B. Recording made on April 28, 1971, Moco-moco Creek, Guyana. Sonograms were prepared on a Kay Electric 7029A Sound Spectrum Analyzer, bandwidth = 19 Hz. (From Hopkins, 1974c.)

Eigenmannia virescens also produce special signals during their breeding season. Unlike Sternopygus, Eigenmannia wait until the rainy season begins so that they may migrate into flooded swamps and grasslands for breeding. No known sexual differences in frequency are known for this species, but sexually mature males respond to other Eigenmannia by giving many discharge interruptions—similar to those used for threat signals—but longer in duration (up to 0.5 sec) and at a much higher delivery rate—up to 50 per minute. Females also give discharge interruptions but do so at a reduced rate.

Fig. 11. A and B. Two examples of discharge variations produced by a female Sternopygus macrurus (f) held captive with her apparent mate (m). The fundamental frequency of the male's and the female's discharges are shown as m! and f_lf respectively. The second harmonics are labeled m₂ and f₂, and so on. Both wide bandwidth (W; bandwidth = 19 Hz) and narrow bandwidth spectrograms (N; bandwidth = 2.8 Hz) are shown for each sequence. (From Hopkins, 1974c.)

When a male and a female in breeding condition are placed together, the male continues to give long discharge interruptions, and then the female begins to produce modulations in her frequency, composed of long series of "rises" strung end to end. The functions of these signals are unknown. Nevertheless, the parallel between aggressive and submissive signals on the one hand, and male or female sexual signals on the other, is interesting and has been noted for many other species (Hinde, 1970).

Electric signals are not always rigidly stereotyped, but rather show considerable variation. The variations in motivational signals can sometimes be correlated with apparent changes in tendencies to perform certain actions. Thus, we can begin to think of a continuum of signals and even a continuum of responses.

The discharge interruptions produced by Eigenmannia vary in two ways: in repetition rate and in duration. During fighting behavior, Eigenmannia produces interruptions, sometimes singly, but more often in clusters or bouts. The median interval between discharge interruptions is 1.5 sec, and if this is arbitrarily taken as a critical interval for defining a bout, we may then speak of single, double, and triple interruptions, and so on. In an analysis of the simultaneous occurrence of electrical and motor actions during fighting between pairs of Eigenmannia, the probability of attack (butt and chase) increased as the number of interruptions in the bout increased. Responses to these graded signals by the interpreter tended to be consistent with the increased probability of attack by the signaler (Hopkins, 1974b).

While discharge interruptions may be an arbitrary signal representing attack, it is clear, as Marler (1961) pointed out for other, similarly repeated signals, that there is a direct physical relationship between the signals and their designata. A signal that is repeated more frequently, is a "stronger" signal. The iconic, not arbitrary, relationship implied here is clear.

Eigenmannia 's discharge interruptions are also graded by duration. During the nonbreeding season, the median duration of interruptions produced by both males and females during fighting was 40 msec. Those produced by males during the breeding season while in the presence of a female had a longer median duration of 90 msec with an occasional male's interruptions lasting 500 msec. This observation suggests that as the tendency to attack or to court changes, so does the associated signaling.

Black-Cleworth (1970) found that SID displays were also graded. Small SIDS (small-frequency excursion and low-peak frequency) differed from largeSIDS (large-frequency excursion and large-peak frequency) not only in the probability of being associated with attackbut also in the responses that they elicited. Small SIDS were less effective in inhibiting approach than large SIDS.

Finally, the dichotomy between discharge arrests and discharge breaks in Gymnarchus and in Gymnotus can only be considered arbitrary. This continuum of signals reflects a continuum of underlying tendency to attack or to flee. As the fish's apparent attack tendency increases, interruptions tend to be briefer, and as the tendency decreases, interruptions tend to be longer, thus making the fish less conspicuous. The nonarbitrary element of conspicuousness introduced into this graded system by the fact that a signal actually is a period of silence, is probably one factor in the evolution of this continuum of displays.

ENVIRONMENTAL SIGNALS

In a few cases, it could be argued that an electric communication signal evokes a response that is consistent with some object or condition of the environment. It is known, for example, that Electrophorus electricus discharges at high frequency and amplitude when capturing prey or when disturbed. This characteristic burst of discharges causes a marked reaction in other electric eels nearby. They approach the disturbed or feeding eel. Bullock (1969) investigated the approach response and found that he could elicit it with a discharging eel in a net, with electrodes connected to a disturbed eel in an adjacent pond, or with electrodes connected to artificial electrical pulses. It is possible that the approach response represents an adaptive response to the presence of food.

Kastoun (1971) explored responses to environmental conditions in Malapterurus electricus by establishing electrical contact through wires between two fish in separate tanks. When one fish was disturbed by tapping on the side of the tank, it gave a short burst of 2 to 5 impulses and the second fish immediately fled to its hiding place. When the first fish was fed, its 14 to 562 strong discharges occurring during feeding caused the second fish to swim rapidly about the tank making circling motions. Finally, when the first fish was irritated with a small stick, its 21 to 113 impulses caused the second fish to approach and attack nearby objects.

NONSOCIAL SIGNALS

While communication involves adaptive responses to signals, some responses apparently have less social importance than others. As two fish approach each other, their electric signals may begin to interfere with each other's elec-trolocating systems. Heiligenberg (1973a, 1973b, 1974) has shown, in studies of the unconditioned following of laterally approaching and receding plastic objects, that the wave species Eigenmannia is jammed by signals having frequencies similar to its own. Eigenmannia discharging at about 500 Hz fail to respond to the approaching plastic objects when an external sine wave, ± 10 Hz from the fish's EOD frequency, is applied to the water. But, in a remarkable adaptive response to jamming signals such as these, discovered by Wantanabe and Takeda (1963) and explored by Bullock (1969) and Bullock et al. (1972a, 1972b), Eigenmannia shifts its own frequency either up or down by an amount sufficient to prevent interference: up to 20 Hz, depending on the amplitude of the stimulus. Apteronotus albifrons shifts up in frequency, but not down.

Pulse fish, like Hypopygus lepturus, are also jammed by extraneous pulses if they occur at the same time as the fish's EOD (Heiligenberg, 1974). Remarkably, several species of pulse fish produce a transient increase in EOD frequency when external stimuli overlap in time with their own pulses (MacDonald and Larimer, 1970; Heiligenberg, 1974). This response appears to help by lowering the probability of coincident pulses that are known to interfere with electrolocation.

While these electrical responses to the presence of jamming signals—conspecific or otherwise—appear to be adapted solely for preventing electrolocation interference, we cannot help noting the similarity between the Jamming Avoidance Response and Long Rises in Eigenmannia, and between phase-sensitive frequency increases and SIDS in pulse fish. It seems likely that these very basic responses could have played a role in the evolution of social signals. When two Eigenmannia with similar frequencies approach each other, for example, a frequency shift that allows continued approach could act as a signal indicating that aggregation will be permitted without conflict.

ELECTRIC COMMUNICATION IN NONELECTRIC FISH?

Lissmann (1958) suggested that electric organs and electroreceptors evolved because of advantages accrued by the ability to electrolocate in murky water. Wynne-Edwards (1962), on the other hand, argued that intraspecific communication could also have been the primary selective force. If the latter hypothesis were true, we might expect to find examples of primitive electrical communication systems in fish that are nonelectric. There are numerous examples of electrical emissions by nonelectric fish. Kalmijn (1972) has made preliminary measurements of electric field strengths surrounding many marine organisms. He has recorded DC potentials of up to 500 μV near the gills and mouth openings of several species of fish. Low-frequency AC fields attaining 500 μV were also strongest near the gills and head and tended to be synchronous with respiratory movements, while high-frequency AC fields were correlated with trunk or tail movements. Kalmijn (1971) also demonstrated that sharks are capable of homing in on fields like those produced by flatfish (Pleuronectes platessa) buried in sand. Other authors (Roth, 1972; Peters and Buwalda, 1972; Peters and Meek, 1973; Kalmijn and Adelman, in prep.) have evidence that similar phenomena occur in fresh water.

But Wynne-Edwards's hypothesis remains untested. While there is a limited knowledge about a variety of very weak electrical emissions that appear to be correlated with certain types of activity in nonelectric African catfish (Lissmann, 1963, and pers. comm.), there are no known cases of responsiveness to such signals.

Summary

Certain teleosts in both marine and freshwater environments are capable of producing and of receiving electric signals, thereby allowing electric communication—a new modality. Our most detailed knowledge concerns the gymnotid fishes of South America and the mormyriform fishes of Africa, two groups that are known to use their electrical capabilities for electrolocation as well. While electric signals are broadcast with little directionality within what appears to be a limited range, their rapid conduction velocity and fade-out make them ideal for encoding messages that fluctuate rapidly in time. Time-varying signals contribute the greatest diversity to the electric modality.

Electric communication serves many functions in the social behavior of these fish, with designators and prescriptors allowing species, sex, age-class, and individual recognition; in facilitating predictions about the motivation of a companion or a rival and about certain aspects of the environment. Motivational signals known for threat, submission, and courtship appear to be the most complex, with some evidence for grading.

References

Altmann, S., 1967. The structure of primate social communication. In: Social Communication Among Primates, S. Altmann, ed. Chicago: University of Chicago Press.

Batteau, D. W., 1967. The role of the pinna in human localizations. Proc Roy. Soc., B 168:158-80.

Bauer, R., 1968. Untersuchungen zur Entladungstätigkeit und zum Beutefangverhalten des Zitterwelses Malapterurus electricus Gemlin 1789 (Siluroidae, Malapteruridae Lacep. 1803). Z. vergl. Physiol., 59:371-402.

Bauer, R., 1970. La décharge électrique pendant le comportement alimentaire de Electrophorus electricus. J. Physiol. (Paris), 62:341-2.

Bauer, R., 1972. High electrical discharge frequency during aggressive behavior in a Mormyrid fish, Gnathonemus petersii. Experientia, 28:669-70.

Belbenoit, P., and Bauer, R., 1972. Video recordings of prey capture behavior and associated electric organ discharge of Torpedo marmorata (Chondrichthyes). Marine Biology, 17:93-99.

Bell, C. C.; Myers, J. P.; and Russell, C. C.; 1974. Electric organ discharge patterns during dominance related behavioral displays in Gnathonemus petersii. J. Comp. Physiol., 92:201-28.

Bennett, M. V. L., 1961. Modes of operation of electric organs. Ann. N.Y. Acad. Sei., 94:458-509.

Bennett, M. V. L., 1968. Neural control of electric organs. In: The Central Nervous System and Fish Behavior, D. Ingel, ed. Chicago: University of Chicago Press, pp. 149-69.

Bennett, M. V. L., 1970. Comparative physiology: electric organs. Ann. Rev. Physiol., 32:471-528.

Bennett, M. V. L., 1971a. Electric organs. In: Fish Physiology, vol. 5, W. S. Hoar and D.J. Randall, eds. New York: Academic Press, pp.347-491.

Bennett, M. V. L., 1971b. Electroreception. In: Fish Physiology, vol. 5, W. S. Hoar and D.J. Randall, eds. New York: Academic Press, pp.493-574.

Black-Cleworth, P., 1970. The role of electric discharges in the non-reproductive social behaviour of Gymnotus carapo. Anim. Behav. Monog., 3:1-77.

Box, H. O., and Westby, G. W. M., 1970. Behaviour of electric fish (Gymnotus carapo) in a group membership experiment. Psychon. Sei., 21:27-28.

Brown, M. V., and Coates, C. W., 1952. Further comparisons of length and voltage in the electric eel Electrophorus electricus. Zoologica, 37:191-97.

Bullock, T. H., 1969. Species differences in effect of electroreceptor input on electric organ pacemakers and other aspects of behavior in Gymnotid fish. Brain, Behav., Evol., 2:85-118.

Bullock, T. H., 1973. Seeing the world through a new sense: electroreception in fish. Amer. Scientist, 61:316-25.

Bullock, T. H.; Hamstra, R. H., Jr.; and Scheich, H.; 1972a. The jamming avoidance response of high frequency electric fish. I. General features. J. Comp. Physiol., 77:1-22.

Bullock, T. H.; Hamstra, R. H., Jr.; and Scheich, H.; 1972b. The jamming avoidance response of high frequency electric fish. II. Quantitative aspects. J. Comp. Physiol., 77:23-48.

Daanje, A., 1950. On the locomotory movements in birds and the intention movements derived from them. Behaviour, 3:48-98.

Darwin, Charles, 1872. The Expression of the Emotions in Man and the Animals. London: John Murray Publishers.

de Oliveria Castro, 1955. Differentiated nervous fibers that constitute the electric organ of Sternarehus albifrons L. Ann. Acad. Basil. Cien., 27:557-60.

Dunham, D. W.; Kortmulder, K.; and Van Iersel, J. J. A.; 1968. Threat and appeasement in Barbus stoliczkanus (Cyprinidae). Behaviour, 30:15-26.

Eisner, T.; Silberglied, R. E.; Aneshansley, D.; Carrel, J. E.; and Howland, H. C.; 1969. Ultraviolet video-viewing: the television camera as an insect eye. Science, 166:1172-74.

Fessard, A., ed., 1974. Handbook of Sensory Physiology.Vol. III, pt. 3: Electroreceptors and Other Specialized Receptors in Lower Vertebrates. New York: Springer-Verlag.

Granath, L. P.; Erskine, F. T. III; Maccabee, B. S.; and Sachs, H. G.; 1968. Electric field measurements on a weakly electric fish. Biophysik, 4:370-72.

Granath, L. P.; Sachs, H. G.; and Erskine, F. T. III; 1967. Electrical sensitivity of a weakly electric fish. Life Sei., 6:2373-77.

Griffin, D. R., 1968. Echolocation and its relevance to communication behavior. In: Animal Communication, T. A. Sebeok, ed. Bloomington: Indiana University Press, pp. 154-64.

Grundfest, H., 1957. The mechanisms of discharge of electric organs in relation to general and comparative electrophysiology. Prog. Biophys., 7:1-85.

Grundfest, H., 1960. Electric fishes. Sei. Amer., October, pp.115-24.

Hagiwara, S., and Morita, H., 1963. Coding mechanisms of electroreceptor fibers in some electric fish. J. Neurophysiol., 26:551-67.

Hagiwara, S., Szabo, T.; and Enger, P. S.; 1965. Electroreceptor mechanisms in a high frequency weakly electric fish, Sternarehus albifrons. J. Neurophysiol., 28:784-99.

Harder, W.; Schief, A.; and Uhlemann, H.; 1964. Zur Funktion des electrischen Organs von Gnathonemus petersii (Gthr. 1862). Z. für vergl. Physiol., 48:302-31.

Harder, W., and Uhlemann, H., 1967. Zum Frequenzverhalten von Gymnarchus niloticus (CUV.) (Mormyriformes, Teleostei). Z. für vergl. Physiol., 54:85-88.

Hazlett, B. A., and Bossert, W. H., 1965. A statistical analysis of the aggressive communications system of some hermit crabs. Anim. Behav., 13:357-73.

Heiligenberg, W. F., 1973a. "Electromotor" response in the electric fish Eigenmannia (Rhamphichthyidae, Gymnotoidei). Nature, 243:301-302.

Heiligenberg, W., 1973b. Electrolocation of objects in the electric fish, Eigenmannia (Rhamphichthyidae, Gymnotoidei). J. Comp. Physiol., 87:137-64.

Heiligenberg, W. F., 1974. Electrolocation and jamming avoidance in a Hypopygus (Rhamphichthyidae, Gymnotoidei), and electric fish with pulse-type discharges. J. Comp. Physiol., 91:223-40.

Heiligenberg, W., 1975. Theoretical and experimental approaches to spatial aspects of electrolocation. J. Comp. Physiol., 103:247-72.

Hinde, R. A., 1970. Animal Behaviour: A Synthesis of Ethology and Comparative Psychology, 2d ed. New York: McGraw-Hill.

Hockett, C. F., 1960. Logical considerations in the study of animal communication. In: Animal Sounds and Communication, W. E. Lanyon and W. N. Tavolga, eds. Washington, D.C.: American Institute of Biological Sciences, pp.392-430.

Hockett, C. F., and Altmann, S. A., 1968. A note on design features. In: Animal Communication, T. A. Sebeok, ed. Bloomington: Indiana University Press, pp.61-72.

Hopkins, C. D., 1972a. Patterns of electrical communication among gymnotid fish. Ph.D. thesis, Rockefeller University.

Hopkins, C. D., 1972b. Sex differences in electric signaling in an electric fish. Science, 176:1035-37.

Hopkins, C., 1973. Lightning as background noise for communication among electric fish. Nature, 242:268-70.

Hopkins, C. D., 1974a. Electric communication in fish. Am. Sci., 62(4):426-37.

Hopkins, C. D., 1974b. Electric communication: Functions in the social behavior of Eigenmannia virescens. Behaviour, 50:270-305.

Hopkins, C. D., 1974c. Electric communication in the reproductive behavior of Sternopygus macrurus. Z. für Tierpsychol., 35:518-35.

Kalmijn, A. J., 1971. The electric sense of sharks and rays. J. Exp. Biol., 55:371-83.

Kalmijn, A. J., 1972. Bioelectric fields in sea water and the function of the ampullae of Lorenzini in elasmo-branch fishes. University of California, Scripps Inst. Oceanography, Ref. Series 72-83.

Kalmijn, A., 1974. The role of electroreceptors in the animal's life. I. The detection of electric fields from inanimate and animate sources other than electric organs. In: Handbook of Sensory Physiology, Vol. III, pt. 3, A. Fessard, ed. New York: Springer-Verlag, pp. 145-200.

Kalmijn, A. J., and Adelman, R., in prep. I. The passive electric sense in siluroid, gymnarchid, and gymnotid fishes. II. The active electric sense in weakly electric fish.

Kastoun, E., 1971. Elektrische felder als Kommunikationsmittel beim Zitterwels. Naturwissenschaften, 58:459.

Keynes, R. D., and Martins-Ferreira, H., 1953. Membrane potentials in the electroplates of the electric eel. J. Physiol., 119:315-51.

Knudsen, E. I., 1974. Behavioral thresholds to electric signals in high frequency electric fish. J. Comp. Physiol., 91:333-53.

Knudsen, E. I., 1975. Spatial aspects of the electric field generated by weakly electric fish. J. Comp. Physiol., 99:103-18.

Konishi, M., 1974. How the owl tracks its prey. Am. Sci., 61:414-24.

Liebermann, L. N., 1956a. Extremely-low-frequency electromagnetic waves, I. Reception from lightning. J. Applied Physics, 27:1473-76.

Liebermann, L. N., 1956b. Extremely-low-frequency electromagnetic waves. II. Propagation properties. J. Applied Physics, 27:1477-83.

Lissmann, H. W., 1958. On the function and evolution of electric organs in fish. J. Exp. Biol., 35:156-91.

Lissmann, H. W., 1961. Ecological studies on gymnotids. In: Bioelectrogenesis, C. Chayas, ed. New York: Elsevier.

Lissmann, H. W., 1963. Electric location by fishes. Sci. Amer., 218:50-59.

Lissmann, H. W., and Machin, K. E., 1958. The mechanism of object location in Gymnarchus niloticus and similar fish. J. Exp. Biol., 35:451-86.

Lissmann, H. W., and Mullinger, A. M., 1968. Organization of ampullary electric receptors in Gymnotidae (Pisces). Proc. Roy. Soc. B 169:345-58.

Lissmann, H. W., and Schwassmann, H. O., 1965. Activity rhythm of an electric fish, Gymnorhamphichthys hypostomus, Ellis. Z. für Vergl. Physiol., 51:153-71.

Lloyd, J. E., 1966. Studies on the flash communication system in Photinus fireflies. Misc. Publ. Mus. of Zool., Univ. Michigan, No. 130.

MacDonald, J. A., and Larimer, J. L., 1970. Phasesensitivity of Gymnotus carapo to low-amplitude electrical stimuli. Z. vergl. Physiol. 70:322-34.

Malcolm, D. S., 1972. Differences in the resting pattern in weakly electric fish, Gnathonemus petersii (abstract only). Bull. Ecol. Soc. Am., 53:(4)27.

Marler, P., 1959. Developments in the study of animal communication. In: Darwin's Biological Work, P. R. Bell, ed. London: Cambridge University Press.

Marler, P., 1961. The logical analysis of animal communication. J. Theoret. Biol., 1:295-317.

Marler, P., and Hamilton, W. J., 1966. Mechanisms of Animal Behavior. New York: John Wiley.

Möhres, F. P., 1957. Elektrische entladungen im Dienste der Revierabgreuzung bei Fischen. Naturwissenschaften, 44:431 -32.

Moller, P., 1970. "Communication" in weakly electric fish, Gnathoneumus niger (Mormyridae). I. Variation of electric organ discharge (EOD) frequency elicited by controlled electric stimuli. Anim. Behav., 18:768-86.

Moller, P., and Bauer, R., 1973. "Communication" in weakly electric fish, Gnathonemus petersii (Mormyridae) II. Interaction of electric organ discharge activities of two fish. Anim. Behav., 21:501-12.

Moynihan, M., 1955. Some aspects of reproductive behavior in the black-headed gull (Larus ridibundus ridibundus L.) and related species. Behaviour Suppl., 4:1-201.

Nelson, K., 1964. The temporal patterning of courtship behavior in the glandulocaudine fishes (Ostariophysi, Characidae). Behaviour, 24:90-146.

Peters, R. C., and Buwalda, R.J. A., 1972. J. Comp. Physiol., 79:29-38.

Peters, R. C., and Meek, J., 1973. Catfish and electric fields. Experientia, 29:299-300.

Pickens, P. E., and McFarland, W. N., 1964. Electric discharge and associated behaviour in the stargazer. Anim. Behav., 12:362-67.

Rommell, S. A., and McCleave, J. D., 1972. Oceanic electric fields: perception by American eels? Science, 176:1233-35.

Roth, A., 1972. Wozu dienen die Elektrorezeptoren der Welse? J. Comp. Physiol., 79:113-35.

Russell, C. J.; Myers, J. P.; and Bell, C. C.; 1974. The echo response in Gnathonemus petersii (Mormyridae). J. Comp. Physiol., 92:181-200.

Scheich, J., 1974. Neuronal analysis of wave form in the time domain: midbrain units in electric fish during social behavior. Science, 185:365-67.

Scheich, J., and Bullock, T. H., 1974. The detection of fields from electric organs. In: Handbook of Sensory Physiology, vol. III, pt. 3, A. Fessard, ed. New York: Springer-Verlag, pp.201-56.

Schwassmann, H. O., 1971. Circadian activity patterns in gymnotid electric fish. In: Biochronometry. Washington, D.C.: National Academy of Sciences, pp. 186-99.

Silberglied, R. E., and Taylor, O. R., 1973. Ultraviolet differences between the sulphur butterflies, Colias eurtheme and C. philodice, and a possible isolating mechanism. Nature, 241:406-408.

Smyth, W. R., 1968. Static and Dynamic Electricity. New York: McGraw-Hill.

Soderberg, E. F., 1969. ELF noise in the sea at depths from 30 to 300 meters. J. Geophys. Res., Space Physics, 74:2376-87.

Steinbach, A. B., 1970. Diurnal movements and discharge characteristics of electric gymnotid fishes of the Rio Negro, Brazil. Biol. Bull., 138:200-10.

Steven, S. S., and Newman, E. B., 1934. The localization of pure tones. Proc. Nat. Acad. Sci., 20:593-96.

Szabo, T., 1965. Sense organs of the lateral line system in some electric fish of the Gymnotidae, Mormyridae and Gymnarchidae. J. Morph., 117:229-50.

Szabo, T., Kalmijn, A. J.; Enger, P. S.; and Bullock, T. H.; 1972. Microampullary organs and a submandibular sense organ in the fresh water ray, Potamotrygon. J. Comp. Physiol., 79:15-27.

Szabo, T., and Suckling, E. E., 1964. L'ârete occasio-nel de la décharge électrique continue du Gymnarchus est-il une réaction naturelle? Naturwissenschaften, 51:92.

Tinbergen, N., 1959. Comparative studies of the behavior of gulls (Laridae); a progress report. Behaviour, 15:1-70.

Valone, J. A., Jr., 1970. Electrical emissions in Gymnotus carapo and their relation to social behavior. Behaviour, 37:1-14.

Wantanabe, A., and Takeda, K., 1963. The change of discharge frequency by AC stimulation in a weak electric fish. J. Exp. Biol., 40:57-66.

Watt, A. D., 1960. ELF electric field from thunderstorms. J. Res. N. B. S., 64D:425-33.

Waxman, S. G.; Pappas, G. D.; and Bennett, M. V. L.; 1972. Morphological correlates of functional differentiation of nodes of ranvier along single fibers in the neurogenic electric organ of the knife fish Sternarehus. J. Cell Biology, 53:210-24.

Westby, G. W. M., 1975. Further analysis of the individual discharge characteristics predicting dominance in the electric fish, Gymnotus carapo. Anim. Behav., 23:249-60.

Westby, G. W. M., and Box, H. O., 1970. Prediction of dominance in social groups of the electric fish, Gymnotus carapo. Psychon. Sci., 21:181-83.

Wilson, E. O., 1970. Chemical communication within animal species. In: Chemical Ecology, E. Sondheimer and J. B. Simeone, eds. New York: Academic Press, pp. 133-56.

Wilson, E. O., and Bossert, W. H., 1963. Chemical communication among animals. Recent Prog. Hormone Res., 19:673-716.

Wynne-Edwards, V. C., 1962. Animal Dispersion in Relation to Social Behavior. New York: Hafner.