preservation: species-specificity and mimicry

Preservation of a signal within a population of animals is primarily due to natural selection, kinship selection and related factors (see Pragmatics section of ch 8). By studying control aspects of communication (see table 1-II, p. 13), one is often able to guess at the adaptive significance of signal-characteristics. A few such commonly employed guesses, however, seem overused, particularly those relating to species-specificity or its absence and those relating to mimicry.

There are definite communicative situations in which one expects to find some species-specificity in optical signals due to their pragmatic uses (ch 8). Among these are signals that serve as behavioral barriers to interspecific hybridization (ethological isolating mechanisms) and signals that promote species-specific group actions of special kinds. As pointed out in ch 8 however, many pairs of closely related species that run a high risk of potential interbreeding differ in only minor, although consistent, ways. I think a more viable research strategy for explaining differences among species is to begin looking at other differences in the communicational situations: differences in the pragmatic uses of the signals (ch 8), differences in the signaling environment (ch 7) and so on. It seems unlikely, for example, that the multiple differences in coloration of male wood warblers are required as simple redundancy to insure recognition of the species. Burtt (1977) has made laudable progress in understanding specific differences in various aspects of coloration in these birds, although much more detailed study will be required to account for all the differences that occur.

Second, the other side of the species-distinctiveness coin is the convergence of some signals of different species to highly similar forms. Cody (1969) and Moynihan (1968) have presented arguments that species often evolve to a common type in order to promote interspecific communication. Certain woodpeckers, for example, are so similar that each species excludes the other from its territory where the ranges overlap. Although species may in such cases evolve toward common types for interspecific communication, this explanation for species-similarities in signals runs the risk of being overused. In some clearcut cases convergence cannot possibly be due to this factor. For example, there occurs in Africa a bird that looks strikingly like the eastern meadowlark of North America, yet is in an entirely different family. The nearly identical color patterns of the two species must be due to convergence for signaling similar things in similar environments, since their ranges do not overlap. Nor does the similarity stop with optical signals. J.T. Emlen (pers. comm.) played tape recordings of the North American bird to the African species and received strong responses: the birds also sound alike. There are many other examples of unrelated look-alikes among birds whose ranges do not overlap. With such striking convergence in allopatric species why should one assume that convergence in sympatric species is due to a different cause? One should look first for similarities in the signaling situation (kinds of environmental noise, kinds of information being communicated, etc.) as an explanation of convergence in signals.

Third, with all due respect to those who have studied the phenomenon carefully, I believe that mimicry is an overused explanation of optical signal-characteristics. As Wickler (1968) has pointed out, especially in his discussion of the "coral-snake" patterns of snakes, mimicry is a complicated phenomenon that is difficult to establish with certainty. Although I am convinced that mimicry does occur, and indeed occurs widely in some groups for some purposes, other off-hand suggestions of mimicry bear closer scrutiny. For example, eye-spot patterns occur widely as markings in animals. In some cases, as with the marks on underwings of certain moths (Blest, 1957), the markings appear as real mimicry designed to frighten birds by true deception. However, one should note that the eye-spot pattern in general also maximizes many properties of visual conspicuousness (ch 7). Especially where such spots do not occur as a pair (e.g., on the goldspotted eel), it seems unlikely that mimicry of eyes has played an important role in evolution, unless specific evidence for eye-mimicry can be found. In general, when there is no compelling and reasonably direct evidence for mimicry, one should seek to evaluate a range of possible explanations for a given color pattern.

These three issues have a common thread: all involve explaining the characteristics of optical signals with reference to other speoies. By off-handedly assigning interspecific explanations to communicational signals that have not been analyzed with respect to the species’ own needs and environment, one becomes blinded to basic factors that dictate the design of signals. The interspecific explanations should emerge as phenomena of last resort, since the commonest problem that faces each species is reliable communication among its own members in a noisy channel.

phylogeny: icons, deception and mimicry

A person trained in semiotics might assert that iconic optical signals often evolve from incipient movements of the referent behavior because those incipient movements resemble the behavior and therefore are iconic. Ethologists have tended to write as if something like the reverse were true: optical signals are iconic because they evolve from incipient movements of their referents. In this second view, the resemblance is virtually accidental, a happy artifact that allows the ethologist to trace probable phylogenetic origins of optical signals. In fact, a combination of both views is probably necessary to understand the phylogeny of most optical signals.

For simplicity, suppose that learning and other forms of experience play a minor role in the ontogeny of signaling behavior (but see next section), and that the evolution of communication depends upon natural selection acting on inter-individual variation in behavioral outputs of the sender and in responsive outputs of the receiver. The most parsimonious evolutionary hypotheses must assume that during the course of phylogeny there is a continuous functional relationship between outputs of the sender and responses of the receiver: no saltatory evolutionary "jumps" are allowed, a restriction necessary in order to keep the hypotheses consistent with current evolutionary thought (e.g., Mayr, 1963). With this background, it is possible to sketch hypotheses about the evolution of iconic signals.

In a simple case, incipient movements of locomotion (e.g., Daanje, 1950) resemble actual locomotion and often precede it temporally. Therefore, any receiver that would benefit from being able to predict locomotion by its companions could do so by cuing on the incipient movements, even though they have not been evolutionarily influenced by factors relating to their role as semiotic signs (ch 2); also see the discussion by Johnston (1976: 58). Natural selection then favors individuals that can react as receivers to the incipient movements. It is likely that they can so react because the incipient movements resemble full locomotion: they are already icons. If the sender benefits from transferring the predictive information about his impending locomotion, then individuals that transfer this information most efficiently will be selected for. In the simplest sense, "most efficiently" means that the receiver notices the movements and responds in a way benefiting the sender (e.g., coordinate group movement). This is the classical process of ritualization, in which the incipient movements are enhanced by factors such as exaggerated amplitude of movement, increased frequency, stereotypy, conspicuous coloration and so on.

Second, the evolution of morphological icons proceeds similarly. For example, ch 8 recounted a whole series of hypotheses concerning factors that promote the use of antlers and horns as optical signals in ungulates. The original evolution of these structures was presumably guided by their use in defense against predators or fighting among conspecifics. Because they are displayed prominently in preparation for fighting, they are automatically signs without further evolution to enhance their signal value. Then, receivers that cue on these structures can better predict impending attack, and ritualization via increasing the conspicuousness of the structures proceeds as in the case of behavioral signals, above. In this case, however, ritualization may include additional morphological elements that resemble horns and therefore create iconic redundancy: manes, ear-shape, ear-markings, facial markings and so on (see figure 22 in Hailman, 1977a). Such additional elements are said to be "automimics" of the basic structures (Guthrie and Petocz, 1970).

The evolution of mimetic patterns in such intraspecific communication proceeds because the mimetic patterns are already iconic signs. For example, the pinnae of the pronghorn are erected to receive acoustic information in social interactions, and hence already have a general shape and orientation that resembles the animal's horns. Selection favors variant shapes of pinnae that more strongly resemble horns (e.g., curled "hook" at the tip) because such shapes are more perfect icons and help increase the redundancy of the optical display. There is no need to believe that receivers actually mistake the pinnae for additional horns.

Not all cases of automimicry need involve redundancy, however. In primate species having optical similarities between the markings and structures of the anogenital region and facial region, the duplication cannot be due to simultaneous redundancy because both sets of signals usually cannot be seen by a receiver at the same time. The presumed sequence of phylogeny in these cases begins with the use of anogenital display as iconic predictors of behavior, which are then made more conspicuous by coloration. With increasing encephalization of signals for the many reasons discussed in previous chapters, elements of the anogenital signals will be duplicated on the facial region because receivers are already cuing on such elements. There is mimicry, but not deception.

Not all cases of reputed mimicry within a species are necessarily mimicry, either. For example, the markings on anal fins of certain mouth-brooding cichlids (see fig 8-2d, p. 253) have been cited as egg-mimics. Wickler (196 8: 226) states that "the female repeatedly attempts to take the round finspots into her mouth as if they were eggs. Since the spots are located next to the male sexual aperture, spermatozoa automatically end up in the female's mouth." Although itis possible that this interpretation is correct, there are reasons to doubt it, and there are alternative explanations for the characteristics of the finspots and their phylogeny. Fish have keen vision and it seems unlikely that the female Haplochromis burtoni cannot discriminate her own eggs from two-dimensional markings on her mate's fins. Furthermore, the female herself possesses similar markings on her fins, although they are not as conspicuous as the male's markings. Other cichlids also possess such markings, but some do not look like eggs at all (to me, anyway), and it seems likely that the species-specificity of these markings is selected for as a reproductive isolating mechanism of last resort: a final species-identification before the sperm are picked up and the eggs fertilized in the female's mouth.

Suppose the female H. burtoni is not actually attempting to pick up markings that she cannot discriminate from eggs. How could the evolution of this signal-system proceed without recourse to explanations involving visual deception? One possibility is as follows. Nipping at the fins of the mate is observed frequently in both sexes of cichlids, and presumably helps stimulate the partner for spawning. Both sexes may therefore evolve markings that index the optimum sites of such tactile stimulation: the area near the genital pore. There are many hypotheses that would account for the circularity of these indexic signals in species in which they are circular: predisposition of receiver-systems to be organized with bullseye-like receptive fields (ch 5), conspicuousness of regular geometric patterns (ch 7), conspicuousness due to outlining, internal contrast of color and brightness, etc. Because the male moves forward while releasing sperm, his anal fin also indexes the position of sperm for the female, and any markings on the fin may enhance the conspicuousness of the index. This explanation is not inconsistent with Wickler's interpretation to this point: it is merely a more complete statement of possible steps in phylogeny.

Wickler (loc. cit.) notes that the female of many species of cichlids pushes the male aside with her mouth to gather his sperm, and he notes that the "egg-dummies may be just an insurance that pushing will be converted to sucking." Therefore, the crux of the question is whether there is further change in the signal to make it more egg-like in order to deceive the female into making a different response. Possibly in H. burtoni there has been, but in cichlids whose indexic markings are not egglike the female takes the sperm anyway, so it is not logically necessary to postulate true deception. If there is no true deception, then it is questionable whether mimicry occurs at all. If one of the attributes of this indexic signal is species-specificity to prevent hybridization, then selection pressures acting on various species to be simply different from all others is likely to produce egg-like markings in at least one species of the group. Egg-mimicry might be involved, but the entire signal-system could have evolved with no deception and no mimicry at all.

Finally, the phylogeny of interspecific mimicry is different yet. Take for example the evolution of eye-like spots on the underwings of certain moths (Blest, 1957). In this case, the behavior of the moth when confronted by a predator was probably initially dependent upon simple startle (ch 6), mediated by incipient flight movements of the wings. The effectiveness of the startlemovement was then enhanced by conspicuous coloration of the usually hidden underwing. Any coloration that increased conspicuousness would be favored, and among those patterns that maximize conspicuousness are bullseye-like spots. Finally, those conspicuous spots that were actually mistaken by predators for eyes were further favored because of the additional frightening effect.

The conclusion of this analysis is simply that not all forms of iconic signals and reputed mimicry evolve in the same way. Interspecific mimicry begins with selection for conspicuousness or other attributes, and proceeds toward convergence with another stimulus when variation allows and the results of the resemblance are favorable to the sender: mimicry is a phylogenetic endpoint selected for true deception. Intraspecific resemblance, on the other hand, may evolve without any deception. They may arise to create redundancy (weapon mimicry), to duplicate signals for use elsewhere on the body (anogenital-facial mimicry) or to enhance conspicuousness by various means. In intraspecific resemblances, the receiver can usually distinguish the iconic signal from the thing it resembles, and in some cases the resemblance may even be accidental. Entire books on mimicry have been written as if all mimicry and the evolutionary processes giving rise to it were essentially identical: that is unlikely.

ontogeny: the role of experience

The role of experience in the development of communicational behavior is understudied, and probably for at least two reasons. First, it is convenient in evolutionary thinking to treat behavioral outputs of animals as if they were fixed by the genetic endowment and showed little variation among individuals due to differences other than genetic differences. Indeed, in order to argue the points in the previous section, I explicitly made such an assumption for purposes of simplifying the evolutionary explanations. Second, the ontogeny of behavior is understudied simply because it is difficult to study--not only technically difficult, requiring special rearing conditions and long-term scrutiny, but also because it is conceptually difficult and experimentally difficult to dissociate all the interacting variables.

It seems likely that the form of many of the speciescommon signals of the kind discussed in this book develop without an appreciable role played by experience. It is not so obvious that the recognition or use of those signals is largely experientially independent. For example, Sackett (1966) found that rhesus macaques reared in isolation from birth responded differentially to photographic transparencies showing different facial expressions of other individuals. This result does not mean, however, that the information transferred by such signals in wild monkey troops is independent of experience. Stephenson (1973) provides five examples of troop-specific use of signals in Japanese macaques. It seems unlikely that these intraspecific differences are due to genetic differences among the troops. More likely, different signals have come to have different referents through individual experience, and these semiotic relationships have persisted in the troops through cultural transmission. If troopspecific communication can be culturally transmitted, species-specific communication can be also: it is only necessary that all individuals have roughly the same experiences during ontogeny.

Experiential factors in communication are by no means restricted to primates, where they are not unexpected. The newly hatched laughing gull chick responds appropriately to the parental signal for feeding: the red bill-tip held in a specific orientation, moved in a specific direction and at a specific speed (see figure 15 in Hailman, 1977a). However, chicks in the wild also respond to many inappropriate stimuli, and then as a function of time since hatching decrease their responsiveness to inappropriate objects (Hailman, 1967a). Furthermore, experiments demonstrate that chicks develop a much fuller perceptual ideal: newly hatched chicks respond to simple models of the parent's bill whereas experienced chicks demand a life-like model of the parent (see figure 18 in Hailman, 1977a). Experiments show that food-reinforcement can promote relevant kinds of perceptual learning in these gull chicks (Hailman, 1967a, 1971). In short, the physical object that elicits responsiveness(i.e., the parent) does not change, but the key sign-vehicle to which chicks respond does change rather markedly during ontogeny, a process I called "perceptual sharpening." Indeed, what Sackett's (1966) study with rhesus macaques may really show is that young monkeys have an initial disposition to respond differently to different facial expressions; perhaps those first responses are then ordinarily channeled ontogenetically by experiences in real social communication with other individuals.

Given that experience can play a role in optical communication, one must still ask how this fact is relevant to the design-characteristics of optical signals. Although this problem-area has been little studied, there are at least two kinds of potential answers. First, the behavioral and morphological elements of the parent gull's signal are designed primarily to elicit the first responses from newly hatched chicks. If the early responses can be assured, the chicks learn more sophisticated sign-vehicles. What they learn may be entirely environmentally dependent, with one configuration being as readily learned as the next. Using food as reward for pecking, for example, I conditioned one species of gull chick to prefer models of a different species (Hailman, 1967a). The physical signal is structured primarily to insure the early responses that initiate the ontogenetic sequences.

A second possibility was suggested years ago (Hailman, 1959b), although it is pure speculation. If signals are learned, at least in part, then their characteristics may be structured to promote the learning. It is not clear yet what optical characteristics one would expect of signals that promote visual learning. I suggested that the crepuscular, hole-nesting wood duck--whose life in the nest is spent in dim illumination and who communicates as an adult primarily in twilight--might be particularly disadvantaged in learning optical signals of the species. These factors might therefore play a role in the evolution of the strikingly colorful and complex display plumage of the male. The example might be wrong, but the possibility remains thatcertain signal-characteristics could be easier to learn than others, and hence selected for to promote perceptual learning.

strategy for future analysis

It appears unlikely that one could predict the signal-characteristics of a species from even a detailed knowledge of the factors shown to be of importance in this volume. One may hold such prediction as an ultimate goal, but there is no way to know if it is an attainable goal unless and until it is attained. In order to make progress, then, some lesser goal is needed that provides a clear method for analysis. I believe that the goal is to be able to predict differences in signal-characteristics among species or other populations. The strategy is to choose groups for comparison that are as similar as possible in all ways except for the signal-characteristic under investigation. For example, if one wants to understand why the male cardinai is so conspicuously red, it seems only sensible to compare it in detail with the closely related pyrrhuloxia, whose male has only a wash of red on the breast and whose female is extremely similar to the female cardinal. One must ask what kinds of differences exist in the communicational needs and communicational environments of the two species. There is, it seems to me, some real prospect of answering this kind of question about species-differences, whereas asking simply why an animal has the signals it does may often prove frustrating.

The analysis is more complicated than simply comparing animals, of course. One must document how communication takes place (control) and how it develops in the individual (ontogeny) before sense can be made of comparative correlates between signals and other variables. The comparative method then focuses on factors that correlate with the ontogenetic systems that lead to the control endpoints, and these factors may be simple variables of the physical signaling environment or complex factors of the cultural environment. Or, the strong correlation may be with taxonomic status, in which case one suspects that historical factors in phylogeny are responsible for the species-differences. Therefore, even though one is using a comparative method, the ultimate analysis involves all four of the classes of behavioral determinants (table 1-II).

In charting the variables for comparison of optical signals between species, one must take into account all the kinds of problems reviewed in this volume: ambient light falling upon the animal, properties and limitations of the sender's abilities to reflect that light, kinds of information being encoded in the signals, optical noise in the communicating environment, and abilities of the receiver to accept and decode the signals. Furthermore, non-signal hypotheses must be considered as explanations for animal coloration and behavior (see chs 4 and 6). Making a comparison between two species is no simple task, but it is one likely to produce new ideas about the optical characteristics of animal signals.

signals and communication

In ch 8 I emphasized that an understanding of what was being communicated could help to understand how it was communicated. A major implication of this book, however, reverses that sequence and hence emphasizes the interrelationships between signals and the information they encode. It seems to me a reasonable assumption that if one begins looking in detail at the characteristics of optical signals, those characteristics will help clarify the kind of information being transferred, and, indeed, the entire communicational process. For example, the study by Thomas Stillwell and me (in prep.) of inciting by female mallards reveals that it is a clear indexic signal, and not vaguely symbolic as previously considered. Although we have not studied the responses of the receivers in the detail required to understand the communication thoroughly, it is clear that merely the form of the signal itself suggests hypotheses about its informational content. If one is to understand communication, I believe one needs to ask after the details of the signals themselves.

man's optical signals

Belonging to Homo sapiens the author and reader alike have some measure of curiosity about this species' optical signals. Since man is an animal, albeit a complex one, he should not be totally free of the factors that dictate the design of signals in other animals. He has clearly elaborated these, and in some ways has freed his signals from the kinds of evolutionary constraints affecting most other species. There have been, numerous attempts to explain some of the basic "body signals" of man that are similar to the intrinsic signals of other animals. Such studies by ethologists in particular have produced many interesting, if not bizarre, hypotheses. This is not the place to review man's signals and hypotheses about them, but it is the place to point out that none of these studies of man has attempted to analyze his signals from the optical viewpoint taken in this volume. Such a viewpoint holds promise for better understanding of facial expressions, gestures and other intrinsic optical signals.

Man, however, has created a whole array of extrinsic optical signals. These range from very straightforward signals such as traffic lights, to complex creations such as dance, architecture, painting and even the notational systems for written language and mathematics. This great range of extrinsic signals cannot be understood in most cases without considerable study of culture and linguistics, but it should be possible to begin examining extrinsic signals from the optical viewpoint.