Sexual Signals in Electric Fishes

Reprinted from Trends in Ecology and Evolution - August 1990 Vol. 5, No. 8 Sexual Signals in Electric Fishes

Electroreceptive bony fishes of Africa (the Bernd Kramer Mormyriforrnes) and South America (the Gymnotiformes) detect and communicate a pulse, and in a few gymnotiform discharge, it is only the frequency of with conspecifics by their continuously fishes, which use either a wave or a peak amplitude that the fish seem discharging electric organs. Laboratory pulse type of discharge. to need for mate recognition16. studies of members of each group are begin- These fish differ from mormyrids in ning to reveal the mechanisms of communi- lntraspecific variability of pulse-type displaying a usually constant, and cating with and finding mates, offering discharges in a mormyrid and a rather high, pulse ratef5. much scope for future studies of the be- gymnotiform havioral ecology of electric fishes. The electric organ discharge Effects of water conductivity on mormyrid (EOD) of the mormyrid Pollimyrus discharge waveform isidori is a brief pulse, with two posi- The EOD waveform of P. isidori The ability to produce and detect tive (P)phasesseparated by a nega- land, because of the general bio- weak electric currents, as a means of tive (N)phase (Fig.2a). This pulse is physical principle^'^^^^ involved, obtaining information about the en- repeated at a low and variable rate probably all mormyrids) depends vironment, is one of the outstanding of about 10-100 pulses per second on the conductivity of surrounding feats of evolution. It is found in two (p.p.s.) when the fish is excited water13 (Fig. 2b). For P. isidori, con- distantly related groups of tropical (a 'buzzer'). The waveform of this ductivity in its natural West African fishes: the elephant-nose fishes pulse-EOD shows great intraspecific habitat is about 20-1 10 KS cm-'. (Mormyriformes) of Africa, and the variabilitys, suggesting that the EOD Water conductivity is generally still knife fishes (Gymnotiformes) of waveform might be a sexual cue9. A lower in the equatorial Zai're region South America (Fig. I ). These fishes similar hypothesis had been ad- where most mormyrid species live, possess a highly sophisticated intel- vanced by Hopkins and Basstofor occasionally extending down to 5 p,S ligence system, the capabilities of an as yet unidentified mormyrid, cm-' (Ref. 3) (i.e. a conductivity which continue to fascinate neuro- Brienomyrus sp.; unfortunately, almost equivalent to that of the physiologists, neuroanatomists and there is no full report (see review deionized water often used in behavioral scientists: thev have in Ref. I I ), and the hypothesis of laboratories). also become attractive subjects for sexual recognition by pulse wave- The dependence of discharge students of evolution and ecology. form needs further experimental waveform on conductivity is weak in Weakly electric bony fishes, all of s~pport'~,~~. some individuals of P. isidori and which seem to be predominantly There is a statistically significant strong in others; therefore, there is nocturnal, test the impedance tendency for female P. isidori to no simple correlation of EOD wave- properties of their environment by show an EOD waveform different form change with conductivity that setting up an electric dipole field for from that of males (Fig. 2a); how- might enable fish to 'read' EOD active electrolocationl. The electric ever, there is considerable overlap waveform under some standard discharges are produced by dis- between the two distributions, and conductivity if the actual water con- crete electric organs (derived from no clear decision concerning the sex ductivity were known to the fish. modified muscle or, in one gym- of an individual on the basis of its Mormyrid electric organs are de- notiform family, specialized motor . EOD waveform is possible for most rived from skeletal muscle, a tissue nerve endings) and are detected by fished3. Furthermore, because two known to respond strongly to specialized lateral line receptors; females with 'wrong' EOD waveform the afferent electrosensory infor- or duration have been observed mation is processed by a brain that spawning repeatedly in the labora- is huge for a teleost (see reviews in tory, it appears that the EOD wave- Ref. 2). form information is not critical for In recent years, interest has pair formation or spawning in P, focused on how electric fish detect isidori 5. mates and on the nature of the An interesting comparison can be signals that are exchanged during made with the gymnotiform Hypo- courtship and spawning (reviewed pomus occidentalis from Panama, in Ref. 3). Progress has been made the discharge of which roughly re- both in the ~orm~ridae*,whose sembles a single-cycle sinisoidal electric discharge takes the form of' pulse. The discharges of males tend to be stronger and to last longer 'Mormyrids also communicate acoustically. A territorial than those of females. Despite con- male defending a nest produces an acoustic song4 that is siderable overlap, the difference strong in the nights preceding courtship; the song also between the distributions for the occurs for about one hour immediately after dark of a spawning night, it wanes rapidly during courtship (the next two sexes is statistically significanti4. two hours), and it is absent during the long subsequent Females give more discharge-rate spawning phase labout four more hoursI5. Mormyrids have responses ('decrement bursts'I5; keen hearingb,'; females may be attracted by these calls . see below) to being stimulated with from a distance. single-cycle sinusoids of male, Fig. 1. (a) The African weakly electric mormyrid Pol- limyrus isidori (up to 9 cm in length). Photograph cour- rather than female, discharge dur- tesy of G. von der Emde. (b)A large male of the South Bernd Kramer is at the Zoological Institute of the ationI6.Of the wide range of spectral American weakly electric knife fish Eigenmannia sp. Universit)i, D-8400 Regensburg, FRG. components of Hypopomus' pulse (males up to 40 cm, females up to 20 cm). 01990. Elsev~erSc~ence Publ~shers Ltd. (UK) 0169-5347190/$02 00 247 TREE vol. 5, no. 8,August 1990

gonadally mature subadults20. The males and females. differ is in a effect of androgen hormones on rapid reflex-like, purely electrical mormyrid electric organs is re- response to a conspecific - the pre- viewed by Bass2'. ferred latency response (PLR) of males and the preferred latency Momyrid courtship and spawning signals avoidance (PLA) of female~~,~~. encoded as inter-pulse interval patterns Males tend to respond to another Recently, the electric discharge fish's EOD by discharging at a la- patterns that accompany the noctur- tency of 12 ms, while females avoid nal courtship and spawning of P. this latency range by delaying their isidori have been recorded in the response EOD for at least 20 ms. The laboratory. Specific EOD interval behavioral significance of this re- patterns were firmly correlated with sponse, also found in other mor- diverse motor acts, suggesting that myrids (mostly in the form of PLR), these EOD patterns are important is unclear. Fig. 2. la) Electric organ discharges, of the pulse type. for mutual acceptance of mates to In summary, while the waveform of a male and a female Pollimyrus isidori, digitized at proceed on to the most critical of the EOD pulse of P. isidori does 2 MHz. The mohead-positive (PI phases are separated stages of reproductive behavior5. not seem to carry critical infor- by the negative (NI phase. In males, the amplitude of the PI phase tends to be lower than that of the P2 phase For example, during courtship and mation despite its great intra- 1i.e. PIP2 < I);in females, the reverse tends to be true spawning bouts the male and specific variability, the inter-EOD (PllP2> I).Ib) A male and a female organ discharge female lock into a regular interval interval patterns clearly do. This is of Pollimyrus isidori, recorded through a range of water pattern of low and constant rate, shown by the stereotyped corre- conductivities: 1.2 or 3.3, 5, 10, 20, 50, 100, 200, 500 pS lation of a wealth of specific patterns cm'.[Conductivity, measured in siemens IS1 percm, is punctuated by a few pauses during the equivalent of the inverse resistivity (Rcm)-'.l Note the moments of vent-to-vent coup- with specific reproductive5and non- that with increasing water conductivity the first P phase ling. Such a pattern is exceedingly reprod~ctive~,~~overt behaviors, decreases in amplitude while the second P phase in- unusual in an interacting pair of P. and the male's predictable switch- creases, and the duration of the N wave becomes isidori, except during courting or ing of pattern whenever the female shorter. From Ref. I?. spawning. Both partners take on enters or leaves his nesting terri- completely different roles, associ- tory. A similar conclusion was de- androgen hormones and their ana- ated with specific interval patterns: rived from a playback study in the bolic effect. This effect is probably the male is territorial and builds a context of agonistic behavior in the the reason for the statistical trend of nest, guarding the eggs and young mormyrid G. peter~ii~~. P. isidori males to have stronger fry; the female is allowed access to The pulse-discharging gymnoti- EODs and a lower P-ratioi3(Fig. 2a). the nest area only during the brief form H. occidentalis also displays Mature males of another mormyrid, courtship and spawning bouts ( 15- discharge-rate modulations during Gnathonemuspetersii, tend to have 25 S each; repeated for hours on a courtship and spawning.The 'decre- a longer EOD than femalesI9, al- spawning night). ment burst' is a very brief acceler- though this trend cannot be seen in Another way in which P. isidori ation of the fish's otherwise steady discharge rateI5; a more detailed description of the relationship with overt behavior is, however, lacking.

lntraspecific variability of a wave-type discharge in a gymnotiform lntraspecific variability in discharge waveform clearly plays a role in the South American gymnotiform Eigenmannia (Sternopygidae), whose EOD is a wave of usually constant frequency (a 'hummer'). Adult females and juveniles have an EOD waveform that resembles a sinusoid much more closely than does that of adult males; the latter produce nar- row head-positive pulses super- Time [msl Frequency I k~rj imposed on a negative baselinei2 (Fig. 3). Accordingly, the intensity of Fig. 3. la, c) Electric organ discharges, of the wave type, of a female and a male Eigenmannia sp. (species range, 250600 Hzl. Note the marked difference in the time pattern of zero-crossing intervals: in the higher harmonics, or overtones, is female, head-positive and negative half-wavesare of almost equal duration, while in the male the head- high in adult males and lower in negative half-wave is much longer than the positive. other individuals. Gottschalk de- (b,dI Fourieramplitude spectra; the fundamental frequency, of 400 Hz in these examples, is expressed scribed a similar waveform variation as 0 dB attenuation. IA decibel is a logarithmic measure of intensity relative to some arbitrary reference in the related Stemopygus (reviewed intensity of 0 dB; compared to the reference intensity, an attenuation of -20 dB is an intensity drop to one tenth, of -40 dB to one hundredth, and so forth.) Note the lower intensity of overtones lof frequency in Ref. 121. 400n, where n is 2,3,4 ...l in the female discharge as compared to the male discharge. From Ref. 12. Food-rewarded Eigenmannia TREE vol. 5, no. 8, August 7990

have been found to discriminate A female will only spawn when a between electronically synthesized male has chirped at her site for at male and female discharge^^^. least one hour. Playback of tape re- Moreover, in a simultaneous choice cordings of a courting male through experiment without reward or a dipole also evoked spawning by a punishment, untrained juvenile and gravid female in the vicinity of the adult fish of both sexes preferred to electrodes26. move to and stay close to an electric In Sternopygus macrurus, males dipole emitting female EODs of and females tend to discharge at a constant frequency (and ampli- frequency difference of an octave27. tude), rather than a dipole emit- Different frequency ranges have ting male EODS~~.This experiment also been observed in the two sexes confirmed that fish can discriminate of Eigenmannia26e'2and of Aptero- male from female EOD waveforms, notus leptorhynchus, a gymnoti- with neither intensity nor frequency form fish that also chirps during Fig. 4. Oscillograms of discharge interruptions, or being factors; the experiment also spawning26.It is, however, not known 'chirps', of a male Eigenmannia virescens. lal A sudden, showed that the difference matters whether a certain minimum fre- almost complete interruption, as observed during agonistic behavior. fbl An incomplete interruption, as to the fish even in a nearly bare quency difference is necessary for observed during courting. Icl A long and complete aquarium, outside a reproductive spawning to occur in a certain pair of interruption following an initially incomplete interrup- context. fish; in Eigenmannia spawning in tion. Adapted from Ref. 26, with permission. Unlike the pulse discharge of the laboratory this does not seem to P. isidori, Eigenmannia's discharge be the case28. waveform is affected very little in both directions, but deviate from by conductivity change (within the Sexual dimorphism of Eigenmannia's the 'model' Eigenmannia either by natural range of water conductiv- jamming avoidance response shifting their discharge frequency in ities, 10-l00 p,S cm-'); the reason A special kind of frequency the 'wrong' direction in response to for the difference is the entirely dif- modulation is the so-called jam- a stimulus frequency slightly higher ferent physiology of their electric ming avoidance response (JAR29,301. than their own, or by a markedly organ^'^,^^ (reviewed in Ref. 25). To avoid 'jamming' of its signal by a asymmetric response strength to Also, temperature change (within conspecific, a 'model' Eigenmannia equal difference frequencies of op- the range 20-33°C) has very little ought to lower its own frequency in posite sign. These two groups prob- effect on Eigenmannia's discharge response to stimuli of slightly higher ably represent the two sexes3' (see waveform, apart from on its fre- frequency, and increase its own Fig. 51. q~ency~~. frequency in response to stimuli Although the explanation of the slightly lower, so that the difference JAR as adaptive for active electro- Signal modulations of 'hummers' during between the two frequencies (the location is plausible and attractive, reproductive behavior beat frequency) increases. How- the evidence is not particularly In Eigenmannia, frequency modu- ever, the reality is more complex strong (seeRef. 3 1 I. An alternative - lations of its usually constant- (Fig. 513'. perhaps not mutually exclusive - frequency wave discharge, and brief The JAR is usually thought to help explanation3' would argue that the interruptions, have been observed maintain a fish's performance of ac- JAR is a frequency modulation that in interacting mature fish and during tive electrolocation in the presence assures a private frequency 'win- spawning in the lab~ratory~~.On of disturbing noise (of similar fre- dow' for all members of a school becoming territorial in a large tank, quency) from a conspecific's dis- (except large males, which do not a large male begins to 'chirp' rou- charge (see Ref. 321, but at the same show the response), so that a fish tinely at night, at a rate between one time some authors suggest a social can better detect and analyse the per minute and five per second. A or communicative role33. sex- and age-correlated EOD wave- 'chirp' is a brief discharge interrup- The recently discovered sexual forms (Fig. 31 and frequency modu- tion (Fig. 4) that is displayed both in dimorphism of JAR strengthens that lations (Figs 4, 51 of other fish, and an agonistic context towards other view3' (Fig. 5). Adult males do not the spatial structure of the group. males27,and during courtship26.In respond to jamming signals by dis- Under the new hypothesis, a fish response to mild attacks of the playing a JAR, but instead pro- would not be trying to minimize, but much larger male, a female will re- duce short interruptions ('chirps') rather to maximize, the effect of peatedly and smoothly raise her or 'short rises'. Adult females, another fish's discharge on its own, discharge frequency by a few to sev- discharging usually at the high- by shifting its own frequency by a eral hertz over a period of tens of frequency end of the species' range, few hertz to an optimal beat fre- seconds ('long rise'). Long rises are only lower their frequency, and do quency. considered a submissive signal in not respond to stimulus frequencies an agonistic context27,but seem to that should evoke a frequency in- Ecology of reproductive cycles play an additional role in courtship crease. Females may, however, in- Kirschbaum (see Ref. 34) found in stimulating the male to increase crease their frequency with a time that by imitating some ecological his rate of chirping. Also observed course resembling a JAR in an ag- changes that precede the reproduc- during the night of spawning are onistic context (submissive signal2'), tion of many Amazonian fishes, 'short rises', displayed by both or during courtship in the presence some weakly electric fishes (such as sexes. of males. Juveniles,finally, respond Eigenmannia virescens and also the 249 TREE vol. 5, no. 8, August 7990

relation, conductivity changes - I.Comp. Physiol. 155, 753-76 1 together with a rise in water level - 8 Liicker, H. and Kramer, B. 11981) Behav. Ecol. Sociobiol. 9, 103-1 09 would work well as stimuli for repro- 9 Westby. G.W.M. and Kirschbaum, F. ( 1982) duction in some, but not all, fish I.Comp. Physiol. 145, 399-403 communities. In some cases there is 10 Hopkins, C.D. and Bass, A.H. 11981 1 no clear pattern of rainy and dry Science 2 12,8547 seasons35;in others, there is such a l l Hopkins, C.D. 11988) in Sensory Biology of Aquatic Animals (Atema, l., Fay, R.R., pattern but no appreciable change Popper, A.N. and Tavolga, W.N., edsl, of water conductivity can be re- pp. 233-268. Springer-Verlag corded throughout the ~ear~~s'~.I2 Kramer, B. 11985) 1. Exp. Biol. 1 19, 41-69 Schwassmann (reviewed in Refs 13 Bratton. B.O. and Kramer, B. (1988) Exp. Biol. 47, 227-238 27,151 has suggested that the dry- 14 Hagedorn. M. and Carr, C.E. (19851 season spawners of the subtropical I.Comp. Physiol. 156, 51 1-523 zones (more than 9-10" N or S) may 15 Hagedorn, M. ( 19881 Copeia 1988, respond to diurnal cues for the con- 324-335 trol of their reproductive cycles, 16 Shumway, C.A. and Zelick, R.D. (1988) I. Comp. Physiol. 163, 465-478 while the wet-season spawners of 17 Bennett, M.V.L. ( 1971 ) in Fish Physiology the more equatorial regions re- (Vol. V1 (Hoar, W.S. and Randall, D.]., eds), spond to physical or chemical cues pp. 347-49 l, Academic Press of the water. 18 Bell, C.C., Bradbury, 1. and Russell, C.I. 119761 1. Cornp. Physiol. 1 10, 65-88 Many tropical fish species per- 19 Landsman, R.E. and Moller, P. ( 1988) haps respond opportunistically to Experientia 44, 900-903 the availability of sufficient high- 20 Kramer. B. and Westby, G.W.M. ( 19851 quality food, and suitable spawning Experientia 41, 1530-1531 grounds, by growth of their ova. 21 Bass, A.H. ( 19861 in Electroreception (Bullock. T.H. and Heiligenberg, W., edsl, A Frequency difference, AF (Hz) more parsimonious explanation pp. 13-70, Wiley such as this does not conflict with 22 Kramer. B. 11978) Behav. Ecol. Sociobiol. Fig. 5. The frequency change (ARI displayed by four most of the available data on the 4,61-74 Eigenmannia individuals in response to stimuli of simi- 23 Kramer, B. 119791 Behav. Ecol. Sociobiol. lar frequency. AR is measured as the fish's response environmental control of reproduc- 6, 67-79 frequency minus its resting frequency; it is plotted here tion in weakly electric fishes (and 24 Kramer, B. and Zupanc, G.K.H. ( 19861 as a function of the frequency difference IAF), i.e. the perhaps other fishes as well). The Naturwissenschaften 73, 679-680 fish's resting frequency minus the constant stimulus occasional spontaneous spawnings 25 Kramer, B. and Otto. B. ( 19881 Behav. frequency. Each point is a mean of 12 measurements, in aquaria at high conductivities Ecol. Sociobiol. 23, 5540 2 I SE. 26 Hagedorn, M. and Heiligenberg, W. Large males (a) barely respond at all, even at in- (that were probably rather rising 119851 Anim. Behav. 33, 254-265 creased stimulus intensities l+10 and +20 dB; not than falling), like the first such ob- 27 Hagedorn, M. 11986) in Electroreception shown). Adult females (b)readily lower their frequency servation by Birkholz (see Ref. 34) IBullock. T.H. and Heiligenberg, W., eds), to negative AFs, but show no or only weak responses to in P. isidori, may also lend some pp. 497-525, Wiley positive AFs. juveniles or subadults respond in both 28 Westby. G.W.M. and Kirschbaum, F. directions; however, those that are presumably females weight to this possibility. In any 1198 11 in Sensory Physiology of Aquatic (cl only give weak frequency increases to positive AFs, case, the laboratory studies de- Lower Vertebrates ( Adv. Physiol. Sci., Vol. while the other type of juvenile Id), which is probably scribed here are now generating 31 1 (Szabo, T. and Czeh. G., eds), pp. 179- the males, gives strong responses to both positive and many interesting questions about 194. Pergamon PresslAkademiai Kiado negative AFs, including AF=O Hz. This type of iuvenile the behavioral ecology of weakly 29 Watanabe, A. and Takeda, K. ( 19631 even increases its frequency when the stimulus fre- I.Exp. Biol. 40, 57-66 quency is slightly higher than its own (negative AFs of electric fishes. 30 Bullock, T.H., Hamstra, R.H. and Scheich, small absolute value up to about I Hz). For further H. ( 1972) 1. Comp. Physiol. 77, 1-48 experimental details, see Ref 31. 31 Kramer, B. ( 19871 1. Exp. Biol. 130, 39-62 References 32 Heiligenberg, W. ( 1986) in I Lissmann, H.W. 119581 1. Exp. Biol. 35, Electroreception (Bullock, T.H. and African P. isidori) can be induced 156-191 Heiligenberg, W., edsl, pp. 613-649, Wiley to spawn in the laboratory. Chief 2 Bullock, T.H. and Heiligenberg, W., eds 33 Scheich.H. 11977) 1. Comp. Physiol. 1 13. among these factors was a slow 1 19861 Electroreception. Wiley 181-255 lowering of water conductivity, as 3 Kramer, B. Electrocornmunication in 34 Kirschbaum, F. ( 1984) Environ. Biol. naturally occurs (following an initial Teleost Fishes: Behavior and Experiments. Fishes 10. 3- 14 Springer-Verlag (in press) 35 Sioli. H.. ed. ( 1984) The Amazon: increase) in the Amazon (Manaus)at 4 Crawford, I.D., Hagedorn, M. and Hopkins. Limnology and Landscape Ecology of a the start of the rainy season. At this C.D. ( 1986) 1. Cornp. Physiol. 159, 297-3 10 Mighty l'ropical River and its Basin, junk time, large nutrient-rich inundation 5 Bratton, B.O. and Kramer, B. ( 19891 Behav Publishers areas become available as spawn- Ecol. Sociobiol. 24. 349-368 36 Westby, G.W.M. 11988) Behav. Ecol. 6 Kramer. B.. Tautz. I. and Markl, H. 11981 1 Sociobiol. 22, 34 1-354 ing grounds for many fish species35. I.Como. Phvsiol. 1 43, 435-44 1 Because of their temporal cor- 7 ~cCbrmi;k. C.A. and Popper. A.N. 119841