Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Out in the open: behavior’s effect on predation risk and thermoregulation by aposematic caterpillars

Out in the open: behavior’s effect on predation risk and thermoregulation by aposematic caterpillars Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 The official journal of the Behavioral ISBE Ecology International Society for Behavioral Ecology Behavioral Ecology (2020), 31(4), 1031–1039. doi:10.1093/beheco/araa048 Original Article Out in the open: behavior’s effect on predation risk and thermoregulation by aposematic caterpillars a,b, a, Matthew E. Nielsen and Johanna Mappes Department of Biological and Environmental Sciences, University of Jyväskylä, Survontie 9 C, FI-40500 Jyväskylä, Finland and Department of Zoology, Stockholm University, Svante Arrhenius väg 18b, SE-114 18 Stockholm, Sweden Received 5 December 2019; revised 31 March 2020; editorial decision 14 April 2020; accepted 6 May 2020. Warning coloration should be under strong stabilizing selection but often displays considerable intraspecific variation. Opposing selec- tion on color by predators and temperature is one potential explanation for this seeming paradox. Despite the importance of behavior for both predator avoidance and thermoregulation, its role in mediating selection by predators and temperature on warning coloration has received little attention. Wood tiger moth caterpillars, Arctia plantaginis, have aposematic coloration, an orange patch on the black body. The size of the orange patch varies considerably: individuals with larger patches are safer from predators, but having a small patch is beneficial in cool environments. We investigated microhabitat preference by these caterpillars and how it interacted with their coloration. We expected caterpillar behavior to reflect a balance between spending time exposed to maximize basking and spending time concealed to avoid detection by predators. Instead, we found that caterpillars preferred exposed locations regardless of their coloration. Whether caterpillars were exposed or concealed had a strong effect on both temperature and predation risk, but caterpillars in exposed locations were both much warmer and less likely to be attacked by a bird predator (great tits, Parus major). This shared optimum may explain why we observed so little variation in caterpillar behavior and demonstrates the important effects of behavior on multiple functions of coloration. Key words: aposematism, Arctia plantaginis, color, microhabitat preference, Parus major, thermoregulation. INTRODUCTION cool environments, can make less effective warming signals, par - ticularly if they reduce internal contrast or contrast with their Warning colors are a classic example of traits that should be under background (Prudic et al. 2006; Lindstedt et al. 2009; Hegna et al. strong stabilizing selection yet often display large amounts of in- 2013; Arenas et al. 2015). traspecific variation (Joron and Mallet 1998; Mappes et  al. 2005; Both warning signaling and thermoregulation depend heavily Stevens and Ruxton 2012). One of the main proposed mechan- on behavior, particularly microhabitat selection, which has received isms underlying variation in aposematic coloration is opposing se- surprisingly little attention in studies of these multiple functions of lection due to the many different functions of color combined with color. A  highly concealed organism will be harder for predators variation in the strength of these sources of selection (Briolat et al. to detect, regardless of coloration, so aposematism has been ar- 2019). In particular, coloration has a key role in thermoregulation, gued to be more likely to evolve in behaviorally exposed organisms particularly by ectotherms. To maximize fitness, ectotherms need (Speed and Ruxton 2005; Grant 2007). Even then, aposematic or- to stay warm without becoming too hot (Huey and Kingsolver ganisms can still face some risk of predation from naïve predators 1989). Color determines how efficiently ectotherms absorb sun- and predators, which can overcome their defenses (Endler and light, with dark colors favored in cool environments to help stay Mappes 2004; Mappes et  al. 2014). Thus, aposematic organisms warm and light colors favored in hot environments (Clusella Trullas should still prefer concealed microhabitats unless the exposed lo- et  al. 2007). At the same time, darker colors, although favored by cations offer some other benefit, such as food availability (Speed et al. 2010). At the same time, microhabitat selection greatly affects many factors that determine an ectotherm’s body temperature, in- cluding exposure to solar radiation (Stevenson 1985; Muñoz et  al. Address correspondence to M. E. Nielsen. E-mail: matthew.nielsen@ zoologi.su.se 2014). Thus, at least in cool environments, microhabitat preference © The Author(s) 2020. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology should experience opposing selection: thermoregulation should instar larva. Under laboratory conditions, A.  plantaginis moths can favor basking in an exposed, sunny microhabitat, but these loca- produce three generations per year and the third generation over- tions should be associated with a greater risk of predation, even for winters. Temperatures in the greenhouse followed the outdoor tem- aposematic animals. peratures and varied between 20 and 30 °C during the day (~20 h) To test how behavior alters both thermoregulation and apo- and decreased to 15–20 °C during the night (~4 h). These temper- sematic signaling, we studied the caterpillars of Arctia plantaginis atures correspond to average temperatures in July in Jyväskylä (data (Linneaus; formerly Parasemia plantaginis; Rönkä et  al. 2016). The available at the Finnish meteorological institute). See Lindstedt aposematic function of the adult moth’s color pattern has been et al. 2016 for more details on general caterpillar rearing. To quan- extensively studied (e.g., Lindstedt et  al. 2011; Nokelainen et  al. tify the length of the orange band, we counted the number of body 2012; 2014), and the degree of melanization in adults has also segments with at least some orange hairs on them (Ojala et  al. been shown to affect their body temperature (Hegna et  al. 2013). 2007). For experiments, caterpillars of the appropriate orange band A  similar relationship appears to exist in the caterpillars. Arctia length were haphazardly selected from among 14 families, with plantaginis caterpillars are black with an orange band, but there is some caterpillars reused across live-caterpillar experiments due to both genetic and plastic variation in the width of this band (Ojala limited availability of individuals with certain orange band lengths. et al. 2007; Lindstedt et al. 2009, 2010, 2016; Galarza et al. 2019). Short-term preference for exposed versus Predatory birds (specifically Parus major) learn to avoid caterpillars concealed locations with broader orange bands more quickly (Lindstedt et  al. 2008), likely due to the greater conspicuousness of this pattern (Mappes To determine the temperature threshold at which caterpillars et  al. 2014). On the other hand, caterpillars with smaller orange switched from preferring exposed to concealed locations, we placed bands and correspondingly more black coloration grow faster in single caterpillars (n  =  50) of varying orange band lengths (3–7 cool temperatures, likely because of improved thermoregulatory segments) under a halogen light (400 W, 8550 lm, 2800  K color ability (Lindstedt et al. 2009). Caterpillar behavior could heavily in- temperature, manufactured by CRX) and measured the time until fluence the optimization of predation risk and thermoregulation, the caterpillars sought shade under a host and the temperature at but the behavior of these caterpillars—like most caterpillars—has which they did so. During the experiment, room temperature was received little attention. Nevertheless, a previous study suggests that held at ~15  °C. At the start of each trial, we placed a caterpillar color pattern may affect the basking behavior of these caterpillars on a potted dandelion (Taraxacum sp.), which had been collected under cool temperatures (Lindstedt et al. 2009). that day from outside (individual plants were reused for multiple We started by investigating the behavior of A.  plantaginis, specif- tests). After the caterpillar stopped moving, we turned on a 400-W ically the caterpillars’ preference for exposed or concealed micro- halogen lampplaced 40  cm directly over the caterpillar and plant, habitats. In our studies, we used caterpillars from central Finland, gently repositioning any leaves that were initially shading the cater- where staying warm should be a priority for growth, but predation pillar (this never led to immediate movement). Each time the cater- rates are also relatively high (at least for adult moths; Nokelainen pillar began moving, we recorded the time elapsed, took a thermal 2014). Thus, we expected behavior to reflect a balance between image of the caterpillar using an infrared camera (FLIR C2, FLIR minimizing exposure to potential predators and maintaining a Systems), and recorded shade temperature using a thermocouple warm body temperature. Then, we investigated the effects of color shaded by a leaf. When the caterpillar stopped moving in a loca- pattern and microhabitat preference on temperature and pred- tion that was at least 20% shaded, we considered the prior move- ator avoidance using wild great tits, P.  major, as a model predator. ment thermoregulatory behavior and used the associated data for Parus major is a common generalist feeder and potential predator of our analysis. After excluding trials that did not occur properly (see A.  plantaginis in the wild, which can easily be trained for a variety Supplementary Methods for more details), final sample sizes were of foraging tasks in the lab and has been used in previous labora- 45, 44, and 45 for time, ambient temperature, and body tempera- tory experiments on A.  plantaginis predation (Lindstedt et  al. 2008, ture respectively. 2011). For both traits, we predicted opposing selection by thermo- To estimate the body temperature of the caterpillar from each regulation and predator avoidance: smaller orange bands (thus, image, we first converted the image to linear-temperature grayscale more melanized caterpillars) would be favored for thermoregula- using FLIR Tools (v6.3, FLIR Systems). Then, we measured the tion but selected against by predation and, similarly, exposed loca- temperature by outlining the caterpillar by hand in imageJ (v1.51; tions would be favored for thermoregulation but selected against by Schneider et  al. 2012) and computing the mean temperature. We predation. evaluated the replicability of this approach by having another person remeasure 10 randomly selected images. These two meas- urements were highly correlated (r  =  0.998), indicating the preci- sion and consistency of this approach. The repeated measurements METHODS differed significantly by a mean of 0.53  °C (paired t-test, standard Rearing of caterpillars and general procedures –5 deviation = 0.23, t = 7.14, P = 5.4 × 10 ), so this approach should Arctia plantaginis caterpillars used in experiments came from a labo- be accurate within 1 °C. We tested whether the body temperature, ratory stock population originally derived from wild females caught shade temperature, and time from the start of heating at which in Central Finland in 2012 and supplemented annually with new the caterpillar sought shade were affected by the length of the wild-caught individuals. The stock was maintained in a greenhouse caterpillar’s orange band using separate linear mixed-effects models at the University of Jyväskylä on a mixed diet of lettuce (Lactuca with the individual plant used in the trial as a random effect. sativa var. crispa) and dandelion (Taraxacum sp.). Experiments were Caterpillar mass (measured posttrial with 0.1 mg precision) was in- conducted in 2015, at which point the lab stock had been main- itially included as an additional fixed effect; however, it had no sig- tained for eight generations. This species usually has one genera- nificant effect on the results for any response variable (P > 0.2), so tion per year in the wild and typically overwinters as third or fourth it was excluded from the final analysis. All statistical analyses were Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism conducted in R (v3.4.0; R Core Team 2018). For this analysis, we cardboard ~30  cm above the ground. During the period of un- used the lme function in the nlme package (Pinheiro et  al. 2018) obscured sunlight, we periodically recorded thermal images of and P-values were generated using likelihood ratio tests between the caterpillars, ultimately using the last pictures taken for each models with and without each fixed effect. condition during full sunlight (after 25  min without shade and after 20  min with half shaded) since these should represent near- Long-term preference for exposed versus equilibrium temperatures. Ambient temperatures (recorded using a concealed locations shaded thermocouple) during these two measurements were within 0.5 °C of each other and, thus, were not factored into the analysis. To test overall preference of caterpillars for exposed versus con- To analyze these images, we used a similar method to the short- cealed positions over longer time periods, including whether color term preference experiment; however, resolution and focal limita- pattern affected this preference, we placed caterpillars (n  =  36) in tions of the thermal camera prevented precise identification of the sets of three (one each of band lengths 4, 5, and 6 segments) on edges of each caterpillar. Instead, we estimated each caterpillar’s host plants (n = 12), and left them there under less intense heat to temperature by using ImageJ to find mean temperature of a 6-pixel observe how much time they spent exposed during the day. As be- diameter circle centered on the warmest point in each caterpillar. fore, recently potted dandelions (Taraxacum sp.) were used as hosts; We also estimated the temperature of a circle at the center of the however, in this case, the dandelion was enclosed in a cylinder of cardboard background in each image to compare its temperature clear acrylic to prevent escape and the leaves were trimmed to fit. in the shade versus sun treatments (Supplementary Methods). We The evening before the trial, the caterpillars were placed on each analyzed the effects of caterpillar orange band length, whether the plant and allowed to adjust to the chambers overnight. Size of the caterpillar was exposed or shaded, and their interaction using a orange patch size was used to distinguish each of the three individ- linear mixed-effects model. We included caterpillar mass as an ad- uals per plant during observations. The next morning, at 9 AM, we ditional fixed effect and the identity of each caterpillar as a random turned on two 400-W halogen lights ~75 cm over the table. At this effect. We also allowed variance to depend on shading treatment. time, 24 out of 36 caterpillars were in exposed positions. Over the We used the lme function in the nlme package for this analysis and next 9 h, we observed the position of each caterpillar every 30 min, P-values were generated using likelihood ratio tests between the full noting whether the caterpillar was fully exposed or at least partially model and models without each fixed effect. To further examine (20%) hidden. Of necessity, enclosures varied in exact distance the large effect of the shading treatment that we found, we also from these lights, producing a range of mean surface temperat- performed an additional likelihood ratio test comparing models ures from 14.5 to 27.3 °C among the enclosures over the course of with no interaction term and with or without shading treatment. the experiment (mean temperature of each enclosure in overhead thermal images taken every 30 min). Caterpillars that showed signs Effect of color and position on predation risk of building a shelter in which to molt or otherwise starting to molt were excluded from analysis from that observation onward (101 ob- To test how exposure and orange band length affected relative servations were excluded out of 708 in total; 11 caterpillars had predation risk, we conducted a laboratory experiment using dead observations excluded, and only one individual was fully excluded). caterpillars and wild-caught P.  major (great tits). Forty-eight birds To test for an effect of orange band length on preference for ex- were trapped between 6 August and 8 December 2015 and kept posed versus hidden positions, we used a generalized linear mixed for up to 2 weeks in captivity for trials with A.  plantaginis, and 16 model with a binomial distribution. We used whether the cater- birds were trapped between 14 February and 1 March for trials pillar was observed exposed as our dependent variable, band length with mealworms. In captivity, they were housed individually in and enclosure surface temperature as fixed effects, and individual plywood cages under an 11:13  h (light:dark) photoperiod and fed nested within pot as a random effect. Models were fit using the with sunflower seeds and vitamin-enriched tall ad libidum. Prior to glmer function in the lme4 package (Bates et al. 2015) in R, and a the experiment, birds had no previous experience in captivity with P-value was generated using a likelihood ratio test between models A. plantaginis (whether they had encountered A. plantaginis in the wild with and without each fixed effect. is unknown but unlikely because there are no known wild popula- tion of A. plantaginis in the vicinity of Konnevesi research station). Effect of color and position on caterpillar Each bird was given an array of 12 caterpillars that were all ei- temperature ther exposed or concealed but varied within array in orange band To test how exposure and orange band length affect the body tem- length from 4 to 6.  This created two overall treatments (exposed perature of caterpillars, we measured the temperature of caterpil- vs. concealed) while also creating within-treatment variation in lars (n  =  48) when placed outside under sunny conditions either warning coloration. Each array contained 12 dead A. plantaginis cat- with or without artificial shading. We tested caterpillars with orange erpillars (killed by freezing but thawed before use): 4 each of band band lengths from 4 to 7 segments, 12 of each size. Caterpillars length 4, 5, and 6.  Each caterpillar was also weighed before the were tethered to a piece of cardboard with 5  cm between indi- trial. Arrays were presented to birds in a 21.0- × 29.7-mm card- viduals using thin string to immobilize them. Caterpillars of each board box with a foam insert covered in brown paper. We filled signal size were evenly divided between two sides but otherwise this box with a 2–3-cm deep layer of dried, dead leaves, predom- placed randomly. Over the course of the experiment, three indi- inantly Betula pendula (silver birch). Each caterpillar was pinned in viduals escaped and were thus excluded from analysis. We con- the box using a black-painted insect pin, 2–3 cm from the edge and ducted the experiment in Jyväskylä (62.230°N, 25.744°E) on 25 8–9 cm from each other, with individuals assigned to specific loca- September 2015. On this day, we placed the tethered caterpillars tions randomly. Depending on treatment, we pinned all caterpillars outside at 2:38 PM and started the experiment when, at 3:31 PM, either below (concealed) or on top of (exposed) the layer of leaves. an extended period of continuous sunlight began. After 28 min of To ensure we had enough caterpillars for all replicates of the ex- unobscured sunlight, half of the array was shaded by a piece of periments, some trials were conducted using caterpillars descended Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology from A. plantaginis originally collected in Estonia (proportion of cat- differences among different orange band lengths using the package erpillars attacked during each trial did not differ significantly be- emmeans (Lenth 2018) and generating P-values from the resulting tween Finnish and Estonian populations and, thus, were pooled for z-scores. all analyses; t-test, t   =  0.275, P  =  0.785). We also repeated both the hidden and exposed treatments using arrays of 12 mealworms RESULTS instead of A. plantaginis caterpillars to test the effect of concealment Effect of color on behavioral exposure—short term on palatable and nonwarningly colored prey. Trials were conducted in a separate wooden box (50-cm wide × We successfully tested 45 caterpillars for conditions under which 50-cm deep × 67-cm high) with a one-way mirror to allow obser- they sought at least partial shade. Caterpillars with larger orange vation and recording (Supplementary Methods). Before each trial, bands sought shade significantly later than those with smaller each bird was food deprived for 2 h total, 90 min in its home box orange bands (β = 0.56  min/segment, standard error [SE] = and 30 min acclimating to the trial box. At that time, the lights in 0.24, = 4.80, P = 0.028; Figure 1a) and they also did so at higher the box were briefly turned off and the array placed in the box. shade temperatures (β = 0.46  °C/segment, SE = 0.22,  =  4.29, Birds that did not begin foraging (as indicated by interacting with P  =  0.038; Figure  1b). Band length did not, however, signifi- a caterpillar) within 45  min after the lights were turned back on cantly affect the body temperature at which caterpillars began were excluded from the study as were birds that foraged for less seeking shade (β = 0.18  °C/segment, SE  =  0.33,  =  0.329, than 30  min total within 115  min with the box. We ultimately P  =  0.57; Figure  1c), which was quite high overall with a mean excluded 12 birds in this way, evenly divided between treatments. of 37.9 °C. Once foraging began, the trial continued until either the bird had attacked all caterpillars or 30  min of foraging had elapsed (ex- cluding any breaks 5 min or longer spent not interacting with cat- erpillars or the box). During the trial, we noted which caterpillars (a) were attacked, specifically, if the caterpillar was pulled off its pin or, in a few cases, consumed while on the pin. We also recorded when the attack occurred to provide information about the order of attack. At the end of the trial, we confirmed which caterpillars were attacked by checking which caterpillars remained on their pins. Except for the first two birds tested, we also estimated the total mass of caterpillar the bird ate by weighing the caterpillar remains of the attacked caterpillars and subtracting from their original weight. In one case, the estimated mass eaten was slightly 2 negative (likely due to measurement error) and was set to 0 be- (b) cause it indicated that the bird had eaten none of the caterpillar despite removing them from the pin. Overall, we successfully tested 36 birds, 18 each with concealed or exposed caterpil- lars, plus 16 more birds with mealworms, 8 each concealed and exposed. We tested how caterpillar exposure affected whether a caterpillar was attacked during a trial using a generalized mixed model with a binomial distribution, with exposure treatment as a fixed effect and individual bird as a random effect. To account for rejection of prey after attack, we also tested the effect of caterpillar exposure on the total mass consumed by each bird using a linear model with (c) exposure as a fixed effect. We repeated both analyses separately for the mealworms. To analyze the effects of color on predation risk, we considered only the first six attacks by each bird (or fewer if 38 the bird attacked less than six caterpillars during the trial) because each bird only had access to three caterpillars of each band length, so the distribution of attacks would necessarily converge toward an equal attack rate on all band lengths if all attacks were con- sidered. We then fit a generalized mixed model using a binomial distribution for whether or not each caterpillar was one of the first six attacked. Exposure treatment, orange band length (as a factor), mass, and row in the array (indicating proximity to the bird’s ini- Orange band length (body segments) tial perch) were used as fixed effects, whereas individual bird was Figure 1 a random effect. The mixed models were fit using the glmer func- Effects of caterpillar orange band length (in number of segments) on the tion in the lme4 package in R and P-values were generated using start of shade-seeking behavior when heated by an overhead halogen a likelihood ratio test between models with and without each fixed light. Lines represent significant (P  <  0.05) fixed effects of band length on effect. All second- and third-degree interactions among fixed effects the response variable in a mixed model. (a) Time (minutes) from start of were nonsignificant (P > 0.2) and excluded from the final analysis. experiment, n  =  45. (b) Surface temperature (degree Celsius) of shaded We performed a Tukey’s post hoc test on the final model for the ground, n =44. (c) Body temperature (degree Celsius) of caterpillar, n = 45. Body temperature (°C) at Shade temperature (°C) at Time (min) at start of shade seeking start of shade seeking start of shade seeking Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism birds did not, however, eat significantly more mealworms (0.123 g, Effect of color on behavioral SE = 0.102 g, t  = 1.20, P = 0.25). exposure—long term 14 To assess how color affected risk of predation, we focused on the We observed 35 caterpillars over a 9-h period. During the vast ma- first six caterpillars attacked by each bird. Larger caterpillars were jority (91.8%) of our observations, caterpillars were fully exposed more likely to be attacked, as were caterpillars closer to the bird’s to the light, spending only 8.2% of their time fully or partially perch (Table  2). Most importantly, orange band length had a sig- concealed. Although caterpillars chose concealed locations infre- nificant effect on risk of predation (Table  2; Figure  4); specifically, quently, orange band length nevertheless affected how often this occurred. Caterpillars with larger orange bands were significantly more likely to be observed fully exposed (β = 0.836 ± 0.276, = 8.43, P = 0.004). Specifically, caterpillars with four orange segments were exposed 84.4% of the time, five orange segments 95.0% of the time, and six orange segments 95.7% of the time. Caterpillars were 25 also significantly more likely to be observed fully exposed when the enclosure was warmer (β = 0.233 ± 0.066, = 12.09, P = 0.0005). Effect of color and position on temperature We measured the temperature of 45 caterpillars under sunny con- ditions twice, once with all in full sun the other with half in full sun and half shaded. We found a significant interaction between the effects of orange band length and shade treatment on temper - Orange band length (body segments) ature (Table  1; Figure  2); however, the relative effect sizes of these Figure 2 two factors differed dramatically. Shading had a large overall effect Body temperature of caterpillars with varying orange band lengths when on caterpillar temperature, cooling caterpillars by 10.7  °C aver- placed outdoors either in full sun or shade. Gray circles and dashed line aged across orange band lengths (  =  142.3, P  <  0.0001). This dif- represent caterpillars in full sun; black squares and solid line represent ference is only partially explained by the effect of shading on the caterpillars in shade. Lines show interacting fixed effects from a mixed caterpillar’s microhabitat (the temperature of the cardboard back- model (P = 0.0365). n = 90 observations of 45 caterpillars. ground differed by 6.6 °C between these treatments). Orange band length, on the other hand, had only a small effect on temperature (a) (b) when in the sun (−0.349  °C/orange segment) and practically no 1.0 effect when shaded (−0.027 °C/orange segment). Caterpillar mass ranged from 0.108 to 0.307  g (mean 0.195  g), and body tempera- ture increased significantly with mass by 9.63 °C/g (Table 1). 0.8 n = 8 Effect of color and position on predation risk 0.6 During 30  min of foraging, P.  major attacked a smaller proportion of A.  plantaginis caterpillars in the exposed than concealed treat- 0.4 n = 8 ment (β = −1.27  ± 0.48,  =  6.53, P  =  0.011; Figure  3). This cor- n = 18 responds to a mean of 2.61 fewer caterpillars (out of 12)  attacked 0.2 when exposed. From the attacked caterpillars, the birds ate, on n = 18 average, 0.372  g less A.  plantaginis caterpillar in the exposed than 0.0 concealed treatment (SE = 0.124  g, t   =  −3.00, P = 0.005). This mass is equivalent to 2.55 caterpillars (mean mass of attacked cat- Concealed Exposed Concealed Exposed erpillars was 0.146 g, SE = 0.0028), indicating that most of the ad- Caterpillar Mealworm ditional caterpillars attacked when concealed were also consumed. Figure 3 In contrast, when foraging on mealworms, birds attacked a greater Boxplot of proportion of (a) A.  plantaginis caterpillars or (b) mealworms proportion of individuals in the exposed than concealed treatment attacked by P. major during foraging trials. Each bird had access to either 12 (β = 1.54  ± 0.50,  =  8.63, P  =  0.003; Figure  3), corresponding to dead caterpillars or mealworms, all of which were pinned in place either a mean of 2.62 more mealworms attacked when exposed. The exposed on top of or concealed beneath a layer of dead leaves. Table 1 Effects of orange band length, position (sun or shade), and mass on temperature (degree Celsius) of A. plantaginis caterpillars Coeff. SE df χ P Orange band length −0.0270 0.090 1 — — Position 13.00 0.879 1 — — Mass (g) 9.63 2.7 1 12.52 0.0004 Orange band length × position −0.322 0.155 1 4.37 0.0365 df, degrees of freedom.  Difference between sun and shade. Proportion attacked Body temperature (°C) Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology Table 2 Effects of exposure treatment, orange band length, mass, and position relative to bird’s perch (rows away) on whether A. plantaginis caterpillars were among the first six caterpillars attacked in an array of 12 by P. major in laboratory trials Coeff. SE df χ P Treatment −0.324 0.213 1 2.33 0.127 Orange band length — — 3 11.28 0.0103 5 body segments −0.097 0.305 — — 6 body segments −0.841 0.326 — — 7 body segments 0.056 0.321 — — Mass (g) 8.75 2.43 1 13.55 0.0002 Distance from perch (rows) −0.980 0.138 1 57.35 <0.0001 Difference between exposed and concealed treatments. b 2 Analyzed as a factor. χ and P-values tested across all levels, coefficients given separately for each level compared with a length of four body segments as a baseline. See Figure 4 for more details. were both much warmer and less likely to be attacked by P.  major 1.0 in our experiments. This parallel selection by both predators and temperature favoring caterpillars in exposed locations helps ex- 0.8 plain the strong preference for exposed positions displayed by the caterpillars in our behavioral experiments. Over 9  h of observa- a a ab 0.6 tions under moderately warm thermal conditions, caterpillars spent the vast majority of their time in the open. This behavior suggests 0.4 that these caterpillars rely on their primary and secondary defense strategies (coloration, hairiness, and chemical defense) for predator 0.2 avoidance. Our findings support “a classic view of aposematism” in which conspicuousness reduces the likelihood of recognition 0.0 errors because predators can detect conspicuous prey at a greater 45 67 distance and, thus, avoid them more reliably (Guilford 1986). There Orange band length (body segments) has, however, been surprisingly few systematic observations of Figure 4 microhabitat selection by aposematic prey in this regard (but see Probability of A.  plantaginis caterpillars with varying orange band length Tabadkani and Nozari 2014; Rößler et al. 2019). being among the first six attacked by P.  major in an array of 12 caterpillars From a thermal perspective, caterpillars did not seek shade until (three of each band length). Bars represent 95% confidence interval, and heated to a surprisingly high mean body temperature of 37.9  °C. lettering indicates significantly differing band lengths in a Tukey’s post hoc This body temperature is comparable to and, in some cases, ex- test (n = 34 birds). ceeds the body temperature at which other species of caterpillar from much warmer environments begin heat-avoidance behavior caterpillars with an orange band 6 segments long were least likely (Sherman and Watt 1973; Nielsen et  al. 2018). This observation to be among the first six caterpillars attacked and significantly less fits with a general trend that the upper thermal limits of both in- likely to be attacked than caterpillars with an orange band 4 or 7 sects and ectotherms decrease little with latitude (Addo-Bediako segments long according to post hoc tests. Given that we limited et al. 2000; Sunday et al. 2014). For high-latitude arthropods, it can our analysis to the first six caterpillars attacked, it is unsurprising be important to capitalize on brief warm, sunny periods to maxi- that exposure had no significant independent effect on attack risk; mize growth (Kukal et al. 1988; Bennett et al. 1999; Birkemoe and however, it also did not significantly interact with orange band Leinaas 2000). length ( = 1.08, P = 0.78). On the other hand, the preference for exposure should represent a risk for animals with warning colors (Endler and Mappes 2004; DISCUSSION Mappes et al. 2014). This risk should be especially great for apose- Coloration of aposematic animals provides an example of op- matic animals like A.  plantaginis that are only somewhat distasteful, posing selection promoting trait variation: trade-offs between so experienced predators will still eat them in the absence of pre- warning signal efficacy and thermoregulatory ability (as well as ferred prey as occurred in our experiment. Nevertheless, we found, other functions) can increase variation in a trait that should other- under laboratory conditions at least, that potential avian predators wise be highly conserved (Briolat et al. 2019). We predicted a sim- (wild-caught, adult P.  major) attacked fewer A.  plantaginis caterpil- ilar opposing selection on microhabitat choice in A.  plantaginis, an lars when the caterpillars were exposed than when they were con- aposematic caterpillar, so we tested how color pattern and behavior cealed under dead leaves, despite the latter requiring more effort interact with body temperature and predation risk in this species. to locate the caterpillars. The opposite pattern occurred with pal- We found that, although the effects of these two traits were not en- atable, nonaposematic mealworms. The greater risk to concealed tirely independent, behavior had a greater effect on both sources of A.  plantaginis is reinforced by the fact that birds attacked similar selection. proportions of concealed caterpillars and concealed mealworms Contrary to our initial expectations, however, we found no during the respective experiments, despite how P. major strongly pre- evidence of opposing selection on microhabitat preference in fers mealworms. Birds may have attacked more hidden caterpillars A.  plantaginis caterpillars. Individuals placed in exposed positions because the leaves obscured the warning signals of the caterpillars. Probability of attack during first 6 attacks Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism If this hypothesis explained our results, however, we would have ex- the effectiveness of the caterpillars’ warning signals (Aronsson and pected concealment to alter the effect of orange band length on Gamberale-Stille 2013; Barnett et  al 2016; although see Aronsson attack risk, which we did not observe. Instead, the greater effort re- and Gamberale-Stille 2009). Regardless of the exact relation- quired to find concealed caterpillars could explain the greater will- ship between band length and predation, the orange band lengths ingness of P.  major to attack them. Birds were frequently observed least likely to be attacked differed from those which were warmest, digging through leaves even in the exposed treatment when nothing indicating the expected opposing selection on color pattern from was present there. In the exposed treatment, the birds may have temperature and predation. initially ignored the exposed caterpillars to search for the poten- Ultimately, we found the expected opposing selection on one tially better prey that could have existed under leaves in the wild. trait (color pattern) but not another (microhabitat preference). The In the concealed treatment, on the other hand, by the time the presence or absence of opposing sources of selection could help ex- birds found the caterpillars hidden under the leaves, there were no plain the differing levels of phenotypic variation observed in these additional locations to search, so the birds may have been more traits. Color pattern in both adult and larval A. plantaginis varies ex- willing to accept them as prey. Notably, this digging behavior was tensively due to both genetic and plastic factors both within and performed by wild-caught birds without training. Indeed, great tits between populations (Ojala et al. 2007; Lindstedt et al. 2009, 2016; have a greater propensity than other tits for feeding on the ground Hegna et al. 2013, 2015), and opposing selection from a variety of (Gosler and Clement 2007). Regardless of the underlying cause, sources, including sexual selection, initial detectability, and thermo- the lower attack risk for exposed caterpillars combined with their regulation, has been argued to enable persistence of this variation warmer temperature helps explain the strong behavioral preference (Lindstedt et  al. 2008, 2009; Nokelainen et  al. 2012; Hegna et  al. we observed in these caterpillars for open, exposed positions. Based 2013; Mappes et  al. 2014). The behavior of A.  plantaginis, on the on these results, we would predict opposing selection by predation other hand, has received little attention, particularly in the cater- and temperature on microhabitat preference in warm environments pillars, but they generally display little activity, at least under lab- instead of cool ones, but the trade-off may not occur in that con- oratory conditions. The minimal variation we find in preferred text either. Battus philenor, a desert-dwelling aposematic caterpillar, microhabitat, and perhaps other aspects of their behavior, could remains fully exposed while still avoiding high temperatures (Nice be caused by having a single preferred microhabitat favored by and Fordyce 2006; Nielsen and Papaj 2017). multiple sources of selection removing the need for temporal or Variation in color, on the other hand, had much weaker effects between-individual variation in habitat preference. than behavior on both body temperature and predation risk. For An important general benefit of aposematism is the opportu- temperature, the effect of orange band length on body temper - nity to occupy microhabitats that are otherwise highly vulnerable ature depended heavily on the position of the caterpillar and, to predation (Speed et  al. 2010). Here, we find that this effect in thus, its behavior: caterpillars with smaller orange bands (more A.  plantaginis is stronger than anticipated despite the fact that the melanin) were warmer as expected when exposed to the sun, but species is only unpalatable and hairy rather than truly toxic. This band length had little effect in the shade. Color and other aspects benefit of aposematism effectively negates the opposing selection of morphology are generally predicted to have a weaker effect on faced by cryptic organisms between behaviors that maximize re- temperature than behavior (Stevenson 1985), and the effect of source use (including favorable thermal environments) and mini- color on temperature has been shown to depend on behavior in mize predator exposure (Speed and Ruxton 2005). Multiple other a wide range of insects (e.g., Kingsolver 1987; Nielsen and Papaj aposematic animals behave in ways that expose them more to 2017). Despite its smaller effect, color pattern did alter thermoreg- predators than comparable cryptic animals (Pinheiro 1996, 2007; ulatory behavior in A.  plantaginis. The body temperature at which Rudh et  al. 2013; Willink et  al. 2013; Tabadkani and Nozari shade-seeking began did not change with band length; however, 2014; Valkonen et  al. 2014; Rößler et  al. 2019). Based on our re- caterpillars with shorter orange bands (more melanic caterpillars) sults, we predict that aposematic organisms may frequently use started shade-seeking behavior sooner and at lower environmental different microhabitats from cryptic organisms and may show re- temperatures. These results indicate that the caterpillars’ internal duced behavioral variation at least in terms of habitat preference or physiology and response to body temperature do not change with antipredator behavior. At the same time, if an aposematic species color pattern and, instead, color pattern alters behavior by chan- is thus specialized for an exposed microhabitat due to parallel se- ging light absorption and, thus, heating rate (Nielsen et  al. 2018). lection from multiple sources, they may be more vulnerable to any This ability of body color to affect thermoregulatory behavior by environmental changes, such as climate change, which reduce their changing body temperature has also been demonstrated in a range fitness in that habitat. Future research could test for selection to of insects (e.g., Kingsolver 1987; Karpestam et  al. 2012; Nielsen prefer exposed microhabitats in other aposematic species and con- et al. 2018). sider additional sources of selection, which may benefit exposure, We confirmed effect of color pattern on predation risk by com- such as sexual selection and foraging. paring the first six caterpillars attacked by P.  major during experi- Regardless, we have shown for an aposematic species that, when ments. Caterpillar with an orange band 6 body segments long were facing wild predators, they are less likely to be attacked when ex- less likely to be attacked, regardless of whether they were hidden or posed. Combined with the thermal benefits of exposure, the greater exposed. This is longer than the average signal size of A. plantaginis safety of caterpillars when exposed helps explain their minimal be- (4.7 segments) but not the largest possible (7 segments). Previous havioral variation. Caterpillars rarely occupied concealed positions work on A.  plantaginis indicates that larger orange bands (>5 seg- except at extreme temperatures. Although variation in color pat- ments) are more effective warning signals than short bands (<4 seg- tern had smaller direct effects than exposure on temperature and ments; Lindstedt et  al. 2008, Mappes et  al. 2014). We add to this predation risk in our experiments, color pattern’s effect on tempera- that warning signal effectiveness might also decrease for the longest ture lead to a corresponding change in thermoregulatory behavior. bands (7 segments). The small black content of these caterpillars Thus, our results reinforce the importance of both behavior and could reduce their internal contrast, which, in turn, could reduce color pattern for the function of aposematic signals not only in the Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology context of predation where they are typically studied but also in a World. Picathartes to tits and chickadees. Vol. 12. Barcelona (Spain): Lynx Edicions. p. 662–709. thermoregulatory context. Grant JB. 2007. Ontogenetic colour change and the evolution of aposema- tism: a case study in panic moth caterpillars. J Anim Ecol. 76:439–447. Guilford T. 1986. How do “warning colours” work? Conspicuousness SUPPLEMENTARY MATERIAL may reduce recognition errors in experienced predators. Anim Behav. Supplementary data are available at Behavioral Ecology online. 34:286–288. Hegna  RH, Galarza  JA, Mappes  J. 2015. Global phylogeography and ge- ographical variation in warning coloration of the wood tiger moth FUNDING (Parasemia plantaginis). J Biogeogr. 42:1469–1481. Hegna  RH, Nokelainen  O, Hegna  JR, Mappes  J. 2013. To quiver or to The work was supported by a Graduate Research Opportunities Worldwide shiver: increased melanization benefits thermoregulation, but re- supplement to National Science Foundation Graduate Research Fellowship duces warning signal efficacy in the wood tiger moth. Proc Biol Sci. Program (award number DGE-1143953) and the Academy of Finland 280:20122812. (grant # 293513). Huey  RB, Kingsolver  JG. 1989. Evolution of thermal sensitivity of ecto- therm performance. Trends Ecol Evol. 4:131–135. We would like to thank Helinä Nisu for help with bird rearing and Tuuli Joron  M, Mallet  JL. 1998. Diversity in mimicry: paradox or paradigm? Salmi for assistance with the mealworm experiments. Aimee Deconinck, Trends Ecol Evol. 13:461–466. Karpestam E, Wennersten L, Forsman A. 2012. Matching habitat choice by Geoffrey Legault, Jee Yun Lee, Juan Galarza, Emily Burdfield-Steel, Sara experimentally mismatched phenotypes. Evol Ecol. 27:893–907. Calhin, and two anonymous reviewers provided valuable feedback on the Kingsolver JG. 1987. Evolution and coadaptation of thermoregulatory be- manuscript. Research using wild P.  major was conducted with permission havior and wing pigmentation pattern in pierid butterflies. Evolution. of Southwest Finland Centre for Economic Development, Transport and 41:472–490. Environment (VARELY/294/2015) and license from the National Animal Kukal  O, Heinrich  B, Duman  JG. 1998. Behavioural thermoregulation in Experiment Board (ESAVI/9114/04.10.07/2014). the freeze-tolerant Arctic caterpillar, Gynaephora groenlandica. J Exp Biol. 138:181–193. Author’s contributions: M.E.N. and J.M. designed the studies. M.E.N. con- Lenth  R. 2018. emmeans: estimated marginal means, aka least-squares ducted the experiments and analyzed the data. M.E.N.  led the writing of means. R package version 1.3.1. Available from: https://CRAN.R- the manuscript. M.E.N.  and J.M.  edited the manuscript and approved the project.org/package=emmeans (accessed 20 December 2018). final version. Lindstedt  C, Eager  H, Ihalainen  E, Kahilainen  A, Stevens  M, Mappes  J. 2011. Direction and strength of selection by predators for the color of Data accessibility: Analyses reported in this article can be reproduced using the aposematic wood tiger moth. Behav Ecol. 22:580–587. the data provided by Nielsen and Mappes (2020). Lindstedt C, Lindström L, Mappes J. 2008. Hairiness and warning colours as components of antipredator defence: additive or interactive benefits? Handling editor: Luke Holman Anim Behav. 75:1703–1713. Lindstedt  C, Lindström  L, Mappes  J. 2009. Thermoregulation constrains effective warning signal expression. Evolution. 63:469–478. Lindstedt  C, Morehouse  N, Pakkanen  H, Casas  J, Christedes  J-P, REFERENCES Kemppainen  K, Lindström  L, Mappes  J. 2010. Characterizing the pig- Addo-Bediako A, Chown SL, Gaston KJ. 2000. Thermal tolerance, climatic ment composition of a variable warning signal of Parasemia plantaginis variability and latitude. Proc Biol Sci. 267:739–745. larvae. Funct Ecol. 24:759–766. Arenas LM, Walter D, Stevens M. 2015. Signal honesty and predation risk Lindstedt  C, Schroderus  E, Lindström  L, Mappes  T, Mappes  J. 2016. among a closely related group of aposematic species. Sci Rep. 5:11021. Evolutionary constraints of warning signals: a genetic trade-off between Aronsson  M, Gamberale-Stille  G. 2009. Importance of internal pattern the efficacy of larval and adult warning coloration can maintain variation contrast and contrast against the background in aposematic signals. in signal expression. Evolution. 70:2562–2572. Behav Ecol. 20:1356–1362. Mappes  J, Kokko  H, Ojala  K, Lindström  L. 2014. Seasonal changes in Aronsson  M, Gamberale-Stille  G. 2013. Evidence of signaling benefits to predator community switch the direction of selection for prey defences. contrasting internal color boundaries in warning coloration. Behav Ecol. Nat Commun. 5:5016. 24:349–354. Mappes  J, Marples  N, Endler  JA. 2005. The complex business of survival Barnett  JB, Scott-Samuel  NE, Cuthill  IC. 2016. Aposematism: balancing by aposematism. Trends Ecol Evol. 20:598–603. salience and camouflage. Biol Lett. 12:20160335. Muñoz  MM, Stimola  MA, Algar  AC, Conover  A, Rodriguez  AJ, Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects Landestoy  MA, Bakken  GS, Losos  JB. 2014. Evolutionary stasis and la- models using lme4. J Stat Softw. 67:1–48. doi: 10.18637/jss.v067.i01. bility in thermal physiology in a group of tropical lizards. Proc Biol Sci. Bennett  VA, Kukal  O, Lee  RE. 1999. Metabolic opportunists: feeding and 281:20132433. temperature influence the rate and pattern of respiration in the high Nice  CC, Fordyce  JA. 2006. How caterpillars avoid overheating: behav- arctic woollybear caterpillar Gynaephora groenlandica (Lymantriidae). J Exp ioral and phenotypic plasticity of pipevine swallowtail larvae. Oecologia. Biol. 202:47–53. 146:541–548. Birkemoe T, Leinaas HP. 2000. Effects of temperature on the development Nielsen  ME, Levin  E, Davidowitz  G, Papaj  DR. 2018. Colour plasticity of an Arctic Collembola (Hypogastrura tullbergi). Funct Ecol. 14:693–700. alters thermoregulatory behaviour in Battus philenor caterpillars by modi- doi: 10.1046/j.1365-2435.2000.00478.x. fying the cue received. Anim Behav. 140:93–98. Briolat  ES, Burdfield-Steel  ER, Paul  SC, Rönkä  KH, Seymoure  BM, Nielsen,  ME, Mappes  J. 2020. Data from: Out in the open: behavior’s ef- Stankowich  T, Stuckert  AMM. 2019. Diversity in warning colora- fect on predation-risk and thermoregulation by aposematic caterpillars. tion: selective paradox or the norm? Biol Rev Camb Philos Soc. 94: Behav Ecol. doi: 10.5061/dryad.95x69p8gk. 388–414. Nielsen  ME, Papaj  DR. 2017. Why have multiple plastic responses? Clusella  Trullas  S, van  Wyk  J-H, Spotila  JR. 2007. Thermal melanism in Interactions between color change and heat avoidance behavior in Battus ectotherms. J Therm Biol. 32:235–245. philenor larvae. Am Nat. 189:657–666. Endler JA, Mappes J. 2004. Predator mixes and the conspicuousness of ap- Nokelainen  O, Hegna  RH, Reudler  JH, Lindstedt  C, Mappes  J. 2012. osematic signals. Am Nat. 163:532–547. Trade-off between warning signal efficacy and mating success in the Galarza JA, Dhaygude K, Ghaedi B, Suisto K, Valkonen J, Mappes J. 2019. wood tiger moth. Proc Biol Sci. 279:257–265. Evaluating responses to temperature during pre-metamorphosis and Nokelainen O, Valkonen J, Lindstedt C, Mappes J. 2014. Changes in pred- carry-over effects at post-metamorphosis in the wood tiger moth (Arctia ator community structure shifts the efficacy of two warning signals in plantaginis). Philos Trans R Soc Lond B Biol Sci. 374:20190295. Arctiid moths. J Anim Ecol. 83:598–605. Gosler  A, Clement  P. 2007. Family Paridae (tits and chickadees). In Ojala  K, Lindström  L, Mappes  J. 2007. Life-history constraints and del  Hoyo  J, Elliott  A, Christie  D, editors. Handbook of the birds of the warning signal expression in an arctiid moth. Funct Ecol. 21:1162–1167. Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism Pinheiro CEG. 1996. Palatability and escaping ability in Neotropical butter- Speed MP, Brockhurst MA, Ruxton GD. 2010. The dual benefits of apose- flies: test with wild kingbirds. Biol J Linn Soc. 59:351–365. matism: predator avoidance and enhanced resource collection. Evolution. Pinheiro  CEG. 2007. Asynchrony in daily activity patterns of butterfly 64:1622–1633. models and mimics. J Trop Biol. 23:119–123. Speed MP, Ruxton GD. 2005. Aposematism: what should our starting point Pinheiro J, Bates D, DebRoy S, Sarkar D; R Core Team. 2018. nlme: linear and be? Proc R Soc B. 272:431–438. nonlinear mixed effects models. R package version 3.1–131.1. Available from: Stevens M, Ruxton GD. 2012. Linking the evolution and form of warning https://CRAN.R-project.org/package=nlme (accessed 20 February 2018). coloration in nature. Proc Biol Sci. 279:417–426. Prudic KL, Skemp AK, Papaj DR. 2006. Aposematic coloration, luminance Stevenson RD. 1985. The relative importance of behavioral and physiolog- contrast, and the benefits of conspicuousness. Behav Ecol. 18:41–46. ical adjustments controlling body temperature in terrestrial ectotherms. R Core Team. 2018. R: a language and environment for statistical com- Am Nat. 126:362–386. puting. Vienna (Austria): R Foundation for Statistical Computing. Sunday JM, Bates AE, Kearney MR, Colwell RK, Dulvy NK, Longino JT, Available from: https://www.R-project.org/ (accessed 1 May 2017). Huey RB. 2014. Thermal-safety margins and the necessity of thermoreg- Rößler  DC, Lötters  S, Mappes  J, Valkonen  JK, Menin  M, Lima  AP, ulatory behavior across latitude and elevation. Proc Natl Acad Sci USA. Pröhl  H. 2019. Sole coloration as an unusual aposematic signal in a 111:5610–5615. Neotropical toad. Sci Rep. 9:1128. doi:10.1038/s41598-018-37705-1. Tabadkani  SM, Nozari  J. 2014. Relaxed predation hinders development Rönkä  K, Mappes  J, Kaila  L, Wahlberg  N. 2016. Putting Parasemia in of anti-predator behaviors in an aposematic beetle. Entomol Exp Appl. its phylogenetic place: a molecular analysis of the subtribe Arctiina 153:199–206. doi: 10.1111/eea.12241. (Lepidoptera). Syst Entomol. 41:844–853. Valkonen  JK, Nokelainen  O, Jokimäki  M, Kuusinen  E, Paloranta  M, Rudh  A, Breed  MF, Qvarnström  A. 2013. Does aggression and explora- Peura  M, Mappes  J. 2014. From deception to frankness: benefits of on- tive behaviour decrease with lost warning coloration? Biol J Linn Soc togenetic shift in the anti-predator strategy of alder moth Acronicta alni 108:116–126. doi: 10.1111/j.1095-8312.2012.02006.x. larvae. Curr Zool. 60:114–122. doi:10.1093/czoolo/60.1.114. Schneider  CA, Rasband  WS, Eliceiri  KW. 2012. NIH image to ImageJ: Willink  B, Brenes-Mora  E, Bolaños  F, Pröhl  H. 2013. Not everything is 25 years of image analysis. Nat Methods. 9:671–675. black and white: color and behavioral variation reveal a continuum be- Sherman  PW, Watt  WB. 1973. The thermal ecology of some Colias but- tween cryptic and aposematic strategies in a polymorphic poison frog. terfly larvae. J Comp Physiol. 83:25–40. Evolution. 67:2783–2794. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavioral Ecology Oxford University Press

Out in the open: behavior’s effect on predation risk and thermoregulation by aposematic caterpillars

Nielsen, Matthew E; Mappes, Johanna
Behavioral Ecology , Volume 31 (4) – Jul 29, 2020
Download PDF
9 pages

Loading next page...
 
/lp/oxford-university-press/out-in-the-open-behavior-s-effect-on-predation-risk-and-0BELURi504

References (114)

Publisher
Oxford University Press
Copyright
Copyright © 2022 International Society of Behavioral Ecology
ISSN
1045-2249
eISSN
1465-7279
DOI
10.1093/beheco/araa048
Publisher site
See Article on Publisher Site

Abstract

Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 The official journal of the Behavioral ISBE Ecology International Society for Behavioral Ecology Behavioral Ecology (2020), 31(4), 1031–1039. doi:10.1093/beheco/araa048 Original Article Out in the open: behavior’s effect on predation risk and thermoregulation by aposematic caterpillars a,b, a, Matthew E. Nielsen and Johanna Mappes Department of Biological and Environmental Sciences, University of Jyväskylä, Survontie 9 C, FI-40500 Jyväskylä, Finland and Department of Zoology, Stockholm University, Svante Arrhenius väg 18b, SE-114 18 Stockholm, Sweden Received 5 December 2019; revised 31 March 2020; editorial decision 14 April 2020; accepted 6 May 2020. Warning coloration should be under strong stabilizing selection but often displays considerable intraspecific variation. Opposing selec- tion on color by predators and temperature is one potential explanation for this seeming paradox. Despite the importance of behavior for both predator avoidance and thermoregulation, its role in mediating selection by predators and temperature on warning coloration has received little attention. Wood tiger moth caterpillars, Arctia plantaginis, have aposematic coloration, an orange patch on the black body. The size of the orange patch varies considerably: individuals with larger patches are safer from predators, but having a small patch is beneficial in cool environments. We investigated microhabitat preference by these caterpillars and how it interacted with their coloration. We expected caterpillar behavior to reflect a balance between spending time exposed to maximize basking and spending time concealed to avoid detection by predators. Instead, we found that caterpillars preferred exposed locations regardless of their coloration. Whether caterpillars were exposed or concealed had a strong effect on both temperature and predation risk, but caterpillars in exposed locations were both much warmer and less likely to be attacked by a bird predator (great tits, Parus major). This shared optimum may explain why we observed so little variation in caterpillar behavior and demonstrates the important effects of behavior on multiple functions of coloration. Key words: aposematism, Arctia plantaginis, color, microhabitat preference, Parus major, thermoregulation. INTRODUCTION cool environments, can make less effective warming signals, par - ticularly if they reduce internal contrast or contrast with their Warning colors are a classic example of traits that should be under background (Prudic et al. 2006; Lindstedt et al. 2009; Hegna et al. strong stabilizing selection yet often display large amounts of in- 2013; Arenas et al. 2015). traspecific variation (Joron and Mallet 1998; Mappes et  al. 2005; Both warning signaling and thermoregulation depend heavily Stevens and Ruxton 2012). One of the main proposed mechan- on behavior, particularly microhabitat selection, which has received isms underlying variation in aposematic coloration is opposing se- surprisingly little attention in studies of these multiple functions of lection due to the many different functions of color combined with color. A  highly concealed organism will be harder for predators variation in the strength of these sources of selection (Briolat et al. to detect, regardless of coloration, so aposematism has been ar- 2019). In particular, coloration has a key role in thermoregulation, gued to be more likely to evolve in behaviorally exposed organisms particularly by ectotherms. To maximize fitness, ectotherms need (Speed and Ruxton 2005; Grant 2007). Even then, aposematic or- to stay warm without becoming too hot (Huey and Kingsolver ganisms can still face some risk of predation from naïve predators 1989). Color determines how efficiently ectotherms absorb sun- and predators, which can overcome their defenses (Endler and light, with dark colors favored in cool environments to help stay Mappes 2004; Mappes et  al. 2014). Thus, aposematic organisms warm and light colors favored in hot environments (Clusella Trullas should still prefer concealed microhabitats unless the exposed lo- et  al. 2007). At the same time, darker colors, although favored by cations offer some other benefit, such as food availability (Speed et al. 2010). At the same time, microhabitat selection greatly affects many factors that determine an ectotherm’s body temperature, in- cluding exposure to solar radiation (Stevenson 1985; Muñoz et  al. Address correspondence to M. E. Nielsen. E-mail: matthew.nielsen@ zoologi.su.se 2014). Thus, at least in cool environments, microhabitat preference © The Author(s) 2020. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology should experience opposing selection: thermoregulation should instar larva. Under laboratory conditions, A.  plantaginis moths can favor basking in an exposed, sunny microhabitat, but these loca- produce three generations per year and the third generation over- tions should be associated with a greater risk of predation, even for winters. Temperatures in the greenhouse followed the outdoor tem- aposematic animals. peratures and varied between 20 and 30 °C during the day (~20 h) To test how behavior alters both thermoregulation and apo- and decreased to 15–20 °C during the night (~4 h). These temper- sematic signaling, we studied the caterpillars of Arctia plantaginis atures correspond to average temperatures in July in Jyväskylä (data (Linneaus; formerly Parasemia plantaginis; Rönkä et  al. 2016). The available at the Finnish meteorological institute). See Lindstedt aposematic function of the adult moth’s color pattern has been et al. 2016 for more details on general caterpillar rearing. To quan- extensively studied (e.g., Lindstedt et  al. 2011; Nokelainen et  al. tify the length of the orange band, we counted the number of body 2012; 2014), and the degree of melanization in adults has also segments with at least some orange hairs on them (Ojala et  al. been shown to affect their body temperature (Hegna et  al. 2013). 2007). For experiments, caterpillars of the appropriate orange band A  similar relationship appears to exist in the caterpillars. Arctia length were haphazardly selected from among 14 families, with plantaginis caterpillars are black with an orange band, but there is some caterpillars reused across live-caterpillar experiments due to both genetic and plastic variation in the width of this band (Ojala limited availability of individuals with certain orange band lengths. et al. 2007; Lindstedt et al. 2009, 2010, 2016; Galarza et al. 2019). Short-term preference for exposed versus Predatory birds (specifically Parus major) learn to avoid caterpillars concealed locations with broader orange bands more quickly (Lindstedt et  al. 2008), likely due to the greater conspicuousness of this pattern (Mappes To determine the temperature threshold at which caterpillars et  al. 2014). On the other hand, caterpillars with smaller orange switched from preferring exposed to concealed locations, we placed bands and correspondingly more black coloration grow faster in single caterpillars (n  =  50) of varying orange band lengths (3–7 cool temperatures, likely because of improved thermoregulatory segments) under a halogen light (400 W, 8550 lm, 2800  K color ability (Lindstedt et al. 2009). Caterpillar behavior could heavily in- temperature, manufactured by CRX) and measured the time until fluence the optimization of predation risk and thermoregulation, the caterpillars sought shade under a host and the temperature at but the behavior of these caterpillars—like most caterpillars—has which they did so. During the experiment, room temperature was received little attention. Nevertheless, a previous study suggests that held at ~15  °C. At the start of each trial, we placed a caterpillar color pattern may affect the basking behavior of these caterpillars on a potted dandelion (Taraxacum sp.), which had been collected under cool temperatures (Lindstedt et al. 2009). that day from outside (individual plants were reused for multiple We started by investigating the behavior of A.  plantaginis, specif- tests). After the caterpillar stopped moving, we turned on a 400-W ically the caterpillars’ preference for exposed or concealed micro- halogen lampplaced 40  cm directly over the caterpillar and plant, habitats. In our studies, we used caterpillars from central Finland, gently repositioning any leaves that were initially shading the cater- where staying warm should be a priority for growth, but predation pillar (this never led to immediate movement). Each time the cater- rates are also relatively high (at least for adult moths; Nokelainen pillar began moving, we recorded the time elapsed, took a thermal 2014). Thus, we expected behavior to reflect a balance between image of the caterpillar using an infrared camera (FLIR C2, FLIR minimizing exposure to potential predators and maintaining a Systems), and recorded shade temperature using a thermocouple warm body temperature. Then, we investigated the effects of color shaded by a leaf. When the caterpillar stopped moving in a loca- pattern and microhabitat preference on temperature and pred- tion that was at least 20% shaded, we considered the prior move- ator avoidance using wild great tits, P.  major, as a model predator. ment thermoregulatory behavior and used the associated data for Parus major is a common generalist feeder and potential predator of our analysis. After excluding trials that did not occur properly (see A.  plantaginis in the wild, which can easily be trained for a variety Supplementary Methods for more details), final sample sizes were of foraging tasks in the lab and has been used in previous labora- 45, 44, and 45 for time, ambient temperature, and body tempera- tory experiments on A.  plantaginis predation (Lindstedt et  al. 2008, ture respectively. 2011). For both traits, we predicted opposing selection by thermo- To estimate the body temperature of the caterpillar from each regulation and predator avoidance: smaller orange bands (thus, image, we first converted the image to linear-temperature grayscale more melanized caterpillars) would be favored for thermoregula- using FLIR Tools (v6.3, FLIR Systems). Then, we measured the tion but selected against by predation and, similarly, exposed loca- temperature by outlining the caterpillar by hand in imageJ (v1.51; tions would be favored for thermoregulation but selected against by Schneider et  al. 2012) and computing the mean temperature. We predation. evaluated the replicability of this approach by having another person remeasure 10 randomly selected images. These two meas- urements were highly correlated (r  =  0.998), indicating the preci- sion and consistency of this approach. The repeated measurements METHODS differed significantly by a mean of 0.53  °C (paired t-test, standard Rearing of caterpillars and general procedures –5 deviation = 0.23, t = 7.14, P = 5.4 × 10 ), so this approach should Arctia plantaginis caterpillars used in experiments came from a labo- be accurate within 1 °C. We tested whether the body temperature, ratory stock population originally derived from wild females caught shade temperature, and time from the start of heating at which in Central Finland in 2012 and supplemented annually with new the caterpillar sought shade were affected by the length of the wild-caught individuals. The stock was maintained in a greenhouse caterpillar’s orange band using separate linear mixed-effects models at the University of Jyväskylä on a mixed diet of lettuce (Lactuca with the individual plant used in the trial as a random effect. sativa var. crispa) and dandelion (Taraxacum sp.). Experiments were Caterpillar mass (measured posttrial with 0.1 mg precision) was in- conducted in 2015, at which point the lab stock had been main- itially included as an additional fixed effect; however, it had no sig- tained for eight generations. This species usually has one genera- nificant effect on the results for any response variable (P > 0.2), so tion per year in the wild and typically overwinters as third or fourth it was excluded from the final analysis. All statistical analyses were Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism conducted in R (v3.4.0; R Core Team 2018). For this analysis, we cardboard ~30  cm above the ground. During the period of un- used the lme function in the nlme package (Pinheiro et  al. 2018) obscured sunlight, we periodically recorded thermal images of and P-values were generated using likelihood ratio tests between the caterpillars, ultimately using the last pictures taken for each models with and without each fixed effect. condition during full sunlight (after 25  min without shade and after 20  min with half shaded) since these should represent near- Long-term preference for exposed versus equilibrium temperatures. Ambient temperatures (recorded using a concealed locations shaded thermocouple) during these two measurements were within 0.5 °C of each other and, thus, were not factored into the analysis. To test overall preference of caterpillars for exposed versus con- To analyze these images, we used a similar method to the short- cealed positions over longer time periods, including whether color term preference experiment; however, resolution and focal limita- pattern affected this preference, we placed caterpillars (n  =  36) in tions of the thermal camera prevented precise identification of the sets of three (one each of band lengths 4, 5, and 6 segments) on edges of each caterpillar. Instead, we estimated each caterpillar’s host plants (n = 12), and left them there under less intense heat to temperature by using ImageJ to find mean temperature of a 6-pixel observe how much time they spent exposed during the day. As be- diameter circle centered on the warmest point in each caterpillar. fore, recently potted dandelions (Taraxacum sp.) were used as hosts; We also estimated the temperature of a circle at the center of the however, in this case, the dandelion was enclosed in a cylinder of cardboard background in each image to compare its temperature clear acrylic to prevent escape and the leaves were trimmed to fit. in the shade versus sun treatments (Supplementary Methods). We The evening before the trial, the caterpillars were placed on each analyzed the effects of caterpillar orange band length, whether the plant and allowed to adjust to the chambers overnight. Size of the caterpillar was exposed or shaded, and their interaction using a orange patch size was used to distinguish each of the three individ- linear mixed-effects model. We included caterpillar mass as an ad- uals per plant during observations. The next morning, at 9 AM, we ditional fixed effect and the identity of each caterpillar as a random turned on two 400-W halogen lights ~75 cm over the table. At this effect. We also allowed variance to depend on shading treatment. time, 24 out of 36 caterpillars were in exposed positions. Over the We used the lme function in the nlme package for this analysis and next 9 h, we observed the position of each caterpillar every 30 min, P-values were generated using likelihood ratio tests between the full noting whether the caterpillar was fully exposed or at least partially model and models without each fixed effect. To further examine (20%) hidden. Of necessity, enclosures varied in exact distance the large effect of the shading treatment that we found, we also from these lights, producing a range of mean surface temperat- performed an additional likelihood ratio test comparing models ures from 14.5 to 27.3 °C among the enclosures over the course of with no interaction term and with or without shading treatment. the experiment (mean temperature of each enclosure in overhead thermal images taken every 30 min). Caterpillars that showed signs Effect of color and position on predation risk of building a shelter in which to molt or otherwise starting to molt were excluded from analysis from that observation onward (101 ob- To test how exposure and orange band length affected relative servations were excluded out of 708 in total; 11 caterpillars had predation risk, we conducted a laboratory experiment using dead observations excluded, and only one individual was fully excluded). caterpillars and wild-caught P.  major (great tits). Forty-eight birds To test for an effect of orange band length on preference for ex- were trapped between 6 August and 8 December 2015 and kept posed versus hidden positions, we used a generalized linear mixed for up to 2 weeks in captivity for trials with A.  plantaginis, and 16 model with a binomial distribution. We used whether the cater- birds were trapped between 14 February and 1 March for trials pillar was observed exposed as our dependent variable, band length with mealworms. In captivity, they were housed individually in and enclosure surface temperature as fixed effects, and individual plywood cages under an 11:13  h (light:dark) photoperiod and fed nested within pot as a random effect. Models were fit using the with sunflower seeds and vitamin-enriched tall ad libidum. Prior to glmer function in the lme4 package (Bates et al. 2015) in R, and a the experiment, birds had no previous experience in captivity with P-value was generated using a likelihood ratio test between models A. plantaginis (whether they had encountered A. plantaginis in the wild with and without each fixed effect. is unknown but unlikely because there are no known wild popula- tion of A. plantaginis in the vicinity of Konnevesi research station). Effect of color and position on caterpillar Each bird was given an array of 12 caterpillars that were all ei- temperature ther exposed or concealed but varied within array in orange band To test how exposure and orange band length affect the body tem- length from 4 to 6.  This created two overall treatments (exposed perature of caterpillars, we measured the temperature of caterpil- vs. concealed) while also creating within-treatment variation in lars (n  =  48) when placed outside under sunny conditions either warning coloration. Each array contained 12 dead A. plantaginis cat- with or without artificial shading. We tested caterpillars with orange erpillars (killed by freezing but thawed before use): 4 each of band band lengths from 4 to 7 segments, 12 of each size. Caterpillars length 4, 5, and 6.  Each caterpillar was also weighed before the were tethered to a piece of cardboard with 5  cm between indi- trial. Arrays were presented to birds in a 21.0- × 29.7-mm card- viduals using thin string to immobilize them. Caterpillars of each board box with a foam insert covered in brown paper. We filled signal size were evenly divided between two sides but otherwise this box with a 2–3-cm deep layer of dried, dead leaves, predom- placed randomly. Over the course of the experiment, three indi- inantly Betula pendula (silver birch). Each caterpillar was pinned in viduals escaped and were thus excluded from analysis. We con- the box using a black-painted insect pin, 2–3 cm from the edge and ducted the experiment in Jyväskylä (62.230°N, 25.744°E) on 25 8–9 cm from each other, with individuals assigned to specific loca- September 2015. On this day, we placed the tethered caterpillars tions randomly. Depending on treatment, we pinned all caterpillars outside at 2:38 PM and started the experiment when, at 3:31 PM, either below (concealed) or on top of (exposed) the layer of leaves. an extended period of continuous sunlight began. After 28 min of To ensure we had enough caterpillars for all replicates of the ex- unobscured sunlight, half of the array was shaded by a piece of periments, some trials were conducted using caterpillars descended Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology from A. plantaginis originally collected in Estonia (proportion of cat- differences among different orange band lengths using the package erpillars attacked during each trial did not differ significantly be- emmeans (Lenth 2018) and generating P-values from the resulting tween Finnish and Estonian populations and, thus, were pooled for z-scores. all analyses; t-test, t   =  0.275, P  =  0.785). We also repeated both the hidden and exposed treatments using arrays of 12 mealworms RESULTS instead of A. plantaginis caterpillars to test the effect of concealment Effect of color on behavioral exposure—short term on palatable and nonwarningly colored prey. Trials were conducted in a separate wooden box (50-cm wide × We successfully tested 45 caterpillars for conditions under which 50-cm deep × 67-cm high) with a one-way mirror to allow obser- they sought at least partial shade. Caterpillars with larger orange vation and recording (Supplementary Methods). Before each trial, bands sought shade significantly later than those with smaller each bird was food deprived for 2 h total, 90 min in its home box orange bands (β = 0.56  min/segment, standard error [SE] = and 30 min acclimating to the trial box. At that time, the lights in 0.24, = 4.80, P = 0.028; Figure 1a) and they also did so at higher the box were briefly turned off and the array placed in the box. shade temperatures (β = 0.46  °C/segment, SE = 0.22,  =  4.29, Birds that did not begin foraging (as indicated by interacting with P  =  0.038; Figure  1b). Band length did not, however, signifi- a caterpillar) within 45  min after the lights were turned back on cantly affect the body temperature at which caterpillars began were excluded from the study as were birds that foraged for less seeking shade (β = 0.18  °C/segment, SE  =  0.33,  =  0.329, than 30  min total within 115  min with the box. We ultimately P  =  0.57; Figure  1c), which was quite high overall with a mean excluded 12 birds in this way, evenly divided between treatments. of 37.9 °C. Once foraging began, the trial continued until either the bird had attacked all caterpillars or 30  min of foraging had elapsed (ex- cluding any breaks 5 min or longer spent not interacting with cat- erpillars or the box). During the trial, we noted which caterpillars (a) were attacked, specifically, if the caterpillar was pulled off its pin or, in a few cases, consumed while on the pin. We also recorded when the attack occurred to provide information about the order of attack. At the end of the trial, we confirmed which caterpillars were attacked by checking which caterpillars remained on their pins. Except for the first two birds tested, we also estimated the total mass of caterpillar the bird ate by weighing the caterpillar remains of the attacked caterpillars and subtracting from their original weight. In one case, the estimated mass eaten was slightly 2 negative (likely due to measurement error) and was set to 0 be- (b) cause it indicated that the bird had eaten none of the caterpillar despite removing them from the pin. Overall, we successfully tested 36 birds, 18 each with concealed or exposed caterpil- lars, plus 16 more birds with mealworms, 8 each concealed and exposed. We tested how caterpillar exposure affected whether a caterpillar was attacked during a trial using a generalized mixed model with a binomial distribution, with exposure treatment as a fixed effect and individual bird as a random effect. To account for rejection of prey after attack, we also tested the effect of caterpillar exposure on the total mass consumed by each bird using a linear model with (c) exposure as a fixed effect. We repeated both analyses separately for the mealworms. To analyze the effects of color on predation risk, we considered only the first six attacks by each bird (or fewer if 38 the bird attacked less than six caterpillars during the trial) because each bird only had access to three caterpillars of each band length, so the distribution of attacks would necessarily converge toward an equal attack rate on all band lengths if all attacks were con- sidered. We then fit a generalized mixed model using a binomial distribution for whether or not each caterpillar was one of the first six attacked. Exposure treatment, orange band length (as a factor), mass, and row in the array (indicating proximity to the bird’s ini- Orange band length (body segments) tial perch) were used as fixed effects, whereas individual bird was Figure 1 a random effect. The mixed models were fit using the glmer func- Effects of caterpillar orange band length (in number of segments) on the tion in the lme4 package in R and P-values were generated using start of shade-seeking behavior when heated by an overhead halogen a likelihood ratio test between models with and without each fixed light. Lines represent significant (P  <  0.05) fixed effects of band length on effect. All second- and third-degree interactions among fixed effects the response variable in a mixed model. (a) Time (minutes) from start of were nonsignificant (P > 0.2) and excluded from the final analysis. experiment, n  =  45. (b) Surface temperature (degree Celsius) of shaded We performed a Tukey’s post hoc test on the final model for the ground, n =44. (c) Body temperature (degree Celsius) of caterpillar, n = 45. Body temperature (°C) at Shade temperature (°C) at Time (min) at start of shade seeking start of shade seeking start of shade seeking Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism birds did not, however, eat significantly more mealworms (0.123 g, Effect of color on behavioral SE = 0.102 g, t  = 1.20, P = 0.25). exposure—long term 14 To assess how color affected risk of predation, we focused on the We observed 35 caterpillars over a 9-h period. During the vast ma- first six caterpillars attacked by each bird. Larger caterpillars were jority (91.8%) of our observations, caterpillars were fully exposed more likely to be attacked, as were caterpillars closer to the bird’s to the light, spending only 8.2% of their time fully or partially perch (Table  2). Most importantly, orange band length had a sig- concealed. Although caterpillars chose concealed locations infre- nificant effect on risk of predation (Table  2; Figure  4); specifically, quently, orange band length nevertheless affected how often this occurred. Caterpillars with larger orange bands were significantly more likely to be observed fully exposed (β = 0.836 ± 0.276, = 8.43, P = 0.004). Specifically, caterpillars with four orange segments were exposed 84.4% of the time, five orange segments 95.0% of the time, and six orange segments 95.7% of the time. Caterpillars were 25 also significantly more likely to be observed fully exposed when the enclosure was warmer (β = 0.233 ± 0.066, = 12.09, P = 0.0005). Effect of color and position on temperature We measured the temperature of 45 caterpillars under sunny con- ditions twice, once with all in full sun the other with half in full sun and half shaded. We found a significant interaction between the effects of orange band length and shade treatment on temper - Orange band length (body segments) ature (Table  1; Figure  2); however, the relative effect sizes of these Figure 2 two factors differed dramatically. Shading had a large overall effect Body temperature of caterpillars with varying orange band lengths when on caterpillar temperature, cooling caterpillars by 10.7  °C aver- placed outdoors either in full sun or shade. Gray circles and dashed line aged across orange band lengths (  =  142.3, P  <  0.0001). This dif- represent caterpillars in full sun; black squares and solid line represent ference is only partially explained by the effect of shading on the caterpillars in shade. Lines show interacting fixed effects from a mixed caterpillar’s microhabitat (the temperature of the cardboard back- model (P = 0.0365). n = 90 observations of 45 caterpillars. ground differed by 6.6 °C between these treatments). Orange band length, on the other hand, had only a small effect on temperature (a) (b) when in the sun (−0.349  °C/orange segment) and practically no 1.0 effect when shaded (−0.027 °C/orange segment). Caterpillar mass ranged from 0.108 to 0.307  g (mean 0.195  g), and body tempera- ture increased significantly with mass by 9.63 °C/g (Table 1). 0.8 n = 8 Effect of color and position on predation risk 0.6 During 30  min of foraging, P.  major attacked a smaller proportion of A.  plantaginis caterpillars in the exposed than concealed treat- 0.4 n = 8 ment (β = −1.27  ± 0.48,  =  6.53, P  =  0.011; Figure  3). This cor- n = 18 responds to a mean of 2.61 fewer caterpillars (out of 12)  attacked 0.2 when exposed. From the attacked caterpillars, the birds ate, on n = 18 average, 0.372  g less A.  plantaginis caterpillar in the exposed than 0.0 concealed treatment (SE = 0.124  g, t   =  −3.00, P = 0.005). This mass is equivalent to 2.55 caterpillars (mean mass of attacked cat- Concealed Exposed Concealed Exposed erpillars was 0.146 g, SE = 0.0028), indicating that most of the ad- Caterpillar Mealworm ditional caterpillars attacked when concealed were also consumed. Figure 3 In contrast, when foraging on mealworms, birds attacked a greater Boxplot of proportion of (a) A.  plantaginis caterpillars or (b) mealworms proportion of individuals in the exposed than concealed treatment attacked by P. major during foraging trials. Each bird had access to either 12 (β = 1.54  ± 0.50,  =  8.63, P  =  0.003; Figure  3), corresponding to dead caterpillars or mealworms, all of which were pinned in place either a mean of 2.62 more mealworms attacked when exposed. The exposed on top of or concealed beneath a layer of dead leaves. Table 1 Effects of orange band length, position (sun or shade), and mass on temperature (degree Celsius) of A. plantaginis caterpillars Coeff. SE df χ P Orange band length −0.0270 0.090 1 — — Position 13.00 0.879 1 — — Mass (g) 9.63 2.7 1 12.52 0.0004 Orange band length × position −0.322 0.155 1 4.37 0.0365 df, degrees of freedom.  Difference between sun and shade. Proportion attacked Body temperature (°C) Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology Table 2 Effects of exposure treatment, orange band length, mass, and position relative to bird’s perch (rows away) on whether A. plantaginis caterpillars were among the first six caterpillars attacked in an array of 12 by P. major in laboratory trials Coeff. SE df χ P Treatment −0.324 0.213 1 2.33 0.127 Orange band length — — 3 11.28 0.0103 5 body segments −0.097 0.305 — — 6 body segments −0.841 0.326 — — 7 body segments 0.056 0.321 — — Mass (g) 8.75 2.43 1 13.55 0.0002 Distance from perch (rows) −0.980 0.138 1 57.35 <0.0001 Difference between exposed and concealed treatments. b 2 Analyzed as a factor. χ and P-values tested across all levels, coefficients given separately for each level compared with a length of four body segments as a baseline. See Figure 4 for more details. were both much warmer and less likely to be attacked by P.  major 1.0 in our experiments. This parallel selection by both predators and temperature favoring caterpillars in exposed locations helps ex- 0.8 plain the strong preference for exposed positions displayed by the caterpillars in our behavioral experiments. Over 9  h of observa- a a ab 0.6 tions under moderately warm thermal conditions, caterpillars spent the vast majority of their time in the open. This behavior suggests 0.4 that these caterpillars rely on their primary and secondary defense strategies (coloration, hairiness, and chemical defense) for predator 0.2 avoidance. Our findings support “a classic view of aposematism” in which conspicuousness reduces the likelihood of recognition 0.0 errors because predators can detect conspicuous prey at a greater 45 67 distance and, thus, avoid them more reliably (Guilford 1986). There Orange band length (body segments) has, however, been surprisingly few systematic observations of Figure 4 microhabitat selection by aposematic prey in this regard (but see Probability of A.  plantaginis caterpillars with varying orange band length Tabadkani and Nozari 2014; Rößler et al. 2019). being among the first six attacked by P.  major in an array of 12 caterpillars From a thermal perspective, caterpillars did not seek shade until (three of each band length). Bars represent 95% confidence interval, and heated to a surprisingly high mean body temperature of 37.9  °C. lettering indicates significantly differing band lengths in a Tukey’s post hoc This body temperature is comparable to and, in some cases, ex- test (n = 34 birds). ceeds the body temperature at which other species of caterpillar from much warmer environments begin heat-avoidance behavior caterpillars with an orange band 6 segments long were least likely (Sherman and Watt 1973; Nielsen et  al. 2018). This observation to be among the first six caterpillars attacked and significantly less fits with a general trend that the upper thermal limits of both in- likely to be attacked than caterpillars with an orange band 4 or 7 sects and ectotherms decrease little with latitude (Addo-Bediako segments long according to post hoc tests. Given that we limited et al. 2000; Sunday et al. 2014). For high-latitude arthropods, it can our analysis to the first six caterpillars attacked, it is unsurprising be important to capitalize on brief warm, sunny periods to maxi- that exposure had no significant independent effect on attack risk; mize growth (Kukal et al. 1988; Bennett et al. 1999; Birkemoe and however, it also did not significantly interact with orange band Leinaas 2000). length ( = 1.08, P = 0.78). On the other hand, the preference for exposure should represent a risk for animals with warning colors (Endler and Mappes 2004; DISCUSSION Mappes et al. 2014). This risk should be especially great for apose- Coloration of aposematic animals provides an example of op- matic animals like A.  plantaginis that are only somewhat distasteful, posing selection promoting trait variation: trade-offs between so experienced predators will still eat them in the absence of pre- warning signal efficacy and thermoregulatory ability (as well as ferred prey as occurred in our experiment. Nevertheless, we found, other functions) can increase variation in a trait that should other- under laboratory conditions at least, that potential avian predators wise be highly conserved (Briolat et al. 2019). We predicted a sim- (wild-caught, adult P.  major) attacked fewer A.  plantaginis caterpil- ilar opposing selection on microhabitat choice in A.  plantaginis, an lars when the caterpillars were exposed than when they were con- aposematic caterpillar, so we tested how color pattern and behavior cealed under dead leaves, despite the latter requiring more effort interact with body temperature and predation risk in this species. to locate the caterpillars. The opposite pattern occurred with pal- We found that, although the effects of these two traits were not en- atable, nonaposematic mealworms. The greater risk to concealed tirely independent, behavior had a greater effect on both sources of A.  plantaginis is reinforced by the fact that birds attacked similar selection. proportions of concealed caterpillars and concealed mealworms Contrary to our initial expectations, however, we found no during the respective experiments, despite how P. major strongly pre- evidence of opposing selection on microhabitat preference in fers mealworms. Birds may have attacked more hidden caterpillars A.  plantaginis caterpillars. Individuals placed in exposed positions because the leaves obscured the warning signals of the caterpillars. Probability of attack during first 6 attacks Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism If this hypothesis explained our results, however, we would have ex- the effectiveness of the caterpillars’ warning signals (Aronsson and pected concealment to alter the effect of orange band length on Gamberale-Stille 2013; Barnett et  al 2016; although see Aronsson attack risk, which we did not observe. Instead, the greater effort re- and Gamberale-Stille 2009). Regardless of the exact relation- quired to find concealed caterpillars could explain the greater will- ship between band length and predation, the orange band lengths ingness of P.  major to attack them. Birds were frequently observed least likely to be attacked differed from those which were warmest, digging through leaves even in the exposed treatment when nothing indicating the expected opposing selection on color pattern from was present there. In the exposed treatment, the birds may have temperature and predation. initially ignored the exposed caterpillars to search for the poten- Ultimately, we found the expected opposing selection on one tially better prey that could have existed under leaves in the wild. trait (color pattern) but not another (microhabitat preference). The In the concealed treatment, on the other hand, by the time the presence or absence of opposing sources of selection could help ex- birds found the caterpillars hidden under the leaves, there were no plain the differing levels of phenotypic variation observed in these additional locations to search, so the birds may have been more traits. Color pattern in both adult and larval A. plantaginis varies ex- willing to accept them as prey. Notably, this digging behavior was tensively due to both genetic and plastic factors both within and performed by wild-caught birds without training. Indeed, great tits between populations (Ojala et al. 2007; Lindstedt et al. 2009, 2016; have a greater propensity than other tits for feeding on the ground Hegna et al. 2013, 2015), and opposing selection from a variety of (Gosler and Clement 2007). Regardless of the underlying cause, sources, including sexual selection, initial detectability, and thermo- the lower attack risk for exposed caterpillars combined with their regulation, has been argued to enable persistence of this variation warmer temperature helps explain the strong behavioral preference (Lindstedt et  al. 2008, 2009; Nokelainen et  al. 2012; Hegna et  al. we observed in these caterpillars for open, exposed positions. Based 2013; Mappes et  al. 2014). The behavior of A.  plantaginis, on the on these results, we would predict opposing selection by predation other hand, has received little attention, particularly in the cater- and temperature on microhabitat preference in warm environments pillars, but they generally display little activity, at least under lab- instead of cool ones, but the trade-off may not occur in that con- oratory conditions. The minimal variation we find in preferred text either. Battus philenor, a desert-dwelling aposematic caterpillar, microhabitat, and perhaps other aspects of their behavior, could remains fully exposed while still avoiding high temperatures (Nice be caused by having a single preferred microhabitat favored by and Fordyce 2006; Nielsen and Papaj 2017). multiple sources of selection removing the need for temporal or Variation in color, on the other hand, had much weaker effects between-individual variation in habitat preference. than behavior on both body temperature and predation risk. For An important general benefit of aposematism is the opportu- temperature, the effect of orange band length on body temper - nity to occupy microhabitats that are otherwise highly vulnerable ature depended heavily on the position of the caterpillar and, to predation (Speed et  al. 2010). Here, we find that this effect in thus, its behavior: caterpillars with smaller orange bands (more A.  plantaginis is stronger than anticipated despite the fact that the melanin) were warmer as expected when exposed to the sun, but species is only unpalatable and hairy rather than truly toxic. This band length had little effect in the shade. Color and other aspects benefit of aposematism effectively negates the opposing selection of morphology are generally predicted to have a weaker effect on faced by cryptic organisms between behaviors that maximize re- temperature than behavior (Stevenson 1985), and the effect of source use (including favorable thermal environments) and mini- color on temperature has been shown to depend on behavior in mize predator exposure (Speed and Ruxton 2005). Multiple other a wide range of insects (e.g., Kingsolver 1987; Nielsen and Papaj aposematic animals behave in ways that expose them more to 2017). Despite its smaller effect, color pattern did alter thermoreg- predators than comparable cryptic animals (Pinheiro 1996, 2007; ulatory behavior in A.  plantaginis. The body temperature at which Rudh et  al. 2013; Willink et  al. 2013; Tabadkani and Nozari shade-seeking began did not change with band length; however, 2014; Valkonen et  al. 2014; Rößler et  al. 2019). Based on our re- caterpillars with shorter orange bands (more melanic caterpillars) sults, we predict that aposematic organisms may frequently use started shade-seeking behavior sooner and at lower environmental different microhabitats from cryptic organisms and may show re- temperatures. These results indicate that the caterpillars’ internal duced behavioral variation at least in terms of habitat preference or physiology and response to body temperature do not change with antipredator behavior. At the same time, if an aposematic species color pattern and, instead, color pattern alters behavior by chan- is thus specialized for an exposed microhabitat due to parallel se- ging light absorption and, thus, heating rate (Nielsen et  al. 2018). lection from multiple sources, they may be more vulnerable to any This ability of body color to affect thermoregulatory behavior by environmental changes, such as climate change, which reduce their changing body temperature has also been demonstrated in a range fitness in that habitat. Future research could test for selection to of insects (e.g., Kingsolver 1987; Karpestam et  al. 2012; Nielsen prefer exposed microhabitats in other aposematic species and con- et al. 2018). sider additional sources of selection, which may benefit exposure, We confirmed effect of color pattern on predation risk by com- such as sexual selection and foraging. paring the first six caterpillars attacked by P.  major during experi- Regardless, we have shown for an aposematic species that, when ments. Caterpillar with an orange band 6 body segments long were facing wild predators, they are less likely to be attacked when ex- less likely to be attacked, regardless of whether they were hidden or posed. Combined with the thermal benefits of exposure, the greater exposed. This is longer than the average signal size of A. plantaginis safety of caterpillars when exposed helps explain their minimal be- (4.7 segments) but not the largest possible (7 segments). Previous havioral variation. Caterpillars rarely occupied concealed positions work on A.  plantaginis indicates that larger orange bands (>5 seg- except at extreme temperatures. Although variation in color pat- ments) are more effective warning signals than short bands (<4 seg- tern had smaller direct effects than exposure on temperature and ments; Lindstedt et  al. 2008, Mappes et  al. 2014). We add to this predation risk in our experiments, color pattern’s effect on tempera- that warning signal effectiveness might also decrease for the longest ture lead to a corresponding change in thermoregulatory behavior. bands (7 segments). The small black content of these caterpillars Thus, our results reinforce the importance of both behavior and could reduce their internal contrast, which, in turn, could reduce color pattern for the function of aposematic signals not only in the Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Behavioral Ecology context of predation where they are typically studied but also in a World. Picathartes to tits and chickadees. Vol. 12. Barcelona (Spain): Lynx Edicions. p. 662–709. thermoregulatory context. Grant JB. 2007. Ontogenetic colour change and the evolution of aposema- tism: a case study in panic moth caterpillars. J Anim Ecol. 76:439–447. Guilford T. 1986. How do “warning colours” work? Conspicuousness SUPPLEMENTARY MATERIAL may reduce recognition errors in experienced predators. Anim Behav. Supplementary data are available at Behavioral Ecology online. 34:286–288. Hegna  RH, Galarza  JA, Mappes  J. 2015. Global phylogeography and ge- ographical variation in warning coloration of the wood tiger moth FUNDING (Parasemia plantaginis). J Biogeogr. 42:1469–1481. Hegna  RH, Nokelainen  O, Hegna  JR, Mappes  J. 2013. To quiver or to The work was supported by a Graduate Research Opportunities Worldwide shiver: increased melanization benefits thermoregulation, but re- supplement to National Science Foundation Graduate Research Fellowship duces warning signal efficacy in the wood tiger moth. Proc Biol Sci. Program (award number DGE-1143953) and the Academy of Finland 280:20122812. (grant # 293513). Huey  RB, Kingsolver  JG. 1989. Evolution of thermal sensitivity of ecto- therm performance. Trends Ecol Evol. 4:131–135. We would like to thank Helinä Nisu for help with bird rearing and Tuuli Joron  M, Mallet  JL. 1998. Diversity in mimicry: paradox or paradigm? Salmi for assistance with the mealworm experiments. Aimee Deconinck, Trends Ecol Evol. 13:461–466. Karpestam E, Wennersten L, Forsman A. 2012. Matching habitat choice by Geoffrey Legault, Jee Yun Lee, Juan Galarza, Emily Burdfield-Steel, Sara experimentally mismatched phenotypes. Evol Ecol. 27:893–907. Calhin, and two anonymous reviewers provided valuable feedback on the Kingsolver JG. 1987. Evolution and coadaptation of thermoregulatory be- manuscript. Research using wild P.  major was conducted with permission havior and wing pigmentation pattern in pierid butterflies. Evolution. of Southwest Finland Centre for Economic Development, Transport and 41:472–490. Environment (VARELY/294/2015) and license from the National Animal Kukal  O, Heinrich  B, Duman  JG. 1998. Behavioural thermoregulation in Experiment Board (ESAVI/9114/04.10.07/2014). the freeze-tolerant Arctic caterpillar, Gynaephora groenlandica. J Exp Biol. 138:181–193. Author’s contributions: M.E.N. and J.M. designed the studies. M.E.N. con- Lenth  R. 2018. emmeans: estimated marginal means, aka least-squares ducted the experiments and analyzed the data. M.E.N.  led the writing of means. R package version 1.3.1. Available from: https://CRAN.R- the manuscript. M.E.N.  and J.M.  edited the manuscript and approved the project.org/package=emmeans (accessed 20 December 2018). final version. Lindstedt  C, Eager  H, Ihalainen  E, Kahilainen  A, Stevens  M, Mappes  J. 2011. Direction and strength of selection by predators for the color of Data accessibility: Analyses reported in this article can be reproduced using the aposematic wood tiger moth. Behav Ecol. 22:580–587. the data provided by Nielsen and Mappes (2020). Lindstedt C, Lindström L, Mappes J. 2008. Hairiness and warning colours as components of antipredator defence: additive or interactive benefits? Handling editor: Luke Holman Anim Behav. 75:1703–1713. Lindstedt  C, Lindström  L, Mappes  J. 2009. Thermoregulation constrains effective warning signal expression. Evolution. 63:469–478. Lindstedt  C, Morehouse  N, Pakkanen  H, Casas  J, Christedes  J-P, REFERENCES Kemppainen  K, Lindström  L, Mappes  J. 2010. Characterizing the pig- Addo-Bediako A, Chown SL, Gaston KJ. 2000. Thermal tolerance, climatic ment composition of a variable warning signal of Parasemia plantaginis variability and latitude. Proc Biol Sci. 267:739–745. larvae. Funct Ecol. 24:759–766. Arenas LM, Walter D, Stevens M. 2015. Signal honesty and predation risk Lindstedt  C, Schroderus  E, Lindström  L, Mappes  T, Mappes  J. 2016. among a closely related group of aposematic species. Sci Rep. 5:11021. Evolutionary constraints of warning signals: a genetic trade-off between Aronsson  M, Gamberale-Stille  G. 2009. Importance of internal pattern the efficacy of larval and adult warning coloration can maintain variation contrast and contrast against the background in aposematic signals. in signal expression. Evolution. 70:2562–2572. Behav Ecol. 20:1356–1362. Mappes  J, Kokko  H, Ojala  K, Lindström  L. 2014. Seasonal changes in Aronsson  M, Gamberale-Stille  G. 2013. Evidence of signaling benefits to predator community switch the direction of selection for prey defences. contrasting internal color boundaries in warning coloration. Behav Ecol. Nat Commun. 5:5016. 24:349–354. Mappes  J, Marples  N, Endler  JA. 2005. The complex business of survival Barnett  JB, Scott-Samuel  NE, Cuthill  IC. 2016. Aposematism: balancing by aposematism. Trends Ecol Evol. 20:598–603. salience and camouflage. Biol Lett. 12:20160335. Muñoz  MM, Stimola  MA, Algar  AC, Conover  A, Rodriguez  AJ, Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects Landestoy  MA, Bakken  GS, Losos  JB. 2014. Evolutionary stasis and la- models using lme4. J Stat Softw. 67:1–48. doi: 10.18637/jss.v067.i01. bility in thermal physiology in a group of tropical lizards. Proc Biol Sci. Bennett  VA, Kukal  O, Lee  RE. 1999. Metabolic opportunists: feeding and 281:20132433. temperature influence the rate and pattern of respiration in the high Nice  CC, Fordyce  JA. 2006. How caterpillars avoid overheating: behav- arctic woollybear caterpillar Gynaephora groenlandica (Lymantriidae). J Exp ioral and phenotypic plasticity of pipevine swallowtail larvae. Oecologia. Biol. 202:47–53. 146:541–548. Birkemoe T, Leinaas HP. 2000. Effects of temperature on the development Nielsen  ME, Levin  E, Davidowitz  G, Papaj  DR. 2018. Colour plasticity of an Arctic Collembola (Hypogastrura tullbergi). Funct Ecol. 14:693–700. alters thermoregulatory behaviour in Battus philenor caterpillars by modi- doi: 10.1046/j.1365-2435.2000.00478.x. fying the cue received. Anim Behav. 140:93–98. Briolat  ES, Burdfield-Steel  ER, Paul  SC, Rönkä  KH, Seymoure  BM, Nielsen,  ME, Mappes  J. 2020. Data from: Out in the open: behavior’s ef- Stankowich  T, Stuckert  AMM. 2019. Diversity in warning colora- fect on predation-risk and thermoregulation by aposematic caterpillars. tion: selective paradox or the norm? Biol Rev Camb Philos Soc. 94: Behav Ecol. doi: 10.5061/dryad.95x69p8gk. 388–414. Nielsen  ME, Papaj  DR. 2017. Why have multiple plastic responses? Clusella  Trullas  S, van  Wyk  J-H, Spotila  JR. 2007. Thermal melanism in Interactions between color change and heat avoidance behavior in Battus ectotherms. J Therm Biol. 32:235–245. philenor larvae. Am Nat. 189:657–666. Endler JA, Mappes J. 2004. Predator mixes and the conspicuousness of ap- Nokelainen  O, Hegna  RH, Reudler  JH, Lindstedt  C, Mappes  J. 2012. osematic signals. Am Nat. 163:532–547. Trade-off between warning signal efficacy and mating success in the Galarza JA, Dhaygude K, Ghaedi B, Suisto K, Valkonen J, Mappes J. 2019. wood tiger moth. Proc Biol Sci. 279:257–265. Evaluating responses to temperature during pre-metamorphosis and Nokelainen O, Valkonen J, Lindstedt C, Mappes J. 2014. Changes in pred- carry-over effects at post-metamorphosis in the wood tiger moth (Arctia ator community structure shifts the efficacy of two warning signals in plantaginis). Philos Trans R Soc Lond B Biol Sci. 374:20190295. Arctiid moths. J Anim Ecol. 83:598–605. Gosler  A, Clement  P. 2007. Family Paridae (tits and chickadees). In Ojala  K, Lindström  L, Mappes  J. 2007. Life-history constraints and del  Hoyo  J, Elliott  A, Christie  D, editors. Handbook of the birds of the warning signal expression in an arctiid moth. Funct Ecol. 21:1162–1167. Downloaded from https://academic.oup.com/beheco/article/31/4/1031/5840992 by DeepDyve user on 19 July 2022 Nielsen and Mappes • Predation, temperature, and aposematism Pinheiro CEG. 1996. Palatability and escaping ability in Neotropical butter- Speed MP, Brockhurst MA, Ruxton GD. 2010. The dual benefits of apose- flies: test with wild kingbirds. Biol J Linn Soc. 59:351–365. matism: predator avoidance and enhanced resource collection. Evolution. Pinheiro  CEG. 2007. Asynchrony in daily activity patterns of butterfly 64:1622–1633. models and mimics. J Trop Biol. 23:119–123. Speed MP, Ruxton GD. 2005. Aposematism: what should our starting point Pinheiro J, Bates D, DebRoy S, Sarkar D; R Core Team. 2018. nlme: linear and be? Proc R Soc B. 272:431–438. nonlinear mixed effects models. R package version 3.1–131.1. Available from: Stevens M, Ruxton GD. 2012. Linking the evolution and form of warning https://CRAN.R-project.org/package=nlme (accessed 20 February 2018). coloration in nature. Proc Biol Sci. 279:417–426. Prudic KL, Skemp AK, Papaj DR. 2006. Aposematic coloration, luminance Stevenson RD. 1985. The relative importance of behavioral and physiolog- contrast, and the benefits of conspicuousness. Behav Ecol. 18:41–46. ical adjustments controlling body temperature in terrestrial ectotherms. R Core Team. 2018. R: a language and environment for statistical com- Am Nat. 126:362–386. puting. Vienna (Austria): R Foundation for Statistical Computing. Sunday JM, Bates AE, Kearney MR, Colwell RK, Dulvy NK, Longino JT, Available from: https://www.R-project.org/ (accessed 1 May 2017). Huey RB. 2014. Thermal-safety margins and the necessity of thermoreg- Rößler  DC, Lötters  S, Mappes  J, Valkonen  JK, Menin  M, Lima  AP, ulatory behavior across latitude and elevation. Proc Natl Acad Sci USA. Pröhl  H. 2019. Sole coloration as an unusual aposematic signal in a 111:5610–5615. Neotropical toad. Sci Rep. 9:1128. doi:10.1038/s41598-018-37705-1. Tabadkani  SM, Nozari  J. 2014. Relaxed predation hinders development Rönkä  K, Mappes  J, Kaila  L, Wahlberg  N. 2016. Putting Parasemia in of anti-predator behaviors in an aposematic beetle. Entomol Exp Appl. its phylogenetic place: a molecular analysis of the subtribe Arctiina 153:199–206. doi: 10.1111/eea.12241. (Lepidoptera). Syst Entomol. 41:844–853. Valkonen  JK, Nokelainen  O, Jokimäki  M, Kuusinen  E, Paloranta  M, Rudh  A, Breed  MF, Qvarnström  A. 2013. Does aggression and explora- Peura  M, Mappes  J. 2014. From deception to frankness: benefits of on- tive behaviour decrease with lost warning coloration? Biol J Linn Soc togenetic shift in the anti-predator strategy of alder moth Acronicta alni 108:116–126. doi: 10.1111/j.1095-8312.2012.02006.x. larvae. Curr Zool. 60:114–122. doi:10.1093/czoolo/60.1.114. Schneider  CA, Rasband  WS, Eliceiri  KW. 2012. NIH image to ImageJ: Willink  B, Brenes-Mora  E, Bolaños  F, Pröhl  H. 2013. Not everything is 25 years of image analysis. Nat Methods. 9:671–675. black and white: color and behavioral variation reveal a continuum be- Sherman  PW, Watt  WB. 1973. The thermal ecology of some Colias but- tween cryptic and aposematic strategies in a polymorphic poison frog. terfly larvae. J Comp Physiol. 83:25–40. Evolution. 67:2783–2794.

Journal

Behavioral Ecology – Oxford University Press

Published: Jul 29, 2020

There are no references for this article.