Importance of internal pattern contrast and contrast against the background in aposematic signals

Transcription

1 Behavioral Ecology doi: /beheco/arp141 Advance Access publication 2 November 2009 Importance of internal pattern contrast and contrast against the background in aposematic signals Marianne Aronsson and Gabriella Gamberale-Stille Department of Zoology, Stockholm University, Stockholm, Sweden Aposematic color patterns that signal prey unprofitability are suggested to work best when there is high contrast within the animal color pattern or between the animal and its background. Studies show that prey contrast against the background increases the signal efficiency. This has occasionally been extended to also explain the presence of internal patterns. We used domestic chicks, Gallus gallus domesticus, to investigate the relative importance for avoidance learning of within-prey pattern contrast and prey contrast against the background. In a series of trials, birds were first trained to avoid artificially made aposematic mealworms that were plain red or red with black stripes, and to discriminate them from palatable brown mealworms, on either a red or a brown background. Second, we investigated how the birds generalized between striped and nonstriped prey. The chicks showed faster avoidance learning when the basic color of the aposematic prey (red) contrasted with the background color (brown). However, there was no similar effect of internal pattern contrast. The generalization test showed a complete generalization between the nonstriped and the striped prey. We conclude that contrasting internal patterns do not necessarily affect predator avoidance learning the same way as shown for prey-to-background contrast in aposematic prey. Key words: avoidance learning, conspicuousness, domestic chick, generalization behavior, internal pattern contrast, signal design, warning coloration. [Behav Ecol 20: (2009)] There are several different strategies used in animal communication, both within and between species. Some use visual signals, whereas others make use of audio and olfactory signals (Guilford 1990; Searcy and Nowicki 2005). Most studies on aposematism have focused on the concept of warning coloration where animals signal their unprofitability to potential predators by having a conspicuous color pattern (Guilford 1990). It has been argued that the function of this conspicuousness is to increase signal efficiency, because warning signals, as all signals, benefit by being easy to detect, discriminate, and memorize by the intended receivers (Guilford and Dawkins 1991). Aposematic animals often have conspicuous markings of bright colors such as red, orange, or yellow, often in combination with a black pattern (Cott 1940). Having such bright hues that contrast strongly against a typical green or brown background makes them easy to recognize and discriminate both from the background and from the palatable and camouflaged prey that the predators usually hunt (Turner 1975; Sherratt and Beatty 2003). An aposematic pattern is suggested to work best when it shows a high contrast within the color pattern itself or a high contrast against the background (Endler 1991). There are many experimental studies suggesting that the conspicuousness of aposematic prey, specifically in the form of contrast against the background, increases signal efficiency (Ruxton et al. 2004). Unpalatable prey items that contrast against the background have been shown to increase initial wariness (Roper and Cook 1989; Lindström et al. 2001), increase the speed and durability of avoidance learning (Gittleman and Harvey 1980; Gittleman et al. 1980; Roper and Wistow 1986; Address correspondence to G. Gamberale-Stille. gabriella. gamberale@zoologi.su.se. Received 18 February 2009; revised 1 September 2009; accepted 19 September Ó The Author Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please journals.permissions@oxfordjournals.org Roper and Redston 1987; Roper and Cook 1989; Riipi et al. 2001), and decrease recognition errors (Gamberale-Stille 2001). However, these effects of background contrast have been suggested to also explain the existence of contrasting internal color patterns (Guilford 1990). Guilford (1990) argues that prey by producing its own internal color boundaries of high contrast will promote conspicuousness and thus the aposematic effect. Still, to our knowledge, there is little experimental evidence for such a corresponding effect of conspicuousness generated by internal pattern contrasts. However, one recent study with domestic chicks as predators implies that the presence of a black striped pattern may improve avoidance learning when presented together with some specific colors (green) but not with others (yellow) (Hauglund et al. 2006). Additionally, a study by Mason (1988) suggests that a more complex stimulus, consisting of both a black pattern and a conspicuous color, may produce a more durable avoidance in wild-caught red-winged blackbirds. That internal pattern contrast is equivalent to increased conspicuousness is an untested conclusion. There are other hypotheses that conflict with this idea. For instance, Papageorgis (1975) suggested that patterns makes an animal look cryptic at longer distance and at the same time remain conspicuous at a shorter distance. In the case of disruptive cryptic coloration, contrasting patterns may be used to camouflage the body shape and thus decrease conspicuousness (Cott 1940; Edmunds 1974), and Kassarov (2003) argues that a prey can reduce its conspicuousness by having a pattern, because stripes, for instance, blur together when looked at from a distance. In snakes, stripes are thought to confuse the predator because they make it difficult to visually focus on any specific point along the animal and thereby making it difficult to tell the direction and speed of the animal when moving (Wolf and Werner 1994; Creer 2005). In connection to this, some recent studies suggest that birds may need color to recognize prey as aposematic and that they may disregard the presence of a black pattern (Exnerová et al. 2006;

2 Aronsson and Gamberale-Stille Contrasting patterns in aposematism 1357 Aronsson and Gamberale-Stille 2008). If this is true, taken together with the fact that there is an effect of signal size on bird avoidance of an aposematic prey (Gamberale and Tullberg 1996, 1998; Forsman and Merilaita 1999; Gamberale- Stille 2000) and that the size of a signaling area affects detectability (Robson and Graham 1981; Polat and Tyler 1999), the presence of a pattern may potentially decrease the aposematic effect, and/or conspicuousness. The aim of the present study is to investigate the separate effects of prey-to-background contrast and within-prey pattern contrast on avoidance learning in domestic chicks (Gallus gallus domesticus). This in order to determine if one can generalize positive aposematic effects of conspicuousness generated by contrast against the background to the presence of internal color patterns. In a series of avoidance learning trials, birds are trained to discriminate aposematic plain red or red-and-black striped mealworms (Tenebrio molitor) from palatable brown mealworms either on a red or on a brown background. This experimental setup is similar to that of Gittleman et al. (1980) in the sense that in both studies, birds were presented with a situation where either the color of the palatable or the unpalatable prey matched or contrasted with the background. Thus, if contrast against the background increases avoidance learning, we expect the chicks presented with unpalatable prey on a contrasting backgroundtolearnfasterthanbirdspresentedwithunpalatable prey on a matching background. Likewise, if internal color patterns increase avoidance learning, we expect birds presented with striped unpalatable prey to learn faster than birds presented with nonstriped unpalatable prey, especially on the background where the color of the unpalatable prey otherwise would match the background. After avoidance learning, we also conduct a generalization test to investigate how the birds generalize between striped and nonstriped prey. In an additional control experiment, we also investigate if the black pattern used in the above mentioned experiment is visible to domestic chicks. It is important for the interpretation of the results to know that the pattern used in the experiment is actually perceivable by the birds. MATERIALS AND METHODS Predators We used a total of 60 male domestic chicks as predators that arrived newly hatched and unfed from a commercial hatchery in batches of about 10 individuals. The chicks were housed in cages ( cm, length 3 width 3 height) with the floor covered with wood chips and heated with a 60 W carbon filament lamp. The housing and experimental room was lit by daylight from the windows, by daylight-strip lights (BIOLUX L36W/72, OSRAM, München, Germany, now very low in UV), and 4 fluorescent strip-lights (L36W/73, OSRAM, emits light between 350 and 420 nm), between 08:00 and 18:00 h. They were fed chick Starter crumbs (Pullfor, Lantmannen, Stockholm, Sweden) and water ad lib. On the day of arrival, the chicks were individually marked with numbers on their backs using a black marker pen. The markings did not cause any increased pecking by other individuals on this area. Prey We used painted dead mealworms as prey, placed on colored paper backgrounds. Prey were painted using nontoxic children s paint (Gouache Tempera, Color & Co, Lefranc and Bourgeois, Le Mans, France). The mealworms were first killed by freezing them for a couple of hours, after which about half of them were painted brown ( Burnt Umber ) and then returned to the freezer. These mealworms were palatable to the chicks. The remaining mealworms were first made distasteful by injecting 0.02 ml of 3% solution of quinine hydrochloride after which they were painted red ( Brilliant Red ) and returned to the freezer. We also painted half of the red mealworms with 2 black stripes ( Black ) across them. The stripes were painted so that the red-black-red-black-red areas were of approximately the same size (Figure 1). Finally, to increase the distastefulness of both the nonstriped and striped mealworms they were painted with 3 drops of Stop n Grow (Mentholatum, East Kilbride, Scotland). Stop n Grow is a bitter tasting, nontoxic, solution used to stop nail-biting and contains the bitter substances denatonium benzoate and quinine sulfate, both of which have been used in previous studies to make prey unpalatable to birds (e.g., Marples and Roper 1997; Skelhorn and Rowe 2005). The mealworms were collected from the freezer about half an hour before each trial to make sure that they were unfrozen and that the paint was dry. The background paper that covered the bottom of the Petri dishes around the runway were of 2 different colors, brown and red, and were chosen to match the colors of the painted mealworms as much as possible (Figure 2). We measured the reflectance of the paints and backgrounds as the percentage of the reflectance of a white standard from the painted mealworms (with Stop n Grow) and background papers using an Ocean Optics (Dunedin, FL) USB 2000 spectrophotometer and PX-2 pulsed xenon light source (Figure 2). This to make sure that the red paint was more similar to the red background paper than to the brown paper as well as the other way around and to control for possible UV reflection. Experimental arena The preexperimental training and the experiment took place in a square shaped arena ( cm high) made of wood, with a 25-cm-wide runway surrounding a smaller enclosure with net walls. This enclosure was used during the preexperimental training for companion chicks. There were 16 circular wells, 4 cm in diameter, sunk about 1.5 cm into the floor, and spaced uniformly along the runway. These wells contained Petri dishes in which the prey were presented on differently colored backgrounds. The experimental arena was situated in the same room as the housing cages. Preexperimental training Before the experiment started the chicks were trained to forage alone for mealworms in the wells along the experimental runway. First, on the day of arrival the chicks received live mealworms in their home cages to get familiarized with the prey. The training in the arena started on the day after arrival and consisted of 5 training sessions distributed more than 2 days. Before the first session, all chicks were divided into 2 groups where half would be trained to forage for prey on brown background and the other half on red background. Throughout the preexperimental training, the avoidance learning trials and the generalization test, each chick only encountered one of the 2 possible background colors. In the first session, groups of 3 chicks at a time were allowed into the arena to forage for live mealworms that were presented 3 in each well. The rest of the birds in the batch that would encounter the same background color acted as companions during this first session and were kept in the companion enclosure of the experimental arena with food and water ad lib. During the afternoon of the first day of arena training, there were 2 consecutive sessions where birds in groups of 5 were allowed to forage

3 1358 Behavioral Ecology alone without companions. Between the morning and afternoon sessions, all chicks were left undisturbed in their home cage for a minimum of 1 h, when no experiments at all took place. Figure 1 The experimental setup. There were 4 treatment groups (column 1), 2 of which received prey on a brown background and 2 on a red background (column 2). The experimental subjects participated in 5 trials of avoidance learning (column 3) with brown palatable prey (S1) and either red nonstriped or red-and-black-striped unpalatable prey (S2). The numbers of prey items encountered in each trial of avoidance learning is given underneath the prey in column 3. After the avoidance learning, birds took part in a generalization test (column 4) where all chicks were presented all 3 kinds of prey (all palatable); brown, red nonstriped, and red and black striped on their respective background color. The numbers of prey present in the generalization test is given to the right of each prey type. together along the runway. During these sessions, and the rest of the training in the arena, the chicks were always presented with single dead, brown painted mealworms in each well. As an addition, at the end of the first training day the chicks also received brown painted mealworms in their home cage to get even more familiarized with the palatable prey used in the experiment. On the second day of arena training there were 2 sessions. During these sessions the chicks were always alone in the runway and in the first session there were 2 chicks in the companion enclosure, but in the last session there were no companions present at all. Additionally, in this last training trial, and for the remainder of the experiment, the walls of the runway that previously had allowed birds to view into the companion enclosure were now covered with plywood boards. After the preexperimental training, all chicks had learned to forage Experimental procedure Before the start of the experiment, the chicks in the 2 background groups (Br and R) were subdivided into 2 new treatment groups each; nonstriped (ns) and striped (S). Thus, there were 4 experimental groups in the experiment, brown/nonstriped (Br/nS), brown/striped (Br/S), red/nonstriped (R/nS), and finally red/striped (R/S) (Figure 1). Each batch of chicks was evenly divided between treatment groups with the final sample sizes of 15 per group (N ¼ 15). The experiment consisted of 2 parts, starting with avoidance learning for 5 trials and then a generalization test under extinction conditions. The experiment started on the fourth day with one trial in the afternoon of each consecutive day. There was no food deprivation before the trials. During the avoidance learning, half of the 16 wells had the positive stimulus with a palatable mealworm, and the other half had a negative stimulus with a distasteful mealworm. The palatable and the distasteful prey were evenly distributed along the wells of the runway, with no more than 2 of the same prey type in succession. Each trial session started at a different position along the runway. The trial ended when the chick passed the 16th well and the number of mealworms attacked, eaten, or pecked at was noted for each trial. After the last avoidance learning trial, all chicks took part in a generalization test, under extinction conditions, to investigate if the presence of a pattern in the aposematic prey affected the chicks generalization behavior. The chicks were presented with both nonstriped and striped mealworms as well as the brown that they knew as palatable; however, inthistestall prey werepalatable. Half of the mealworms were brown, and half were red with one quarter nonstriped and one quarter striped. The background colors were the same as experienced during avoidance learning. Chicks were allowed one round around the arena, and we recorded the number of each prey type that was attacked. Detection test To make sure that the chicks were able to detect the black stripes on the mealworms an additional test was performed with 11 new chicks. The test took part in a small arena made out of a screened-off part of a housing cage measuring cm. The floor of the cage was covered with white paper, and in the middle of the floor was a cm red painted rectangle situated. The paint was the same as was used on the mealworms in the experiment. First, we accustomed the chicks to the arena by presenting them in pairs with a few live mealworms placed on the red rectangle. This was done twice for all chicks. After that, we added 5 pairs of black stripes distributed evenly over the area of the red rectangle. The pairs of stripes were made to mimic the stripes of the experimental mealworms and were of the same size and at the same distance to each other. The chicks were tested singly and were first presented 2 live mealworms placed on the rectangle, to get them to start foraging. For the following 2 minutes, we registered the number of pecks inside the red area, and we noted especially whether the pecks were directed to the black patterns or not. Figure 2 The reflectance spectra from the brown and red background papers and paints that we used in the experiment. The measurements from the paints are when covered with Stop n Grow. Data analysis For the avoidance learning, the proportion palatable prey attacked in each trial was calculated as (palatable prey attacked)/(all prey attacked). The data were transformed

4 Aronsson and Gamberale-Stille Contrasting patterns in aposematism 1359 (X^[20.947]) to fit the assumptions of parametric tests and were compared using repeated measures analysis of variance (ANOVA), with trial number as within factor and background color and prey pattern as between factors. For the generalization test, we calculated the proportion of the presented prey of each type that were attacked as (number of brown attacked)/8, (number of nonstriped attacked)/4, or (number of striped attacked)/4. We used Friedman tests to investigate general differences in attack probability on the 3 prey types within each treatment group, followed by Wilcoxon signed-rank tests for more detailed comparisons. We also used a Mann Whitney U test to compare attacks between the 2 background colors. Additionally, we compared the attack probability of the novel prey pattern in the generalization test with the attack probability of the same pattern during the first trial of avoidance training (i.e., how the birds that were to be trained with that same pattern reacted to it on the first encounter when it was novel to them). These data were compared with a Mann Whitney U test. For the detection test, we used a Sign Test to compare the number of pecks delivered on the stripes to the number of pecks on other areas within the red rectangle to determine the visibility of the black pattern used in the experiment. RESULTS Although both types of aposematic prey in the experiment (red and red with stripes) were initially novel to the birds and had classical warning colors, all birds except one (from group R/ ns) attacked the prey on first encounter. Thus, the birds in this experiment showed no initial aversion toward the aposematic prey. Furthermore, chicks in all groups, irrespective of background color or the presence or absence of stripes on the prey, learned to some degree to avoid the unpalatable prey and to discriminate it from the palatable prey (Figure 3, Table 1). The repeated measures ANOVA shows a significant effect of background color, indicating a difference in the degree of avoidance learning; chicks presented with prey on a brown background, matching the palatable prey and contrasting with the unpalatable prey, learned faster compared with those trained on the red background color that matched the unpalatable prey (Figure 3 and Table 1). There was, however, no general effect of pattern, suggesting that the presence or absence of stripes did not affect the degree to which the birds learned to avoid the unpalatable prey in the experiment. There was also a significant effect of trial, showing that the birds in general learned to avoid attacking the unpalatable prey, and a significant interaction between trial and pattern (Table 1). The latter suggests different shapes of the learning curves depending on the presence/absence of a black pattern. Figure 3 shows that the major difference between the learning curves occurs in the second trial. There, birds trained with red prey without a black pattern (especially on brown background) show a greater avoidance of the aposematic prey than do birds trained with striped prey, and this difference is significant (Post hoc test, Fisher Least Significant Difference, P ¼ 0.017). However, this benefit for the nonstriped prey is already gone in the next trial. The result from the generalization test is shown in Figure 4, and there is a clear difference in all treatment groups in attack probability of the 3 prey types (Freidman tests, degrees of freedom ¼ 2; Br/nS, v 2 ¼ 22.53, P, ; Br/S, v 2 ¼ 22.8, P, ; R/nS, v 2 ¼ 8.93, P ¼ 0.012; R/S, v 2 ¼ 9.03, P ¼ 0.011). In all 4 groups, there is no significant difference between the 2 warningly colored prey (Wilcoxon signed-rank test, Br/nS, Z ¼ 20.54, P ¼ 0.593; Br/S, Z ¼ 21.21, P ¼ 0.225; R/nS, Z ¼ , P ¼ 0.515; R/S, Z ¼ , P ¼ 0.753), suggesting a complete Figure 3 The proportion of all attacked prey that were palatable in each of 5 trials of avoidance learning for the 4 treatment groups (mean 6 standard error). Circles represent brown background color, and triangles represent red background color. Solid lines and symbols represent nonstriped prey, and broken lines and empty symbols represent striped prey. generalization over pattern in this study. There is, however, a difference between the proportion palatable brown prey attacked compared with the warningly colored prey (Wilcoxon signed-rank test, Br/(nS 1 S) pooled, Z ¼ 24.78, P, ; R/ (ns 1 S) pooled, Z ¼ 24.21, P, ), and there is a difference in avoidance of the warningly colored prey between the 2 background colors (Mann Whitney U test, Br vs. R: U ¼ , P, ). This means that the generalization test not only supports the results from the avoidance learning with respect to the higher effect of background contrast in avoidance of aposematic prey but also adds the information that chicks did not differentiate between the prey types with or without the contrasting internal pattern (Figure 4). It could be argued that the avoidance of the novel warningly colored prey in the generalization test may not be due to generalization but to neophobia. To investigate this possibility, we Table 1 Results of a repeated measures ANOVA where prey background (red or brown) and pattern (red prey with or without stripes) were the between factors and trial number (1 5) were the within factors Effects SS df MS F P Background < Pattern Background 3 Pattern Error Trial < Trial 3 background Trial 3 pattern Trial 3 background 3 pattern Error Bold text indicates significance level P, 0.05; df, degrees of freedom; SS, sum of squares; MS, mean square.

5 1360 Behavioral Ecology Figure 4 The proportion of presented prey of each type that was attacked (mean 6 standard error) during the generalization test by chicks in the different treatment groups. White bars represent attacks on brown prey, black bars represent attacks on red prey, and gray bars represent attacks on striped prey. All prey were palatable in the generalization test. compare the behavior of birds in the first trial of avoidance learning to the behavior of birds in the generalization test with respect to the proportions attacked of presented aposematic prey with or without a black pattern on the same respective background. The attack probability of the novel aposematic prey in the first learning trial was much higher than in the generalization test ( as compared with ; mean 6 1 standard error, Mann Whitney U Test, U ¼ , P, ), indicating that the avoidance in the generalization test was due to generalization after avoidance learning. In the detection test, 10 of 11 birds pecked at the black stripes whereas one bird did not peck at all. Only one of the birds pecked (once) outside of the black stripes, but this bird also repeatedly pecked at the stripes. Thus, all the 10 pecking birds pecked predominantly on the black stripes, and this is significant (Sign Test, N ¼ 10, P, 0.001). From this we conclude that the black stripes were visible to the birds. DISCUSSION The chicks in our experiment showed faster avoidance learning when the unpalatable prey had a basic coloration (red) that contrasted with the background color. This is true independently of the presence or absence of internal patterns. This finding is supported by previous studies (Gittleman and Harvey 1980; Roper and Wistow 1986; Roper and Redston 1987; but see Sillén- Tullberg 1985). We found no similar effect of internal pattern contrast because birds in general avoided striped and nonstriped prey to a similar degree on their respective backgrounds. However, there was a significant effect of internal pattern on the shape of the learning curves (Figure 3), but contrary to the suggestion that internal contrasts may improve avoidance learning, the nonstriped prey were avoided more than the striped prey. This difference could imply that learning is faster for prey without internal contrasts but if so the effect is very small in the present experiment and already gone in the following trials. In any case, this experiment provides no evidence to any benefits in predator avoidance learning from having an internally contrasting black pattern. Also in the generalization test, the chicks that were presented prey on a contrasting brown background attacked significantly less of the warningly colored prey compared with the chicks presented prey on the matching red background. This is in accordance with a greater avoidance learning on the brown background. Additionally, the chicks completely generalized their previous experience of either plain red prey or red prey with black stripes to the respective novel color pattern where black stripes were either present or absent. There is a possibility that avoidance of novel color patterns can be due to other factors such as innate avoidance or neophobia (e.g., Marples and Roper 1996; Jetz et al. 2001) rather than generalization from learned avoidance. However, we find it very unlikely that this is the case in the present experiment. First, because the chicks did not show any avoidance of these color patterns when first encountered in the experiment. Second, the detailed comparison of attack probabilities between trial 1 and the generalization test of each prey type on respective background (when the prey were novel to birds in both situations) showed a considerably greater avoidance in birds that had undergone the avoidance learning trials. Thus, because the avoidance of the novel color pattern, plain red or striped, only occurs after avoidance learning of the other respective color pattern and not before, we feel quite safe to ascribe the avoidance to generalization. However, a noxious experience may enhance unlearned biases against specific warning colors (e.g., Rowe and Skelhorn 2005), and we cannot rule out the possibility that this may have affected birds behavior in the generalization test. Nevertheless, we conclude that the birds did not bias their behavior based on the presence or absence of internal contrasting patterns in the present experiment. As mentioned in the introduction, a pattern consisting of contrasting elements may actually reduce conspicuousness and is an important component in disruptive crypsis, especially when some color components match colors in the background (Cott 1940; Edmunds 1974; Schaefer and Stobbe 2006; Stevens et al. 2006). Generally, and in the present experiment, it is difficult to evaluate what pattern characteristics birds find conspicuous in aposematic prey. For instance, the presence of a black pattern may increase conspicuousness due to internal color and intensity contrasts but at the same time reduce prey-to-background contrasts. This may have played a role in the present experiment if the birds found the black color of the striped prey less contrasting than the red color to the brown background and therefore found the striped prey in total less contrasting to the background in that trial. This could possibly explain the lack of positive effects on avoidance learning from internal patterns on the brown background, but it cannot explain the lack of effect of internal contrasts on the red background. On this background, the striped prey were, if anything, more contrasting both with respect to background contrast and to internal contrast and neither here did the pattern positively affect the avoidance learning. In addition, the detection test showed that the stripes are quite visible to the birds because they were pecked at repeatedly in a foraging situation. As mentioned above, the positive antipredatory effects of conspicuousness due to background contrast in aposematic prey has been hypothetically extended to conspicuousness generated by internal pattern contrasts (Guilford 1990). We found no support for this in the present experiment, and our results suggest that this extension of conspicuousness in aposematic prey, to include within-prey pattern contrast, is not so straightforward. When discussing their experiments on initial wariness toward aposematic prey, Roper and Cook (1989) argue that neither striped prey nor bicolored prey (both containing color boundaries) were shown to be more aversive

6 Aronsson and Gamberale-Stille Contrasting patterns in aposematism 1361 than a plain black prey, and many color combinations were even preferred compared with control prey. Neither were the striped prey more aversive than the bicolored prey which, according to Roper and Cook (1989), hypothesis of Guilford (1990) predicts. Several studies with birds as predators have shown that they often use only parts of a warning signal as cues, whereas other components are ignored (Schmidt 1960; Terhune 1977; Evans et al. 1987; Aronsson and Gamberale-Stille 2008), and when learning to avoid prey, their attention is often biased toward the most salient component within a color pattern. Less salient components seem to be attended to only if they solely predict the unprofitability of the prey (Gamberale-Stille and Guilford 2003). Rowe et al. (2004) investigated the importance of pattern similarity between Müllerian mimics and found no effect on learning in great tits when there were 2 different patterns signaling unpalatability compared with one and that an increase in pattern similarity between the mimics did not affect their mortality. This suggests a quite broad generalization over pattern in that study. That the chicks in this present study generalize completely between prey with or without the black contrasting pattern suggests that their main attention was on the color component within the color pattern. Similarly, a study using 4 species of wild birds and color mutants of a normally red and black aposematic heteropteran bug shows that white mutants (otherwise equally defended) with the same black pattern but without the red color are not at all avoided and generalized (Exnerová et al. 2006; see also Svádová et al. 2009). Taken together, these studies suggest that birds often show selective attention to some characteristics of the color pattern and disregard others. However, the fact that there are many cases of mimicry where pattern similarity is very exact indicates that at least some important predators generalize very narrowly over many color pattern components. Especially in situations where Batesian mimicry is very common, it would be advantageous for a predator to learn exact color patterns when they alone predict unpalatability of the prey. There are many hypotheses about the benefits to the contrasting elements in aposematic color patterns, both related to signal efficacy and to other benefits. For instance, one reason to evolve a contrasting internal pattern would be if the predators perceive the presence of a pattern as more symmetrical than the absence. Symmetric signals have been suggested to positively influence prey detection and the association with unpalatability (Forsman and Herrström 2004). For instance, it has been shown that pigeons have easier to detect, to learn, and to reproduce patterns from their memory when they are symmetric compared with asymmetric patterns (Delius and Nowak 1982), and an otherwise strong signal could be weakened by asymmetries in the pattern, the color, or the shape of the signaling pattern elements (Forsman and Merilaita 1999; Forsman and Herrström 2004). Another possibility is that regular patterns may make the prey animal stand out from the chaotic background environment and thereby improve recognition and detection (Kenward et al. 2004). Also, it has been suggested that contrasting patterns within warning colors may facilitate the learning of a specific hue, so that a possible function of patterns (i.e., black) might be to direct the attention of the predator to the warning color itself (i.e., red, Rowe and Guilford 2000). According to Osorio et al. (1999) different parts of a prey s color pattern may play different roles in a predator s foraging behavior; the contrasting pattern attracts the attention, whereas the other colors give specific information and are remembered more accurately. They also suggest that it may be beneficial to have a high contrasting internal pattern due to predators supernormal responses to such patterns (Osorio et al. 1999). In this study, we have only aimed to investigate one specific hypothesis explaining the common occurrence of contrasting black patterns in colorful aposematic displays; the idea that the internal pattern contrasts promotes a conspicuousness that have the same effect as prey color contrast against the background in that it improves avoidance learning in predators (Guilford 1990). From this experiment, we conclude that internal pattern contrasts do not necessarily promote predator avoidance learning in the same way as prey contrast against the background, and the positive signal effects shown for preyto-background contrast cannot simply be expanded to also explain the presence of contrasting internal patterns in aposematic prey. As mentioned above, there are a vast number of hypotheses explaining the role of contrasting patterns in aposematic signal design (see e.g., Kenward et al. 2004; Aronsson and Gamberale-Stille 2008, for short reviews), but there is little experimental support for any of them. Thus, more studies are warranted to investigate if there is a general explanation to why aposematic prey often have evolved these contrasting patterns. In this study, the focus has been on warning coloration and the question of how different components of a complex signal contribute to signal efficacy. However, our results may also be relevant to the general understanding of other biological signaling systems where learning is involved. FUNDING The Swedish Research Council grants (D-numbers & to G.G.-S.). We thank Olof Leimar, Birgitta S. Tullberg, and 3 anonymous referees for their valuable comments on the manuscript and Carolina Bragée for helping out with the experiments. This work was carried out under permit from Stockholms djurförsöksetiska nämnd (D-nr 191/07). REFERENCES Aronsson M, Gamberale-Stille G Domestic chicks primarily attend to colour, not pattern, when learning an aposematic coloration. Anim Behav. 75: Cott HB Adaptive coloration in animals. London: Methuen. Creer DA Correlations between ontogenetic change in color pattern and antipredator behavior in the racer, Coluber constrictor. Ethology. 111: Delius JD, Nowak B Visual symmetry recognition by pigeons. Psychol Res. 44: Edmunds M Defence in animals: a survey of anti-predator defences. Harlow (UK): Longman. Endler JA Interactions between predators and prey. In: Krebs JR, Davies N, editors. Behavioural ecology: an evolutionary approach. Oxford: Blackwell. p Evans DL, Castoriades N, Badruddine H The degree of mutual resemblance and its effect on predation in young birds. Ethology. 74: Exnerová A, Svádová K, Stŷs P, Barcalová S, Landová E, Prokopová M, Fuchs R, Socha R Importance of colour in the reaction of passerine predators to aposematic prey: experiments with mutants of Pyrrhocoris apterus (Heteroptera). Biol J Linn Soc. 88: Forsman A, Herrström J Asymmetry in size, shape, and color impairs the protective value of conspicuous color patterns. Behav Ecol. 15: Forsman A, Merilaita S Fearful symmetry: pattern size and asymmetry affects aposematic signal efficacy. Evol Ecol. 13: Gamberale G, Tullberg BS Evidence for a peak-shift in predator generalization among aposematic prey. Proc R Soc B Biol Sci. 263:

7 1362 Behavioral Ecology Gamberale G, Tullberg BS Aposematism and gregariousness: the combined effect of group size and coloration on signal repellence. Proc R Soc B Biol Sci. 265: Gamberale-Stille G Decision time and prey gregariousness influence attack probability in naive and experienced predators. Anim Behav. 60: Gamberale-Stille G Benefit by contrast: an experiment with live aposematic prey. Behav Ecol. 12: Gamberale-Stille G, Guilford T Contrast versus colour in aposematic signals. Anim Behav. 65: Gittleman JL, Harvey PH Why are distasteful prey not cryptic? Nature. 286: Gittleman JL, Harvey PH, Greenwood PJ The evolution of conspicuous coloration: some experiments in bad taste. Anim Behav. 28: Guilford T The evolution of aposematism. In: Evans DL, Schmidt JO, editors. Insect defenses: adaptive mechanisms and strategies of prey and predators. Albany (NY): State University of New York Press. p Guilford T, Dawkins MS Receiver psychology and the evolution of animal signals. Anim Behav. 42:1 14. Hauglund K, Hagen SB, Lampe HM Responses of domestic chicks (Gallus gallus domesticus) to multimodal aposematic signals. Behav Ecol. 17: Jetz W, Rowe C, Guilford T Non-warning odors trigger innate color aversions as long as they are novel. Behav Ecol. 12: Kassarov L Are birds the primary selective force leading to evolution of mimicry and aposematism in butterflies? An opposing point of view. Behaviour. 140: Kenward B, Wachtmeister CA, Ghirlanda S, Enquist M Spots and stripes: the evolution of repetition in visual signal form. J Theor Biol. 230: Lindström L, Rowe C, Guilford T Pyrazine odour makes visually conspicuous prey aversive. Proc R Soc Lond B Biol Sci. 268: Marples NM, Roper TJ Effects of novel colour and smell on the response of naive chicks towards food and water. Anim Behav. 51: Marples NM, Roper TJ Response of domestic chicks to methyl anthranilate odour. Anim Behav. 53: Mason JR Direct and observational learning by redwinged blackbirds (Agelaius poeniceus): the importance of complex visual stimuli. In: Zentall TR, Galef BGJ, editors. Social learning: psychological and biological perspectives. Hillsdale (NJ): Lawrence Erlbaum. p Osorio D, Jones CD, Vorobyev M Accurate memory for colour but not pattern contrast in chicks. Curr Biol. 9: Papageorgis C Mimicry in neotropical butterflies. Am Sci. 63: Polat U, Tyler CW What pattern the eye sees best. Vision Res. 39: Riipi M, Alatalo RV, Lindström L, Mappes J Multiple benefits of gregariousness cover detectability costs in aposematic aggregations. Nature. 413: Robson JG, Graham N Probability summation and regional variation in contrast sensitivity across the visual-field. Vision Res. 21: Roper TJ, Cook SE Responses of chicks to brightly colored insect prey. Behaviour. 110: Roper TJ, Redston S Conspicuousness of distasteful prey affects the strength and durability of one-trial avoidance-learning. Anim Behav. 35: Roper TJ, Wistow R Aposematic coloration and avoidancelearning in chicks. Q J Exp Psychol. 38: Rowe C, Guilford T Aposematism: to be red or dead. Trends Ecol Evol. 15: Rowe C, Lindström L, Lyytinen A The importance of pattern similarity between Müllerian mimics in predator avoidance learning. Proc R Soc Lond B Biol Sci. 271: Rowe C, Skelhorn J Colour biases are a question of taste. Anim Behav. 69: Ruxton GD, Sherratt TN, Speed MP Avoiding attack. The evolutionary ecology of crypsis, warning signals and mimicry. Oxford: Oxford University Press. Schmidt RS Predator behaviour and the perfection of incipient mimetic resemblances. Behaviour. 16: Searcy WA, Nowicki S The evolution of animal communication: reliability and deception in signaling systems. Princeton: Princeton University Press. Schaefer HM, Stobbe N Disruptive coloration provides camouflage independent of background matching. Proc R Soc Lond B Biol Sci. 273: Sherratt TN, Beatty CD The evolution of warning signals as reliable indicators of prey defense. Am Nat. 162: Sillén-Tullberg B The significance of coloration per se, independent of background, for predator avoidance of aposematic prey. Anim Behav. 33: Skelhorn J, Rowe C Tasting the difference: do multiple defence chemicals interact in Müllerian mimicry? Proc R Soc Lond B Biol Sci. 272: Stevens M, Cuthill IC, Windsor AMM, Walker HJ Disruptive contrast in animal camouflage. Proc R Soc Lond B Biol Sci. 273: Svádová K,ExnerováA,Štys P, Landová E,ValentaJ,Fučíková A,SochaR Role of different colours of aposematic insects in learning, memory and generalization of naïve bird predators. Anim Behav. 77: Terhune EC Components of a visual stimulus used by scrub jays to discriminate a batesian model. Am Nat. 111: Turner JRG A tale of two butterflies. Nat Hist. 84: Wolf M, Werner YL The striped color pattern and striped nonstriped polymorphism in snakes (Reptilia, Ophidia). Biol Rev. 69: