CHAPTER 13

FISHES

WILLIAM N. TAVOLGA

It is well known that fishes interact with each other in a variety of ways, especially during many phases of reproductive behavior, schooling, and even in predator-prey relations. Some of these interactions can be called “communications,” whereas the status of others is doubtful. Although fishes comprise a highly diverse group, they are considered lower vertebrates, and their behavioral capacities certainly place them on a different level of behavioral organization from the majority of tetrapods. The relative simplicity and stereotypy of fish behavior, as compared with the behavior of birds and mammals, makes the problem of analysis of cause and effect somewhat easier. Furthermore, the study of the factors involved in interactions among fishes provides an insight into the evolution and development of comparable situations in the higher forms. For these reasons, a study of the kinds of communications among fishes is of interest.

Another reason for interest is that fishes are aquatic, thus the medium in which communication takes place is different in its physical characteristics from that used by terrestrial animals. Adaptation and specialization for communication in different media would be expected to be qualitatively different.

Primitive in some aspects and advanced in others, interactions among fishes occur at various levels of complexity and organization. An understanding of these interactions must depend on an appréciation of the level involved in each case, and a theoretical framework for the study of animal interaction and communication is needed. It is appropriate, therefore, to explore the general concept of communication and its applicability to interactions among fishes.

COMMUNICATION DEFINED—STATEMENT OF THE PROBLEM

I am sure that many, if not all, of the contributors to this volume will be dealing with the problem of defining communication. I should like to add some ideas. It is particularly appropriate to do so in connection with fishes because of their nexus-like position morphologically and psychologically in the phylogeny of vertebrates.

Most animals influence each others behavior by visual, acoustic, chemical, and other means, and some of these interactions have been categorized as communication. Frings and Frings (1964) stated that “Communication between animals involves the giving off by one individual of some chemical or physical signal, that, on being received by another, influences its behavior.” This is obviously a broad approach, and the concept of communication is extended thereby to include virtually all phases of animal behavior. For practical purposes Frings and Frings introduced some limitations on this definition, such as the necessity for specialized structures involved in the production and reception of the signal and the primary consideration of intraspecies interactions.

Various attempts have been made to approach the study of animal communication by looking first at human communication and language and then seeking out factors common to both human and other animal groups (e.g., Marler, 1961; Sebeok, 1965). Many such approaches have made cross-phyletic comparisons with little regard for the differences in phyletic position and level of organization occupied by the organisms being compared. The social organization of ants and bees is obviously different from that of human societies and is based on qualitatively different endocrine, stimulative, neural, genetic, developmental, and other factors. It is, therefore, clearly inappropriate to apply the complex of interacting factors that control the behavior patterns of social insects to a theoretical analysis of human society. It is equally erroneous to use the methods and theory developed for the study of human language in the investigation of a kind of communication found in another species at a different organizational level.

The concept of levels is basic to the biological sciences, and Schneirla (1953) has postulated a hierarchical arrangement of levels of behavioral integration. These levels represent a series of progrèssive advances from unicellular to multicellular, radially symmetrical to bilaterally symmetrical, diffuse to centralized nervous organization, biosocial to psychosocial. Any one level of integration in behavior, although similar in some respects to other levels, differs sufficiently to warrant its separate consideration. These levels are qualitatively different and, as such, require distinctive instrumentation, experimental operations, and theoretical approaches.

The dilemma of defining communication as it exists among animals, including man, can be resolved by applying the concept of levels. Communication does not exist as a single phenomenon; there are, instead, different kinds of interactions characteristic of different levels of organization and only some of these levels involve what can be called communication in a strict sense.

LEVELS OF INTERACTION IN ANIMAL COMMUNICATION

The following is a tentative schema for the classification of levels of interaction and communication according to phyletic position and behavioral organization. The levels are not always clearly separable, but they are nevertheless discrete. More important than their specific definitions is the concept of levels of interaction. This is presented in more detail elsewhere (Tavolga, in press).

Vegetative Level

An organism can affect the behavior of other organisms by its physical presence alone. The growth pattern, tropisms, shape, and mass of an organism can serve as a stimulus. This level of stimulus emitter is characteristic of plants, and interactions among plants are on this level. This would include the simple release of pollen or gametes. Possibly some protozoans and primitive metazoans, such as sponges, interact on this level.

Tonic Level

Continuous or, at times, irregularly episodic processes that are basic to species-typical development and function, e.g., homeostasis, are defined by Schneirla (1965) as tonic. Such processes exist at all levels of integration and contribute to all stages of the development of an organism. When tonic processes result in external physiological and behavioral manifestations, they can affect the behavior of other organisms and, therefore, they represent stimuli in the interactions among organisms. Interactions in which tonic processes predominate are considered to be on the tonic level.

Included in the stimuli resulting from these processes are (1) chemical exudates produced as a by-product of normal metabolism and (2) basic locomotor movements and color patterns. Symbiotic relationships, including mutualism, commensalism, and parasitism, depend on interactions at this level. The stimuli provided by many prey organisms and at least some types of trophallactic and trail-following behavior involve tonic interactions.

Phasic Level

In the sense used by Schneirla (1959, 1965), phasic processes are those involved in discontinuous, more or less regular stages or events in the development of the organization of an animal. This is in contrast to the more continuous, basic tonic processes. Interactions in which phasic processes predominate can be allocated to the phasic level. Interactions at this level usually involve broad and multichannel stimuli with some specialization on the part of both the emitter and receiver. The receiver, therefore, can respond in a discriminatory fashion, i.e., in species or sex discrimination.

The well-known hive dances of returning forager bees, stimulated by food, emit stimuli in a wide range of visual, olfactory, gustatory, acoustic, and tactual channels. Sex discrimination in many teleost fishes is based upon phasic interactions. Feeding responses of many predatory insects, fishes and anuran amphibians depend on a simple discrimination of food vs. nonfood.

Signal Level

Emitters on this level are characterized by possessing specialized structures that produce energy outputs (stimuli) along a single channel and usually within a narrow band of the spectrum of that channel, e.g., narrow frequency range, specific duration, or temporal patterning. Such stimuli that function in interactions are termed signals. The production of specific chemical attractants, i.e., pheromones (Wilson, 1965), by some insects represents a channeling of the output into signals. In mechanical channels, specialized sonic organs, as in cicadas and many orthopterans (Alexander, 1961; Dumortier, 1963), emit specific signals. Color patterns in many animals, especially among fishes and birds, are also signals in the sense of occupying narrow portions of the visual channel, e.g., shape, color, and movement (flicker).

Reception at the signal level is characterized by specialized receptor mechanisms and specific, often stereotyped, responses. Echolocation (see general review by Vincent, 1963) is a special case in which the emitter and receiver are the same animal. Elaborate emitting organs and highly developed responses are associated with echolocation in bats and the toothed whales. The mechanism of electric location and prey detection by certain fishes (Lissmann, 1963) is similarly a signal level interaction.

The development of social organization in ants and bees is based upon exchange of specific, narrow-channel stimuli. The development of this type of social bond is dominated by organic processes and has been termed biosocial (Schneiria and Rosenblatt, 1961). By contrast, the development of socialization in mammals involves additional, qualitatively different forms of reciprocal stimulation, and this level has been called psychosocial. Although superficially similar, the levels of interaction in these two groups are clearly different. The development of song in the higher passerine birds is, to some extent, psychosocial in that there are stages at which acoustic patterns from other birds can be absorbed into the repertoire (Lanyon, 1961).

Symbolic Level

The numerous studies on social behavior in infrahuman primates, especially those of Yerkes and Yerkes (1935), Köhler (1925), and, more recently Washburn et al. (1965), have shown that interactions among these animals are qualitatively different from the psychosocial level of interaction, and primitive forms of symbolism can be observed in their gestures, facial expressions, and even vocalizations.

Language Level

The term “language” has been used for several levels of communication, usually including “speech.” G. G. Simpson (1966) recently reemphasized the error of using this term for interactions in such forms as bees. In agreement with Simpson, I prefer to restrict the term to that form of unique, primarily vocal communication characteristic of man. Here symbols are used in a teleological sense, abstract ideas are communicated, and not only present conditions but also past and future events are communicated and symbolized. The development of human communication is also unique and closely associated with social context (Carmichael, 1964).

COMMUNICATION DEFINED IN TERMS OF LEVELS

Returning to the original problem of the definition of communication, this term could include any one or more levels in the above hierarchy of interactions. Restriction of the scope of the term now becomes practical and meaningful. I suggest that the signal, symbolic, and language levels comprise communication, whereas the vegetative, tonic, and phasic levels include primitive forms of interactions. This separation can be made from the point of view of the emitter, the channel, or the receiver. For communication to take place, the emitter must possess a specialized stimulus-producing mechanism (chemical, morphological, or behavioral). The stimulus must occupy a narrow portion of the available spectrum of the channel (frequency range, duration, patterning, chemical specificity). The receiver must possess specialized receptors and respond in a specific manner.

The emitter and the receiver need not necessarily operate on the same level. Thus, forager bees, returning to the hive, provide stimuli on a broad-channel basis (on the phasic level), whereas the receiver bees respond in a specific, highly oriented manner (on the signal level). Predatory animals usually respond to narrow-channel stimuli (signals), whereas the prey organism is an emitter on the phasic or tonic level. A given animal can also interact at different levels depending upon its developmental stage and the immediate circumstances.

INTERACTIONS IN AN AQUATIC MEDIUM

The channels available for animal interactions can be listed as photic (visual), mechanical (including tactual and acoustic), thermal, chemical (gustatory and olfactory), and electrical (limited to certain fish). Information transmitted along any of these channels can be broadly classified as long vs. short range and directional vs. nondirectional. Each of these channels has certain limitations in an aquatic medium as compared with a terrestrial environment.

The photic channel is severely limited in water (Dietrich, 1963), especially in sea water in regions of high planktonic concentration. In pure water, extinction of a light beam increases logarithmically with increased wavelength from its minimum in the blue-green range (0.40 to 0.75 /x). In addition, suspended particles produce extinction by scattering. Ocean water in which visibility exceeds 15 meters is unusual indeed, and at depths below 500 meters there is virtually no illumination from sunlight. The probability is that the range for effective vision in the marine environment is generally less than one meter, and in areas of high turbidity this effective range may be reduced to only a few centimeters.

The chemical channel is potentially an effective one in water because of the large range of substances that are easily suspended or dissolved in water. The studies of Hasler and Wisby (1951) on discrimination of stream odors in fishes, and of Kleerekoper and Mogensen (1959, 1963) on the detection of trout odors (specific amines) by lampreys, have shown that the olfactory sense in fishes and fish-like vertebrates is extremely sensitive and capable of being a long-range interaction channel. This channel, however, is slow and nondirectional, and the source of the stimulus can be located only by means of some kinesis like movements in which the animal simply moves about until it finds areas of progressively higher stimulus concentrations.

Although useful in a terrestrial environment, the thermal channel is virtually unavailable to aquatic animals, especially to ectothermic forms such as fishes. Water absorbs heat rapidly, and a thermal gradient attenuates much too fast for any effective reception as a stimulus in an interaction. Both ac and dc electric fields are readily set up in water, especially in salt water, and many species of fish, in addition to the well-known electrical forms, are now known to be able to detect electric potentials produced by other organisms (Dijkgraaf, in press). The sensitivity of the ampullae of Lorenzini in sharks has been estimated at the level of only a few microvolts by Dijkgraaf. Although Lissmann (1963) has demonstrated the utilization of electrical detection of obstacles and prey by some electric fishes, the function of these recently discovered electrical receptors in other species has not yet been fully investigated. It is likely that the effective range of such receptors does not exceed a few centimeters, but the distribution of the receptors on the body makes it possible for the receiving animal to locate the source of the stimulus accurately. How widely this energy channel is used in interaction among fishes still remains to be studied.

Aside from the short-range, direct-contact function of tactile receptors, the mechanical channel offers several advantages for interactions and communication among fishes. Acoustic energy under water is effective in two forms: as pressure and as displacement. Both the inner ear and the lateral line are essentially displacement-sensitive, but the inner ear receives near-field displacements from the nearby swim bladder. The swim bladder acts as a transducer for pressure waves and transforms them into local near-field effects (Harris and van Bergeijk, 1962; van Bergeijk, 1964). Pure pressure waves are efficiently propagated in water, and this form of acoustic energy is probably the most rapid and effective channel for long-range interactions. Pressure waves are essentially nondirectional, except at close range and high intensity, and there is, as yet, no evidence that fish can directionalize in a farfield. In effect, fishes have a single ear, since the two inner ears are not only close together but coupled to a single swim bladder. Moulton (1967) presented some data that could indicate far-field orientation in the goldfish, but his experiments involve high-intensity sounds of such levels that the pressure wave alone could produce displacements over the threshold of the lateral-line organs. Van Bergeijk (1964) postulated that directional orientation to a sound source by fishes could take place only in the near-field, at distances under ʎ/2π, and that such orientation is possible by means of lateral-line reception. The majority of sounds produced by swim bladder mechanisms in fishes range from 50 to 100 Hz, and this would provide an effective near-field range of from about 50 to 3 meters, respectively. Marshall (1967) concluded, on the basis of anatomical studies, that benthopelagic fishes may be more dependent on acoustic channels than either the higher-level mesopelagic or the bottom-dwelling benthonic species. The acoustic channel, therefore, offers the most effective range and optimum information content for interactions among aquatic animals, especially fishes.

LEVELS OF INTERACTIONS AMONG FISHES

There are four principal behavioral situations in which interactions take place among fishes: feeding, predator-prey relations, reproductive behavior, and schooling.

Feeding

When one animal is feeding, it is clear that others are attracted to the site. Some of the cues are olfactory, e.g., chemicals released from the food material itself; some are visual, e.g., the sight of feeding animals; and some are acoustic, e.g., sounds of gnashing teeth. From the point of view of the emitter, these stimuli would be on the phasic level of interaction. The energy output is nonspecific and multichannel and covers a broad spectrum of the channel; thus, according to the definition proposed here, these interactions cannot be considered as communication, but are on the phasic level of interaction.

The response of the receiver animal in this situation is also nonspecific and may result in an approach to the feeding site. Such responses are unlikely to be on the signal level, and they probably represent phasic interactions.

Predator-Prey Relations

Stimuli that are emitted by a prey organism can be chemical, photic, or mechanical. Such stimuli are usually generated by primarily tonic processes, such as normal locomotion, excretions, and even just body shapes on the vegetative level. The broader and less specific the energy outputs, the greater is the selective advantage for the survival of the prey organism. Coloration and body shape are often specialized for concealment, and this can be achieved either by cryptic or disruptive coloration or sometimes by mimicry of an inanimate object or another species. Breder (1946) reviewed the remarkably large number of instances in which fishes resemble plants or plant parts, and the various forms of camouflage among shore and reef fishes are well known (Breder, 1946, 1949). The energy outputs are basically broad-channel and primitive, but in some instances, as in mimicry, the signal level of interaction occurs: a species provides visual cues that are characteristic of another species.

The predator, on the other hand, can operate on the signal level. Specific characteristics of the prey organism are detected and reacted to. Many predatory fishes, however, are nonspecific in their dietary requirements, at least in captivity, and will attack any object of about the right size and mobility. It is possible that the visual system of fishes operates in a manner like that of frogs, in which a simple discrimination is made between food vs. nonfood (Lettvin, et al., 1961). It is this nonspecific attack response among some fishes that enables fishermen to use such a large variety of artificial lures successfully, although the majority of species are more rigid in their feeding responses.

The utilization of the visual channel by predatory fishes can become quite complex. Observations made at Marine Studios, St. Augustine, Florida, by F. G. Wood, Jr. (personal communication) showed that predatory fishes such as barracuda (Sphyraena) and jacks (Caranx), in a huge community tank, will attack newly introduced animals primarily. Some minor, yet discernible, differences in behavior of the prey organism are apparently detected by the predator.

Chemical factors may play a role in inducing attack behavior. A “frightened” (one under stress) fish produces substances in its body mucus that stimulate “fright” behavior in other animals of the same species (von Frisch, 1942; Verheijen, 1956), and it is quite possible that a predator can also detect the presence of this Schreckstoff.

In addition, predators also use the acoustic channel. Sharks have been shown to be attracted to the source of low-frequency sound, such as would be produced by an injured or abnormally swimming animal (Nelson and Gruber, 1963).

Schooling

Aggregations and schools of fishes constitute a primitive form of social behavior and are based essentially on the maintenance of more or less constant distances and orientations among the members of a group of animals of the same species and size. Shaw (1960, 1961) has shown that the development of schooling begins at an early stage in many species and may start as a nonspecific optomotor reaction. The sensory cues used by schooling fishes that have been studied are primarily visual, but hydrodynamic, acoustic, and chemical cues may also play a role (Breder, 1959, 1965). The cues are generally broad and multichannel, yet the final result gives the appearance of a closely ordered, precisely spaced, and oriented, organized moving mass of animals. The interactions in schooling are primarily on the tonic or primitive phasic level, i.e., involving normal locomotor patterns and basic body shapes, and the visual and lateral-line systems appear to be prominent as controlling sensory mechanisms.

Reproductive Behavior

As considered here, reproductive behavior in a broad sense includes not only prespawning and spawning, but territorial defense, parental care, and various forms of aggressive behavior, i.e., any forms of behavior that are related to the process of sexual reproduction (see review by Aronson, 1957). In their monograph on reproduction in fishes, Breder and Rosen (1966) pointed out that of the some 20,000 known species of fishes, only about 300 have been well described as to breeding habits. Only a small number of these have been investigated experimentally. The exact mechanisms and sensory cues involved in reproductive behavior are actually known in a very few species. Purely descriptive reports, although useful as a starting point, are inadequate for an analysis of interactions and communication.

It is clear that chemical factors function in the attraction of the sexes in prespawning behavior in many species. Tavolga (1956b) showed that ovarian secretions from gravid gobies (Bathygobius soporator) stimulated courtship behavior in males in the absence of any visual stimulus. Amouriq (1964) found that males of the guppy (Poecilia reticulata) were stimulated by water in which females had been kept. It is evident that these chemicals are species-specific, although no chemical analyses have been performed. Certainly the specificity of the stimulus substance and the responses of the receiver animals places this phenomenon on a level of organization distinctly above that of the general effects of excretory and food substances. The well-known, sometimes specific “conditioning” of aquarium water (Allee et al., 1946) is on the tonic level, from the viewpoint of both emitter and receiver organism, since the effects are primarily physiological rather than behavioral. The fright substance (Schreckstoff) liberated by some fishes (von Frisch, 1942; Verheijen, 1956) would be on the phasic level of interaction. It is likely that the substances involved in reproductive behavior will, upon analysis, turn out to be more specialized, and certainly the behavioral responses are on a higher integrative level. Such interactions, therefore, would be ineluded on the signal level, and, according to the definition proposed here, they would constitute forms of communication. Whether all these chemicals can be called “pheromones” is dependent upon one’s definition, and it seems undesirable to extend the term to include all cases of chemical interactions on the same grounds as it is not useful to extend the term “communication” to all levels of interaction.

Visual factors in reproductive behavior certainly operate on the tonic and phasic levels of interaction. Unfortunately, only a few species have been experimentally investigated, but those that have been show indications of operating on the signal level. The case most frequently cited is that of the stickleback (ter Pelkwijk and Tinbergen, 1937), in which the interaction of specific visual stimuli plays an essential role in the sequence of events leading toward spawning. Several other instances (cited by Breder and Rosen, 1966) also apparently utilize signals in many stages of reproductive behavior. A large number of descriptions of aggressive and territorial defense behavior can be explained on the basis of simple, nonspecific, broad-channel cues which could be tonic or phasic in interaction level. On the basis of the law of parsimony, it would be desirable to accept explanations of interacting behavior at lower levels of integration unless experimental evidence demonstrates the existence of specialized emitting structures, narrow-channel stimuli, and specific responses.

In the acoustic channel the possibilities for signal level interactions are considerable, for the reasons outlined earlier. Numerous specializations are found among fishes for sound production, especially those utilizing the swim bladder (Schneider, 1961, 1967; Tavolga, 1964, 1965). Winn (1964) proposed that, in addition to the parameters of frequency and duration, temporal patterning of sound pulses could function as specific signals in sex and species discrimination. The auditory capacities of fishes are well adapted for the reception of most fish sounds, since their sensitivities are highest in the low-frequency range below 500 Hz (Tavolga and Wodinsky, 1963). In spite of these data, experimental information on the specificity of acoustic interactions is sparse indeed. Tavolga (1958) showed that in the goby (Bathygobius soporator), both males and females would respond to playback of courtship sounds of males but would also respond to crude imitations of the sounds as long as the frequency range and repetition rate were in the same order of magnitude. Winn (1967) found that toadfish would respond vocally to playbacks of their boat-whistle calls and that the repetition rate of the playback was important in enhancing the response. Playback experiments are essential in determining the specificity of the responses of fishes to acoustic stimuli to find if these interactions involve signals in the strict sense of the term as defined here. Until such experimental evidence is available, we must assume that, with the possible exception of the toadfish, these acoustic interactions among fishes are primarily on the phasic level.

EVOLUTION OF COMMUNICATION IN FISHES

As in any aspect of behavior, it is difficult to make any positive statements on evolution, since no paleontological evidence is available. Based upon morphology, however, the phylogeny of fishes is reasonably well known, particularly for the major groupings (Greenwood et al., 1966). Combining this phylogeny with the concept of levels of interaction can permit one to conjecture on the evolution of interactions among fishes.

Most fishes can react with each other on the tonic level. Excretory wastes, normal mucous secretions, carbon dioxide output, and other metabolic by-products are readily diffused through water and can affect other animals in some physiological fashion. Body shapes, basic color patterns, normal locomotion, and other manifestations of homeostasis and maintenance behavior can result in both intraand interspecies interactions.

Phasic level interactions are also common in fishes. The reaction to Schreckstoff is an example, even though it can sometimes be speciesspecific. Many predator-prey interactions, in addition to much aggressive and territorial behavior, are on this level. Schooling is primarily a phasic level interaction, and some aspects of reproductive behavior are also phasic.

Although there is little evidence of signal level interactions in fish, indications are that the exchange of specific signals may be important in reproductive behavior, especially during the prespawning and parental care stages. The courtship stage is probably most crucial, since species and sex discrimination takes place at this time, and a continuing series of attracting interactions is necessary to hold the pair in proximity until the proper physiological state for spawning is reached. Such interactions can also stimulate the members of the pair to develop into spawning condition. Selective pressure for the evolution of specific signal mechanisms is likely to be high at this stage. Specialized reproductive activities, therefore, are the main interaction types among fish that fall into the category of communication as defined here. There is no evidence, however, that any of these signal interactions in fish approach the level of social organization and social communication found in birds and mammals.

The details for most of the following discussion are derived from the recent extensive monograph by Breder and Rosen (1966). Among the most primitive chordates, interactions appear to be primarily on the tonic level. Breder and Rosen reported that there appears to be no sex discrimination or courtship and mating in the amphioxus (Branchiostoma). The behavior of adult urochords (tunicates) seems to be on the vegetative level.

Behavior of the cyclostomes (class Agnatha) is known mostly from studies on fresh-water lampreys. Although nest-building and mating activities can become quite complex, the interactions in sex discrimination are based on simple, phasic-level responses.

The majority of living elasmobranchs are viviparous, and most sharks and rays possess intromittent organs. Beyond the fact that copulation does occur in these forms, little is known of the actual mating behavior or of the behavior that precedes copulation. Libby and Gilbert (1960) described mating behavior in the clear-nose skate (an egg-laying species, Raja eglanteria) in which the male is reported to nibble on the pectoral fins of the female. It appears highly doubtful that any of the elasmobranchs possess interactions on the signal level.

Experimental data for electric location are available for a few species of teleosts only (Lissmann, 1963). One suborder of rays is well known to possess electric powers, e.g., Torpedo and Narcine, but no electric location function has been established for these. It is likely that the powerful electric field is produced simply as a defense mechanism or in stunning prey.

Among the primitive actinopterygian fishes, only the bowfin (Amia) appears to have well-defined prespawning behavior, although no experimental study has ever been made. The male constructs a tunnellike nest in a dense patch of weeds and, after spawning, guards the eggs in a manner characteristic of many of the higher teleost fishes.

According to the recently proposed classification of teleostean fish (Greenwood et al., 1966), the primitive living teleosts belong in the superorder Elopomorpha (including, in part, the families Elopidae, Megalopidae, and Albulidae and the order containing most eel-like fishes, the Anguilliformes) and the superorder Clupeomorpha (including the typical herring-like forms, e.g., families Clupeidae and Engraulidae). On the basis of the available information, all of these are school or group spawners. Breder and Rosen (1966) referred to this type of behavior as a connuhium confusum, although they pointed out that these so-called mass spawnings actually consist of small spawning groups of a single female and several males. It is probable that none of the interactions among these species are beyond the phasic level.

By contrast, the more advanced order Salmoniformes includes many species in which courtship and nest-building appear to be highly developed. Unfortunately, there is no experimental evidence on the specific cues that these fish use in sex discrimination, and it is impossible to determine the level on which these interactions occur. It is likely, just as a conjecture, that chemical and visual stimuli predominate and that some of these could reach the specificity of signals, in the sense used here.

It is clear, from the earlier discussion, that numerous representatives of the higher teleosts have attained some degree of communication via signals and that this level of interaction is particularly well-developed in the stages of reproductive behavior leading toward spawning, although, as in some cichlid fishes (Kühme, 1963), brood care may also involve responses to signals. The reader is referred to the review by Breder and Rosen (1966) for details.

A brief review of the studies on the gobiid fish Bathygobius soporator seems in order here. After an initial descriptive report on the reproductive behavior (Tavolga, 1954), the sequential nature and the variability of this sequence of events leading to spawning was presented (Tavolga, 1956a). The hormonal controls over parts of this behavior were then investigated, with the finding that castrated males still continued to exhibit courtship but their territorial defense and sex discriminatory behavior broke down (Tavolga, 1955). Ovarian secretions from gravid females were found to be specific olfactory stimuli for courtship in males, whereas acoustic stimuli emitted by courting males were found to attract both females and males (Tavolga, 1956b, 1958). Visual cues of color, body shape, and movement were also important in bringing the mating pair together and holding them together in the nesting site until spawning (Tavolga, 1956a, b). In brood care, the synchronous hatching of the larvae is the result of an interplay of mechanical and visual stimuli between the male and the larvae (Tavolga, 1954). The entire pattern of reproductive activity is complexly influenced by stimuli in the visual, chemical, tactile, and acoustic channels, as well as by the recent experiences of the interacting individuals and their hormonal condition. Controlling this behavior are also the environmental conditions such as proper salinity, temperature, and availability of suitable nesting sites. Vegetative, tonic, phasic, and even some signal level interactions are all involved and complexly interrelated. Unlike the simple sequential situation of stimuli and counterstimuli as proposed for the stickleback (Gasterosteus) behavior by Tinbergen (1951) on the basis of visual factors only, there appears to be considerable variability and room for adaptive adjustment by virtue of a multiple, interrelated form of reciprocal stimulation, with some redundancy in the interaction system.

Throughout this discussion, however, it becomes evident that information of an analytic type is not available, even for species whose behavior has been described in considerable detail. For a causal analysis and an understanding of the behavioral mechanisms of a given set of species-typical interactions, a description of events, no matter how detailed, is only a first step. The description needs to be quantitative; i.e., a demonstration of the range of variability is necessary. A next step must be a study of the specific cues involved, and this should be coupled with information on the physiological and behavioral mechanisms of the emitting system and the receivers sensory capacities. In the study of the response system, the level of integration at which the response operates must be understood, and the phyletic position of the species under investigation must be taken into consideration. A further step that, to date, has been rarely undertaken, is a developmental study. In this the questions are asked: How does this interaction develop in the individual? What are the factors of maturation and experience that lead to the formation of a species-typical interaction? If approached properly and objectively, such questions should lay the “nature-nurture” controversy to rest, since one need not make any a priori assumptions as to the innateness of the behavior patterns (Moltz, 1965). An example of such a developmental approach is demonstrated by the study of the ontogeny of schooling behavior by Shaw (1960, 1961). Unfortunately, developmental investigations in behavior are not easy, and the requirements of being able to raise the animals under controlled, yet optimal, conditions tend to eliminate many species. Nevertheless, it seems to me that the limitations of cross-sectional descriptions should be recognized, and that approaching the study of animal communication from the concept of integration levels and from longitudinal principles, i.e., developmental analysis as proposed by Schneirla (1959, 1965), would be more fruitful avenues of research.

REFERENCES

Alexander, R. D., 1961. Sound communication in Orthoptera and Cicadidae. In Animal Sounds and Communication (W. E. Lanyon and W. N. Tavolga, eds.), Amer. Inst. Biol. Sei., Publ. no. 7, pp. 38-92.

Allee, W. C., P. Frank, and M. Berman, 1946. Homotypic and heterotypic conditioning in relation to survival and growth of certain fishes. Physiol. Zool., 19:243-258.

Amouriq, L., 1964. L’activité et le phénomène chez Lehistes reticulatus (Poeciliidae, Cyprinodontiformes). C. R. Acad. Sei. Paris, 259:2701-2702.

Aronson, L. R., 1957. Reproductive and parental behavior. In Physiology of fishes (M. E. Brown, ed.), vol. 2, Academic Press, New York, pp. 271-304.

Bergeijk, W. A. van, 1964. Directional and nondirectional hearing in fish. In Marine Bio-Acoustics (W. N. Tavolga, ed.), Pergamon Press, Oxford, pp. 281-299.

Breder, C. M., Jr., 1946. An analysis of the deceptive resemblances of fishes to plant parts, with critical remarks on protective coloration, mimicry and adaptation. Bull. Bing. Oceanogr. Coll., 10(2):1-49.

——, 1949. On the relationship of social behavior to pigmentation in tropical shore fishes. Bull. Amer. Mus. Nat. Hist., 94:83-106.

——, 1959. Studies on social groupings in fishes. Bull. Amer. Mus. Nat. Hist., 117:393-482.

——, 1965. Vortices and fish schools. Zoologica, 50:97-114.

——, and D. E. Rosen, 1965. Vortices and fish schools. Zoologica, 50:97-114.

——, and D. E. Rosen, 1966. Modes of reproduction in fishes. Natural History Press, Garden City, New York.

Carmichael, L., 1964. The early growth of language capacity in the individual. In New Directions in the Study of Language (E. H. Lenneberg, ed.), M.I.T. Press, Cambridge, pp. 1-22.

Dietrich, G., 1963. General oceanography. Wiley, New York.

Dijkgraaf, S., in press. The biological significance of the lateral line organs. In Proc. of Conf. on Lateral Line Detectors (P. Cahn and W. A. van Bergeijk, eds.), Indiana University Press, Bloomington.

Dumortier, B., 1963. Ethological and physiological study of sound emissions in Arthropoda. In Acoustic Behaviour of Animals (R.-G. Busnel, ed.), Elsevier, Amsterdam, pp. 583-654.

Frings, H., and M. Frings, 1964. Animal Communication. Blaisdell, New York.

Frisch, K. von, 1942. Über einen Schreckstoff der Fischhaut und seine biologische Bedeutung. Z. Vergl. Physiol, 29:46-145.

Greenwood, P. H., D. E. Rosen, S. H. Weitzman, and G. S. Myers, 1966. Phyletic studies of teleostean fishes, with a provisional classification of living forms. Bull Amer. Mus. Nat. Hist., 131:339-456.

Harris, G. G., and W. A. van Bergeijk, 1962. Evidence that the lateral line organ responds to near-field displacements of sound sources in water. J. Acoust. Soc. Amer., 34:1831-1841.

Hasler, A. D., and W. J. Wisby, 1951. Discrimination of stream odors by fishes and its relation to parent stream behaviour. Amer. Naturalist, 85:223-238.

Kleerekoper, H., and J. A. Mogensen, 1959. The chemical composition of scent of fresh water fish with special reference to amines and amino acids. Z. vergl Physiol, 42:492-500.

——, and J. A. Mogensen, 1963. Role of olfaction in the orientation of Petromyzon marinus. I. Response to a single amine in prey’s body odor. Physiol Zool., 36:347-360.

Köhler, W., 1925. The mentality of apes. Harcourt Brace, New York.

Kühme, W., 1963. Chemische ausgelöste Brutpflegeund Schwarmreaktionen bei Hemichromis bimaculatus (Pisces). Z. Tierpsychol, 20:688-704.

Lanyon, W. E., 1961. The ontogeny of vocalization in birds. In Animal Sounds and Communication (W. E. Lanyon and W. N. Tavolga, eds.), Amer. Inst. Biol. Sei., Publ. no. 7, pp. 321-347.

Lettvin, J. Y., H. R. Maturana, W. H. Pitts, and W. S. McCullough, 1961. Two remarks on the visual system of the frog. In Sensory Communication (W. A. Rosenblith, ed.), M.I.T. and Wiley, New York, pp. 757-776.

Libby, E. L., and P. W. Gilbert, 1960. Reproduction in the clear-nosed skate, Raja eglanteria. Anat. Ree., 138:365.

Lissmann, H. W., 1963. Electric location by fishes. Sei. American, 208(3): 50-59.

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

Marshall, N. B., 1967. Sound-producing mechanisms and the biology of deepsea fishes. In Marine Bio-Acoustics, II (W. N. Tavolga, ed.), Pergamon Press, Oxford, pp. 123-133.

Moltz, H., 1965. Contemporary instinct theory and the fixed action pattern. Psychol. Rev., 72:27-47.

Moulton, J. M., 1967. Directional hearing in fishes. In Marine Bio-Acoustics, II (W. N. Tavolga, ed.), Pergamon Press, Oxford, pp. 187-232.

Nelson, D. R., and S. H. Gruber, 1963. Sharks: attraction by low-frequency sounds. Science, 142:975-977.

Pelkwijk, J. J. ter, and N. Tinbergen, 1937. Eine reizbiologische Analyse einiger Verhaltensweisen von Gasterosteus aculeatus L. Z. Tierpsychol., 1:193-200.

Schneider, H., 1961. Neuere Ergebnisse der Lautforschung bei Fischen. Naturwiss., 48:513-518.

——, in press. Morphology and physiology of sound-producing mechanisms in teleost fishes. In Marine Bio-Acoustics, II (W. N. Tavolga, ed.), Pergamon Press, Oxford.

Schneirla, T. C., 1953. The concept of levels in the study of social phenomena. In Groups in Harmony and Tension (M. Sherif and C. Sherif, eds.), Harper, New York, pp. 52-75.

——, 1959. An evolutionary and developmental theory of biphasic processes underlying approach and withdrawal. In Nebraska Symposium on Motivation, 1959 (M. R. Jones, ed.), Univ. Nebraska Press, Lincoln, pp. 1-42.

——, 1965. Aspects of stimulation and organization in approach/withdrawal processes underlying vertebrate behavioral development. In Advances in the Study of Behavior, vol. 1 (D. L. Lehrman, R. Hinde and E. Shaw, eds.), Academic Press, New York, pp. 1-74.

——, and J. S. Rosenblatt, 1961. Behavioral organization and genesis of the social bond in insects and mammals. Amer. J. Orthopsychiatry, 31:223-253.

Sebeok, T. A., 1965. Animal communication. Science, 147:1006-1014.

Shaw, E., 1960. The development of schooling behavior in fishes. Physiol. Zool., 33:79-86.

——, 1961. The development of schooling in fishes. II. Physiol. Zool., 34:263-272.

Simpson, G. G., 1966. The biological nature of man. Science, 152:472-478.

Tavolga, W. N., 1954. Reproductive behavior in the gobiid fish, Bathygobius soporator. Bull. Amer. Mus. Nat. Hist., 104:431-459.

——, 1955. Effects of gonadectomy and hypophysectomy on pre-spawning behavior in males of the gobiid fish, Bathygobius soporator. Physiol. Zool., 28:218-233.

——, 1956a. Pre-spawning behavior in the gobiid fish, Bathygobius soporator. Behaviour, 9:53-74.

——, 1956b. Visual, chemical and sound stimuli as cues in the sex discriminatory behavior of the gobiid fish, Bathygobius soporator. Zoologica, 41: 49-64.

——, 1958. The significance of underwater sounds produced by males of the gobiid fish, Bathygobius soporator. Physiol. Zool., 31:259-271.

——, 1964. Sonic characteristics and mechanisms in marine fishes. In Marine Bio-Acoustics (W. N. Tavolga, ed.), Pergamon Press, Oxford, pp. 195-211.

——, 1965. Review of marine bio-acoustics: State of the art: 1964. Tech. Rep. NAVTRADEVCEN 1212-1, U.S. Naval Training Device Center, Port Washington, N.Y.

——, in press. Levels of interaction in animal communication. In Development and evolution of behavior (L. R. Aronson, E. Shaw, and D. S. Lehrman, eds.), Freeman Press, New York.

——, and J. Wodinsky, 1963. Auditory capacities in fishes. Pure tone thresholds in nine species of marine teleosts. Bull. Amer. Mus. Nat. Hist., 126: 179-239.

Tinbergen, N., 1951. The study of instinct. Clarendon Press, Oxford.

Verheijen, F. J., 1956. Transmission of a fright reaction amongst a school of fish and the underlying sensory mechanisms. Experientia, 12:202-204.

Vincent, F., 1963. Acoustic signals for auto-information or echolocation. In Acoustic Behaviour of Animals (R.-G. Busnel, ed.), Elsevier, Amsterdam, pp. 183-227.

Washburn, S. L., P. C. Jay, and J. B. Lancaster, 1965. Field studies of Old World monkeys and apes. Science, 150:1541-1547.

Wilson, E. O., 1965. Chemical communication in the social insects. Science, 149:1064-1071.

Winn, H. E., 1964. The biological significance of fish sounds. In Marine Bio-Acoustics (W. N. Tavolga, ed.), Pergamon Press, Oxford, pp. 213-231.

——, 1967. Vocal facilitation and the biological significance of toadfish sounds. In Marine Bio-Acoustics, II (W. N. Tavolga, ed.), Pergamon Press, Oxford, pp. 283-304.

Yerkes, R. M., and A. W. Yerkes, 1935. Social behavior in infrahuman primates. In Handbook of Social Psychology (C. Murchison, ed.), Clark Univ. Press, Worcester, pp. 973-1033.