movement

Movement is the most obvious principle of conspicuousness. When an object moves within a stationary visual field, it sets itself off immediately. There are even special cells in the visual system of some animals that respond only to certain kinds of movement within their receptive fields (ch 5). Movement is a chief component of intrinsic optical signals (fig 4-1), and along with body shape (including expressions) has received the most attention by ethologists (Hailman, 1977a). The task is therefore not to document the importance of movement in conspicuousness, but rather to see if certain attributes of movement are especially conspicuous.

The first point is that movement must be within a certain range of Speed in order to have maximum conspicuousness. Very slow movements (e.g., locomotion of sloths) will not be detected readily, and very rapid movements may exceed the critical fusion frequency of the eye (fig 5-13). There is a broad range within these limits in which movement will be easily detected, and since certain visual cells appear to have optimum detection at certain speeds, it seems possible that there are optimum speeds of movement for conspicuous optical signals. For example, parent gulls move their bills in front of their young chicks, and if hungry, the chicks beg for food by pecking at the bill. Chicks have a preferred speed of movement, found in experiments to be about 12 cm/s, and the mean speed of the parental bill in the signal-movement measured from movie film was 14.5 cm/s (Hailman, 1967a). The optimum speeds may be relative to the size of the object. For example, Schleidt (1961) found that turkeys discriminate flying predators from other entities overhead by the number of body-lengths the hawks move per unit time. Discrimination is thus independent of distance from the hawk and is coded more complexly than simple speed of the image across the retina.

Directionality of movement is a second variable that deserves experimental attention. For example, the parent gull's bill is held vertically and moved horizontally in the usual signaling situation. When a horizontally held, vertically moved bill is presented to chicks, their response rate is much lower (Hailman, 1967a). In fact, experiments using models of both orientations, as well as motionless and moved in both directions, yielded significant interactions among the variables. The results suggested that movement across the long axis of the bill and movement parallel to the visual horizon both play a part in the characteristics of the optimum signal.

An obvious principle of conspicuousness in movement is repetition. Repeated movements are a form of temporal redundancy (ch 2) and are exhibited by many animals. For example, the male turkey performs strutting movements throughout the day (Schleidt, 1964). The tail-flashing of the dark-eyed junco is more eye-catching as a repeated spreading of the tail to flash the white outer feathers than would be a chronic spread. Besides adding conspicuousness, of course, repetition of a signal and the interval between signals can be used to encode specific information (ch 8).

Sudden onset of a signal also makes it conspicuous by virtue of emphasizing the change from motionlessness to movement. This principle is used protectively in startie-deception (ch 6), where conspicuousness is followed by immediate concealment as a mechanism promoting confusion. In social communication, sudden onset followed by other elements of conspicuousness make a signal more obvious. Simpson (1973) emphasizes the role of surprise in conspicuousness, a notion similar to that of "news" mentioned in ch 2.

A full classification of movements and the variables that make them especially conspicuous has not been devised, but it seems likely that other factors contribute. In repeated movement, rhythmicity may promote conspicuousness to visual systems particularly tuned to certain frequencies, whereas in other cases arhythmic, jerky movements may be more conspicuous, as in claw-waving displays of fiddler crabs (see figure 21 in Hailman, 1977a). Indeed, if visual systems are tuned to particular speeds, directions and frequencies of movement, generally low variance in repeated performance--movement stereotypy--is a factor of conspicuousness itself (Marler, 1974: 36). More is said about stereotypy in ch 8, below.

These sorts of variables in movement must also be considered relative to the environmental context in which the motion occurs. A leaf fish (fig 6-3) moves, but is concealed because it moves by drifting with the current. Therefore, conspicuous movement must be at a different speed or in a different direction from movements of the relevant optical environment. Movement in relation to specific environments is considered later in this chapter.

Finally, movement may be exaggerated by extrinsic optical signals. For example, the branch-shaking of many monkeys creates a far more conspicuous threat display than would movement of the animal alone. Many kinds of movements carry specific information (next chapter) and hence are structured for other than mere conspicuousness, but most movement helps render the sender more conspicuous. Indeed, Noble and Curtis (1939) felt that the primary role of signal-coloration in the jewel fish was to accentuate movement, so I now turn to conspicuousness of coloration.

Reversed counter-shading

As noted in ch 6, animals cast a shadow upon the substrate and also shadow the lower parts of their own bodies. I know of no social signals that utilize the substrate shadow. An animal could, however, become visually conspicuous by emphasizing its body-shadow through reversed counter-shading: the "with-shadow" pattern of Albrecht (1962) or ^“inverse counter-shading" of G.W. Barlow (1974). The previous chapter mentioned examples of animals that possess reversed counter-shading because they habitually pose upside-down, thus making the pattern concealing. There are, in addition, animals leading rightside-up lives that are light dorsally and dark ventrally, and hence quite visually conspicuous. A striking example is the male bobolink, an icterid blackbird of hayfields that is largely white and cream-colored above. Few animals are chronically counter-shaded in reverse, although in male eider ducks (Somateria) and among fishes in some of the grunts (Pomadasyidae) the pattern is permanent. More usually, reversed counter-shaded plumages like that of the male bobolink or certain plovers (Pluvialis) are assumed only for the duration of the breeding season. In fishes, the coloration may be assumed for a shorter period of some specific behavior requiring conspicuousness, as in parental defense (Keenleyside, 1972; Baylis, 1974a); see also Morris (1958).

More common than reversed counter-shading of the entire animal is reversed counter-shading on a local part of the body used in display. Among birds, for example, several shorebirds assume black throat, breast or belly patches for the breeding season (e.g., ruddy turnstone, rock sandpiper, dunlin) and in several orioles (Icterus) the black throat and upper breast permanently contrast with the orange crown and neck. Indeed, there are so many cases of breeding plumages of birds and courtship colorations of fishes that involve at least small areas of dark coloration below lighter coloration that one suspects reversed counter-shading is an extremely widespread attribute of optical signals in animals. Some examples of reversed counter-shading are shown in fig 7-1.

Hamilton and Peterman (1971) discuss counter-shading in coral reef fishes, and if I understand their point correctly, emphasize the temporal contrast in coloration when a fish changes rapidly from the concealingly counter-shaded pattern to display coloration. The display coloration itself, in the examples discussed, is not counter-shaded in reverse, but derives its conspicuousness by contrast with the counter-shaded pattern, so that the latter is selected for because of its role in signaling. They state that "countershading hides a signal rather than the fish" (p. 363), although it is not clear to me how it does one without doing the other.

fig 7-1. Some examples of reversed counter-shading that render animals visually conspicuous. Some animals are wholly reverse counter-shaded or nearly so, whereas others; such as the male hooded oriole, have restricted regions of conspicuous shading. Male spectacled eiders are habitually shaded whereas male bobolinks are shaded only during the breeding season, and some fishes may intensify shading for relatively brief periods.

enhancement of shape

The third general principle of conspicuousness is enhancement of shape. In this case, however, conspicuousness of signals cannot be deduced by simple reversal of concealment of entire animals. Animal outlines are suppressed by mechanisms such as transparent appendages or disruptive coloration (ch 6), but opaque appendages and non-disruptive coloration can hardly be considered specific adaptations for conspicuousness. One kind of nondisruptive coloration--uniform coloration--is indeed conspicuous in providing a clear appreciation of an animal's outline, when the color contrasts with the background (see below on surface-contrast. Thus, the all-blue male indigo bunting, all-red cardinal, all-black crow and other uniformly colored animals provide good visual outlines. Another method of enhancing shape is to outline or edge it in a contrasting border, as in the monarch and other butterflies. This strategy is not used often in vertebrates, but certain fish having distinctive fin shapes may emphasize the shapes with black edging (e.g.,the palometa). T. Johnston (pers. comm.) has suggested that depth perception allows a solid, three-dimensional animal or object to be detected more readily than a flattened, two-dimensional one. Therefore, one might expect flattened body parts (lepidopteran wings, fish fins, etc.) to show bordering contrast more than other body parts. Further-more, a flat surface has an unambiguous border that can be outlined, whereas a solid object may be seen from different angles and hence would favor enhancement by uniform coloration.

Enhancement of outline is more commonly employed locally on an animal's body to emphasize the shape of a signal-patch. Here, black edging of a shape is commonly employed, as in the edging of the golden-crowned sparrow's and the golden-crowned kinglet's golden crowns, the blue jay's and barn owl's white faces, the hooded merganser's white crest, and so on in many birds. Fishes commonly employ the same principle of outlining a signal-patch in black, as with the gold spots on the goldspotted eel, but more often have a dark spot outlined in white or other light color, as in certain frogfishes (Antennarius), young of certain d amselfishes (Eupomacentrus) and butterflyfishes (Chaetodon), young rock beauty, silver porgy and many others. Birds sometimes have signal coloration outlined in white (e.g., the russet belly-patch of the gray partridge), but convincing examples are fewer. Examples of both polarities are common among lepidopterans. As in the case with local reversed counter-shading, local outlining of shape in less dramatic examples is probably widespread.

Another mechanism for enhancement of outline is to create a signal-patch of unusual shape, so that it is conspicuous by virtue of its novelty. This suggestion is simply a manifestation of the general principle of surprisal or news (ch 2). The strategy of unusual shape, however, is difficult to identify in practice. Animal body shapes may be changed by disguising appendages or by other deceptive means that obliterate tell-tale shapes (ch 6), but it is difficult to reverse this principle because unusualness of shape is hard to specify in the abstract. One attribute of unusualness of shape in nature might be geomtertc regularity: for example, circles, triangles and rectangles are rare both as body shape or as any other pattern in nature--except as optical signals. Eyes, of course, are circular, and round patches of animal coloration are often explained as eye-mimics (ch 6). Nevertheless, round signal-patches seem far too widespread to be explained simply as eye-mimics, and one element promoting circular patches may be the unusualness of their shape. Many examples from fishes were cited above.

Triangular patterns are surprisingly recurrent in avian plumages, particularly as color-patches assumed by the male for the breeding season. Some display triangles are nearly equilateral, as in the white triangle of the hooded merganser's crest (when fanned), the paired triangles (dark or light) in tails of several ptarmigans (Lagopus), the black triangular patches on the wing-tips of the black-legged kittiwake, or the russet facial patch of the Cape May warbler. Other triangles may be long and thin, but still strikingly noticeable in shape, as in the black wing-marks of Sabine's gull, white wing-marks of Bonaparte's gull, mustache marks (red or black) of the male common flicker, the red crest of the pileated woodpecker, or the black facial marks of golden-winged and olive warbiers.

Other regular geometric shapes, like rectangles, as well as certain improbable shapes, may also be selected for as signals because of their unusualness and hence compelling visual conspicuousness. Rectangles, or parallelograms, with quite straight edges occur in the speculum patches on the wings of many ducks (see figure 20 in Hailman, 1977a), but are not otherwise common in birds. Similarly, a marine fish, the black margate, possesses a compelling black parallelogram on its ventro-lateral surface. Quite improbable shapes of coloration are found on some birds--e.g. the white crescent on the male blue-winged teal and male Barrow's goldeneye (see figure 9 in Hailman, 1977a)--but as noted above, the notion of unusualness is a difficult one outside of regular geometric shapes known to be rare in nature. Some examples of enhancement of shape are shown in fig 7-2.

Fig 7-2. Enhancement of shape exemplified by various animals. An animal may be uniformly colored like the cardinal, to contrast with its background, or may outline critical (particularly flattened) parts with special coloration, like the palometa. Light colors may be outlined in black (monarch) or dark outlined in white (mourning clock). Conspicuousness may also be achieved by shapes that are infrequent in nature, such as geometrically regular shapes: circles (frogfish), triangles (merganser) and rectangles (margate).

Extrinsic signals may also utilize unusual shapes to increase conspicuousness. For example, the bower of Amblyornis macgregoriae is a triangular pile of sticks surrounded by a circular ring of moss on the ground (Firth, 1970). Other species of bowerbirds have even more elaborate bowers, but many have triangular and circular elements.

Surface-contrast

An obvious principle of conspicuousness is visual contrast, but the question is: what is contrast, how is it measured and what does it predict about signal-coloration? Chapter 5 divided apparent surface coloration into the phenomenal dimensions of brightness, hue and saturation, and developed the colorimetric notions of luminance, dominant wavelength and excitation purity as methods for measuring the phenomenal dimensions (table 5-I). Consider each in turn as a component of visual contrast.

The most straightforward kind of contrast depends simply upon the amount of visible light reflected from two surfaces: brightness contrast. The extreme brightness contrast occurs when one surface reflects nearly all incident light and the other reflects almost none of it, so that the surfaces appear achromatically white and black, respectively. Brightness contrast is employed commonly in emphasis of shape by outlining an animal or signal-patch in either black or white (see examples cited in the previous section). Furthermore, brightness contrast is also employed in homogeneous coloration to emphasize animal shape, which may partly explain the black or dark gray coloration of various birds that live in open, well-lighted habitats such as fields and marshes (e.g., crow, blackbirds, juncos, male lark bunting, male phainopepla, etc.).

An animal may use brightness contrast for other than outlining a signal-patch or emphasizing body shape through homogeneous coloration: signal-patches themselves may be black orwhite. One may expect achromatic coloration of signal-patches anywhere the background coloration ofthe patch is rather light or dark. For example, the black male lark bunting possesses boldly contrasting white wing-patches and the nearly all-black white-headed woodpecker has, of course, a contrasting head. Or, reversing the polarity of contrast, the snowy owl is marked with black scalloping and black about the face, terns (Sterninae) of many species are white or light gray with contrasting black caps, and the wood stork is white with a dark gray head and black wing-patches. Most signal-spots on otherwise silvery fishes are contrastingly black. Similarly, black patches and spots that are presumed social signals may be found on birds having chromatic but quite light ground colors such as the male dickcissel's black throat on its yellow breast, the male cardinal's black face on its red body, and so on. Of course, black markings on birds likely have other functions as well: resistance to abrasion, reduction of glare, etc. (see ch 4).

The second dimension of visual contrast is that of hue, correlated primarily with the spectral peak of refleeted light (table 5-I). Before attempting to define contrast in hue per se, however, it is worth noting that differences in hue may actually operate as brightness contrast in special cases. The most convincing example comes from plant-animal communication (fig 2-2), where it has been observed since the turn of the century that hummingbird-pollinated flowers are overwhelmingly red. This is not due, I think, to any ancestral predisposition for hummingbirds to be attracted to red, although a red-preference could have evolved subsequently. Rather, flowers adapted for hummingbird pollination must produce large quantities of nectar to be attractive to a large animal; therefore, it is not adaptive for such flower to attract insects that might deplete the nectar or pollen without effecting pollination. Because insects, by and large, have spectral sensitivities shifted toward shorter wavelengths (ch 5), they see well in the UV but poorly in the "red" part of the spectrum, where vertebrate sensitivity is generally good. Therefore, a red flower looks bright (and hence contrasting with its background) to a hummingbird, but dark (and hence not so contrasting) to an insect: red is the only color that is conspicuous to birds without being so to insects (Pijl and Dodson, 1966; Raven, 1972).

Contrast of hue in a more general sense is difficult to define. Non-uniformity of spectral discriminability defines phenomenally similar blocks of wavelengths in the visible spectrum (fig 5-10, p. 133), but does not reveal whether blue is equally discriminable from, say, red and yellow. It seems intuitively likely that complementary colors provide the best contrast in hue. Complementary wavelengths are those monochromatic lights that combine to yield an achromatic sensation. Figure 7-3 plots representative boundaries of hues on the outside of the spectral locus of the C.I.E. chromaticity diagram. A line has been extended from each boundary through the white-point to the interior of the opposite boundary in order to define the boundaries of complementary hues on the interior of the diagram in fig 7-3.

Fig 7-3. Complementary dominant wavelengths may be defined by the C.I.E. ohromaticity diagram for equal-energy white illumination. (See fig 5-12 on p. 137 for explanation of the diagram and the soale of wavelengths.) Hues have been labeled on the outside periphery of the spectral locus, with approximate hue-boundaries marked by arrows pointing toward the white-point (W). When extended through the white-point, the arrows define boundaries of complementary hues on the inside periphery of the diagram (broken arrows). Therefore, complementary hues are found on either side of the spectral locus (unbroken line), so the diagram may be read from inside-out or outside-in. For examples beginning at the lower left! the complement of violet is always a green, but the complement of blue may be a green, yellow or orange. Exactly the same relations may be found at the opposite side of the diagram.

The approach of fig 7-3 to the problem of complementary colors provides some instructive cautions. Some complements are straightforward: the complement of yellow is always some sort of blue and that of red, violet and purple some sort of green. All other relations are more complicated. For example, depending on the dominant wavelength of a green stimulus, its complement may be orange, red, purple, violet or even blue. Similarly, the complement of a blue may be green, yellow or orange. Figure 7-3 shows that meaningful simplification of the notion of complementary hues is not possible, despite statement to the contrary in popular books on color. In order to find the "ideal" complementary hue for a given stimulus, one must know the spectral distribution of light in order to plot the stimulus-point on a chromaticity diagram for calculation of the complementary dominant wavelength. One should also recall that the entire approach via the C.I.E. chromaticity diagram is based on human vision, and that the particular use here is based on the white standard of equal-energy. The approach is therefore of use only so far as animal color-vision of interest resembles human color-vision in metameric properties (see ch 5), and when the relevant white-point can be determined as applicable to a given perceptual situation.

The final component of visual contrast is saturation, the dimension of surface coloration given least attention in the perception literature. Like brightness, the contrast between two surfaces differing in saturation is greatest when the saturations differ by the maximum amount, the range being from achromatic to monochromatic (ch 5). Only one useful generalization seems extractable: since there are few highly saturated stimuli in the natural environment other than flowers (which are actually optical signals of plants to animals), one may expect animal signal coloration to be highly saturated. Because of the broad bands of wavelengths reflected by any non-fluorescing surface that is not dark, reflecting surfaces can never be really monochromatic. Nevertheless, within the constraints provided by biochromes and schemochromes (ch 4), one expects "pure" colors rather than pastel shades from animal signals. This expectation so obviously obtains in birds, fishes and many other animals that an explicit check for reasonableness is unnecessary.

The final component of visual contrast is saturation, the dimension of surface coloration given least attention in the perception literature. Like brightness, the contrast between two surfaces differing in saturation is greatest when the saturations differ by the maximum amount, the range being from achromatic to monochromatic (ch 5). Only one useful generalization seems extractable: since there are few highly saturated stimuli in the natural environment other than flowers (which are actually optical signals of plants to animals), one may expect animal signal coloration to be highly saturated. Because of the broad bands of wavelengths reflected by any non-fluorescing surface that is not dark, reflecting surfaces can never be really monochromatic. Nevertheless, within the constraints provided by biochromes and schemochromes (ch 4), one expects "pure" colors rather than pastel shades from animal signals. This expectation so obviously obtains in birds, fishes and many other animals that an explicit check for reasonableness is unnecessary.

The problem, then, is to put contrast of brightness, hue and saturation together in a framework that integrates these simultaneously occurring properties of surface coloration. Unfortunately, no ideal color-space has yet been devised, as noted in ch 5. The model required is one of a three-dimensional space in which one distance between two plotted stimuli expresses the contrast between them. Ideally, the mode should be based on empirically derived functions of discriminability over the three-dimensional space, but I have been unable to find such data, even for human perception. Various color-spaces have been proposed, as mentioned in ch 5, but each has certain drawbacks, so that a coherent and predictive model of surface-contrast remains to be devised; one attempt is mentioned at the end of this chapter.

Finally, one may note that the conspicuousness of bright, complementary, saturated colors is not restricted to intrinsic signals. Bowerbird decorate their bowers with colorful berries, flower petals, leaves and other objects they can find (Marshall, 1953; Gilliard, 1969).

miscellaneous principles

A few factors promoting conspicuousness do not fall readily into the above categories: image-size, repetition of coloration in space and signal-rarity.

It is obvious that, in general, the greater the image-Size, the more conspicuous an object--at least up to a point. Size of an animal is due to factors other than image-size in optical communication, to be sure, but animals can vary the imaginal areas of their optical signals by the distance at which they display from intended receivers and by the size of any signal-patterns of coloration. On expects relatively large color-patterns to be correlated with long-distance signaling, as in territorial advertisement. Short-distance communication, as in precopulatory display, might involve smaller movements or color-patches. The prediction is difficult to check without detailed investigation of the communication in the species of interest.

Size of an image has upper limits, however. If signal-conspicuousness depends in part on visual contrast with the background, then one predicts that there should be some upper limit to the size of a signal. A case can be made for this principle in the contrasting red color-patch on the bills of parent gulls, which are signals that elicit and direct the begging responses of their chicks (e.g., Hailman, 1967a). Species with relatively thin bills have entirely red lower parts, whereas in species with thick bills the red signal-patch is confined to the bill-tip (where mandibles converge to a narrower width) or to a red spot on the lower mandible, as shown in fig 7-4.

Fig 7-4. Optimum size of signals exemplified by bills of gull species. Small bills are wholly red, but larger species have yellow bills with red restricted to the narrower tip or to a narrower spot on the lower mandible (after Hailman 1967a).

Repetition in space is similar in effect to temporal repetition. Both principles of conspicuousness derive their advantage from redundancy (ch 2). Spatial repetition works by creating a regularity that is uncommon in nature and hence attractive to the eye; in this sense, it is related to unusualness of shape (see above). Repetitive patterns are often found in birds, as a series of tail-spots in cuckoos (Coccyzus) and trogons (Trogon), a series of head-stripes in sparrows (e.g., Zonotrichia) and certain wood warblers (Parulidae), repetitive wing-bars, breast-spots, breast-stripes, etc. The principle is no less common among fishes, with repetitive spotting in such species as the great barracuda, graysby, spotted goatfish and others; repetitive vertical barring in schoolmasters, spadefish, banded butterflyfish, sergeant major and others; repetitive horizontal striping in sharksucker, several grunts (Haemulon), slippery dick, several parrotfishes (Scarus) and others. Some examples of spatial-repetition are shown in fig 7-5.

Fig 7-5. Spatial repetition of coloration,as in white spots on the cuckoo's black tail,black spots on the goatfish's silvery body, black bards on the spadefish's light gray body or white stripes on the remora's dark body.

Finally, the notion of surprise in signals, mentioned above (see also Simpson, 1973) was developed earlier by Moynihan (1970) into a general evolutionary theory of displays. Noting that rare signals individually transfer more information than common ones (the notion of "news" discussed in ch 2), Moynihan points out that signal-rarity is a principle of conspicuousness. He goes on to speculate that being conspicuous and effective, such rare signals will be selected for during phylogeny, but in becoming more common lose their rarity and hence effectiveness, and so eventually disappear from the repertoire. That phylogenetic theory is not the concern here, but it may be noted that it is a difficult theory to evaluate empirically.

Table 7-I

Some Principles of Visual Conspicuousness

The principles of conspicuousness that I have been able to identify are summarized in table 7-1. No doubt the list is incomplete, but it does serve as a starting point for attempting to see how specific conditions of the communicating environment may help structure the characteristics of animal signals.

extremely dim illumination

In most cases the nocturnal environment is not devoid of ambient light in the way in which a deep cave is, for instance. Experienced campers know that starlight alone on a clear night provides the human eye with considerable information, and full moonlight seems bright to the darkadapted eye. Because such nocturnal ambient light is below photopic operating levels of our cones (ch 5), we see the world in black and white. Based on this introspection we expect nocturally signaling animals to have primarily white signals; any other color would reflect less light and hence reduce the brightness contrast. It is not surprising, then, that nocturnal birds such as most goatsuckers (Caprimulgidae) have white patches on the tail, neck or wings, surrounded by black to enhance the contrast. In interkingdom communication, moth- and bat-pollinated flowers are white. It seems worth pointing out, however, that there has been no systematic attempt to show empirically that animals in general resemble us in lacking color vision at nocturnal levels of ambient light. Data such as the presumed rod/cone break in dark-adaptation curves (fig 5-4, p. 124) do suggest that nocturnal color-vision is unlikely, but the evidence is scarce and there seems to be no theoretical argument compelling such belief.

The alternative to white, maximumly reflecting coloration is of course bioluminescence. As expected, deep-sea and nocturnal organisms have been the primary animals to evolve bioluminescent social signals.

sun-angle

Everyone knows from experience that the spectral distribution of direct sunlight differs through the day: sunlight is quite white near midday and orange as sunset approaches. (Early risers know the morning sunlight is similarly orange, but I have only rare personal confirmation.) The change in light quality is due to Rayleigh scattering (eq 3.9), which is greatest at lowest angles of the sun, when sunlight must travel a longer path through the atmosphere (ch 3). For a given time of day and year, the sun's angle also decreases as one moves along a line of longitude from the equator toward a pole on the earth's surface; latitudinal differences in ambient light are due to the same principal factor as hourly differences. Finally, because the earth is tilted on its axis, the north pole is angled farther from the sun than is the south pole in December, whereas in June the relations are reversed because the earth has traveled half its orbit around the sun while maintaining the same tilt on its axis. Seasonal differences in ambient light quality are therefore also due to sun-angle. The interaction of time of day, latitude and season in determining sun-angle may be shown graphically as in fig 7-6, which simplifies the problem of estimating general quality of ambient light.

Two important aspects of available light change with sun-angle: as the sun becomes higher in the sky, the absolute irradiance falling on the earth increases and the spectrum shifts. Figure 7-7 on p. 212 shows the major effects, simulated by the SOLREF program of Dr. W.P. Porter for Death Valley, California on 21 June (summer solstice). Just after sunrise (0530 hours, local standard time), when the sun's angle is less than 5° above the horizon, the spectrum is a monotonically increasing function of wavelength. A half-hour later the total irradiance is much higher and the spectrum has flattened to a considerable extent. An hour after this there is still some increase in irradiance-level, but the spectrum has changed very little. After five more hours a similar increase in irradiance has occurred, but again, little change in the spectral distribution of light. The changes following noon are symmetrical with the morning functions. From this simulation, one can see that the primary effects of color quality (spectral distribution) occur only very near sunrise and sunset; i.e., at quite low sun-angles. Because of details not important here, the latitudinal effects are slightly different, but in general only the lowest sun-angles (say, less than 10°) have important effects on color quality of ambient irradiance, making it decidedly orange in appearance.

Fig 7-6. The sun's angle (altitude above horizon) depends upon the time of day and the latitude, as shown by the three-dimensional plots. Furthermore, these relations change through the seasons, as shown by differences in plots for the winter and summer solstices.

Fig 7-7. Spectral irradiance depending upon sun-angle exemplified by time-of-day changes for one date at one location (computer simulation).

The angle from which an object such as an animal is viewed relative to the sun's rays makes a larger difference in the object's appearance when the sun-angle is low than when it is high. At low sun-angles a viewer with his back to the sun sees an animal illuminated by the orange sunlight, and if the animal is against the sky, that sky will be intensely blue because of Rayleigh back-scattering. Looking toward the sun, however, the animal will be illuminated primarily by dimmer back-scattered plus refleeted light, the former being bluish in emphasis, the latter being reddish or otherwise colored depending upon the reflecting surfaces in the neighborhood. At high sun-angles, on the other hand, the animal will be similarly illuminated by direct and indirect sunlight whatever the angle of observation.

It seems likely that these differences in ambient light that result from differences in sun-angle will have some influence on the evolution of signal-coloration. In particular, an animal that signals at low sun-angle may be expected to utilize long-wavelength-reflecting signals (red, orange, yellow) when signals are colored rather than being achromatically white. This expectation is based on the fact that total irradiance is relatively low, so that a bright signal is conspicuous. Signals that reflect most strongly in the part of the spectrum where ambient light levels are high (i.e., long wavelengths) will appear brightest. Furthermore, against blue sky or green foliage, such long-wavelength-reflecting signals will show high color-contrast (fig 7-3, p. 203).

The difficulty of checking these expectations concerning signal-coloration used at low sun-angles resides in lack of information about the timing of animal communication. Most temperate and polar animals breed during the summer, when the sun-angles at their latitudes are relatively greatest.

A body of information available, however, is latitudinal distribution of birds, a major group employing optical signaling. One expects that the more northerly the range of a north temperate species, the more likely it is to show red or other long-wavelength signal coloration. A comparative check may be made by holding phylogenetic relationships relatively constant and looking for colorational differences that correlate with latitude among species within the same taxonomic group.

Species of the Subfamily Carduelinae range over a large latidudinal spectrum. In North America, the most northerly finches of this group are the redpolls (Acanthis), which breed in Canada and Alaska and range south in winter only to the northern tier of American states. The primary signal-coloration of these northerly birds is a bright red cap; there is also black around the bill, and the hoary redpoll has a white rump. Nearly as northern in distribution are the crossbills (Loxia) and the pine grosbeak, in which the males are rosy to orange-red dorsally. The purple finch and its relatives (Carpodacus) are more southerly and have the same male red coloration, but these birds also frequent the mountainous areas of western North America where the added problem of altitude occurs (see next section). The most southern carduelines are the goldfinches (Spinus), which all range into Mexico; only the pine siskin from this group breeds a significant distance north into Canada. All these birds lack red entirely, and have signal-coloration mainly of yellow and black. Roughly the same relations apply in Europe, except that the European goldfinch ranges very far north into Scandinavia, and unlike the southerly New World Spinus species has a bright red face. In sum, those species that remain at high latitudes during winter months when the sun-angle is low have long-wavelength signals as expected, and their southerly relatives lack them. I do not think this informal check is a very accurate test of the prediction, but the result is consistent with the expectation and so encourages further investigation.

altitude

The effect of altitude upon available light is not the same as that of latitude, although there are some similarities. As one goes up a mountain, the patch of sunlight through the earth's atmosphere becomes shorter, but the effects are not the same as those of moving toward the equator. Variations in sun-angle merely alter the length of the light-path through the entire atmosphere, whereas elevation determines which layers of the atmosphere light must traverse: high on a mountain sunlight simply does not traverse the lower part of the atmosphere before striking an animal. Since the atmosphere becomes denser closer to the earth, relatively small changes in altitude cause noticeable changes in ambient irradiance. Furthermore, the composition of the atmosphere differs at different heights above sea level, so the filtering effects change.

The major changes in ambient light with increased altitude appear to be higher irradiance levels and a subtle shift toward a shorter-wavelength emphasis in light quality. Figure 7-8 shows SOLREF simulations for the latitude of Death Valley, California at three elevations. The floor of Death Valley is slightly below sea level (lowest place on earth) and to the west lies Mt. Whitney at 14,495 feet (highest place in the contiguous United States). The top of this mountain is therefore about 4.4 km in elevation, lying between the two upper curves in fig 7-8. From the figure two effects of elevation may be noted. First, irradiance levels increase with altitude, the change being less pronounced the higher one goes. And second, there is a shift from an essentially flat spectrum to one that peaks about 480 nm, that region of the spectrum we see as blue. Whether these effects are quantitatively sufficient to influence animal coloration remains an empirical question.

Fig 7-8. Spectral irradiance depending upon altitude above sea level (computer simulation).

The consequences of these changes are not easy to predict. One expectation might be that with short and long wavelengths available for reflection, animals confined to high altitudes could evolve purple signals (ch 5), a color quite rare in most animals. Birds most evidently confined to high elevations are the rosy finches (Leucostiate) of the high Rocky Mountains. All three species have pinkish-purple wing- and rump-patches. Nevertheless, the most purple bird in North America is the varied bunting of Mexico, whose range avoids the high Mexican piateau, so the possible effects of altitude on signal-coloration remain an open question.

general habitat

The foregoing discussion of available light for signaling disregards any influence of the habitat, and so applies only to animals signaling in quite open areas. As was pointed out in ch 3, absorption and reflection by plants and other objects in specific habitats may have a large effect upon light available for signaling (e.g., fig 3-13, p. 79). For present purposes, one may divide surface habitat into (a) open, direct-sunlit areas, and (b) vegetated, indirectly lighted areas, and then ask what differences in signal-coloration might be expected between them. Following this, the underwater habitat is considered.

There are three major kinds of open area habitats: field and prairies, deserts, and the surface of bodies of water. Naturally, what constitutes an open, directly sunlit habitat depends upon the size of the animal being considered: a short-grass prairie may be open to a deer but heavily vegetated from a mouse's viewpoint. The major light characteristics found in open areas are high levels of irradiance and directness of the radiation. These are the two requirements for effective use of signals that depend upon specular reflectance, dichromatism and iridescence (ch 4), so one might expect such signal-coloration to occur in these open habitats and not in vegetated ones.

The expectation is difficult to check, but iridescent colors are particularly common in ducks, especially the puddle-ducks (Anatinae), in hummingbirds (Trochilidae) and in sunbirds (Nectariniidae). Ducks, of course, do frequent open-water areas, and even the tree-nesting wood duck displays on open water, sometimes in bright sunlight. Hummingbirds live in vegetated areas, but males display their iridescent colors in bright sunlight by special aerial courtship flights before the female (Hamilton, 1965). On the other hand, many birds that live in open habitats have not evolved iridescent coloration, so there is no precise correlation. Open areas provide the opportunity for the evolution of iridescent coloration but do not demand it.

Vegetated habitats have the opposite characteristics with respect to ambient light: the absolute levels are low and the light is diffuse, due to filtration and reflection (fig 3-14, p. 80). Furthermore, the light quality is shifted toward a distinctly greenish illumination (fig 3-13, p. 79). Even in forested areas, however, some direct sunlight penetrates, often to ground level, so there is a continuum of lighting conditions between totally open areas and the deepest, darkest forests. D.H. Janzen (pers. comm.) has pointed out that near sunrise and sunset in stratified tropical forests a good deal of direct sunlight streams in nearly parallel with the ground, so that forest lighting is qualitatively different at different times of day. These problems make it difficult to check any expectations without quite detailed information concerning where and when animals actually display their coloration.

The expectations themselves are relatively straightforward: one expects highly reflecting signals such as white or yellow to provide maximum reflection in dark forest. Dark colors, such as violet, blue, green and even red, should be rare. A good check on this expectation is provided by Burtt's (1977) data on wood warblers (Parulidae). In general, warblers living in dense spruce and other northern forests have yellow or white signal-coloration, enhanced by contrasting black, whereas species displaying in more open areas have evolved other coloration. For example, the male blackburnian warbler is a treetop species that has a bright orange throat, the chestnutsided warbler is a scrub species that has a greenish cap and chestnut sides and the redstarts (Setophaga), which often display in bright habitats, have red or orange. Nevertheless, the black-throated blue warbler is all-blue above and apparently lives in quite forested habitats, so simple correlations without specific knowledge of displaysites are only suggestive. More will be said about the coloration of wood warblers in a later section, when other factors can be taken into account.

Table 7-II

Available Light and Signal Coloration

As was noted in ch 3, two primary factors determine available light in aquatic habitats: depth in the water and reflection from the bottom (figs 3-15 and 3-16, pp. 82 and 83). Turbidity (eq 3.14) also plays a role in determining available light, but its effects as transmission noise are possibly even more pronounced, as noted in a later section. As depth in the water increases, the light falls in level, shifts to shorter-wavelength emphasis and becomes more uniform with respect to angle of view (fig 3-15). These effects are somewhat offset by reflection from the bottom (fig 3-16). Unfortunately, consideration of available light for reflection in aquatic habitats generates no clear expectations of signal-coloration, except for the probable need of high reflectance. This need is admirably met by the total reflectance produced by guanine plates in fish scales (see ch 4), which may reflect all ambient light or be modified as interference filters to reflect selective colors.

Table 7-II (opposite) summarizes some of the variations in light available for optical communication under various environmental circumstances and the possible effects upon the variables of signal-coloration.

translucency

Noise in a translucent medium is due to small particles suspended in the medium between the sender and the receiver. For convenience of reference I shall call this "translucency noise." Its principal effects are to blur the sharpness of visual images, reduce the amount of reflected light reaching the eye, and selectively absorb or scatter spectral components of the transmitted signal.

Two principal types of translucency noise occur. In terrestrial environments, water vapor in the form of mist, fog, etc. is the primary source, whereas in aquatic environments turbidity due to suspended particulate matter is the important source. Their effects are similar, except in spectral transmission. Water vapor appears to cause primarily Mie scattering (eq 3.10), the spectral effects of which are complicated, whereas absorption due to at least some kinds of turbidity increases with spectral frequency (eq 3.13).

There appears to be no easy way to combat the first two effects of translucency noise (blurring and low light level) except by decreasing the distance between sender and receiver when the noise occurs. One might expect optical signals in translucent media to consist of large patches of color and rather gross movements to combat the loss of image-sharpness, but I can find no critical evidence bearing on this hypothesis. Reduction in the total amount of light received is probably not serious compared with the loss of image, and in the extreme case both factors cause total opacity. The solution to this problem is simply to use another channel. Wootton (1971) notes that a stickleback living in tea-colored water has less welldeveloped optical signals than its congeners living in transparent media.

Animals should be able to combat spectral effects of translucent noise by reflecting those wavelengths that penetrate best. When the noise is due in part to scattering rather than simple absorption, then the back-scattering may create an optical background of the opposite spectrai extreme and thus increase the signal-to-noise ratio. Baylis (1974a) notes that the yellow signal-coloration of a cichlid fish may be an adaptation for getting the signal through its turbid environment. G.W. Barlow (1974) notes that cichlid species living in turbid waters generally have yellow, orange or red markings (especially about the eyes), whereas those living in clear waters have blue and green as characteristic coloration.

Spectral effects of terrestrial water vapor require more study in relation to animal signals. Depending upon the size-composition of droplets in various kinds of fog and mist, the Mie scattering could create various spectral effects (ch 3). Fog lights on automobiles are ordinarily of long wavelength, possibly an empirical choice based on what light penetrates best and back-scatters least toward the driver. Back-scattering is not as important in Mie scattering as in Rayleigh scattering, and in any case an animal sender communicating by reflected light is primarily concerned with penetration to the receiver. It could be that the signal-coloration of marsh birds like the yellow-headed blackbird and red-winged blackbird shows long-wavelength reflectance to penetrate early morning mist. Minnaert (1954: 257) states that the sun is usually white when seen through fog or mist, but may be red--an occurrence he attributes to Rayleigh scattering by very small droplets. Foglights and animal signals may therefore be adaptations to rare conditions of Rayleigh scattering, as well as more usual conditions of Mie scattering. Empirical studies in specific signaling habitats and conditions would be most useful.

opacity

Opacity noise is due to any physical object that partly blocks the receiver's view of the signal-object. The principal source of opacity noise of transmission in both terrestrial and aquatic habitats is plant material. The chief effect of the noise is to obscure parts or all of the signal-object, and the obvious way to combat this noise is to avoid it by communicating in a clear environment, or at least at a short distance. Thus, many songbirds perch on the outer branches of trees or at the tops of trees, or fly while displaying; many mammals seek the crest of a hill or the top of a boulder for optical communication, etc.

Given that an animal lives in and must display in a vegetated environment, how could its optical signals be designed to combat opacity noise? Apparently this question has been given little attention, so it is not even clear as yet how it should be approached. In ch 3 I suggested that a beginning might be made by modeling the simple case of circular "holes" in the vegetation through which circular signals are viewed. Considering the case of a single "hole" in the vegetation, the receiver can view the greatest part of the signal by minimizing the distance to the hole and maximizing the distance to the signal (eq 3.17). This beginning appears to predict a possible spatial arrangement for optical communication in small forest songbirds. MacArthur (1958) noted that in some species of boreal wood warblers the female forages low whereas the male displays from the tops of trees. This spatial arrangement often is such as to combat opacity noise, as shown in fig 7-10.

Extension of the kind of model represented by eq (3.17) would have to deal with at least four complicating factors. First, the size of the "viewing holes" in the environment, their geometric pattern in the plane normal to the line of view, and their spacing in that plane all affect the probability that a signal-object of a given size will be seen in its entirety. Second, the number of "planes with holes" between the sender and receiver, as well as their distances from the communicating animals, must be taken into account. Third, the shape of the "holes" relative to the shape of the signal-object must be considered. And last, one must deal with the problem of whether or not is is necessary for the receiver to see the entire signal-object. This decision depends upon the nature of the information-carrying sign vehicle (see Johnston, 1976, for a theoretical discussion). It is premature to consider all these ramifications here, but a few comments relative to them can be added.

Fig 7-10. An optimum strategy for observation when opaque objects are in the line of sight is to minimize the observer-to-object distance (d_h) and maximize the object-to-sender distance (d_Q). See eq (3.17), p. 86.

One need not see an entire signal-object in order to recognize it. Often, the view of part or several disconnected parts of the object is sufficient to construct the object perceptually. One strategy of animal signaling, then, may be to create rather large, homogeneous signals that allow visual reconstruction when only parts of them are seen. The opposing strategy is to create a small signal, the entirety of which is viewed.

In order to test comparatively the hypothesis of opposing strategies one requires predictions of factors that favor the strategies differentially. Body-size of the sender is one such factor. If a woodland bird is small, it cannot use the strategy of "perceptual reconstruction˙" even if the entire bird were homogeneously colored with bright signal coloration, not enough of it could be seen for reliable reconstruction. Therefore, one may expect small woodland birds to use small signal-patches of coloration that may be seen in their entirety, whereas large woodland birds may be homogeneously colored for signaling. However, once again this depends upon the nature of the sign vehicle involved. The expectation does, though, appear to have some merit. Male wood warblers are rarely homogeneously colored, having instead small patches of presumed signal-coloration on various parts of their bodies (caps, facial marks, breast-bands, etc.). Large woodland birds, however, are often homogeneously colored, as in many jays and the blue-gray tanager (blue color), summer tanager and cardinal (red), and so on. Other large woodland birds are boldly marked in only two colors, such as many species of orioles (orange and black), scarlet tanager (red and black), evening grosbeak (yellow and black), etc. There may be other reasons for this correlation between body-size and coloration, however, such as differences in mean signaling distance.

More attention needs to be paid to the systematic optical inhomogeneities of environments relative to opacity noise. For example, the branches of coniferous trees tend to radiate horizontally from a single trunk, whereas branches of broadleaved trees tend to grow upward at an oblique angle and then divide into smaller branches with irregular angular orientations. In essence, a coniferous forest approaches a layered optical environment that may allow better viewing in the horizontal direction than would a broadleaved forest. Another woodland example of structual inhomogeneity is caused by the fact that leaves are in favorable places for absorbing light. One can often see well vertically near the trunk of a tree, but not toward its periphery. A related inhomogeneity is due to the lowest branches being the oldest, and hence the thickest. By reasoning similar to that behind the development of eq (3.17), one might expect that it is easier to view something from below than from above. A squirrel hunter told me that a strategy for detecting his prey is to lie on the ground looking up through the trees, and this strategy may be another factor favoring the differential heights of wood warbler sexes (fig 7-10).

Similar optical inhomogeneities occur in environments other than forests. For example, marsh vegetation tends to be vertical arrays of long-stemmed grasses and sedges. This array causing opacity noise might, again, be combatted by different strategies: either by use of long, thin, vertically aligned signals that could be viewed entirely; or by wide, horizontally aligned signals that would have to be perceptually reconstructed. In this case, the problem with the first strategy is that the signal might be mistaken for its background (part of the problem of detection noise, considered previously), as when the bittern "freezes" vertically as a deceptive posture to escape detection. Aquatic environments also have optical inhomogeneities, such as kelp beds, layering of coral heads and so forth.

Table 7-IV

Optical Transmission and Signal Characteristics

In sum, one can state that opacity noise of transmission is an obviously important consideration of both the spatial arrangements of communicants and the physical structure of optical signals. However, it is just as obviously a complex matter that requires considerable attention before hypotheses can be generated from knowledge of the environment that predict design characteristics of optical communication.

Table 7-IV summarizes some aspects of signal-transmission and their possible effects on optical communication. It represents a bare beginning of an area that hopefully will receive more experimental attention.

timimg

The obvious solution to the concealment-vs-conspicuousness problem is simply to be concealed generally and conspicuous only for brief periods of communication. This strategy is used so generally among animals that one may not stop to consider that optimum communication in the absence of opposing needs might dictate that animals be habitually conspicuous. Ernst Mayr (pere. comm.) pointed out to me that birds on South Pacific islands, having relatively fewer predators than on continental areas, are generally more brightly colored. Furthermore, larger North American songbirds--such as orioles (Icterus), tanagers (Piranga) and various large finches (e.g., cardinal, rosebreasted grosbeak, evening grosbeak, blue grosbeak)--may be less subject to predation than smaller songbirds, and the males of all these species are quite brightly colored. Smaller songbirds, presumably suffering greater predation, are more concealingly colored (e.g., emberizine sparrows, tyrannid flycatchers, parid tits). And when the needs of optical communication dictate bright coloration in male songbirds, as in the parulid warblers, the birds molt back into concealing plumage after the breeding season, unlike larger songbirds that retain their bright coloration the year-round.

In general, then, an animal that communicates optically will probably be as conspicuous as predation pressure allows. In each case, some balance must be struck between failure of reproduction because of the inability to attract a mate or hold a territory on the one hand, and failure of reproduction due to death or injury by predation on the other. Animals solve this dilemma by timing conspicuousness in two general ways: either by color-changes or by display of normally hidden conspicuous coloration. Many male birds change into special conspicuous coloration for the breeding season, molting back into concealing plumage thereafter, whereas many fishes, some cephalopods and a few other animals can change color over shorter periods of time (ch 4) to assume conspicuous coloration for communicational purposes. The fact that animals have hidden coloration displayed only for communication is so well known (e.g., Tinbergen, 1951) that one needs only to be reminded of rump and tail colors of mammals; wing-stripes, tail-spots and rump-patches of birds, etc.

We ordinarily think of display coloration in terms of a hidden color-patch that is revealed by movement, but B.D. Sustare and E.H. Burtt (pers. comm.) have suggested that something like the reverse is also possible. Some animal movements are so rapid that visually conspicuous coloration at rest could become concealing in motion. For example, the wingbeat frequencies of hummingbirds and some insects exceed the critical fusion frequency (ch 5), at least at moderate illuminance levels, so that a bold pattern of black-and-white might fade to a more concealing gray in flight. It is also possible that rapid alternation of bright colors could lead to a perceptually fused color that is concealing. However, I know of no cogent examples of this theoretically possible strategy.

T. Johnston (pers. comm.) points out that in many situations predators can more easily capture stationary prey. Therefore, one may expect stronger selection for concealment in the resting postures. Furthermore, motionlessness per se is concealing (ch 6), so in general one expects more optical signaling in moving than in stationary situations.

background

Without changing coloration in any way an animal can be concealed against one background and conspicuous against another. This solution to the opposing needs is probably utilized to a much greater extent than has been noted in the literature, although Moynihan (1975) makes a specific mention of this principle in relation to cephalopod signals. For example, the brown thrasher is generally colored concealingly, being counter-shaded brown above with dark streaks on a light belly below. On the ground, the thrasher is quite inconspicuous, but the male almost always sings from the very top of a tree, where his mate and neighboring males see him as a dark silhouette against the open sky. Even in this conspicuous position, a hawk from above will not see the thrasher as clearly as will conspecifics because the optical backgrounds differ. Therefore, two principles concerning background may operate in animal communication: choice of a different background for display than for other activities, and choice of a site in which the background is different for social observers and potential predators.

There is an interesting example among fishes that interrelates background, body shape and coloration. As noted previously, two principles of optical conspicuousness are brightness contrast and spatial repetition (table 7-I). Vertically barred and horizontally striped fishes are equally conspicuous in this regard, so one may ask what factors dictate the pattern to be assumed by a particular species. The answer, most probably, relates to the backgrounds that make the patterns concealing under non-display circumstances. A long-bodied fish that swims rapidly in open water will likely be tracked readily by predators, and so is longitudinally striped as a mechanism of motion-deception (ch 6). Yet stationary, it is conspicuous due to factors mentioned above. Conversely, a deep-bodied fish adapted for twisting slowly through vegetation will be concealingly colored with vertical bars by the principle of matching coloration (ch 6). Displaying in the open, moving rapidly through vegetation to create a flicker effect (ch 5), or orienting its body vertically all render the fish conspicuous. For example, the slender remora, an open-water fish associated with sharks, has black and white horizontal stripes, whereas the laterally flattened spadefish has vertical barring (fig 7-5, p. 207). The most convincing cases come within individual species, as among certain wrasses (Halichoeres) and parrotfishes (Scarus) where the young animals or females (or both) are longer and thinner than the males, and also tend to be clearly horizontally striped, in contrast to the uniform coloration or indistinct vertical barring of the males.

Exceptions test the rules, and it would be instructive to know why the pelagic pilotfish (a slender jack that follows sharks) is vertically barred. Magnuson and Gooding (1971) report that a pilotfish apparently defends a shark as a moving territory, so the conspicuous coloration might be involved in intraspecific aggressive display. However, when chasing another pilotfish, the chaser usually blanches to a subtly counter-shaded, much less conspicuous coloration. The authors suggest that the barred coloration might be aposematic: coloration to warn their potential predators away. Questions still remain, however, because the remora, which also accompanies sharks, is also black and white, but has the expected longitudinal stripes of a pelagic fish (fig 7-5). The example illustrates how difficult it may be to account for the exact display coloration of an animal, even when something is known of its habitat and behavior.

addressee-specificity

In special cases, it is possible that the sender can orient so that it is conspicuous to a particular addressee and less so to other observers, even against the same kind of background. This phenomenon has received little attention and probably depends on various optical mechanisms, so a few examples seem worth exploring. In some cases, addressing a signal is a fairly straightforward matter not involving any special optical principles. For example, a displaying peacock orients his raised tail with the colorful surface facing the female. From behind, the coloration is noticeably duller and more concealing, and because the tail is planar, from the side it presents a small image. In other cases, addressee-specificity may be more involved.

In ch 6 the ability of fishes to reflect a bright pulse of light from their guanine-containing scales was noted in the context of startle-deception. The fish is momentarily conspicuous, but immediately becomes concealed by swimming in a new direction that may minimize the body area seen by the potential predator. An analogous optical ploy is utilized for social communication, in which the display of bright specular reflectance is followed by lateral or otherwise conspicuous orientation of the sender. Male fishes of many species, for example, perform such a display before females at a possible spawning site. It is not totally clear how fish control the specular flash of light, but it seems likely from considering the nature of their scales that orientation perpendicular to the sender-receiver axis could play a role. It may be that in usual postures the orientation of each scale is a little different, promoting no overall specular flash, and that special body postures that align the planes of many scales create the conspicuous signal.

A somewhat analogous case may occur in the male mallard, whose head appears green, purple or black in various situations. The exact basis of the coloration has not been scrutinized (A. Brush, pers. comm.): it could be dichromatic but most likely is an interference phenomenon (ch 4). In any case, I have observed that viewing the head feathers of specimens normal to the feather-plane maximizes the green coloration, whereas viewing them end-on down the shaft maximizes the purple coloration. In most diffuse light the feathers appear nearly black. Mallards change their head colors during display by piloerection, and perhaps also by specific orientation with respect to sunlight. It is evident, for example, that in the "head-round" courtship display posture (Lorenz, 1941; Weidman, 1956) the feathers are erected and the head appears purple to the viewing female. During other displays, such as the "grunt-whistle" the feathers appear to be depressed, flashing green coloration toward the female. In the display known as "showing the back of the head to the female" the male leads the female, turning his head from side to side, and hence possibly flashing purple and green alternately. (It is difficult to observe this display from the female's viewpoint, so I have few critical notes as yet.) The point is that the male mallard is able to address conspicuous coloration toward the intended receiver, and to some extent appear less conspicuously colored to other observers.

Finally, there is the phenomenon discovered by Hamilton (1965) in which the male Anna's hummingbird orients his courtship display with respect to the sun's rays and to the female, so that she sees his iridescent cap, throat and back. To observers at other angles the bird may simply look dark.

general orientation

The orientation of the sender with respect to an intended conspecific receiver can help to separate concealment and conspicuousness without specific recourse to different backgrounds or different colorations of the same body part, as discussed in the foregoing sections. In those cases, the undesired and desired observers see the same body parts of the sender, but perceive them differently because of background or special reflection. A more general phenomenon of display-orientation is analogous to display-timing (above): the sender orients so that the intended receiver simply sees some different part of the sender's body than does the general observer. Like timing, this principle of orientation is so well documented (e.g., Tinbergen, 1951) that one needs do little more than point it out for sake of completeness. What does seem worth exploring is that these orientational aspects of signaling predict certain accompanying adaptations of morphology, particularly shape and coloration.

An excellent example of an adaptation of shape is the display dewlap of certain small lizards. A male orients laterally with respect to his intended receiver, and lowers the brightly colored dewlap. Seen from the side, the position of another male or female to which he is displaying, the male's signal is quite conspicuous. However, because the dewlap is very thin, the male viewed from above, from in front, or from behind shows no bright coloration and remains concealed. This example is closely related to that of the peacock mentioned previously, except that the bird specifically orients his tail per se toward the female: the signal is addressed by moving specific body parts. There is therefore a probable continuum between kinds of orientations that render a signal conspicuous to some observers and not to others.

Burtt (1977) attempted to see if the expected correlation between coloration and use (concealment or conspicuousness) could be documented quantitatively in wood warblers (Parulidae). He characterized the coloration of warblers and their optical backgrounds by means of a three dimensional color-space having axes of dominant-wavelength, excitation purity and relative luminance (table 5-I, p. 138). The exact scaling considerations are too detailed to recount here; see Burtt (1977). He then measured the reflectances of leaves and warbler colorations spectrophotometrically and also measured spectroradiometrically irradiances in various forest habitats where warblers occur. From the spectral products of these two elements, he calculated the average radiance of various warbler colors and various leaf colors, and then plotted all in the three-dimensional color-space, as shown in fig 7-11. The diagram shows that warbler coloration of blues, white, yellow and orange plot apart from leaf colors, whereas warbler coloration of green, browns, gray, black and chestnut plot among the leaf colors. One therefore expects these latter colors to be large areas of dark dorsal coloration for counter-shading and matching coloration (the concealing part of coloration, ch 6), whereas the blue, white, yellow and orange colors on warblers should be smaller areas used for optical signaling. With a few exceptions, this is just the pattern found in nearly 50 species of warblers. The result goes only one step beyond intuition, but does illustrate a method for objectively assessing which colors should be concealing and which conspicuous, and hence predicts to some extent the spatial arrangement of these colors on the animals.

Fig 7-11. A three-dimensional space in which each leaf and warbler color plots as a point under specified conditions of illumination. In this example, some plumage colors plot near leaf colors, whereas others (blues, white, yellow and orange) plot farther from the leaves, and hence are more conspicuous. (After Burtt, 1977.)

distance

Because an animal is usually detected visually by predators from afar and by conspecifics from nearby, it would be potentially useful to be concealed from afar while being simultaneously conspicuous nearby. There appear to be at least three ways in which this goal could be accomplished with special coloration.

First, an animal could be colored so as to engage in the deception I call element-matching (ch 6) when seen from afar, but not when seen closeby. I was led to this hypothesis by watching mountain bluebirds in western North America. The male is often described as "sky-blue" (e.g., Robbins et al., 1966: 234), but he obviously does not blend into the sky when seen against it. However, this species frequents open high meadows, where it perches in trees that, when seen against the sky, have irregular patches of blue sky seen through the green foliage. At least under certain conditions of viewing, male mountain bluebirds perched in such trees are difficult to detect because they appear as just another patch of blue in the foliage. An exact match in brightness with the sky is unnecessary because the tree is such a mosaic of dark and light patches that brightness-contrast is less important in conspicuousness than is contrast in hue. Directly against the open blue sky the male looks noticeably dark and contrasting, of course. Furthermore, up close the male appears very contrasting in both color and brightness with the darker green foliage, and its avian shape becomes evident so that the bird is highly conspicuous. The male mountain bluebird is in fact conspicuous nearby and at least sometimes highly concealed at a distance.

Another way in which an animal could be concealed from afar but conspicuous nearby depends upon spatial fusion of color (ch 5). An array of bright colors perceptually fuses at a distance to achromatic, desaturated coloration, given the proper selection of colors. Possible examples of this strategy occur in various parrotfishes (Scarus and Sparisoma), where each scale may be edged in a color that contrasts with the general ground color, or where some scales are one color and some are a different color. These fishes never appear brightly colored to me from a distance, but when I dive close to them, they appear brightly colored and highly conspicuous. Some quail (Callipepla and Lophortyx) show ventral scaly coloration with must contrast nearby that blurs to a general gray in the distance. The yellow warbler has bright red streaks on its breast that become virtually invisible at several meters distance.

A third mechanism relating to distance concerns conspicuousness of barred or striped patterns. If the contrast between dark and light stripes is not too great, the pattern will be detected only at some optimum distance (ch 5). Furthermore, since contrast-sensitivity optima differ in spatial frequency among species (fig 5-15, p. 144), a barred animal could be conspicuous to one species (say, its conspecifics) while remaining homogeneously colored and hence less conspicuous to another species (say, its predators). How important this factor really is remains to be determined, but a large number of coral-reef fishes are either longitudinally striped or vertically barred (e.g., Chaplin, 1972).

Some of the possible solutions to the opposing needs of concealment and conspicuousness are summarized in table 7-V. It is surely an incomplete list, but demonstrates how consideration of optical principles in ecologically relevant situations may explain certain characteristics of animal signals.