- Email: [email protected]
Size variability effects on visual detection are influenced by colour pattern and perceived size
Animal Behaviour 143 (2018) 131e138
Contents lists available at ScienceDirect
Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav
Size variability effects on visual detection are influenced by colour pattern and perceived size Einat Karpestam a, Sami Merilaita b, c, Anders Forsman a, * a
Ecology and Evolution in Microbial Model Systems, EEMIS, Department of Biology and Environmental Science, Linnaeus University, Kalmar, Sweden Behavioural and Evolutionary Ecology Group, Environmental and Marine Biology, Department of Biosciences, Åbo Akademi University, Turku, Finland c Department of Biology, University of Turku, Turku, Finland b
a r t i c l e i n f o Article history: Received 21 December 2017 Initial acceptance 8 January 2018 Final acceptance 29 June 2018 MS. number: 18-00002 Keywords: body size camouflage cognition colour pattern polymorphism crypsis detection and perception predation visual stimuli
Most animals including humans use vision to detect, identify, evaluate and respond to potential prey items in complex environments. Theories predict that predators' visual search performance is better when targets are similar than when targets are dissimilar and require divided attention, and this may contribute to colour pattern polymorphism in prey. Most prey also vary in size, but how size variation influences detectability and search performance of predators that utilize polymorphic prey has received little attention. To evaluate the effect of size variability on prey detection we asked human subjects to search for images of black, grey and striped pygmy grasshoppers presented on computer screens in sizevariable (large, medium and small) or in size-invariable (all medium) sequences (populations) against photographs of natural grasshopper habitat. Results showed that size variability either increased or reduced detection of medium-sized targets depending on colour morph. To evaluate whether bias in perceived size varies depending on colour pattern, subjects were asked to discriminate between two grasshopper images of identical size that were presented in pairs against a monochromatic background. Subjects more often incorrectly classified one of the two identical-sized targets as being larger than the other in colour-dimorphic than in monomorphic presentations. The distinctly patterned (striped) morph elicited stronger size perception biases than the dorsally grey or black morphs, and striped grasshoppers were incorrectly classified more often as smaller than grey grasshoppers. The direction of the effect of size variability on detection changed across colour patterns as the bias in perceived size increased. Such joint effects of variation in size and colour pattern on detection and perception can impact the outcome of behavioural and evolutionary interactions between visually oriented predators and their camouflaged prey. This may have consequences for population dynamics, evolution of polymorphisms, community species composition and ecosystem functioning. © 2018 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
For individuals in search of mates or food, the ability to detect and recognize objects and compare among them is of utmost importance. For instance, within the context of mate choice, females and males must be able to recognize individuals that belong to the same species, discriminate between potential mates, and evaluate the quality of potential mates and competitors. Discrimination is often based on visual cues such as variation in size, coloration, shape and minor deviations from bilateral symmetry of ornaments (Andersson, 1994). Within the context of foraging, predators commonly search visually for preferred prey that are similar in size but differ in appearance from nonpreferred prey and inedible objects, or for prey that are similar in appearance but of
* Correspondence: A. Forsman, Ecology and Evolution in Microbial Model Systems, EEMIS, Department of Biology and Environmental Science, Linnaeus University, SE-391 82 Kalmar, Sweden. E-mail address: [email protected] (A. Forsman).
variable sizes owing to sexual size dimorphism and age differences. This is likely to have favoured recognition and detection strategies that rely on multiple visual features that are diagnostic of object identity (Rosselli, Alemi, Ansuini, & Zoccolan, 2015). Visual perception concerns both sensory stimulation elicited by the target's physical properties (‘bottom-up’ processes) and cognitive processes such as expectations or specific knowledge about the target (‘top-down’ processes; Corbetta & Shulman, 2002; Rosselli et al., 2015; Wolfe, Oliva, Horowitz, Butcher, & Bompas, 2002). For example, detection of salient targets is little influenced by the observer's previous knowledge (Wolfe, Butcher, Lee, & Hyle, 2003). On the other hand, when the search task is difficult, for example due to camouflage, top-down processes including previous knowledge and experience with targets and backgrounds may , 2012). Because the become more important (Chen & Hegde amount of information that can be processed at any given time is limited, search efficiency for a target category is expected to be
https://doi.org/10.1016/j.anbehav.2018.07.013 0003-3472/© 2018 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
132
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
negatively affected both by diversity within the target category (Karpestam, Merilaita, & Forsman, 2014b, 2016; Langley, Riley, Bond, & Goel, 1996; Pietrewicz & Kamil, 1979; Plaisted, 1997; Schmidt, Sebastian, Wilder, & Rypstra, 2012; Werner & Hall, 1974) and by the degree to which targets differ from the visual background (Karpestam, Merilaita, & Forsman, 2012, 2013). Accordingly, it can be predicted that target variability weakens the performance of searchers. It can further be hypothesized that variability in some target features (e.g. coloration, shape, orientation, size or movement) may be more important than others, depending on the utility of the selection criteria used by searchers (Duncan & Humphreys, 1989; Zelinsky, 2005). For example, Kazemi, Gamberale-Stille, Tullberg, and Leimar (2014) investigated the role of stimulus salience for predator learning, and found that prey colour had a stronger influence on the rate of predator learning than the pattern and the shape of colour pattern elements. However, little is known about whether, and how, phenotypic diversity among potential prey may affect the visual search performance of predators. Yet, illuminating this issue is important because any effects of variation in size and colour pattern on detection and perception can impact behavioural and evolutionary interactions between visually oriented predators and their camouflaged prey. In a series of previous studies, we have used computer-based detection experiments with human subjects acting as ‘predators’ searching for ‘prey’ in the form of photographs of colour polymorphic Tetrix subulata grasshoppers inserted into images of their natural background. In brief, these previous experiments have demonstrated that (1) probability of detection depends on target colour pattern (Karpestam et al., 2014b, 2013), (2) the relative protective value of colour patterns changes with the visual background (Karpestam et al., 2012) and with target size (Karpestam, Merilaita, & Forsman, 2014a) and (3) variation in colour pattern among targets (colour polymorphism) impairs search performance and decreases detection rate (Karpestam et al., 2014b, 2016). Because T. subulata males are larger than females and colour morph frequencies differ between the sexes (Forsman, 2018), we investigated in a recent study the effect of body size on detection risk of colour morphs and on colour morph frequencies. We found that prey size can differently influence detection of different colour patterns (Karpestam et al., 2014a). It is known that shapes of a specific size presented during a learning phase are more easily remembered and recognized if they remain the same size during the test phase (Jolicoeur, 1987; Milliken & Jolicoeur, 1992). Therefore, in the present study we focused on the effect of variation in size: the question of whether variation in size per se among different prey (as often exists in prey populations) impairs detection (as does variation in colour pattern, reviewed in Karpestam, Merilaita, & Forsman, 2016) has not been previously addressed. Accurate assessment of prey body size allows predators to evaluate both the rewards and the costs associated with initiating an attack. Larger prey contain more nutrients and energy, but may also be more likely to inflict costs associated with a more difficult capture or greater risk of injury (e.g. Cooper & Stankowich, 2010; Forsman, 1996). The size of a prey and whether it deviates from the size of other potential prey individuals in a group may also influence predation risk through the oddity effect (i.e. a deviating prey being singled out in a group and targeted by a predator, Landeau & Terborgh, 1986; Morse, 1970; Rodgers, Ward, Askwith, & Morrell, 2011; Theodorakis, 1989). Correct assessment of prey size is therefore important for predators' decision making and foraging success. Conversely, biased decisions due to incorrect assessment of size may favour the prey by decreasing probability of attack. Psychological experiments have shown that colour and pattern can influence perception of the size of visual targets (Gundlach &
Macoubrey, 1931; Ling & Hurlbert, 2004; McClain et al., 2014). This raises the possibility that biases in perceived size have the potential to modify the degree to which detection of prey is impaired by true size variability. If such moderating effects on perception depend on colour pattern, this can be of importance for ecological and evolutionary interactions and processes. Yet, it has not been established whether natural colour patterns of prey differently influence size perception. The knowledge gap regarding the interactive effect of body size and colour pattern on perception is unsatisfying, because natural predators commonly encounter prey of various sizes owing to sexual size dimorphism, age differences and feeding history (Forsman, 1995; Forsman & Appelqvist, 1999; King, 1987). On a more general note, improved knowledge of how variation in different phenotypic dimensions impacts the perception of and interactions between different individuals can further the understanding of the evolution of biodiversity. Here, we present two experiments addressing these questions. Our first aim in the present study was to test the hypothesis that size variability reduces detection of otherwise similar prey in serial encounters. To achieve this, we compared the performances of human subjects that searched for images of pygmy grasshoppers presented on computer screens, in size-variable (large, medium and small) sequences or in size-invariable (all medium) sequences. Our second aim was to test whether and how the colour patterns of pygmy grasshoppers influenced perception of size. To that end, we conducted a second experiment in which human subjects were presented with cards showing pairs of grasshoppers that were of identical size and either belonged to the same or to different colour morphs. Our third aim was to explore whether the effects of size variability on detection were influenced by biases in perceived size. To that end, the outcomes of experiment 1 and 2 were considered together. METHODS Equidistant presentation of the prey patterns was crucial for the testing of our hypotheses concerning detectability of prey (experiment 1) and perception of prey size (experiment 2). Further, it was essential that estimates of detectability or size differences were not confounded by differences between predators in motivation, hunger levels, experience, or any acquired preferences or aversions that might influence the decision to attack or ability to successfully capture the prey. For the evaluation of how colour pattern influences perceived size it was also crucial that different prey could be presented simultaneously (experiment 2). We therefore used two-dimensional images to represent the prey and we presented the images to human subjects that represented the ‘predators’, in two separate experiments. Experiment 1: Effect of Size Variability on Detection Methods To examine whether size variability reduces search efficiency of visually oriented predators and thereby improves the survival prospects of prey individuals that are members of size-variable populations we asked human subjects to search for images of grasshoppers inserted into photographs of their natural habitat. The approach used for the computer-based experiment was similar to what we have used in our previous studies designed to address related questions (Karpestam et al., 2012, 2013, 2014a, 2014b, 2016). Using humans as substitutes for real predators in detection experiments is an increasingly common approach to studying the function and evolution of protective coloration (see Fig. 1 in Karpestam, Merilaita, & Forsman, 2013). One reason for this is that direct observations of natural predation events are typically rare.
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
Furthermore, the outcome of predatoreprey interactions in the wild can be influenced by variation in predator motivation and in the rate or efficiency by which predators encounter, detect, recognize, attack, capture and kill or consume the prey. The computer-based approach used in this and previous studies enabled good control of prey presentations. It avoided the risks that comparisons of prey survival were influenced by variation in predator motivation or encounter rate and allowed for high levels of replication (see below). A potential problem with using humans as substitutes for natural predators (e.g. passerine birds) is that their visual system and cognitive abilities can differ. However, we have established previously that, in our study system, detection rates of grasshopper colour morphs as estimated by humans with this computer-based approach offer reliable predictors of selection imposed by natural visual predators and conform well with spatiotemporal shifts in relative frequencies of alternative morphs in natural populations (Karpestam et al., 2013). Photographs of natural T. subulata grasshopper habitats were taken in summer 2010 with a digital camera (Panasonic Lumix DMCTZ7, the 35 mm film camera equivalent focal length was 37 mm) using ‘macro’ mode at a vertical distance of approximately 30 cm from above. Pictures were taken at heterogeneous postfire environments at three locations in southeast Sweden that were at different stages of recovery and represented a mixture of burnt and unburnt habitat patches in similar proportions. Postfire environments are often populated by pygmy grasshoppers, which thrive in € & Forsman, 2006; Forsman, 2018; disturbed habitats (Ahnesjo Forsman, Karlsson, Wennersten, Johansson, & Karpestam, 2011; € , & Forsman, 2008). A full description Karlsson, Caesar, Ahnesjo and images are available elsewhere (Karpestam et al., 2014a, 2013). Adult T. subulata grasshoppers that belonged to the black, grey or striped colour morph (Fig. 1) were collected at the same sites and brought to the laboratory. Seven individuals of each morph were photographed dorsally against brown cardboard in diffuse natural daylight conditions to avoid shadows. To minimize any differences that photography might generate between the morphs, all photographs of the grasshoppers were taken on one occasion under the same conditions and using the same fixed settings. Grasshopper images were cropped from their original background and saved on a transparent background to enable embedding in natural background images using Adobe Photoshop CS4 (Karpestam et al., 2012). For presentation of the images we used a purpose-written program written in MATLAB 2011 (Mathworks, Natick, MA, U.S.A.)
100
Percentage detected
80
60
40
20
0
MI
MV
MI
MV
MI
MV
Figure 1. The effect of size variability on detectability of medium-sized grasshopper images. Percentage (mean ± SE) of medium-sized black (black circles), grey (grey circles) and striped (open circles) Tetrix subulata grasshoppers detected by human subjects when presented on computer screens against photographs of natural grasshopper habitat in either size-invariable (MI) or size-variable (MV) sequences.
133
and a 1500 screen (Fujitsu Lifebook e series; Karpestam et al., 2012, 2013). Because the screen of the laptop had not been calibrated to minimize possible colour and luminance biases, the perceived colours and luminance of the displays may have deviated to some extent from the perceived colours and luminance of the actual grasshoppers and backgrounds. The combinations of background and grasshopper images, as well as the position and rotation angle of the grasshopper in the background image, were randomly selected by the program. In the size-invariable populations, grasshopper images were scaled to medium size (approximately 11 mm from head to tip of the pronotum when measured on the presentation screen, corresponding to a large male or a small female). In the size-variable population, grasshoppers were scaled to three size categories within the natural size range: large (14 mm on screen, corresponding to large females), medium (11 mm on screen) and small (8 mm on screen, corresponding to small males; Karpestam et al., 2014a). Before the start of an experimental session, the subject received a verbal explanation of the task and short practice blocks. The purpose of the practice block was to familiarize the subject with the grasshopper colour morphs and sizes. This consisted of 20 s of presentation of grasshoppers against a uniform grey background. This was followed by two images each with one medium-sized grasshopper for the size-invariable group, and three images with one grasshopper that was small, medium or large (one for each size) for the size-variable group. The positions of the grasshoppers varied between the images in the practice blocks but was identical for all subjects. All explanations and instructions to subjects were identical and given by E.K. The subjects were instructed to search for one grasshopper in the image and to use the mouse to click on it. The average viewing distance between subject and screen was about 55 cm. After the training sessions the subject was left alone to carry out the experimental session. If a subject clicked on the image, the presentation program recorded the result (‘correct’ if on the grasshopper, otherwise ‘wrong’), after which the next combination of background and grasshopper images was presented. If a test subject did not click the mouse within 60 s from the beginning of a presentation, the presentation program recorded the result of the trial as ‘wrong’, and the next combination of images was presented. In total, 69 subjects took part in the detection experiments. Eighteen (nine women, nine men, age range 20e53 years, comprising academic staff and students at Linnaeus University) performed three size-invariable experimental blocks each (one for each of the three colour morphs), in a fully balanced design. Each size-invariable block consisted of 10 sequential presentations of medium-sized grasshoppers belonging to a single colour morph. Another 51 subjects (27 women, 24 men, age range 17e52 years, comprising academic staff and students at Linnaeus University) performed one size-variable block each, with 17 subjects for each of the three colour morphs. Each size-variable block consisted of 15 sequential presentations of grasshopper images belonging to a single colour morph but representing the three size categories, five presentations of each size category in random sequence for each subject (Karpestam et al., 2014a). All subjects had normal or corrected normal vision. Statistical analyses In the first analytical step, we asked whether subjects more easily detected grasshoppers that were presented in size-invariable sequences than when they were presented in the size-variable sequences, whether subjects detected black, grey and striped grasshopper morphs differently, and whether the effect of size variability on detection depended on colour morph. To that end, we applied generalized linear models (GLMs) to binary data on detection with the procedure GLIMMIX in SAS (Bolker et al., 2009;
134
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
Collett, 1991; Littell, Milliken, Stroup, Wolfinger, & Schabenberger, 2006). The response variable was the number of detected prey out of the number of presented prey in the trial. This response variable was modelled as having a binomial distribution, with a logit-link power function. Colour morph, size variability treatment and the interaction between colour morph and size variability were included as fixed factors. Statistical significance was assessed using the Type III test of fixed effects. Additionally, the Akaike information criterion (AIC) was used to compare the fit of models with and without the interaction effect and identify the one that provided the best fit to the data (lowest AIC). See ‘Effects of Coloration and Size Variability on Detection’ for the results of these analyses. In a second analytical step, we asked whether, from the prey's perspective, being part of a size-variable group or population provided protection in terms of reduced detection, and whether the effect of size variability on detection depended on colour pattern. To that end, we compared detection of medium-sized grasshoppers between size-invariable and size-variable sequences (excluding data for large and small grasshoppers). For this analysis we also used GLMs for binary data, implemented with the procedure GLIMMIX, treating colour, size variability and their interaction as explanatory factors. A significant effect of the interaction would indicate that the magnitude and/or the direction of the difference between colour morphs varied according to presentation treatment. See ‘Interactive Effects of Size Variability and Coloration’ for the results of these analyses. We estimated the protective effect of size variability for medium-sized grasshoppers of each colour morph by subtracting the percentage of detected images in size-variable sequences from the percentage of detected images in size-invariable sequences (i.e. the morph-specific difference between results for MI and MV in Fig. 1). These estimates were used together with data on the effects of colour pattern on size perception from experiment 2 to evaluate whether the effects of size variability on detection were associated with size perception biases. Experiment 2: Effect of Coloration on Size Perception Methods To evaluate the hypothesis that colour patterns differently influence perceived size of grasshoppers, human subjects (96 students and staff of Linnaeus university, age span 19e64 years) were presented with four cards in sequence, each of which showed photographic images of two grasshoppers in a uniform, light green (Natural Colour System code B90G:1020) background (Appendix Fig. A1). The paired grasshoppers were of identical size and either belonged to the same or to different colour morphs. Each subject was presented with one (out of three) monomorphic pair (same morph) and three cards with colour pattern-dimorphic pairs (one card for each morph combination). Presentation cards measured 6 5.5 cm and were laminated with transparent plastic for easier handling and better preservation. Grasshoppers of each pair were 1.2 cm long, parallel to each other and 2.5 cm apart, leaving a 1.5 cm margin on each side. To ensure that all images were of identical size and shape regardless of colour morph, we only used the image of a single grasshopper and manipulated its colour to create the light grey, black and striped natural colour morphs. We used 12 cards of which six showed pairs of the same morph, two each of the black, striped and grey colour morphs. To make the task more difficult, grasshoppers were not perfectly aligned but displaced 2 mm along the longitudinal axis. In one card the right grasshopper was slightly shifted towards the anterior direction, and in the other card the left grasshopper was shifted. The remaining six cards showed mixed morph pairs, two cards each of grey-black, grey-striped and striped-black morph
combinations. The two cards of each dimorphic pair were lateral mirror images. For example, in one card the grey grasshopper was on the left side and slightly shifted forwards and the black grasshopper was on the right side, while on the other card the black grasshopper was on the left side and slightly shifted forwards and the grey grasshopper on the right. There were 48 unique combinations of presentation order of the cards, and each sequence was only used twice. Cards were presented at eye height at a 50 cm distance, but subjects could approach or move away from the card if they wanted to. Subjects were asked to verbally indicate, for each card, whether either, and if so which, of the two grasshoppers (left or right) was larger than the other. The response is potentially influenced by how the question is formulated, and subjects might have responded differently if the question had been more neutral. However, subjects were not forced/did not have to indicate that either the left or the right grasshopper was larger but could instead respond that neither was larger than the other. Because the instructions were the same for all subjects and presentation cards, any effect of how the question was formulated on the response should not bias the results from our comparisons of colour morph combinations. Observers (E.K. or A.F) were oblivious to the colour morph combination that was presented but cards were marked with numbers on the back to enable scoring of results. All subjects (age span 19e64 years, comprising academic staff and students at Linnaeus University) had normal acuity with or without glasses and none were colour blind. Statistical analyses Data from the card presentations were analysed using generalized linear mixed-effects models (GLMM). This was implemented with the procedure GLIMMIX in SAS (Bolker et al., 2009; Littell et al., 2006). In this approach, the answer (different size or same size) was modelled as a binary response variable, colour morph combination was treated as a fixed factor (with six levels) and subject was included as a random factor to account for the fact that each subject contributed four repeated measures. For each of the three types of colour-dimorphic presentations, the chi-square goodness-of-fit test (Sokal & Rohlf, 1981) was used to evaluate the null hypothesis that the observed frequency distribution of outcomes (left larger, right larger, no difference) did not differ from an even expected frequency distribution of outcomes (i.e. 33% of each of the three outcomes). Ethical Note The experiments were performed in accordance with the established ethical guidelines of the Declaration of Helsinki and adhered to the Swedish legislation and ethical regulations for experimental research (http://www.codex.vr.se/en/manniska5. shtml) and to all institutional guidelines. As our research did not involve ‘a physical intervention affecting a person who is participating in the research, or is conducted in accordance with a method intended to physically or mentally influence a person who is participating in the research’, it did not require ethical vetting according to the Central Ethical Review Board (http://www.epn.se/ en/start/) and the Swedish Act concerning the Ethical Review of Research Involving Humans (2003:460, http://www.epn.se/media/ 1205/the_ethical_review_act.pdf). The experimental protocol was approved by the Linnaeus University Faculty of Health and Life Sciences and the Ethical Advisory Board in Southeast Sweden. Informed consent was obtained from all subjects who volunteered to participate after they had received a description of the set-up. There was no reward for those who did participate nor any repercussion for those who did not want to participate in the experiment.
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
RESULTS Experiment 1 Effects of coloration and size variability on detection When comparing the detection of medium-sized individuals in size-invariable (medium only) sequences with detection of individuals of all sizes (large, medium and small) presented in the size-variable sequences, there was a significant main effect of colour morph (F2,99 ¼ 15.88, P < 0.0001). Comparisons of leastsquares means showed that black individuals were detected at the lowest rate (lsmeans ± SE ¼ 0.42 ± 0.036), striped individuals at an intermediate rate (0.52 ± 0.036) and grey individuals at the highest rate (0.62 ± 0.036). There was no significant main effect on detection of size variability (F1,99 ¼ 0.05, P ¼ 0.82). The interaction between size variability and colour morph fell short of statistical significance (F2,99 ¼ 2.88, P ¼ 0.061), but the model with the interaction provided a slightly better fit to the data (AIC ¼ 532.75) than a model without the interaction (534.53). Interactive effects of size variability and coloration Detection of medium-sized grasshoppers depended on the interaction between presentation treatment (size-invariable or sizevariable) and colour morph (interaction: F2,99 ¼ 5.51, P ¼ 0.005; coloration: F2,99 ¼ 8.70, P ¼ 0.0003; size variability: F1,99 ¼ 0.09, P ¼ 0.76). Comparisons between different models confirmed that the best-fitting model included the interaction (AIC for model with interaction ¼ 456.52 and without interaction ¼ 463.84). The significant interaction effect demonstrated that body size variability influenced detectability differently in the three colour morphs. Specifically, the interaction reflected that in the striped morph a larger proportion of the medium-sized individuals were detected when they were presented in size-variable sequences than when presented in size-invariable sequences (F1,33 ¼ 8.01, P ¼ 0.008; Fig. 1), whereas medium-sized black (F1,33 ¼ 2.43, P ¼ 0.13) and medium-sized grey (F1,33 ¼ 0.65, P ¼ 0.42) individuals tended to be detected at slightly, but not significantly, lower rates in size-variable than in size-invariable sequences (Fig. 1). Experiment 2: Effect of Coloration on Size Perception When grasshopper images of identical size were presented in the same or mixed colour pattern pairs against a uniform background, the propensity to incorrectly classify one of two targets as larger than the other depended on the colour pattern combination (GLMM, implemented with procedure GLIMMIX, Type III test of fixed effect of treatment: F 5,378 ¼ 2.76, P ¼ 0.018; the covariance parameter estimate indicated a significant random effect of human subject: estimate ± SE ¼ 2.82 ± 0.712). Correct classifications were more common when the two targets were of a similar colour pattern (monomorphic) than when they were of different colour patterns (F 1,382 ¼ 9.05, P ¼ 0.003; Fig. 2). Participants were more likely to correctly classify the two grasshoppers in monomorphic pairs as being of similar size when presented with uniform nonpatterned morphs (black 50% or grey 38%) than when presented with the distinctly patterned striped morph (29%; Fig. 2). The effect of size variability on detection of medium-sized grasshopper images (as demonstrated in the first experiment, see Fig. 1) changed qualitatively (from protective to elevating risk of detection) as the bias in size perception increased (Fig. 3). Subjects
60
Percentage scored as similar size
Grasshoppers were treated in a way that minimized adverse impacts during photography, and were kept outdoors in 10-litre plastic rearing buckets with soil and moss (Forsman et al., 2011) before and after photography (Karpestam et al., 2012).
135
50
40
30
20
10 0
Black
Grey
Striped StripedBlack
Same colour
GreyBlack
StripedGrey
Different colours
Figure 2. Perception of size in relation to colour pattern. Percentages of presentations of paired grasshopper images that were correctly classified as being of similar size by human subjects for each colour pattern combination. There were 96 subjects each of whom was presented with one of the monomorphic pairs (same colour) and all three dimorphic combinations with different-coloured grasshoppers.
were most prone to misjudge the relative size of striped grasshoppers in colour pattern-monomorphic presentations, and medium-sized striped grasshoppers were detected at higher rates if presented in size-variable than in size-invariable sequences (Fig. 3). Results for the three types of dimorphic presentations showed that subjects were significantly more prone to incorrectly classify the grey morph as larger when presented next to the striped morph (Fig. 4). Subjects also tended to classify the grey morph as larger when presented next to the black morph (Fig. 4). The apparent size of the striped morph was not influenced by the morph of the other grasshopper (Fig. 4). DISCUSSION Our findings demonstrate that the effects of size variability on visual detection are moulded by colour pattern and that there are biases in perceived size that also vary according to colour pattern. Results from the computer-based detection experiment demonstrate that size variability affected the capacity of human subjects to detect images of pygmy grasshoppers that were presented on computer screens against photographs of complex natural grasshopper habitats. The effect of size variability on detection of medium-sized grasshoppers depended on colour pattern (as evidenced by the interaction). The interaction reflected that mediumsized striped grasshoppers were more easily detected in sizevariable sequences that included small, medium and large images than in size-invariable (all medium) sequences, whereas mediumsized grey and black grasshoppers were not. In the size perception experiment we evaluated the propensity of human subjects that were presented with two grasshopper images of identical size to incorrectly classify one of them as being larger than the other. The results demonstrated that the distinctly patterned striped grasshopper colour morph elicited stronger size perception biases than the more uniform grey or black morphs. This indicates that some colour patterns are more likely to induce a bias in size perception than others. This effect may influence selection on colour pattern, particularly in gregarious prey when the predator's choice of prey is influenced by size preferences. The results from the size perception experiment also suggest that the moderating impact of colour pattern on the effect that size
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
15
60
10
50
5
(a)
40
0
30
–5
20
–10
10
–15 0 –20 –25
60 50
55 60 65 Size perception bias (%)
70
75
Figure 3. Protective effect of size variability on detection as a function of size perception bias, for black, grey and striped pygmy grasshopper colour morphs. The effect of size variability on detection of medium-sized grasshopper images was calculated as the difference between size-invariable (MI) and size-variable (MV) sequences for each colour morph in Fig. 1. High positive values on the y-axis correspond to a higher rate of escaping detection, i.e. that human subjects detected medium-sized targets on a computer screen at a lower rate when presented in size-variable than in size-invariable sequences. Negative values indicate that they were detected at a higher rate in size-variable sequences. Greater size perception bias values on the x-axis indicate that subjects were more likely to incorrectly classify one of two grasshopper images as being larger than the other when presented with cards showing two images that were of identical size.
variability had on detection was due in part to biases with regard to perception of object size. Perhaps the most important and novel key message of our study emerged when the outcomes of the detection and perception experiments were considered together. Combining the results from the two experiments suggested that the effects of size variability on detection of medium-sized grasshoppers (as demonstrated in the first experiment, Fig. 1) tended to go from positive to negative as the bias in perceived size increased (Fig. 3). This result thus suggests that the effect of size variability changed qualitatively from protective to elevating risk of detection as the bias in size perception increased. Perception bias with regard to object size thus seems to resolve the unexpected positive effect of size variability on detection of the striped grasshopper colour morph. To identify the mechanism behind this finding (Fig. 3) is beyond the scope of our study. However, a possible explanation is that the longitudinal and distinct yellowish stripe in the striped pattern constituted a more salient stimulus (Kazemi et al., 2014), and that it was therefore more easily recognized and generalized across different sizes. Conversely, the black morph which was most difficult to detect (Fig. 1) may be more difficult to recognize and generalize across different sizes. Additionally, different colour patterns may reduce detection and recognition via alternative camouflage mechanisms (Merilaita, Scott-Samuel, & Cuthill, 2017), and these can in turn be differently influenced by size (Karpestam et al., 2014a). The sizes of the grasshoppers corresponded to different inanimate objects in the natural background images, such as mosses, twigs and burnt or dry spruce needles. The black and grey grasshopper morphs may have gained protection via background matching coloration. The striped morph may instead have been mistakenly taken for one of the dry yellowish spruce needles on the ground and gained protection from masquerade. Importance of Variability for Detection The detection experiment uncovered no significant effect of size variability as such on overall detection, because the proportion of detected grasshopper images did not differ between size-variable
Judged as bigger (%)
Protective effect of size variability
136
Striped
Black
None
Grey
Black
None
Grey
Striped
None
(b)
50 40 30 20 10 0 60
(c)
50 40 30 20 10 0
Figure 4. Effect of colour pattern combination on perception of size differences. Frequency distribution of responses by human subjects presented with cards showing photographic images of two grasshoppers of identical size but different colour patterns. Subjects were asked to indicate whether either, and if so which, of the two grasshoppers was larger. Three different dimorphic combinations of (a) stripedeblack, (b) greyeblack and (c) greyestriped colour morphs were shown. Goodness-of-fit tests: (a) c22 ¼ 2.39, P ¼ 0.30; (b) c22 ¼ 5.29, P ¼ 0.07; (c) c22 ¼ 20.8, P < 0.001.
and size-invariable sequences, for any of the three colour morphs. This outcome differs from the findings of previous detection experiments which have tested for effects of variation in colour pattern (rather than size), and demonstrate that colour pattern polymorphism negatively affects search performance and may afford protection against predators for both individuals and populations (reviewed in Karpestam et al., 2016). Taken together, this suggests that variability in some target features may be more important than variability in others. It is known that searchers may rely more on certain selection criteria (Duncan & Humphreys, 1989; Zelinsky, 2005). There is also evidence that searchers use multifeature discrimination processing strategies (e.g. Rosselli et al., 2015), but we are not aware of any previous study that has investigated whether size variability reduces detection. However, the available evidence seems to suggest that variation in colour pattern is more important than variation in size. One may speculate that apparent size is a less reliable diagnostic feature because objects are rarely seen at a constant distance
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
(Tsurui, Honma, & Nishida, 2013; Tullberg, Merilaita, & Wiklund, 2005), and because natural predators commonly encounter potential prey items of different sizes owing to sexual size dimorphism and age differences (Forsman, 1995; Forsman & Appelqvist, 1999; King, 1987). This may further reduce the utility of size in recognition, memory, perception and detection within the context of predatoreprey interactions, and perhaps in other settings. Our present findings add to the body of evidence that colour and pattern can influence perception of size (Gundlach & Macoubrey, 1931; Ling & Hurlbert, 2004; McClain et al., 2014). Our results showed that subjects incorrectly classified one target as being larger than the other more often when presented with pairs in which the two identical-sized grasshoppers had different colour patterns than when the grasshoppers belonged to the same colour morph. Subjects also more often incorrectly classified images of striped grasshoppers as being smaller than images of identical-sized grey grasshoppers. Size perception biases (or size congruency effects, Jolicoeur, 1987; Milliken & Jolicoeur, 1992), combined with size variability effects on detection, might have consequences for behavioural decisions and fitness of prey individuals. This can translate into correlational selection that favours certain combinations of traits over others (Brodie, 1992; Forsman & Appelqvist, 1998). This, in turn, may ultimately mould the evolution of phenotypic integration and genetic correla€ & Forsman, tions between colour pattern and body size (Ahnesjo 2003; Forsman, 2018), promote the evolution of sexual dichromatism (Andersson, 1994) and ontogenetic irreversible shifts in coloration associated with growth (Booth, 1990), favour the evolution of rapid and reversible morphological or physiological colour change (Bagnara & Hadley, 1973; Edelaar, Jovani, & Gomez-Mestre, 2017; Stuart-Fox & Moussalli, 2009), and impact on the dynamics and maintenance of colour polymorphisms (Forsman, 2018; Karpestam et al., 2016, 2014a; Pietrewicz & Kamil, 1979; Tsurui et al., 2013). Conclusions In conclusion, search performance of human subjects and detection of grasshopper images were not affected to any significant degree by variation in target size as such. This outcome is in contrast to previous studies demonstrating that variation in colour pattern affords protection against predation for both individuals and populations, suggesting that colour pattern is a more important cue than size. From the point of view of the targeted ‘prey’, probability of detection of medium-sized individuals depended on whether the population was presented in a size-variable or a size-invariable sequence. Our findings further demonstrated that size variability effects on visual detection were influenced by colour pattern, and that different colour patterns differently affected size perception. Importantly, the effect of size variability on detection changed qualitatively, from protecting against detection to elevating the risk of detection, as the bias in perceived size increased. Such joint effects on detection and perception can impact the outcome of behavioural and evolutionary interactions between visually oriented predators and their camouflaged prey and may also have farreaching consequences for population dynamics, community species composition and ecosystem functioning. Author Contributions A.F., E.K. and S.M. designed the detection experiments; E.K. photographed animals and backgrounds, wrote the presentation program and undertook the computer-based detection experiments; A.F. and E.K. designed and undertook the size perception experiments; E.K. and A.F. analysed data. A.F., E.K. and S.M. contributed to the writing and read and approved the final manuscript.
137
Data Availability Data are available from the Dryad Digital Repository https://doi. org/10.5061/dryad.17vf831. Acknowledgments We thank the many colleagues, students and friends who volunteered to act as predators in this research, Z. Kons and B. Karpestam for their assistance in the field and laboratory, and two anonymous referees for comments on the manuscript. This work was funded by Linnaeus University (grants to A.F.) and the Academy of Finland (S.M.). References €, J., & Forsman, A. (2003). Correlated evolution of colour pattern and body Ahnesjo size in polymorphic pygmy grasshoppers, Tetrix undulata. Journal of Evolutionary Biology, 16(6), 1308e1318. https://doi.org/10.1046/j.1420-9101.2003.00610.x. €, J., & Forsman, A. (2006). Differential habitat selection by pygmy grasshopper Ahnesjo color morphs; interactive effects of temperature and predator avoidance. Evolutionary Ecology, 20, 235e257. https://doi.org/10.1007/s10682-006-6178-8. Andersson, M. (1994). Sexual selection. Princeton, NJ: Princeton University Press. Bagnara, J. T., & Hadley, M. E. (1973). Chromatophores and color change: The comparative physiology of animal pigmentation. Englewood Cliffs, NJ: Prentice-Hall. Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., et al. (2009). Generalized linear mixed models: A practical guide for ecology and evolution. Trends in Ecology & Evolution, 24(3), 127e135. Booth, C. L. (1990). Evolutionary significance of ontogenetic colour change in animals. Biological Journal of the Linnean Society, 40, 125e163. Brodie, E. D. (1992). Correlational selection for colour pattern and antipredator behaviour in the garter snake Thamnophis ordinoides. Evolution, 46, 1284e1298. , J. (2012). Learning to break camouflage by learning the backChen, X., & Hegde ground. Psychological Science, 23(11), 1395e1403. https://doi.org/10.1177/ 0956797612445315. Collett, D. (1991). Modelling binary data. London, U.K.: Chapman & Hall. Cooper, J. W. E., & Stankowich, T. (2010). Prey or predator? Body size of an approaching animal affects decisions to attack or escape. Behavioral Ecology, 21(6), 1278e1284. https://doi.org/10.1093/beheco/arq142. Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention in the brain. Nature Reviews Neuroscience, 3(3), 201e215. https:// doi.org/10.1038/nrn755. Duncan, J., & Humphreys, G. W. (1989). Visual-search and stimulus similarity. Psychological Review, 96(3), 433e458. https://doi.org/10.1037//0033-295x.96.3.433. Edelaar, P., Jovani, R., & Gomez-Mestre, I. (2017). Should I change or should I go? Phenotypic plasticity and matching habitat choice in the adaptation to environmental heterogeneity. American Naturalist,190(4), 506e520. https://doi.org/10.1086/693345. Forsman, A. (1995). Opposing fitness consequences of color pattern in male and female snakes. Journal of Evolutionary Biology, 8(1), 53e70. Forsman, A. (1996). Body size and net energy gain in gape-limited predators: A model. Journal of Herpetology, 30(3), 307e319. Forsman, A. (2018). On the role of sex differences for evolution in heterogeneous and changing fitness landscapes: Insights from pygmy grasshoppers. Philosophical Transactions of the Royal Society B: Biological Sciences, 20170429. https://doi.org/10.1098/rstb.2017.0429. Forsman, A., & Appelqvist, S. (1998). Visual predators impose correlational selection on prey color pattern and behavior. Behavioral Ecology, 9(4), 409e413. Forsman, A., & Appelqvist, S. (1999). Experimental manipulation reveals differential effects of colour pattern on survival in male and female pygmy grasshoppers. Journal of Evolutionary Biology, 12(2), 391e401. https://doi.org/10.1046/j.14209101.1999.00041.x. Forsman, A., Karlsson, M., Wennersten, L., Johansson, J., & Karpestam, E. (2011). Rapid evolution of fire melanism in replicated populations of pygmy grasshoppers. Evolution, 65, 2530e2540. https://doi.org/10.1111/j.1558-5646.2011.01324.x. Gundlach, C., & Macoubrey, C. (1931). The effect of color on apparent size. American Journal of Psychology, 43, 109e111. https://doi.org/10.2307/1414243. Jolicoeur, P. (1987). A size-congruency effect in memory for visual shape. Memory & Cognition, 15(6), 531e543. https://doi.org/10.3758/bf03198388. € , J., & Forsman, A. (2008). Dynamics of colour Karlsson, M., Caesar, S., Ahnesjo polymorphism in changing environments: Fire melanism and then what? Oecologia, 154, 715e724. Karpestam, E., Merilaita, S., & Forsman, A. (2012). Reduced predation risk for melanistic pygmy grasshoppers in post-fire environments. Ecology and Evolution, 2, 2204e2212. Karpestam, E., Merilaita, S., & Forsman, A. (2013). Detection experiments with humans implicate visual predation as a driver of colour polymorphism dynamics in pygmy grasshoppers. BMC Ecology, 13, 17. https://doi.org/10.1186/1472-6785-1113-1117. Karpestam, E., Merilaita, S., & Forsman, A. (2014a). Body size influences differently the detectabilities of colour morphs of cryptic prey. Biological Journal of the Linnean Society, 113(1), 112e122. https://doi.org/10.1111/bij.12291.
138
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
Karpestam, E., Merilaita, S., & Forsman, A. (2014b). Natural levels of colour polymorphism reduce performance of visual predators searching for camouflaged prey. Biological Journal of the Linnean Society, 112, 546e555. https://doi.org/10.1111/bij.12276. Karpestam, E., Merilaita, S., & Forsman, A. (2016). Colour polymorphism protects prey individuals and populations against predation. Scientific Reports, 6, 22122. https://doi.org/10.1038/srep22122. Kazemi, B., Gamberale-Stille, G., Tullberg Birgitta, S., & Leimar, O. (2014). Stimulus salience as an explanation for imperfect mimicry. Current Biology, 24(9), 965e969. https://doi.org/10.1016/j.cub.2014.02.061. King, R. B. (1987). Colour pattern polymorphism in the Lake Erie water snake, Nerodia sipedon insularum. Evolution, 41, 241e255. Landeau, L., & Terborgh, J. (1986). Oddity and the ‘confusion effect’ in predation. Animal Behaviour, 34(5), 1372e1380. https://doi.org/10.1016/S0003-3472(86)80208-1. Langley, C. M., Riley, D. A., Bond, A. B., & Goel, N. (1996). Visual search for natural grains in pigeons (Columba livia): Search images and selective attention. Journal of Experimental Psychology: Animal Behavior Processes, 22(2), 139e151. Ling, Y. Z., & Hurlbert, A. (2004). Color and size interactions in a real 3D object similarity task. Journal of Vision, 4(9), 721e734. https://doi.org/10.1167/4.9.5. Littell, R. C., Milliken, G. A., Stroup, W. W., Wolfinger, R. D., & Schabenberger, O. (2006). SAS for mixed models (2nd ed.). Cary, NC: SAS Institute Inc. McClain, A. D., van den Bos, W., Matheson, D., Desai, M., McClure, S. M., & Robinson, T. N. (2014). Visual illusions and plate design: The effects of plate rim widths and rim coloring on perceived food portion size. International Journal of Obesity, 38(5), 657e662. Merilaita, S., Scott-Samuel, N. E., & Cuthill, I. C. (2017). How camouflage works. Philosophical Transactions of the Royal Society B: Biological Sciences, 372(1724). https://doi.org/10.1098/rstb.2016.0341. Milliken, B., & Jolicoeur, P. (1992). Size effects in visual recognition memory are determined by perceived size. Memory & Cognition, 20(1), 83e95. https:// doi.org/10.3758/bf03208257. Morse, D. H. (1970). Ecological aspects of some mixed-species foraging flocks of birds. Ecological Monographs, 40(1), 119e168. https://doi.org/10.2307/1942443. Pietrewicz, A. T., & Kamil, A. C. (1979). Search image formation in the blue jay (Cyanocitta cristata). Science, 204, 1332e1333. Plaisted, K. (1997). The effect of interstimulus interval on the discrimination of cryptic targets. Journal of Experimental Psychology: Animal Behavior Processes, 23(2), 248e259. https://doi.org/10.1037/0097-7403.23.2.248. Rodgers, G. M., Ward, J. R., Askwith, B., & Morrell, L. J. (2011). Balancing the dilution and oddity effects: Decisions depend on body size. PLoS One, 6(7), e14819. https://doi.org/10.1371/journal.pone.0014819.
Rosselli, F. B., Alemi, A., Ansuini, A., & Zoccolan, D. (2015). Object similarity affects the perceptual strategy underlying invariant visual object recognition in rats. Frontiers in Neural Circuits, 9, 10. https://doi.org/10.3389/fncir.2015.00010. Schmidt, J. M., Sebastian, P., Wilder, S. M., & Rypstra, A. L. (2012). The nutritional content of prey affects the foraging of a generalist arthropod predator. PLoS One, 7(11), e49223. https://doi.org/10.1371/journal.pone.0049223. Sokal, R. S., & Rohlf, F. J. (1981). Biometry. New York, NY: Freeman. Stuart-Fox, D., & Moussalli, A. (2009). Camouflage, communication and thermoregulation: Lessons from colour changing organisms. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1516), 463e470. https://doi.org/ 10.1098/rstb.2008.0254. Theodorakis, C. W. (1989). Size segregation and the effects of oddity on predation risk in minnow schools. Animal Behaviour, 38(3), 496e502. https://doi.org/ 10.1016/S0003-3472(89)80042-9. Tsurui, K., Honma, A., & Nishida, T. (2013). Size-dependent predation risk partly explains the sex-related marking polymorphism in the sexually size-dimorphic pygmy grasshopper Tetrix japonica. Entomological Science, 16(2), 136e144. https://doi.org/10.1111/j.1479-8298.2012.00543.x. Tullberg, B. S., Merilaita, S., & Wiklund, C. (2005). Aposematism and crypsis combined as a result of distance dependence: Functional versatility of the colour pattern in the swallowtail butterfly larva. Proceedings of the Royal Society B: Biological Sciences, 272(1570), 1315e1321. https://doi.org/10.1098/rspb.2005.3079. Werner, E. E., & Hall, D. J. (1974). Optimal foraging and the size selection of prey by the Bluegill Sunfish (Lepomis macrochirus). Ecology, 55(5), 1042e1052. https:// doi.org/10.2307/1940354. Wolfe, J. M., Butcher, S. J., Lee, C., & Hyle, M. (2003). Changing your mind: On the contributions of top-down and bottom-up guidance in visual search for feature singletons. Journal of Experimental Psychology: Human Perception and Performance, 29(2), 483e502. https://doi.org/10.1037/0096-1523.29.2.483. Wolfe, J. M., Oliva, A., Horowitz, T. S., Butcher, S. J., & Bompas, A. (2002). Segmentation of objects from backgrounds in visual search tasks. Vision Research, 42(28), 2985e3004. https://doi.org/10.1016/s0042-6989(02)00388-7. Zelinsky, G. J. (2005). Specifying the components of attention in a visual search task. In G. R. L. Itti, & J. Tsotsos (Eds.), Neurobiology of attention (pp. 395e400). New York, NY: Elsevier Academic Press.
Appendix
Figure A1. Example of presentation cards used to test for effects of colour pattern on size perception bias.
Contents lists available at ScienceDirect
Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav
Size variability effects on visual detection are influenced by colour pattern and perceived size Einat Karpestam a, Sami Merilaita b, c, Anders Forsman a, * a
Ecology and Evolution in Microbial Model Systems, EEMIS, Department of Biology and Environmental Science, Linnaeus University, Kalmar, Sweden Behavioural and Evolutionary Ecology Group, Environmental and Marine Biology, Department of Biosciences, Åbo Akademi University, Turku, Finland c Department of Biology, University of Turku, Turku, Finland b
a r t i c l e i n f o Article history: Received 21 December 2017 Initial acceptance 8 January 2018 Final acceptance 29 June 2018 MS. number: 18-00002 Keywords: body size camouflage cognition colour pattern polymorphism crypsis detection and perception predation visual stimuli
Most animals including humans use vision to detect, identify, evaluate and respond to potential prey items in complex environments. Theories predict that predators' visual search performance is better when targets are similar than when targets are dissimilar and require divided attention, and this may contribute to colour pattern polymorphism in prey. Most prey also vary in size, but how size variation influences detectability and search performance of predators that utilize polymorphic prey has received little attention. To evaluate the effect of size variability on prey detection we asked human subjects to search for images of black, grey and striped pygmy grasshoppers presented on computer screens in sizevariable (large, medium and small) or in size-invariable (all medium) sequences (populations) against photographs of natural grasshopper habitat. Results showed that size variability either increased or reduced detection of medium-sized targets depending on colour morph. To evaluate whether bias in perceived size varies depending on colour pattern, subjects were asked to discriminate between two grasshopper images of identical size that were presented in pairs against a monochromatic background. Subjects more often incorrectly classified one of the two identical-sized targets as being larger than the other in colour-dimorphic than in monomorphic presentations. The distinctly patterned (striped) morph elicited stronger size perception biases than the dorsally grey or black morphs, and striped grasshoppers were incorrectly classified more often as smaller than grey grasshoppers. The direction of the effect of size variability on detection changed across colour patterns as the bias in perceived size increased. Such joint effects of variation in size and colour pattern on detection and perception can impact the outcome of behavioural and evolutionary interactions between visually oriented predators and their camouflaged prey. This may have consequences for population dynamics, evolution of polymorphisms, community species composition and ecosystem functioning. © 2018 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
For individuals in search of mates or food, the ability to detect and recognize objects and compare among them is of utmost importance. For instance, within the context of mate choice, females and males must be able to recognize individuals that belong to the same species, discriminate between potential mates, and evaluate the quality of potential mates and competitors. Discrimination is often based on visual cues such as variation in size, coloration, shape and minor deviations from bilateral symmetry of ornaments (Andersson, 1994). Within the context of foraging, predators commonly search visually for preferred prey that are similar in size but differ in appearance from nonpreferred prey and inedible objects, or for prey that are similar in appearance but of
* Correspondence: A. Forsman, Ecology and Evolution in Microbial Model Systems, EEMIS, Department of Biology and Environmental Science, Linnaeus University, SE-391 82 Kalmar, Sweden. E-mail address: [email protected] (A. Forsman).
variable sizes owing to sexual size dimorphism and age differences. This is likely to have favoured recognition and detection strategies that rely on multiple visual features that are diagnostic of object identity (Rosselli, Alemi, Ansuini, & Zoccolan, 2015). Visual perception concerns both sensory stimulation elicited by the target's physical properties (‘bottom-up’ processes) and cognitive processes such as expectations or specific knowledge about the target (‘top-down’ processes; Corbetta & Shulman, 2002; Rosselli et al., 2015; Wolfe, Oliva, Horowitz, Butcher, & Bompas, 2002). For example, detection of salient targets is little influenced by the observer's previous knowledge (Wolfe, Butcher, Lee, & Hyle, 2003). On the other hand, when the search task is difficult, for example due to camouflage, top-down processes including previous knowledge and experience with targets and backgrounds may , 2012). Because the become more important (Chen & Hegde amount of information that can be processed at any given time is limited, search efficiency for a target category is expected to be
https://doi.org/10.1016/j.anbehav.2018.07.013 0003-3472/© 2018 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
132
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
negatively affected both by diversity within the target category (Karpestam, Merilaita, & Forsman, 2014b, 2016; Langley, Riley, Bond, & Goel, 1996; Pietrewicz & Kamil, 1979; Plaisted, 1997; Schmidt, Sebastian, Wilder, & Rypstra, 2012; Werner & Hall, 1974) and by the degree to which targets differ from the visual background (Karpestam, Merilaita, & Forsman, 2012, 2013). Accordingly, it can be predicted that target variability weakens the performance of searchers. It can further be hypothesized that variability in some target features (e.g. coloration, shape, orientation, size or movement) may be more important than others, depending on the utility of the selection criteria used by searchers (Duncan & Humphreys, 1989; Zelinsky, 2005). For example, Kazemi, Gamberale-Stille, Tullberg, and Leimar (2014) investigated the role of stimulus salience for predator learning, and found that prey colour had a stronger influence on the rate of predator learning than the pattern and the shape of colour pattern elements. However, little is known about whether, and how, phenotypic diversity among potential prey may affect the visual search performance of predators. Yet, illuminating this issue is important because any effects of variation in size and colour pattern on detection and perception can impact behavioural and evolutionary interactions between visually oriented predators and their camouflaged prey. In a series of previous studies, we have used computer-based detection experiments with human subjects acting as ‘predators’ searching for ‘prey’ in the form of photographs of colour polymorphic Tetrix subulata grasshoppers inserted into images of their natural background. In brief, these previous experiments have demonstrated that (1) probability of detection depends on target colour pattern (Karpestam et al., 2014b, 2013), (2) the relative protective value of colour patterns changes with the visual background (Karpestam et al., 2012) and with target size (Karpestam, Merilaita, & Forsman, 2014a) and (3) variation in colour pattern among targets (colour polymorphism) impairs search performance and decreases detection rate (Karpestam et al., 2014b, 2016). Because T. subulata males are larger than females and colour morph frequencies differ between the sexes (Forsman, 2018), we investigated in a recent study the effect of body size on detection risk of colour morphs and on colour morph frequencies. We found that prey size can differently influence detection of different colour patterns (Karpestam et al., 2014a). It is known that shapes of a specific size presented during a learning phase are more easily remembered and recognized if they remain the same size during the test phase (Jolicoeur, 1987; Milliken & Jolicoeur, 1992). Therefore, in the present study we focused on the effect of variation in size: the question of whether variation in size per se among different prey (as often exists in prey populations) impairs detection (as does variation in colour pattern, reviewed in Karpestam, Merilaita, & Forsman, 2016) has not been previously addressed. Accurate assessment of prey body size allows predators to evaluate both the rewards and the costs associated with initiating an attack. Larger prey contain more nutrients and energy, but may also be more likely to inflict costs associated with a more difficult capture or greater risk of injury (e.g. Cooper & Stankowich, 2010; Forsman, 1996). The size of a prey and whether it deviates from the size of other potential prey individuals in a group may also influence predation risk through the oddity effect (i.e. a deviating prey being singled out in a group and targeted by a predator, Landeau & Terborgh, 1986; Morse, 1970; Rodgers, Ward, Askwith, & Morrell, 2011; Theodorakis, 1989). Correct assessment of prey size is therefore important for predators' decision making and foraging success. Conversely, biased decisions due to incorrect assessment of size may favour the prey by decreasing probability of attack. Psychological experiments have shown that colour and pattern can influence perception of the size of visual targets (Gundlach &
Macoubrey, 1931; Ling & Hurlbert, 2004; McClain et al., 2014). This raises the possibility that biases in perceived size have the potential to modify the degree to which detection of prey is impaired by true size variability. If such moderating effects on perception depend on colour pattern, this can be of importance for ecological and evolutionary interactions and processes. Yet, it has not been established whether natural colour patterns of prey differently influence size perception. The knowledge gap regarding the interactive effect of body size and colour pattern on perception is unsatisfying, because natural predators commonly encounter prey of various sizes owing to sexual size dimorphism, age differences and feeding history (Forsman, 1995; Forsman & Appelqvist, 1999; King, 1987). On a more general note, improved knowledge of how variation in different phenotypic dimensions impacts the perception of and interactions between different individuals can further the understanding of the evolution of biodiversity. Here, we present two experiments addressing these questions. Our first aim in the present study was to test the hypothesis that size variability reduces detection of otherwise similar prey in serial encounters. To achieve this, we compared the performances of human subjects that searched for images of pygmy grasshoppers presented on computer screens, in size-variable (large, medium and small) sequences or in size-invariable (all medium) sequences. Our second aim was to test whether and how the colour patterns of pygmy grasshoppers influenced perception of size. To that end, we conducted a second experiment in which human subjects were presented with cards showing pairs of grasshoppers that were of identical size and either belonged to the same or to different colour morphs. Our third aim was to explore whether the effects of size variability on detection were influenced by biases in perceived size. To that end, the outcomes of experiment 1 and 2 were considered together. METHODS Equidistant presentation of the prey patterns was crucial for the testing of our hypotheses concerning detectability of prey (experiment 1) and perception of prey size (experiment 2). Further, it was essential that estimates of detectability or size differences were not confounded by differences between predators in motivation, hunger levels, experience, or any acquired preferences or aversions that might influence the decision to attack or ability to successfully capture the prey. For the evaluation of how colour pattern influences perceived size it was also crucial that different prey could be presented simultaneously (experiment 2). We therefore used two-dimensional images to represent the prey and we presented the images to human subjects that represented the ‘predators’, in two separate experiments. Experiment 1: Effect of Size Variability on Detection Methods To examine whether size variability reduces search efficiency of visually oriented predators and thereby improves the survival prospects of prey individuals that are members of size-variable populations we asked human subjects to search for images of grasshoppers inserted into photographs of their natural habitat. The approach used for the computer-based experiment was similar to what we have used in our previous studies designed to address related questions (Karpestam et al., 2012, 2013, 2014a, 2014b, 2016). Using humans as substitutes for real predators in detection experiments is an increasingly common approach to studying the function and evolution of protective coloration (see Fig. 1 in Karpestam, Merilaita, & Forsman, 2013). One reason for this is that direct observations of natural predation events are typically rare.
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
Furthermore, the outcome of predatoreprey interactions in the wild can be influenced by variation in predator motivation and in the rate or efficiency by which predators encounter, detect, recognize, attack, capture and kill or consume the prey. The computer-based approach used in this and previous studies enabled good control of prey presentations. It avoided the risks that comparisons of prey survival were influenced by variation in predator motivation or encounter rate and allowed for high levels of replication (see below). A potential problem with using humans as substitutes for natural predators (e.g. passerine birds) is that their visual system and cognitive abilities can differ. However, we have established previously that, in our study system, detection rates of grasshopper colour morphs as estimated by humans with this computer-based approach offer reliable predictors of selection imposed by natural visual predators and conform well with spatiotemporal shifts in relative frequencies of alternative morphs in natural populations (Karpestam et al., 2013). Photographs of natural T. subulata grasshopper habitats were taken in summer 2010 with a digital camera (Panasonic Lumix DMCTZ7, the 35 mm film camera equivalent focal length was 37 mm) using ‘macro’ mode at a vertical distance of approximately 30 cm from above. Pictures were taken at heterogeneous postfire environments at three locations in southeast Sweden that were at different stages of recovery and represented a mixture of burnt and unburnt habitat patches in similar proportions. Postfire environments are often populated by pygmy grasshoppers, which thrive in € & Forsman, 2006; Forsman, 2018; disturbed habitats (Ahnesjo Forsman, Karlsson, Wennersten, Johansson, & Karpestam, 2011; € , & Forsman, 2008). A full description Karlsson, Caesar, Ahnesjo and images are available elsewhere (Karpestam et al., 2014a, 2013). Adult T. subulata grasshoppers that belonged to the black, grey or striped colour morph (Fig. 1) were collected at the same sites and brought to the laboratory. Seven individuals of each morph were photographed dorsally against brown cardboard in diffuse natural daylight conditions to avoid shadows. To minimize any differences that photography might generate between the morphs, all photographs of the grasshoppers were taken on one occasion under the same conditions and using the same fixed settings. Grasshopper images were cropped from their original background and saved on a transparent background to enable embedding in natural background images using Adobe Photoshop CS4 (Karpestam et al., 2012). For presentation of the images we used a purpose-written program written in MATLAB 2011 (Mathworks, Natick, MA, U.S.A.)
100
Percentage detected
80
60
40
20
0
MI
MV
MI
MV
MI
MV
Figure 1. The effect of size variability on detectability of medium-sized grasshopper images. Percentage (mean ± SE) of medium-sized black (black circles), grey (grey circles) and striped (open circles) Tetrix subulata grasshoppers detected by human subjects when presented on computer screens against photographs of natural grasshopper habitat in either size-invariable (MI) or size-variable (MV) sequences.
133
and a 1500 screen (Fujitsu Lifebook e series; Karpestam et al., 2012, 2013). Because the screen of the laptop had not been calibrated to minimize possible colour and luminance biases, the perceived colours and luminance of the displays may have deviated to some extent from the perceived colours and luminance of the actual grasshoppers and backgrounds. The combinations of background and grasshopper images, as well as the position and rotation angle of the grasshopper in the background image, were randomly selected by the program. In the size-invariable populations, grasshopper images were scaled to medium size (approximately 11 mm from head to tip of the pronotum when measured on the presentation screen, corresponding to a large male or a small female). In the size-variable population, grasshoppers were scaled to three size categories within the natural size range: large (14 mm on screen, corresponding to large females), medium (11 mm on screen) and small (8 mm on screen, corresponding to small males; Karpestam et al., 2014a). Before the start of an experimental session, the subject received a verbal explanation of the task and short practice blocks. The purpose of the practice block was to familiarize the subject with the grasshopper colour morphs and sizes. This consisted of 20 s of presentation of grasshoppers against a uniform grey background. This was followed by two images each with one medium-sized grasshopper for the size-invariable group, and three images with one grasshopper that was small, medium or large (one for each size) for the size-variable group. The positions of the grasshoppers varied between the images in the practice blocks but was identical for all subjects. All explanations and instructions to subjects were identical and given by E.K. The subjects were instructed to search for one grasshopper in the image and to use the mouse to click on it. The average viewing distance between subject and screen was about 55 cm. After the training sessions the subject was left alone to carry out the experimental session. If a subject clicked on the image, the presentation program recorded the result (‘correct’ if on the grasshopper, otherwise ‘wrong’), after which the next combination of background and grasshopper images was presented. If a test subject did not click the mouse within 60 s from the beginning of a presentation, the presentation program recorded the result of the trial as ‘wrong’, and the next combination of images was presented. In total, 69 subjects took part in the detection experiments. Eighteen (nine women, nine men, age range 20e53 years, comprising academic staff and students at Linnaeus University) performed three size-invariable experimental blocks each (one for each of the three colour morphs), in a fully balanced design. Each size-invariable block consisted of 10 sequential presentations of medium-sized grasshoppers belonging to a single colour morph. Another 51 subjects (27 women, 24 men, age range 17e52 years, comprising academic staff and students at Linnaeus University) performed one size-variable block each, with 17 subjects for each of the three colour morphs. Each size-variable block consisted of 15 sequential presentations of grasshopper images belonging to a single colour morph but representing the three size categories, five presentations of each size category in random sequence for each subject (Karpestam et al., 2014a). All subjects had normal or corrected normal vision. Statistical analyses In the first analytical step, we asked whether subjects more easily detected grasshoppers that were presented in size-invariable sequences than when they were presented in the size-variable sequences, whether subjects detected black, grey and striped grasshopper morphs differently, and whether the effect of size variability on detection depended on colour morph. To that end, we applied generalized linear models (GLMs) to binary data on detection with the procedure GLIMMIX in SAS (Bolker et al., 2009;
134
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
Collett, 1991; Littell, Milliken, Stroup, Wolfinger, & Schabenberger, 2006). The response variable was the number of detected prey out of the number of presented prey in the trial. This response variable was modelled as having a binomial distribution, with a logit-link power function. Colour morph, size variability treatment and the interaction between colour morph and size variability were included as fixed factors. Statistical significance was assessed using the Type III test of fixed effects. Additionally, the Akaike information criterion (AIC) was used to compare the fit of models with and without the interaction effect and identify the one that provided the best fit to the data (lowest AIC). See ‘Effects of Coloration and Size Variability on Detection’ for the results of these analyses. In a second analytical step, we asked whether, from the prey's perspective, being part of a size-variable group or population provided protection in terms of reduced detection, and whether the effect of size variability on detection depended on colour pattern. To that end, we compared detection of medium-sized grasshoppers between size-invariable and size-variable sequences (excluding data for large and small grasshoppers). For this analysis we also used GLMs for binary data, implemented with the procedure GLIMMIX, treating colour, size variability and their interaction as explanatory factors. A significant effect of the interaction would indicate that the magnitude and/or the direction of the difference between colour morphs varied according to presentation treatment. See ‘Interactive Effects of Size Variability and Coloration’ for the results of these analyses. We estimated the protective effect of size variability for medium-sized grasshoppers of each colour morph by subtracting the percentage of detected images in size-variable sequences from the percentage of detected images in size-invariable sequences (i.e. the morph-specific difference between results for MI and MV in Fig. 1). These estimates were used together with data on the effects of colour pattern on size perception from experiment 2 to evaluate whether the effects of size variability on detection were associated with size perception biases. Experiment 2: Effect of Coloration on Size Perception Methods To evaluate the hypothesis that colour patterns differently influence perceived size of grasshoppers, human subjects (96 students and staff of Linnaeus university, age span 19e64 years) were presented with four cards in sequence, each of which showed photographic images of two grasshoppers in a uniform, light green (Natural Colour System code B90G:1020) background (Appendix Fig. A1). The paired grasshoppers were of identical size and either belonged to the same or to different colour morphs. Each subject was presented with one (out of three) monomorphic pair (same morph) and three cards with colour pattern-dimorphic pairs (one card for each morph combination). Presentation cards measured 6 5.5 cm and were laminated with transparent plastic for easier handling and better preservation. Grasshoppers of each pair were 1.2 cm long, parallel to each other and 2.5 cm apart, leaving a 1.5 cm margin on each side. To ensure that all images were of identical size and shape regardless of colour morph, we only used the image of a single grasshopper and manipulated its colour to create the light grey, black and striped natural colour morphs. We used 12 cards of which six showed pairs of the same morph, two each of the black, striped and grey colour morphs. To make the task more difficult, grasshoppers were not perfectly aligned but displaced 2 mm along the longitudinal axis. In one card the right grasshopper was slightly shifted towards the anterior direction, and in the other card the left grasshopper was shifted. The remaining six cards showed mixed morph pairs, two cards each of grey-black, grey-striped and striped-black morph
combinations. The two cards of each dimorphic pair were lateral mirror images. For example, in one card the grey grasshopper was on the left side and slightly shifted forwards and the black grasshopper was on the right side, while on the other card the black grasshopper was on the left side and slightly shifted forwards and the grey grasshopper on the right. There were 48 unique combinations of presentation order of the cards, and each sequence was only used twice. Cards were presented at eye height at a 50 cm distance, but subjects could approach or move away from the card if they wanted to. Subjects were asked to verbally indicate, for each card, whether either, and if so which, of the two grasshoppers (left or right) was larger than the other. The response is potentially influenced by how the question is formulated, and subjects might have responded differently if the question had been more neutral. However, subjects were not forced/did not have to indicate that either the left or the right grasshopper was larger but could instead respond that neither was larger than the other. Because the instructions were the same for all subjects and presentation cards, any effect of how the question was formulated on the response should not bias the results from our comparisons of colour morph combinations. Observers (E.K. or A.F) were oblivious to the colour morph combination that was presented but cards were marked with numbers on the back to enable scoring of results. All subjects (age span 19e64 years, comprising academic staff and students at Linnaeus University) had normal acuity with or without glasses and none were colour blind. Statistical analyses Data from the card presentations were analysed using generalized linear mixed-effects models (GLMM). This was implemented with the procedure GLIMMIX in SAS (Bolker et al., 2009; Littell et al., 2006). In this approach, the answer (different size or same size) was modelled as a binary response variable, colour morph combination was treated as a fixed factor (with six levels) and subject was included as a random factor to account for the fact that each subject contributed four repeated measures. For each of the three types of colour-dimorphic presentations, the chi-square goodness-of-fit test (Sokal & Rohlf, 1981) was used to evaluate the null hypothesis that the observed frequency distribution of outcomes (left larger, right larger, no difference) did not differ from an even expected frequency distribution of outcomes (i.e. 33% of each of the three outcomes). Ethical Note The experiments were performed in accordance with the established ethical guidelines of the Declaration of Helsinki and adhered to the Swedish legislation and ethical regulations for experimental research (http://www.codex.vr.se/en/manniska5. shtml) and to all institutional guidelines. As our research did not involve ‘a physical intervention affecting a person who is participating in the research, or is conducted in accordance with a method intended to physically or mentally influence a person who is participating in the research’, it did not require ethical vetting according to the Central Ethical Review Board (http://www.epn.se/ en/start/) and the Swedish Act concerning the Ethical Review of Research Involving Humans (2003:460, http://www.epn.se/media/ 1205/the_ethical_review_act.pdf). The experimental protocol was approved by the Linnaeus University Faculty of Health and Life Sciences and the Ethical Advisory Board in Southeast Sweden. Informed consent was obtained from all subjects who volunteered to participate after they had received a description of the set-up. There was no reward for those who did participate nor any repercussion for those who did not want to participate in the experiment.
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
RESULTS Experiment 1 Effects of coloration and size variability on detection When comparing the detection of medium-sized individuals in size-invariable (medium only) sequences with detection of individuals of all sizes (large, medium and small) presented in the size-variable sequences, there was a significant main effect of colour morph (F2,99 ¼ 15.88, P < 0.0001). Comparisons of leastsquares means showed that black individuals were detected at the lowest rate (lsmeans ± SE ¼ 0.42 ± 0.036), striped individuals at an intermediate rate (0.52 ± 0.036) and grey individuals at the highest rate (0.62 ± 0.036). There was no significant main effect on detection of size variability (F1,99 ¼ 0.05, P ¼ 0.82). The interaction between size variability and colour morph fell short of statistical significance (F2,99 ¼ 2.88, P ¼ 0.061), but the model with the interaction provided a slightly better fit to the data (AIC ¼ 532.75) than a model without the interaction (534.53). Interactive effects of size variability and coloration Detection of medium-sized grasshoppers depended on the interaction between presentation treatment (size-invariable or sizevariable) and colour morph (interaction: F2,99 ¼ 5.51, P ¼ 0.005; coloration: F2,99 ¼ 8.70, P ¼ 0.0003; size variability: F1,99 ¼ 0.09, P ¼ 0.76). Comparisons between different models confirmed that the best-fitting model included the interaction (AIC for model with interaction ¼ 456.52 and without interaction ¼ 463.84). The significant interaction effect demonstrated that body size variability influenced detectability differently in the three colour morphs. Specifically, the interaction reflected that in the striped morph a larger proportion of the medium-sized individuals were detected when they were presented in size-variable sequences than when presented in size-invariable sequences (F1,33 ¼ 8.01, P ¼ 0.008; Fig. 1), whereas medium-sized black (F1,33 ¼ 2.43, P ¼ 0.13) and medium-sized grey (F1,33 ¼ 0.65, P ¼ 0.42) individuals tended to be detected at slightly, but not significantly, lower rates in size-variable than in size-invariable sequences (Fig. 1). Experiment 2: Effect of Coloration on Size Perception When grasshopper images of identical size were presented in the same or mixed colour pattern pairs against a uniform background, the propensity to incorrectly classify one of two targets as larger than the other depended on the colour pattern combination (GLMM, implemented with procedure GLIMMIX, Type III test of fixed effect of treatment: F 5,378 ¼ 2.76, P ¼ 0.018; the covariance parameter estimate indicated a significant random effect of human subject: estimate ± SE ¼ 2.82 ± 0.712). Correct classifications were more common when the two targets were of a similar colour pattern (monomorphic) than when they were of different colour patterns (F 1,382 ¼ 9.05, P ¼ 0.003; Fig. 2). Participants were more likely to correctly classify the two grasshoppers in monomorphic pairs as being of similar size when presented with uniform nonpatterned morphs (black 50% or grey 38%) than when presented with the distinctly patterned striped morph (29%; Fig. 2). The effect of size variability on detection of medium-sized grasshopper images (as demonstrated in the first experiment, see Fig. 1) changed qualitatively (from protective to elevating risk of detection) as the bias in size perception increased (Fig. 3). Subjects
60
Percentage scored as similar size
Grasshoppers were treated in a way that minimized adverse impacts during photography, and were kept outdoors in 10-litre plastic rearing buckets with soil and moss (Forsman et al., 2011) before and after photography (Karpestam et al., 2012).
135
50
40
30
20
10 0
Black
Grey
Striped StripedBlack
Same colour
GreyBlack
StripedGrey
Different colours
Figure 2. Perception of size in relation to colour pattern. Percentages of presentations of paired grasshopper images that were correctly classified as being of similar size by human subjects for each colour pattern combination. There were 96 subjects each of whom was presented with one of the monomorphic pairs (same colour) and all three dimorphic combinations with different-coloured grasshoppers.
were most prone to misjudge the relative size of striped grasshoppers in colour pattern-monomorphic presentations, and medium-sized striped grasshoppers were detected at higher rates if presented in size-variable than in size-invariable sequences (Fig. 3). Results for the three types of dimorphic presentations showed that subjects were significantly more prone to incorrectly classify the grey morph as larger when presented next to the striped morph (Fig. 4). Subjects also tended to classify the grey morph as larger when presented next to the black morph (Fig. 4). The apparent size of the striped morph was not influenced by the morph of the other grasshopper (Fig. 4). DISCUSSION Our findings demonstrate that the effects of size variability on visual detection are moulded by colour pattern and that there are biases in perceived size that also vary according to colour pattern. Results from the computer-based detection experiment demonstrate that size variability affected the capacity of human subjects to detect images of pygmy grasshoppers that were presented on computer screens against photographs of complex natural grasshopper habitats. The effect of size variability on detection of medium-sized grasshoppers depended on colour pattern (as evidenced by the interaction). The interaction reflected that mediumsized striped grasshoppers were more easily detected in sizevariable sequences that included small, medium and large images than in size-invariable (all medium) sequences, whereas mediumsized grey and black grasshoppers were not. In the size perception experiment we evaluated the propensity of human subjects that were presented with two grasshopper images of identical size to incorrectly classify one of them as being larger than the other. The results demonstrated that the distinctly patterned striped grasshopper colour morph elicited stronger size perception biases than the more uniform grey or black morphs. This indicates that some colour patterns are more likely to induce a bias in size perception than others. This effect may influence selection on colour pattern, particularly in gregarious prey when the predator's choice of prey is influenced by size preferences. The results from the size perception experiment also suggest that the moderating impact of colour pattern on the effect that size
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
15
60
10
50
5
(a)
40
0
30
–5
20
–10
10
–15 0 –20 –25
60 50
55 60 65 Size perception bias (%)
70
75
Figure 3. Protective effect of size variability on detection as a function of size perception bias, for black, grey and striped pygmy grasshopper colour morphs. The effect of size variability on detection of medium-sized grasshopper images was calculated as the difference between size-invariable (MI) and size-variable (MV) sequences for each colour morph in Fig. 1. High positive values on the y-axis correspond to a higher rate of escaping detection, i.e. that human subjects detected medium-sized targets on a computer screen at a lower rate when presented in size-variable than in size-invariable sequences. Negative values indicate that they were detected at a higher rate in size-variable sequences. Greater size perception bias values on the x-axis indicate that subjects were more likely to incorrectly classify one of two grasshopper images as being larger than the other when presented with cards showing two images that were of identical size.
variability had on detection was due in part to biases with regard to perception of object size. Perhaps the most important and novel key message of our study emerged when the outcomes of the detection and perception experiments were considered together. Combining the results from the two experiments suggested that the effects of size variability on detection of medium-sized grasshoppers (as demonstrated in the first experiment, Fig. 1) tended to go from positive to negative as the bias in perceived size increased (Fig. 3). This result thus suggests that the effect of size variability changed qualitatively from protective to elevating risk of detection as the bias in size perception increased. Perception bias with regard to object size thus seems to resolve the unexpected positive effect of size variability on detection of the striped grasshopper colour morph. To identify the mechanism behind this finding (Fig. 3) is beyond the scope of our study. However, a possible explanation is that the longitudinal and distinct yellowish stripe in the striped pattern constituted a more salient stimulus (Kazemi et al., 2014), and that it was therefore more easily recognized and generalized across different sizes. Conversely, the black morph which was most difficult to detect (Fig. 1) may be more difficult to recognize and generalize across different sizes. Additionally, different colour patterns may reduce detection and recognition via alternative camouflage mechanisms (Merilaita, Scott-Samuel, & Cuthill, 2017), and these can in turn be differently influenced by size (Karpestam et al., 2014a). The sizes of the grasshoppers corresponded to different inanimate objects in the natural background images, such as mosses, twigs and burnt or dry spruce needles. The black and grey grasshopper morphs may have gained protection via background matching coloration. The striped morph may instead have been mistakenly taken for one of the dry yellowish spruce needles on the ground and gained protection from masquerade. Importance of Variability for Detection The detection experiment uncovered no significant effect of size variability as such on overall detection, because the proportion of detected grasshopper images did not differ between size-variable
Judged as bigger (%)
Protective effect of size variability
136
Striped
Black
None
Grey
Black
None
Grey
Striped
None
(b)
50 40 30 20 10 0 60
(c)
50 40 30 20 10 0
Figure 4. Effect of colour pattern combination on perception of size differences. Frequency distribution of responses by human subjects presented with cards showing photographic images of two grasshoppers of identical size but different colour patterns. Subjects were asked to indicate whether either, and if so which, of the two grasshoppers was larger. Three different dimorphic combinations of (a) stripedeblack, (b) greyeblack and (c) greyestriped colour morphs were shown. Goodness-of-fit tests: (a) c22 ¼ 2.39, P ¼ 0.30; (b) c22 ¼ 5.29, P ¼ 0.07; (c) c22 ¼ 20.8, P < 0.001.
and size-invariable sequences, for any of the three colour morphs. This outcome differs from the findings of previous detection experiments which have tested for effects of variation in colour pattern (rather than size), and demonstrate that colour pattern polymorphism negatively affects search performance and may afford protection against predators for both individuals and populations (reviewed in Karpestam et al., 2016). Taken together, this suggests that variability in some target features may be more important than variability in others. It is known that searchers may rely more on certain selection criteria (Duncan & Humphreys, 1989; Zelinsky, 2005). There is also evidence that searchers use multifeature discrimination processing strategies (e.g. Rosselli et al., 2015), but we are not aware of any previous study that has investigated whether size variability reduces detection. However, the available evidence seems to suggest that variation in colour pattern is more important than variation in size. One may speculate that apparent size is a less reliable diagnostic feature because objects are rarely seen at a constant distance
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
(Tsurui, Honma, & Nishida, 2013; Tullberg, Merilaita, & Wiklund, 2005), and because natural predators commonly encounter potential prey items of different sizes owing to sexual size dimorphism and age differences (Forsman, 1995; Forsman & Appelqvist, 1999; King, 1987). This may further reduce the utility of size in recognition, memory, perception and detection within the context of predatoreprey interactions, and perhaps in other settings. Our present findings add to the body of evidence that colour and pattern can influence perception of size (Gundlach & Macoubrey, 1931; Ling & Hurlbert, 2004; McClain et al., 2014). Our results showed that subjects incorrectly classified one target as being larger than the other more often when presented with pairs in which the two identical-sized grasshoppers had different colour patterns than when the grasshoppers belonged to the same colour morph. Subjects also more often incorrectly classified images of striped grasshoppers as being smaller than images of identical-sized grey grasshoppers. Size perception biases (or size congruency effects, Jolicoeur, 1987; Milliken & Jolicoeur, 1992), combined with size variability effects on detection, might have consequences for behavioural decisions and fitness of prey individuals. This can translate into correlational selection that favours certain combinations of traits over others (Brodie, 1992; Forsman & Appelqvist, 1998). This, in turn, may ultimately mould the evolution of phenotypic integration and genetic correla€ & Forsman, tions between colour pattern and body size (Ahnesjo 2003; Forsman, 2018), promote the evolution of sexual dichromatism (Andersson, 1994) and ontogenetic irreversible shifts in coloration associated with growth (Booth, 1990), favour the evolution of rapid and reversible morphological or physiological colour change (Bagnara & Hadley, 1973; Edelaar, Jovani, & Gomez-Mestre, 2017; Stuart-Fox & Moussalli, 2009), and impact on the dynamics and maintenance of colour polymorphisms (Forsman, 2018; Karpestam et al., 2016, 2014a; Pietrewicz & Kamil, 1979; Tsurui et al., 2013). Conclusions In conclusion, search performance of human subjects and detection of grasshopper images were not affected to any significant degree by variation in target size as such. This outcome is in contrast to previous studies demonstrating that variation in colour pattern affords protection against predation for both individuals and populations, suggesting that colour pattern is a more important cue than size. From the point of view of the targeted ‘prey’, probability of detection of medium-sized individuals depended on whether the population was presented in a size-variable or a size-invariable sequence. Our findings further demonstrated that size variability effects on visual detection were influenced by colour pattern, and that different colour patterns differently affected size perception. Importantly, the effect of size variability on detection changed qualitatively, from protecting against detection to elevating the risk of detection, as the bias in perceived size increased. Such joint effects on detection and perception can impact the outcome of behavioural and evolutionary interactions between visually oriented predators and their camouflaged prey and may also have farreaching consequences for population dynamics, community species composition and ecosystem functioning. Author Contributions A.F., E.K. and S.M. designed the detection experiments; E.K. photographed animals and backgrounds, wrote the presentation program and undertook the computer-based detection experiments; A.F. and E.K. designed and undertook the size perception experiments; E.K. and A.F. analysed data. A.F., E.K. and S.M. contributed to the writing and read and approved the final manuscript.
137
Data Availability Data are available from the Dryad Digital Repository https://doi. org/10.5061/dryad.17vf831. Acknowledgments We thank the many colleagues, students and friends who volunteered to act as predators in this research, Z. Kons and B. Karpestam for their assistance in the field and laboratory, and two anonymous referees for comments on the manuscript. This work was funded by Linnaeus University (grants to A.F.) and the Academy of Finland (S.M.). References €, J., & Forsman, A. (2003). Correlated evolution of colour pattern and body Ahnesjo size in polymorphic pygmy grasshoppers, Tetrix undulata. Journal of Evolutionary Biology, 16(6), 1308e1318. https://doi.org/10.1046/j.1420-9101.2003.00610.x. €, J., & Forsman, A. (2006). Differential habitat selection by pygmy grasshopper Ahnesjo color morphs; interactive effects of temperature and predator avoidance. Evolutionary Ecology, 20, 235e257. https://doi.org/10.1007/s10682-006-6178-8. Andersson, M. (1994). Sexual selection. Princeton, NJ: Princeton University Press. Bagnara, J. T., & Hadley, M. E. (1973). Chromatophores and color change: The comparative physiology of animal pigmentation. Englewood Cliffs, NJ: Prentice-Hall. Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., et al. (2009). Generalized linear mixed models: A practical guide for ecology and evolution. Trends in Ecology & Evolution, 24(3), 127e135. Booth, C. L. (1990). Evolutionary significance of ontogenetic colour change in animals. Biological Journal of the Linnean Society, 40, 125e163. Brodie, E. D. (1992). Correlational selection for colour pattern and antipredator behaviour in the garter snake Thamnophis ordinoides. Evolution, 46, 1284e1298. , J. (2012). Learning to break camouflage by learning the backChen, X., & Hegde ground. Psychological Science, 23(11), 1395e1403. https://doi.org/10.1177/ 0956797612445315. Collett, D. (1991). Modelling binary data. London, U.K.: Chapman & Hall. Cooper, J. W. E., & Stankowich, T. (2010). Prey or predator? Body size of an approaching animal affects decisions to attack or escape. Behavioral Ecology, 21(6), 1278e1284. https://doi.org/10.1093/beheco/arq142. Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention in the brain. Nature Reviews Neuroscience, 3(3), 201e215. https:// doi.org/10.1038/nrn755. Duncan, J., & Humphreys, G. W. (1989). Visual-search and stimulus similarity. Psychological Review, 96(3), 433e458. https://doi.org/10.1037//0033-295x.96.3.433. Edelaar, P., Jovani, R., & Gomez-Mestre, I. (2017). Should I change or should I go? Phenotypic plasticity and matching habitat choice in the adaptation to environmental heterogeneity. American Naturalist,190(4), 506e520. https://doi.org/10.1086/693345. Forsman, A. (1995). Opposing fitness consequences of color pattern in male and female snakes. Journal of Evolutionary Biology, 8(1), 53e70. Forsman, A. (1996). Body size and net energy gain in gape-limited predators: A model. Journal of Herpetology, 30(3), 307e319. Forsman, A. (2018). On the role of sex differences for evolution in heterogeneous and changing fitness landscapes: Insights from pygmy grasshoppers. Philosophical Transactions of the Royal Society B: Biological Sciences, 20170429. https://doi.org/10.1098/rstb.2017.0429. Forsman, A., & Appelqvist, S. (1998). Visual predators impose correlational selection on prey color pattern and behavior. Behavioral Ecology, 9(4), 409e413. Forsman, A., & Appelqvist, S. (1999). Experimental manipulation reveals differential effects of colour pattern on survival in male and female pygmy grasshoppers. Journal of Evolutionary Biology, 12(2), 391e401. https://doi.org/10.1046/j.14209101.1999.00041.x. Forsman, A., Karlsson, M., Wennersten, L., Johansson, J., & Karpestam, E. (2011). Rapid evolution of fire melanism in replicated populations of pygmy grasshoppers. Evolution, 65, 2530e2540. https://doi.org/10.1111/j.1558-5646.2011.01324.x. Gundlach, C., & Macoubrey, C. (1931). The effect of color on apparent size. American Journal of Psychology, 43, 109e111. https://doi.org/10.2307/1414243. Jolicoeur, P. (1987). A size-congruency effect in memory for visual shape. Memory & Cognition, 15(6), 531e543. https://doi.org/10.3758/bf03198388. € , J., & Forsman, A. (2008). Dynamics of colour Karlsson, M., Caesar, S., Ahnesjo polymorphism in changing environments: Fire melanism and then what? Oecologia, 154, 715e724. Karpestam, E., Merilaita, S., & Forsman, A. (2012). Reduced predation risk for melanistic pygmy grasshoppers in post-fire environments. Ecology and Evolution, 2, 2204e2212. Karpestam, E., Merilaita, S., & Forsman, A. (2013). Detection experiments with humans implicate visual predation as a driver of colour polymorphism dynamics in pygmy grasshoppers. BMC Ecology, 13, 17. https://doi.org/10.1186/1472-6785-1113-1117. Karpestam, E., Merilaita, S., & Forsman, A. (2014a). Body size influences differently the detectabilities of colour morphs of cryptic prey. Biological Journal of the Linnean Society, 113(1), 112e122. https://doi.org/10.1111/bij.12291.
138
E. Karpestam et al. / Animal Behaviour 143 (2018) 131e138
Karpestam, E., Merilaita, S., & Forsman, A. (2014b). Natural levels of colour polymorphism reduce performance of visual predators searching for camouflaged prey. Biological Journal of the Linnean Society, 112, 546e555. https://doi.org/10.1111/bij.12276. Karpestam, E., Merilaita, S., & Forsman, A. (2016). Colour polymorphism protects prey individuals and populations against predation. Scientific Reports, 6, 22122. https://doi.org/10.1038/srep22122. Kazemi, B., Gamberale-Stille, G., Tullberg Birgitta, S., & Leimar, O. (2014). Stimulus salience as an explanation for imperfect mimicry. Current Biology, 24(9), 965e969. https://doi.org/10.1016/j.cub.2014.02.061. King, R. B. (1987). Colour pattern polymorphism in the Lake Erie water snake, Nerodia sipedon insularum. Evolution, 41, 241e255. Landeau, L., & Terborgh, J. (1986). Oddity and the ‘confusion effect’ in predation. Animal Behaviour, 34(5), 1372e1380. https://doi.org/10.1016/S0003-3472(86)80208-1. Langley, C. M., Riley, D. A., Bond, A. B., & Goel, N. (1996). Visual search for natural grains in pigeons (Columba livia): Search images and selective attention. Journal of Experimental Psychology: Animal Behavior Processes, 22(2), 139e151. Ling, Y. Z., & Hurlbert, A. (2004). Color and size interactions in a real 3D object similarity task. Journal of Vision, 4(9), 721e734. https://doi.org/10.1167/4.9.5. Littell, R. C., Milliken, G. A., Stroup, W. W., Wolfinger, R. D., & Schabenberger, O. (2006). SAS for mixed models (2nd ed.). Cary, NC: SAS Institute Inc. McClain, A. D., van den Bos, W., Matheson, D., Desai, M., McClure, S. M., & Robinson, T. N. (2014). Visual illusions and plate design: The effects of plate rim widths and rim coloring on perceived food portion size. International Journal of Obesity, 38(5), 657e662. Merilaita, S., Scott-Samuel, N. E., & Cuthill, I. C. (2017). How camouflage works. Philosophical Transactions of the Royal Society B: Biological Sciences, 372(1724). https://doi.org/10.1098/rstb.2016.0341. Milliken, B., & Jolicoeur, P. (1992). Size effects in visual recognition memory are determined by perceived size. Memory & Cognition, 20(1), 83e95. https:// doi.org/10.3758/bf03208257. Morse, D. H. (1970). Ecological aspects of some mixed-species foraging flocks of birds. Ecological Monographs, 40(1), 119e168. https://doi.org/10.2307/1942443. Pietrewicz, A. T., & Kamil, A. C. (1979). Search image formation in the blue jay (Cyanocitta cristata). Science, 204, 1332e1333. Plaisted, K. (1997). The effect of interstimulus interval on the discrimination of cryptic targets. Journal of Experimental Psychology: Animal Behavior Processes, 23(2), 248e259. https://doi.org/10.1037/0097-7403.23.2.248. Rodgers, G. M., Ward, J. R., Askwith, B., & Morrell, L. J. (2011). Balancing the dilution and oddity effects: Decisions depend on body size. PLoS One, 6(7), e14819. https://doi.org/10.1371/journal.pone.0014819.
Rosselli, F. B., Alemi, A., Ansuini, A., & Zoccolan, D. (2015). Object similarity affects the perceptual strategy underlying invariant visual object recognition in rats. Frontiers in Neural Circuits, 9, 10. https://doi.org/10.3389/fncir.2015.00010. Schmidt, J. M., Sebastian, P., Wilder, S. M., & Rypstra, A. L. (2012). The nutritional content of prey affects the foraging of a generalist arthropod predator. PLoS One, 7(11), e49223. https://doi.org/10.1371/journal.pone.0049223. Sokal, R. S., & Rohlf, F. J. (1981). Biometry. New York, NY: Freeman. Stuart-Fox, D., & Moussalli, A. (2009). Camouflage, communication and thermoregulation: Lessons from colour changing organisms. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1516), 463e470. https://doi.org/ 10.1098/rstb.2008.0254. Theodorakis, C. W. (1989). Size segregation and the effects of oddity on predation risk in minnow schools. Animal Behaviour, 38(3), 496e502. https://doi.org/ 10.1016/S0003-3472(89)80042-9. Tsurui, K., Honma, A., & Nishida, T. (2013). Size-dependent predation risk partly explains the sex-related marking polymorphism in the sexually size-dimorphic pygmy grasshopper Tetrix japonica. Entomological Science, 16(2), 136e144. https://doi.org/10.1111/j.1479-8298.2012.00543.x. Tullberg, B. S., Merilaita, S., & Wiklund, C. (2005). Aposematism and crypsis combined as a result of distance dependence: Functional versatility of the colour pattern in the swallowtail butterfly larva. Proceedings of the Royal Society B: Biological Sciences, 272(1570), 1315e1321. https://doi.org/10.1098/rspb.2005.3079. Werner, E. E., & Hall, D. J. (1974). Optimal foraging and the size selection of prey by the Bluegill Sunfish (Lepomis macrochirus). Ecology, 55(5), 1042e1052. https:// doi.org/10.2307/1940354. Wolfe, J. M., Butcher, S. J., Lee, C., & Hyle, M. (2003). Changing your mind: On the contributions of top-down and bottom-up guidance in visual search for feature singletons. Journal of Experimental Psychology: Human Perception and Performance, 29(2), 483e502. https://doi.org/10.1037/0096-1523.29.2.483. Wolfe, J. M., Oliva, A., Horowitz, T. S., Butcher, S. J., & Bompas, A. (2002). Segmentation of objects from backgrounds in visual search tasks. Vision Research, 42(28), 2985e3004. https://doi.org/10.1016/s0042-6989(02)00388-7. Zelinsky, G. J. (2005). Specifying the components of attention in a visual search task. In G. R. L. Itti, & J. Tsotsos (Eds.), Neurobiology of attention (pp. 395e400). New York, NY: Elsevier Academic Press.
Appendix
Figure A1. Example of presentation cards used to test for effects of colour pattern on size perception bias.