Behavioral Ecology Vol. 15 No. 1: 141–147 DOI: 10.1093/beheco/arg092 Asymmetry in size, shape, and color impairs the protective value of conspicuous color patterns Anders Forsmana and Joakim Herrströmb Department of Biology and Environmental Science, Kalmar University, SE-391 82 Kalmar, Sweden, and bSyngenta Seeds AB, Box 302, SE-261 23, Landskrona, Sweden a The received view of protective coloration in animals is that conspicuous colors and patterns have evolved because they elicit avoidance behavior in potential predators. In the present study, we examine the spontaneous response of naive predators (Gallus gallus domesticus) to artificial prey to test the hypothesis that deviations from bilateral symmetry of signaling pattern elements may negatively influence the avoidance-inducing effect of conspicuous color patterns. Chicks displayed stronger aversions to artificial ‘‘butterfly’’ prey items possessing symmetric color pattern elements than to those possessing asymmetric signals with pattern elements of different color or shape. Although they attacked signals with a size asymmetry of 5% at the same rate as symmetric signals, signals with a size asymmetry of 7.5% or more were attacked more often than were symmetric signals. These results suggest that the protective value of conspicuous color patterns is impaired by asymmetry in color, shape, and size of color pattern elements. Our findings also argue against the notion that animals have inherent preferences for symmetric over asymmetric objects, and demonstrate the existence of a threshold for asymmetry detection, beyond which further increments in asymmetry have no influence on signal efficacy. Key words: asymmetry, communication, perception, predation risk, signaling. [Behav Ecol 15:141–147 (2004)] onspicuous and simple color patterns (often red, yellow, or white in combination with black) are common among animals that are distasteful, noxious, or otherwise potentially dangerous to their predators (see Cott, 1940; Summers and Clough, 2001; Wallace, 1867). The common view is that conspicuousness has evolved because it constitutes a strong visual signal that is easy for receiving predators to detect, learn, and associate with unpalatability (Alatalo and Mappes, 1996; Endler, 1991; Gittleman and Harvey, 1980; Guilford and Dawkins, 1991; Lindström et al., 1999; Yashi and Higashi, 1998). However, conspicuous coloration may provide protection against predators even if the prey lacks chemical or structural defense mechanisms, because coloration may elicit spontaneous avoidance behaviors in naive predators (see Coppinger, 1969, 1970). It has been suggested that bilateral asymmetry also may play a role in communication, but this has been studied primarily within the context of mate choice (Møller, 1993; Swaddle, 1999a). In the present study, we examine the spontaneous response of naive chicks toward artificial prey to test the idea (Forsman and Merilaita, 1999; Kirkpatrick and Rosenthal, 1994; Møller and Swaddle, 1997) that asymmetry may negatively influence the protective value of avoidance-inducing color patterns. Several lines of evidence suggest that deviations from bilateral symmetry of signaling color pattern elements may influence the protective value of prey coloration. Thus, animals appear to be perceptually sensitive to very small deviations from bilateral symmetry (Møller, 1993; Schwabl and Delius, 1984; Swaddle, 1999a,b). Experiments on humans (Attneave, 1954), pigeons (Delius and Nowak, 1982), and honeybees (Horridge, 1996) further have shown that symmetrical patterns are easier to detect, learn, and reproduce C Address correspondence to A. Forsman. E-mail: anders.forsman@ hik.se. Received 23 August 2002; revised 3 February 2003; accepted 13 March 2003. from memory than are asymmetric ones. On the one hand, this raises the possibility that symmetry of visual warning signals may improve avoidance learning by predators (Forsman and Merilaita, 1999; Kirkpatrick and Rosenthal, 1994; Møller and Swaddle, 1997). On the other, if asymmetric signals are more difficult to perceive and elicit a weaker neural response compared with symmetric signals (Osorio, 1996), one would expect naive predators to show stronger unconditioned avoidance behaviors toward prey with symmetric color patterns. It has also been suggested that organisms may exhibit inherent general preferences for symmetric over asymmetric objects, although the empirical evidence is inconclusive (for review, see Swaddle, 1999a). Such general symmetry preferences would render predators more inclined to attack prey with symmetric than with asymmetric color patterns. To discriminate between these competing hypotheses, we conducted a series of experiments to study the effects of signal color and asymmetry on unconditioned aversions to conspicuously colored prey by presenting naive domestic chickens (Gallus gallus domesticus) to pairs of artificial paper ‘‘butterflies’’ (Forsman and Merilaita, 1999). Visual stimuli may be said to resemble ‘‘multicomponent’’ signals (Rowe, 1999) in the sense that different aspects of the stimulus are handled separately and may differently affect the perception of and response to the signal (Osorio et al., 1999). We therefore performed separate tests for effects of asymmetry in color, shape, and size of the signaling pattern elements. Several previous studies have shown that color influences the protective value of warning signals (see Coppinger, 1969, 1970; Osorio et al., 1999), but few attempts have been made to examine the role of color in symmetry perception (but see Morales and Pashler, 1999; Zhang and Gerbino, 1992). There is experimental evidence to suggest that the response of females to male secondary sexual ornaments is influenced by color asymmetry. For instance, Swaddle and Cuthill (1994) showed that female zebra finches choose symmetrically leg-banded males over asymmetrically banded ones. To our knowledge, Behavioral Ecology vol. 15 no. 1 International Society for Behavioral Ecology 2004; all rights reserved. Behavioral Ecology Vol. 15 No. 1 142 asymmetry detection that did incorporate an element of individual learning, Swaddle (1999b) showed that European starlings could accurately discriminate 1.8% asymmetry from symmetry but was unable to discriminate 1.25% asymmetric stimuli from symmetric patterns. In the present study, we test the hypothesis that there may exist an asymmetry threshold beyond which further increments in size asymmetry levels do not influence the magnitude of the initial response of naive predators to avoidance-inducing signals. METHODS Predators and artificial prey items Figure 1 Artificial paper butterflies presented to chicks in the six experiments used to test for effects on unconditioned predator aversions of color, shape, and asymmetry in color, shape, and size of pattern elements of conspicuously colored prey. The isolated figure at the right is an inset showing the dimensions of the butterflies. The size and shape of the butterflies and the position of the food crumb was identical in all experiments and trials. The figure shows the color, shape, and size of the signaling pattern elements of the two types of butterflies presented simultaneously to birds in each experiment and trial, and the number of birds used in each treatment. Numbers running along the base of the butterflies in experiment 6 represent the diameter (in mm) of the left and right spot. See text for details. however, no study has tested the hypothesis that color asymmetry may negatively influence the protective value of warning signals. With few exceptions (Merilaita, 1998; Summers and Clough, 2001), animal color patterns show a high degree of bilateral symmetry in shape of pattern elements (Wallace, 1889). This may reflect either developmental constraints or selection against asymmetric phenotypes (Forsman and Merilaita, 1999). In the present study, we test if the protective value of avoidance-inducing signals is impaired by shape asymmetry. A central question in understanding the role of asymmetry in the evolution of signaling is the perceptual capability of predators to detect and discriminate symmetric from asymmetric signals (Swaddle, 1999a,b). Because the mean asymmetry in natural populations typically constitutes only 1–2% of trait size, it has been suggested that most individuals may be perceived as symmetric (Swaddle, 1999a). In a recent experiment designed specifically to investigate the limits to As predators naive domestic chicks (G. gallus domesticus) were used on days 5–7 after hatching. Before trials, birds were housed by a commercial breeder in a 1200-m2 hangar at 30 C, 60% relative humidity, 21 h light/3 h dark photo-period, stocking density of 17 individuals/m2, and with food and water ad lib. The experimental arena consisted of a beige 40 3 30 3 40 cm plastic cage, the bottom of which was covered with sawdust. Artificial paper butterflies were used as prey (see Forsman and Merilaita, 1999). Butterflies were made from black paper and measured 40 mm wide, 30 mm high, and 20 mm ‘‘girth’’ (Figure 1). Each butterfly was affixed under a plastic Petri dish (diameter ¼ 55 mm). Inside the dish, one chick-starter food crumb (approximately 3–4 mm in diameter and beige to yellowish brown in color) was placed at the middle of the butterfly. The food crumbs used in the experiments were identical to those provided by the breeder as food for the birds when maintained in the hangar. Because the food crumb was always placed over the midbody of the black butterfly (Figure 1), it was equally conspicuous across the paired butterflies tested within each treatment. At the onset of a trial, two individual birds (to avoid stressing the birds) were transferred to the experimental cage containing the two types of butterflies (monochromatic black versus black with signaling colored pattern elements or symmetric versus asymmetric) placed side by side. A trial was terminated when one of the two birds had eaten or pecked at the food crumb on one of the two presented butterflies, in which case both birds were replaced to avoid confounding effects owing to social learning (Sherwin et al., 2002). Preliminary observations revealed that birds that had made no attempt to peck at the food crumb within the first 60 s were very unlikely to do so even if remaining in the experimental cage for up to 10 min. Trials were therefore also terminated if none of the birds had attacked the butterfly within 3 min, in which case one of the birds was replaced. The exact duration of trials was not recorded but ranged from 5–120 s. Each bird was allowed to choose only once and presented only two different butterflies. After the experiments, the birds were returned to the hangar. In the series of experiments described below each part was designed in the light of the results of the previous one. Experiment 1: testing for effects of spot color The first experiment was designed to test if the protective value of conspicuous coloration depends on the color of the signaling pattern elements. Birds were offered a choice between a monochrome black butterfly and a black butterfly with two signaling circular spots (diameter ¼ 9.49 mm). Birds were divided into four treatments according to spot color (Figure 1). The spot colors used in the four treatments were blue (Natural Colour System code B:1060), red (Y90R:1070), yellow (Y10R:1070), and white (pure white). For each treatment, 30 naive birds were used. We expect butterflies Forsman and Herrström • Signal asymmetry and predation risk with colored spots to suffer a lower rate of food-crumb removal compared with monochromatic butterflies, under the hypothesis that birds have stronger unconditioned aversions to more conspicuous color patterns. Experiment 2: testing for effects of spot color asymmetry To test the hypothesis that the protective value of warning signals is influenced by color asymmetry, a new set of birds were presented with two butterflies simultaneously, one with two spots of the same color and another with two spots of different color (Figure 1). Two different treatments were used, and 50 naive birds were used for each of the two treatments. In the first treatment, a butterfly with two red spots was presented together with a butterfly with one red spot and one yellow spot. In the second treatment, a butterfly with two white spots was presented together with a butterfly with one white spot and one yellow spot. These particular color combinations were chosen because the results from experiment 1 revealed that the birds showed significant aversions only toward yellow spots but tended to be attracted to white spots (see below). Thus, if the avoidance-inducing effect of yellow is stronger than the avoidance-weakening effect of color asymmetry, one would expect birds in the first treatment to avoid the asymmetric butterfly with one yellow spot and instead peck first at the symmetric butterfly with two red spots. By the same line of argument, one would expect birds in the second treatment to peck first at the symmetric butterfly with two white spots (to which they are attracted) and avoid the asymmetric butterfly with one yellow spot. In contrast, the hypothesis that avoidance-inducing signals are impaired by color asymmetry predicts symmetric butterflies to suffer a lower rate of food-crumb removal. Experiment 3: testing for effects of pattern shape This experiment was designed to test the hypothesis that efficacy of conspicuous coloration is influenced by the shape of the signaling pattern elements (Figure 1). Because the results of experiment 1 suggested that birds had significant unconditioned aversions only to black butterflies with yellow signals, we used yellow pattern elements in this and subsequent experiments. We tested the reactions of birds presented with two butterflies with signals consisting either two yellow circular spots (diameter ¼ 9.49 mm) or two yellow squares (8.41 3 8.41 mm), using 20 naive birds. To avoid confounding effects of signaling area (Forsman and Merilaita, 1999) the area of the yellow spots and squares was identical. Experiment 4: testing for effects of pattern shape asymmetry This experiment was designed to test the hypothesis that the efficacy of conspicuous coloration is negatively influenced by asymmetry in shape of the signaling pattern elements. For this purpose, the reactions of birds when offered a choice between a signal consisting of two yellow squares and a signal consisting of one yellow square and one yellow spot were tested, using 20 naive birds (Figure 1). This particular combination of shapes was chosen because the results from experiment 3 revealed that birds showed stronger aversions to circular spots than to squares (see below). Thus, if the stronger avoidance-inducing effect of spots compared with squares is more important than is the avoidance-weakening effect of shape asymmetry, one would expect birds to avoid the asymmetric butterfly with one spot and one square and peck first at the symmetric butterfly with two squares. In contrast, the hypothesis that shape asymmetry impairs the 143 protective value of warning signals predicts asymmetric butterflies to have the highest rate of food-crumb removal. Experiment 5: testing for effects of pattern shape asymmetry using novel signals Although predators may be naive in the sense that they are inexperienced, they are not naive in an evolutionary sense (Alatalo and Mappes, 1996). To reduce the problems associated with unlearned preferences and biases against certain colors and patterns, we tested for effects of shape asymmetry by using butterflies with novel pattern elements (crosses and bars) not normally found in nature. Two different treatments were used. In the first treatment, a butterfly with a signal consisting of one cross and one bar was presented together with a butterfly with a signal consisting of two crosses (Figure 1). In the second treatment, a signal consisting of one cross and one bar was presented with a signal consisting of two bars. In each treatment, 40 naive birds were used. Butterflies with one bar and one cross were expected to have the highest rate of food-crumb removal in both treatments, under the hypothesis that shape asymmetry impairs the protective value of conspicuous coloration. Experiment 6: testing for effects of pattern size asymmetry This final experiment was designed to test the hypothesis that the protective value of conspicuous coloration is impaired by asymmetries in the size of pattern elements, and to test the idea that there may exist a perceptual asymmetry threshold beyond which further increments in asymmetry levels do not influence the magnitude of the initial response of naive predators to avoidance-inducing signals. For this purpose a new set of naive birds was used. Each bird was simultaneously offered one butterfly with a symmetric signal consisting of two equal-sized yellow spots (diameter ¼ 9.49 mm) and one butterfly with an asymmetric signal consisting of two different-sized spots (Figure 1). Five different degrees of size asymmetry were used, ranging from 5–33% of mean spot size. To avoid confounding effects of signaling area (Forsman and Merilaita, 1999), spot sizes were chosen so that the combined area of the two yellow spots was identical in symmetric and asymmetric signals and in all treatments. The size (diameter, in mm) of large and small spots in the five treatments were as follows: 5% (9.6/8.7), 7.5% (9.9/8.4), 10% (10.0/8.2), 15% (10.5/7.7), and 33% (11.5/5.9). For each treatment, 80 naive birds were used. We expect butterflies with size-asymmetric signals to experience a higher overall rate of food-crumb removal. Moreover, if the relative aversion to symmetry is influenced by the difference in level of asymmetry between alternative prey, one would expect the relative rate of food-crumb removal from asymmetric butterflies to increase progressively across treatments, from the lowest to the highest degree of spot-size asymmetry. In contrast, the hypothesis of an asymmetry detection and response threshold predicts the variation in relative rate of food-crumb removal among treatments to follow a stepfunction. RESULTS Although the behavioral response varied among birds presented with different kinds of butterflies, all birds seemed intimidated and hesitated before pecking at the food crumb, regardless of the color and shape of the signaling pattern elements of the butterfly. Despite the large numbers of individuals used, all birds pecked first at the food crumb and never at the signaling pattern elements. Behavioral Ecology Vol. 15 No. 1 144 Figure 2 The proportion of artificial black butterflies with visual warning signals of different colors (blue, red, white, or yellow) attacked first when presented together with a monochromatic black butterfly in experiment 1. The dotted reference line indicates equal rate of foodcrumb removal from monochromatic and spotted butterflies. Symbols above bars indicate results from two-tailed binomial test of a difference from the randomly expected proportion eaten. n.s. indicates not significant; s, .05 , p , .10; *p , .05; **p , .01. Figure 3 Comparison of rate of food-crumb removal for color asymmetric butterflies with two signaling spots of different color presented together with color symmetric butterflies with a signal consisting of two spots of the same color. Results are from experiment 2. The symbols along the abscissa indicate the shape and color (R indicates red; W, white; and Y, yellow) of signaling pattern elements on each of the two prey types presented together. Symbols above bars indicate results from two-tailed binomial tests for large samples with correction for continuity. n.s. indicates not significant. *p , .05. Effects of spot color When birds were offered a choice between a monochromatic black butterfly presented and a black butterfly with two signaling spots, there were clear differences in the response among birds presented with signals of different colors (v2 ¼ 13.53, df ¼ 3, p ¼ .004) (Figure 2). Butterflies with blue or red signaling spots suffered the same rate of food-crumb removal as that of monochromatic butterflies (two-tailed binomial test of a difference from the randomly expected proportion eaten, that is, 50%: blue, p . .9; red, p ¼ .86), whereas butterflies with white spots suffered a somewhat, but not significantly, higher rate of removal than did monochromatic butterflies without spots (p ¼ .098). Birds significantly avoided only butterflies with two yellow spots (p ¼ .002). Effects of color asymmetry When birds were offered a choice between a color-symmetric butterfly with two red spots versus an asymmetric butterfly with one red and one yellow spot, both kinds experienced the same rate of food-crumb removal (two-tailed binomial test for large samples with correction for continuity, z ¼ 0.16, p ¼ .87) (Figure 3). In contrast, butterflies with one white and one yellow spot had a significantly higher rate of removal than did butterflies with two white spots (z ¼ 2.12, p ¼ .034) (Figure 3). Effects of shape and shape asymmetry Results from experiment 3 and 4 revealed that birds responded differently to the butterflies depending on the shape of the signaling pattern elements, and suggest that the protective value of conspicuous coloration is impaired by shape asymmetry. Thus, butterflies with a signal consisting of two circular pattern elements experienced a lower rate of food-crumb removal than did butterflies with two squares (two-tailed binomial test, p ¼ .012) (Figure 4A). Despite the stronger aversion induced by circular compared with squared pattern elements, however, shape-asymmetric butterflies with one spot and one square had a significantly higher rate of food-crumb removal than did symmetric butterflies with two squares (p ¼ .002) (Figure 4A). Effects of shape asymmetry as revealed by using novel signals A negative effect of shape asymmetry on the avoidanceinducing effect of the signal was also evident in the response of birds presented butterflies with signals not normally found in nature. Thus, butterflies with asymmetric signals consisting of one cross and one bar suffered a higher rate of food crumb removal compared with that of symmetric butterflies with either two crosses (binomial test, z ¼ 2.06, p ¼ .039) or two bars (z ¼ 2.37, p ¼ .0017) (Figure 4B). Perceptual size asymmetry threshold The response of birds presented with one butterfly with two equal-sized spots and one butterfly with an asymmetric signal consisting of two different-sized spots suggests that the protective value of avoidance-inducing signals is impaired by asymmetry in the size of pattern elements. In the treatment with a mean spot-size asymmetry of 5%, butterflies with asymmetric signals were attacked at the same rate as those with symmetric signals (two-tailed binomial test, z ¼ 0.22, p ¼ .91), whereas the size-asymmetric butterflies suffered a significantly higher rate of food-crumb removal than did symmetric butterflies in all treatments with a spot-size asymmetry of 7.5% or more (7.5%: z ¼ 2.80, p ¼ .0038; 10%: z ¼ 3.24, p ¼ .0014; 15%: z ¼ 3.46, p ¼ .0006; 33%: z ¼ 3.02, p ¼ .0026) (Figure 5). The relative rate of food-crumb removal from butterflies with size-asymmetric signals did not increase progressively across treatments with increasing degree of asymmetry (rs ¼ .45, n ¼ 5 treatments, p ¼ .45) but instead followed a step function (Figure 5), a finding consistent with expectations from the hypothesis of a threshold for asymmetry detection and response beyond which further increments in asymmetry levels have no influence on signal efficacy. DISCUSSION In summary, birds showed stronger spontaneous aversions to artificial butterflies possessing symmetric coloration than to Forsman and Herrström • Signal asymmetry and predation risk 145 Figure 5 Proportion of asymmetric signaling butterflies with one large and one small yellow spot from which the food crumbs were first removed when presented together with symmetric signaling butterflies with two equal-sized yellow spots in experiment 6. The values along the abscissa are degree of bilateral spot-size asymmetry, expressed as a percentage of spot diameter. They represent a continuum from nearly symmetric (far left) to very asymmetric (far right). The dotted reference line indicates equal rate of food-crumb removal for the symmetric and asymmetric signal type. Symbols above bars indicate results from twotailed binomial tests for large samples. n.s. indicates not significant. **p , .01; ***p , .001. Figure 4 Rate of food-crumb removal for signaling butterflies with different shapes of the signaling pattern elements. The symbols along the abscissa indicate the shapes of the two signaling pattern elements on each of the two prey types presented together. (a) The two bars to the left indicate the comparison between two symmetric signal types used in experiment 3. The two right bars indicate the comparison between one symmetric and one asymmetric signal used in experiment 4. (b) Comparison of rate of food-crumb removal for prey with asymmetric signals consisting of one cross and one bar when presented together with symmetric signals consisting either of two crosses (two left bars) or two bars (right bars). Results are from experiment 5. Symbols above bars indicate results from two-tailed binomial tests. *p , .05; **p , .01. butterflies with asymmetries in color, shape, or size of the signaling pattern elements. These findings are consistent with the hypothesis (Forsman and Merilaita, 1999; Kirkpatrick and Rosenthal, 1994; Møller and Swaddle, 1997) that the protective value of conspicuous prey color patterns is impaired by asymmetry. The response of birds presented simultaneously with a monochromatic butterfly and a butterfly with signaling spots differed depending on the color of the spots. Although birds showed no aversion to butterflies with blue or red spots, they avoided yellow spots and showed a weak attraction (albeit not statistically significant) to white spots. These findings highlight the variable influence of different colors on unconditioned aversions (see also Coppinger, 1969; 1970; Osorio et al., 1999). The unconditioned avoidance of butterflies with yellow spots further suggests that a deviant mutant, such as a conspicuously colored prey, does not necessarily suffer high predation from naive predators as a result of an increased probability of detection (Endler, 1991; Götmark, 1994; Lindström et al., 1999). Although experiment 1 uncovered variable effects on unconditioned aversion of color per se, color symmetry seems to be even more important. Thus, in the second treatment of experiment 2, butterflies with one white spot and one yellow spot suffered a significantly higher rate of removal than did butterflies with two white spots. Because the results from experiment 1 revealed that birds significantly avoided yellow but not white spots, the latter result must reflect a negative effect of color asymmetry on avoidance-inducing signaling, rather than an effect of color per se. In the first treatment, butterflies with one red spot and one yellow spot were attacked at the same rate as were color symmetric butterflies with two red spots. Viewed alone, this result may be owing to equal aversion (or attraction) to red and yellow. However, given that the birds in experiment 1 avoided yellow but not red spots, the finding that birds in experiment 2 showed no aversion to butterflies with one yellow spot and one red spot is consistent with the hypothesis that color asymmetry impairs the protective value of conspicuous coloration. In a previous investigation of the role of color asymmetry on sexual signaling, Swaddle and Cuthill (1994) found that female zebra finches chose color-symmetric leg-banded males over color-asymmetric banded ones (see also Fiske and Amundsen, 1997). Thus, although color symmetry seems to make warning signals more aversive, it appears to increase the attractiveness of secondary sexual ornaments. Taken together, these findings argue against the notion (for review, see Swaddle, 1999a) that organisms have inherent general preferences for symmetry over asymmetry, and suggest instead that receivers’ responses to color-asymmetric signals are context specific. The finding that butterflies with a signal consisting of two circular pattern elements experienced a lower rate of foodcrumb removal than did butterflies with two squares may reflect a stronger avoidance of more eyelike objects (Coppinger, 1969, 1970; Tinbergen, 1974). Despite the stronger aversion induced by circular compared with squared pattern elements, however, shape-asymmetric butterflies with one spot and one square suffered a significantly higher rate of foodcrumb removal than did symmetric butterflies with two squares. A similar negative effect of shape asymmetry on avoidance was evident among birds presented with novel pattern elements not normally found in nature. It thus seems that the protective value of conspicuous coloration is impaired not only by color asymmetry but also by asymmetries Behavioral Ecology Vol. 15 No. 1 146 in shape of the signaling pattern elements. This latter finding suggests that the high degree of bilateral symmetry in shape of color pattern elements characteristic of most species of animals (Wallace 1889) may reflect not only developmental constraints but also selection against asymmetric phenotypes imposed by visually oriented predators (Forsman and Merilaita, 1999). The mean asymmetry in natural populations typically constitutes only 1–2% of trait size, and it has been suggested that most individuals may be perceived as symmetric and that asymmetry may play no role in signaling (Swaddle, 1999a). However, the degree of asymmetry varies both among characters and individuals. For instance, analyses of asymmetry in color pattern elements on the wings of three species of moths (Arctia caja L., Noctua orbona (L., Smerinthus ocellata L.) showed that mean asymmetries constituted 4.3% (range ¼ 2.1–7.0%) of trait size, whereas individual asymmetry levels reached as high as 26% (Forsman and Merilaita, 2003). Swaddle (1999b) showed that European starlings could accurately discriminate 1.8% asymmetry from symmetry, but were unable to discriminate 1.25% asymmetric stimuli from symmetric patterns. Similarly, we found that birds attacked butterflies with a spot-size asymmetry of 5% at the same rate as those with symmetric signals, whereas the size-asymmetric butterflies suffered a significantly higher rate of food-crumb removal than did symmetric butterflies in all treatments with a spot-size asymmetry of 7.5% or more. Interestingly, birds’ avoidance of symmetric signals did not increase progressively across treatments as a function of the degree of spot-size asymmetry, but the variation among treatments instead followed a step-function. These results are consistent with the idea (Schwabl and Delius, 1984; Swaddle, 1999b) of a perceptual asymmetry threshold beyond which increased asymmetry levels have no or little influence on signal efficacy. Because our experiments measured unconditioned avoidance only, with no learning being involved (unlike the experimental design used by Swaddle [1999b], which did incorporate learning), the observed threshold in our experiment may be an overestimate of the true detection threshold and represent instead a response threshold. Collectively, these results nevertheless suggest that at least in some species of butterflies and moths, the magnitude of individual differences in size asymmetry of color patterns exceeds the asymmetry detection threshold in birds and, hence, may be influenced by selective predation. Conspicuous coloration typically occurs in animals that are distasteful, noxious, or otherwise dangerous to potential predators (Cott, 1940; Endler, 1991; Gittleman and Harvey, 1980; Guilford and Dawkins, 1991; Summers and Clough, 2001; Wallace, 1867). However, we measured the spontaneous and initial response of naive predators, with no learning or unpalatability being involved. The stronger aversions toward symmetric color patterns as revealed by our experiments thus suggest that a high degree of symmetry in color, shape, and size of color pattern elements may decrease susceptibility of conspicuous prey (even if they have no chemical or structural defense mechanisms) to inexperienced predators, as well as enhance the survival prospects of rare mutant aposematic phenotypes. The influence of asymmetry on the evolution of conspicuous color patterns will be even greater if it proves to negatively influence also the ability of predators to learn and associate visual warning signals with unpalatability. We thank Ingemar J. for immense help with the birds, and J. Ahnesjö, E. Civantos, L. Lindström, and S. Ulfstrand for helpful comments on an earlier version of the manuscript. The study was supported by the Swedish Natural Science Research Council (grant to A.F.), Växjö University, and Kalmar University. REFERENCES Alatalo RV, Mappes J, 1996. Tracking the evolution of warning signals. Nature 382:708–709. Attneave F, 1954. Some informational aspects of visual perception. Psychol Rev 61:183–193. Coppinger RP, 1969. The effect of experience and novelty on avian feeding behaviors with reference to the evolution of warning coloration in butterflies, part I: reactions of wild-caught adult blue jays to novel insects. Behaviour 35:45–49. Coppinger RP, 1970. The effect of experience and novelty on avian feeding behaviors with reference to the evolution of warning coloration in butterflies, part II: reactions of naive birds to novel insects. Am Nat 104:323–335. Cott HB, 1940. Adaptive coloration in animals. London: Methuen. Delius JD, Nowak B, 1982. Visual symmetry recognition by pigeons. Psychol Res 44:199–212. Endler JA, 1991. Interactions between predators and prey. In: Behavioural ecology: an evolutionary approach, 3rd ed (Krebs JR, Davies NB, eds). Oxford: Blackwell; 169–196. Fiske P, Amundsen T, 1997. Female bluethroats prefer males with symmetric colour bands. Anim Behav 54:81–87. Forsman A, Merilaita S, 1999. Fearful symmetry: pattern size and asymmetry affects aposematic signal efficacy. Evol Ecol 13:131– 140. Forsman A, Merilaita S, 2003. Fearful symmetry? Intra-individual comparisons of asymmetry in cryptic vs. signaling colour patterns in butterflies. Evol Ecol 17:491–507. Gittleman JL, Harvey PH, 1980. Why are distasteful prey not cryptic? Nature 286:149–150. Götmark F, 1994. Does a novel bright color patch increase or decrease predation? Red wings reduce predation risk in European blackbirds Proc R Soc Lond B 256:83–87. Guilford T, Dawkins MS, 1991. Receiver psychology and the evolution of animal signals. Anim Behav 42:1–14. Horridge GA, 1996. The honeybee (Apis mellifera) detects bilateral symmetry and discriminates its axis. J Insect Phys 42: 755–764. Kirkpatrick M, Rosenthal GG, 1994. Symmetry without fear. Nature 372:134–135. Lindström L, Alatalo RV, Mappes J, Riipi M, Vertainen L, 1999. Can aposematic signals evolve by gradual change? Nature 397:249– 251. Merilaita S, 1998. Crypsis through disruptive coloration in an isopod. Proc R Soc Lond B 265:1059–1064. Møller AP, 1993. Female preference for apparently symmetrical male sexual ornaments in the barn swallow Hirundo rustica. Behav Ecol Sociobiol 32:371–376. Møller AP, Swaddle JP, 1997. Asymmetry, developmental stability and evolution. Oxford: Oxford University Press. Morales D, Pashler H, 1999. No role for colour in symmetry perception. Nature 399:115–116. Osorio D, 1996. Symmetry detection by categorization of spatial phase: a model. Proc R Soc Lond B 263:105–110. Osorio D, Miklosi A, Gonda Z, 1999. Visual ecology and perception of coloration patterns by domestic chicks. Evol Ecol 13:673– 689. Rowe C, 1999. Receiver psychology and the evolution of multicomponent signals. Anim Behav 58:921–931. Schwabl U, Delius JD, 1984. Visual bar length discrimination threshold in the pigeon. Bird Behav 5:118–121. Sherwin CM, Heyes CM, Nicol CJ, 2002. Social learning influences the preferences of domestic hens for novel food. Anim Behav 63: 933–942. Summers K, Clough ME, 2001. The evolution of coloration and toxicity in the poison frog family (Dendrobatidae). Proc Natl Acad Sci USA 98:6227–6232. Swaddle JP, 1999a. Is fluctuating asymmetry a visual signal? In: Animal signals: signalling and signal design in animal communication (Espmark Y, Amundsen T, Rosenqvist G, eds). Trondheim: Tapir Academic Press; 383–403. Forsman and Herrström • Signal asymmetry and predation risk Swaddle JP, 1999b. Limits to length asymmetry detection in starlings: implications for biological signalling. Proc R Soc Lond B 266: 1299–1303. Swaddle JP, Cuthill IC, 1994. Preference for symmetrical males by female zebra finches. Nature 367:165–166. Tinbergen N, 1974. Curious naturalists. Harmondsworth: Penguin Education. 147 Wallace AR, 1867. Transactions of the Entomological Society of London. 3rd. Series 3, lxxx–lxxxi. Wallace AR, 1889. Darwinism. London: Macmillan. Yashi S, Higashi M, 1998. The evolution of warning signals. Nature 394:882–884. Zhang L, Gerbino W, 1992. Symmetry in opposite-contrast dot patterns. Perception 21(suppl. 2):95.
© Copyright 2026 Paperzz